Photocatalysis in TiO2 aqueous suspension: Effects of mono- or di-hydroxyl substitution of butanedioic acid on the disappearance and mineralisation rates

Article abstract of DOI:10.1016/j.cattod.2011.07.010

The photocatalytic degradation of butanedioic acid (BDA), hydroxybutanedioic acid (OH-BDA) and 2,3-dihydroxybutanedioic acid (diOH-BDA) (succinic acid, malic acid and tartaric acid, respectively) in a TiO₂ aqueous suspension at pH 3 was investigated to determine the effects of hydroxyl substitution. Temporal variations in the concentrations of the main intermediate products are reported. The use of a set of reactions involving decarboxylation and formation of alkylperoxy and alkoxy radicals has allowed us to account for the occurrence of these products, and to suggest the preferential decarboxylation of the COOH/COO^- group adjacent to the OH group in OH-BDA. The adsorbed amounts in the dark increased with the number of OH groups, while the irradiation time necessary for total disappearance of the initial diacid, decreased. However, the evolution rate of CO₂, which was initially the same for the three acids, was then lower for the hydroxylated diacids for a significant period of irradiation time. This difference is tentatively attributed predominantly to the effect of hydroxyl substitution on the adsorption mode. One of the carboxyl groups would remain away from the surface (and hence not be available for direct electron transfer) due to the restricted mobility of the adsorbed diacid caused by hydrogen-bonding of TiO₂ with the alcohol group adjacent to the other carboxyl group. This hypothesis is also effective to qualitatively interpret how the zeta potential varied in the course of the degradation of each diacid. The present study further illustrates the essential role of distinct adsorption modes in photocatalytic events.

Full text of DOI:10.1016/j.cattod.2011.07.010

Catalysis Today 178 (2011) 51–57
Contents lists available at ScienceDirect
Catalysis Today
journal homepage: www.elsevier.com/locate/cattod
Photocatalysis in TiO aqueous suspension: Effects of mono- or di-hydroxyl
2
substitution of butanedioic acid on the disappearance and mineralisation rates
a
a
a
b
a,∗
Wenny Irawaty , Donia Friedmann , Jason Scott , Pierre Pichat , Rose Amal
a
ARC Centre of Excellence for Functional Nanomaterials, School of Chemical Engineering, The University of New South Wales, Sydney, NSW 2052, Australia
Photocatalyse et Environnement, CNRS/Ecole Centrale de Lyon (STMS), 69134 Ecully CEDEX, France
b
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 13 May 2011
Received in revised form 12 July 2011
Accepted 12 July 2011
Available online 9 August 2011
The photocatalytic degradation of butanedioic acid (BDA), hydroxybutanedioic acid (OH-BDA) and 2,3-
dihydroxybutanedioic acid (diOH-BDA) (succinic acid, malic acid and tartaric acid, respectively) in a
TiO2 aqueous suspension at pH 3 was investigated to determine the effects of hydroxyl substitution.
Temporal variations in the concentrations of the main intermediate products are reported. The use
of a set of reactions involving decarboxylation and formation of alkylperoxy and alkoxy radicals has
allowed us to account for the occurrence of these products, and to suggest the preferential decarboxy-
Keywords:
Photocatalysis
−
lation of the COOH/COO group adjacent to the OH group in OH-BDA. The adsorbed amounts in the
dark increased with the number of OH groups, while the irradiation time necessary for total disappear-
ance of the initial diacid, decreased. However, the evolution rate of CO2, which was initially the same
for the three acids, was then lower for the hydroxylated diacids for a significant period of irradiation
time. This difference is tentatively attributed predominantly to the effect of hydroxyl substitution on
the adsorption mode. One of the carboxyl groups would remain away from the surface (and hence not
be available for direct electron transfer) due to the restricted mobility of the adsorbed diacid caused by
hydrogen-bonding of TiO2 with the alcohol group adjacent to the other carboxyl group. This hypothesis
is also effective to qualitatively interpret how the zeta potential varied in the course of the degrada-
tion of each diacid. The present study further illustrates the essential role of distinct adsorption modes
in photocatalytic events.
Butanedioic acid
Hydroxyl functional group
Adsorption mode
Degradation pathway
Mineralisation rate
©
2011 Elsevier B.V. All rights reserved.
1
. Introduction
Aliphatic acids are important raw materials for the manufacture
of textiles, soaps, metallic materials and alcoholic or non-alcoholic
beverages [11,12]. Carboxylic acids are often-encountered inter-
mediate products of the oxidation of organic molecules en route
to mineralisation of large molecules such as those contained in for
example natural organic matter, pharmaceuticals, pesticides and
dyes, and they are resistant to ordinary chemical oxidation. In addi-
tion, the ability of carboxylic acids to form complexes with toxic
heavy metal ions is also of concern since this has been observed to
increase the bioavailability of the heavy metal ions in the envi-
ronment [13]. Furthermore, the presence of those acids at high
concentration during chlorination in drinking water treatment
could lead to generation of disinfection byproducts (DBPs) such as
haloacetic acids that have been reported to be carcinogenic [14–16].
Therefore, the presence of any dissolved organic carbons (DOCs)
including carboxylic acid molecules needs to be controlled before
discharging to receiving waters.
Numerous articles and patents about TiO2 photocatalysis have
been published. Many reviews (for example [1–8] for 2009 and
010 only) have also been issued concerning both the fundamen-
2
tals and particular aspects of this field. The high bearing the target
organic can have on governing photocatalyst performance, means
recognising and understanding the factors influencing this aspect
are important. For example, studies by Tran et al. [9] and Denny
et al. [10] showed that even within an organic class, minor changes
to the basic molecule structure can strongly influence photocat-
alyst performance. That is, changes to the molecule in terms of
carbon chain length, chain branching and the varied presence of
functional groups, as well as the nature, number and positions of
substituents on benzene and other arene rings, could, in some cases,
provide substantial differences in removal and mineralisation rates.
In the present study, the mineralisation rate is defined as the CO2
evolution rate measured during the photocatalytic reaction of the
diacid.
This study is designed to elucidate the role that functional
groups within a chosen aliphatic acid play in governing the removal
and mineralisation rates using TiO2 photocatalysis. The model
molecules selected for the mechanistic investigation reported here
are butanedioic acid and its mono-hydroxylated and symmetrical
di-hydroxylated derivatives. These are commonly found in winery
∗ Corresponding author. Tel.: +61 2 9385 4361; fax: +61 2 9385 5966.
E-mail address: r.amal@unsw.edu.au (R. Amal).
0
920-5861/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.cattod.2011.07.010

5
2
W. Irawaty et al. / Catalysis Today 178 (2011) 51–57
wastewaters with typical pH between 3.65 and 5.40 [11,17]. It was
thought that the presence, in addition to the carboxyl groups, of
hydroxyl group(s) capable of forming hydrogen-bonds with TiO2,
should change the adsorption mode and, consequently, allows one
to determine its role. To help validate our interpretation of the
photocatalytic events, measurements of zeta potential were car-
ried out before the irradiation was started and in the course of
the photocatalytic degradation as these measurements provide
information on the adsorption of charged molecules as previously
demonstrated [18]. The reaction systems used in this study are, as
usual in laboratory experiments, much simpler than those found
in real wastewaters; however, the results from such experiments
should help to identify important factors affecting the degradation
of organic compounds in photocatalytic systems.
equilibrium at which time samples were collected for HPLC and
zeta potential analyses.
2.4. Analytical methods
HPLC (Waters 2695) with a Photodiode Array Detector (Waters
996) was used to identify organic compounds in solution. Samples
2
were taken from the reactor and subsequently filtered through a
.20 ␮m Millipore PTFE filter into small glass vials, capped and anal-
0
ysed by HPLC. An Atlantis T3 column (5 ␮m, 4.6 mm × 250 mm) was
◦
adopted to separate the compounds at 30 C. The mobile phase was
an aqueous solution of ammonium dihydrogen phosphate (pH 2.5)
for identification of BDA, diOH-BDA and their intermediates, while
sodium dihydrogen phosphate (pH 2.7) was used for OH-BDA and
its intermediates. The separation of organics was performed under
2
. Experimental
−
1
an isocratic condition with a mobile phase flow rate of 0.5 mL min
and injection volume of 50 ␮L. Unknown organics were identi-
fied by comparing the retention times and UV spectra with pure
standards. Linear calibration curves were obtained for organic con-
centrations in the range 0–12 ppm. TOC analysis was performed
on a Shimadzu TOC-VCSH analyser. Prior to analysis, the samples
were filtered through the PTFE filter. Zeta potential analysis was
2
.1. Catalyst and chemicals
®
Aeroxide TiO P25 (primary particle size ∼25 to 30 nm, surface
2
2
−1
area ∼50 m g , anatase to rutile ratio of 4:1) was used as the
photocatalyst since this TiO specimen is most often used in pho-
2
tocatalytic studies. Chemicals were analytical or HPLC grade and
used as supplied: butanedioic acid (>99%, Sigma–Aldrich ),
hydroxybutanedioic
dihydroxybutanedioic acid (>99%, Sigma–Aldrich ), propanedioic
acid (99%, Sigma–Aldrich ), ethanoic acid (99%, Ajax), hydrox-
yethanoic acid (99%, Sigma–Aldrich ), sodium dihydrogen
phosphate (98%, M&B) and ammonium dihydrogen phosphate
used to monitor the presence of organics on the TiO surface as the
®
2
mineralisation reaction proceeded. Samples taken at selected time
intervals were measured immediately using a Brookhaven Zeta-
PALS analyser. The measurements were repeated in the absence of
the parent acids to observe the effect of UV irradiation on the zeta
potential of the photocatalyst.
®
acid
(≥99%,
Sigma–Aldrich ),
2,3-
®
®
®
(
98%, Ajax). Perchloric acid (70%, w/w in water, Frederick Chem-
ical) and phosphoric acid (85%, w/w in water, Fluka) were used
to control the pH of TiO2 suspension and HPLC mobile phase,
respectively. All solutions were prepared using ultra pure water
3. Results and discussion
3.1. Results
(
Millipore Milli-Q water).
3.1.1. Photocatalytic oxidation of the diacids
2.2. Photocatalytic oxidation
Control experiments (not shown) demonstrated no mineralisa-
tion of the diacids in the presence of TiO2 without UV irradiation.
Photolysis of the diacids was also not observed. Fig. 1(a, c and e)
show CO2 detected during the TiO2 photocatalytic oxidation of BDA,
OH-BDA and diOH-BDA, respectively, as a function of irradiation
time (tUV). Complete mineralisation of BDA, OH-BDA and diOH-
BDA was achieved in approximately 30, 38 and 36 min, respectively.
Additionally, TOC analysis employed at the end of the photocat-
alytic oxidation (results not shown) confirmed that the amount
of organics remaining in solution was negligible. Also apparent in
Fig. 1(a, c and e) is the three diacids have different photocatalytic
mineralisation profiles. The BDA mineralisation profile follows a
typical organic degradation profile, whereby the mineralisation
rate gradually decreases with time, while OH-BDA and diOH-BDA
have distinctive regions where the rates were constant for a period
of time. Comparing the changing CO2 generation rates (Fig. 1(a, c
and e)) with intermediate product generation profiles (Fig. 1(b, d
and f)) for each diacid demonstrates a relationship exists between
the two, as expected. What is apparent is that the rate changes align
with the appearance and disappearance of each particular inter-
mediate product generated during degradation. Consequently, the
regions depicted in Fig. 1(a–f) were selected based on the change
in CO2 generation rate. Three intermediate products (A1, C1 and
C2) in Fig. 1(b, d and f) could not be confirmed by comparison with
commercial standards; therefore, they are presented as HPLC peak
areas.
Photocatalytic oxidation of BDA, OH-BDA and diOH-BDA were
assessed in a 250 mL slurry spiral reactor as described previously
−
19]. Briefly, a 0.2 g L TiO suspension (200 mL) was irradiated for
2
1
[
3
0 min to remove organic impurities from the TiO2 surface. Then,
the system was air-equilibrated prior to injecting 100 ␮L of the
parent solution, corresponding to 2000 ␮g carbon (equivalent to
4
2 ␮mol of the parent acid). The solution was circulated for 20 min
to allow for dark adsorption of the parent acid on the surface of TiO2
before the light (Sylvania 18 W blacklight blue lamp with maximum
emission of 365 nm) was switched on. The photocatalytic minerali-
sation reaction of the three diacids was monitored by recording the
amount of carbon dioxide (CO ) generated by means of an online
2
conductivity meter (Jenway 4330). The experiments were repeated
three times to ensure the reproducibility of the results. A TOC (total
organic carbon) analyser was used to confirm complete minerali-
sation by measuring the concentration of total organic remaining
in solution. HPLC (high performance liquid chromatography) was
also used to analyse and identify the organic compounds left in the
solution as the reaction progressed, while a zeta potential analyser
was used to study the organics on the photocatalyst surface (details
of the methods are given in Section 2.4).
2
.3. Dark adsorption study
Dark adsorption experiments were performed in a 250 glass bot-
tle at room temperature. BDA, OH-BDA and diOH-BDA solutions
Complete mineralisation corresponded to the absence of any
detected intermediate product for BDA and OH-BDA. This was not
the case for diOH-BDA, for which no intermediate products were
detected, under our analytical conditions, between 26 and 43 min
irradiation time while CO2 was still being generated. Complete
comprising 42 ␮mol (of each organic) were added to 200 mL of a
−
1
0
.2 g L
TiO2 suspension (pH adjusted to 3 ± 0.05) and agitated
by magnetic stirring for 20 min to promote adsorption/desorption

W. Irawaty et al. / Catalysis Today 178 (2011) 51–57
53
1
1
1
80
50
20
50
200
160
120
80
O
a
b
I
HO
OH
40
30
20
10
0
II
III
IV
O
9
6
3
0
0
0
0
40
I
II
III
IV
0
0
4
8
12 16 20 24 28 32 36 40 44
0
4
8
12 16 20 24 28 32 36 40 44
1
80
50
20
50
200
160
120
80
O
c
d
I
HO
OH
1
1
40
30
20
10
0
O
OH
II
III
IV
9
6
3
0
0
0
0
40
I
II
III
IV
0
0
4
8
12 16 20 24 28 32 36 40 44
0
4
8
12 16 20 24 28 32 36 40 44
1
1
1
80
50
20
50
200
160
120
80
OH
O
e
f
HO
OH
40
30
20
10
0
I
II
III
IV
V
O
OH
9
6
3
0
0
0
0
40
I
II
III
IV
V
0
0
4
8
12 16 20 24 28 32 36 40 44
Irradiation time (min)
0
4
8
12 16 20 24 28 32 36 40 44
Irradiation time (min)
Fig. 1. (a, c, e) CO2 generation profiles during photomineralisation of BDA, OH-BDA and diOH-BDA, respectively; (b, d, f) evolution and removal of intermediate compounds
during irradiation of BDA, OH-BDA and diOH-BDA, respectively. (ꢀ) BDA; (ꢁ) OH-BDA; (♦) diOH-BDA; ( ) propanedioic acid; (ꢀ) ethanoic acid; (×) hydroxyethanoic acid;
(
•) intermediate A1, believed to be 3-oxopropanoic acid [19]; (ꢁ) intermediate C1; (ꢂ) intermediate C2. Unidentified intermediate profiles (dashed lines) are based on peak
−1
areas from HPLC chromatograms. Experimental conditions: TiO2: 0.2 g L ; diacid initial amount: 42 ␮mol; initial pH: 3 ± 0.05.
disappearance of the diacid in solution corresponded to a change in
60
60
50
40
30
20
10
0
the slope of the curve of the CO amount against time for BDA only,
2
5
4
3
0
0
0
possibly because this disappearance occurred when the remaining
amounts of intermediate products were either nil or low (ethanoic
acid).
3.1.2. Dark adsorption study
The adsorption on TiO2 of BDA, OH-BDA and diOH-BDA in the
dark was studied. The results (Fig. 2) indicate that the adsorbed
amount is enhanced by additional OH functional group numbers on
20
10
0
the carbon chain, as expected. Also shown in Fig. 2 is the TiO zeta
2
potential before and after dark adsorption of the diacids; it clearly
shows that diOH-BDA imparted the greatest reduction to the TiO2
surface charge, followed by OH-BDA, and that the variation due to
BDA was low. This is proposed to reflect the greater adsorption of
the carboxylate anions.
In this study, the initial suspension pH was set at 3 ± 0.05. At
this pH, the TiO2 surface is positively charged. The pH variation
after the diacid adsorption was negligible. The amount of dissoci-
ated diacids in solution at pH 3, based on the pKa values [20], is
presented in Table 1. Approximately 50% of diOH-BDA exists in the
BDA
OH-BDA
diOH-BDA
Fig. 2. The amounts of the diacid adsorbed on TiO2 particles and the changes in TiO2
zeta potential after organic adsorption. ( ) The amount of the adsorbed diacid on the
surface of TiO ; (ꢂ) the initial TiO zeta potential before adsorption; (ꢁ) the TiO zeta
potential after organic adsorption. Experimental conditions: TiO2: 0.2 g L ; diacid
initial amount: 42 ␮mol; initial pH: 3 ± 0.05; adsorption period: 20 min. Note that
the TiO2 has not undergone carbon burn off. While this is expected to affect the
amount of adsorbed organics, and hence the measured zeta potential, the general
trend shown here is expected to be upheld when comparing between the different
organics.
2
2
2
−
1

5
4
W. Irawaty et al. / Catalysis Today 178 (2011) 51–57
Table 1
time there was a gradual return of the zeta potential to that of
Dissociated structures and corresponding amounts for BDA, OH-BDA and diOH-BDA
in water at 25 C and pH 3.
neat TiO . The profile suggests either low levels of mildly adsorbed
◦
2
organics (e.g. intermediate A ) and/or weakly adsorbed organics
1
Substance
BDA
Dissociated species structures
Amount (%)
(e.g. ethanoic acid) are present on the TiO2 surface during BDA
−
−
photodegradation.
HOOC–CH2–CH2–COO
5.85
0.01
25.83
0.21
47.93
2.02
−
OOC–CH2–CH2–COO
HOOC–CH2–CHOH–COO
A detailed description of OH-BDA photodegradation in rela-
tion to zeta potential is described in a previous publication [19].
Briefly, there is little change in the zeta potential upon irradia-
tion until around the 22 min mark where a significant increase is
observed. This represents the time at which intermediate A1 and
propanedioic acid are removed from the system and ethanoic acid is
dominant (Fig. 1(d)). Despite similar intermediates being detected
as in the BDA system, extended lower value of the zeta potential
for OH-BDA may be due to the higher concentrations observed for
these intermediates during irradiation.
−
−
OH-BDA
−
OOC–CH2–CHOH–COO
HOOC–CHOH–CHOH–COO
−
diOH-BDA
−
−
OOC–CHOH–CHOH–COO
60
50
40
30
20
10
0
On irradiation of the diOH-BDA system, the zeta potential
remained at ∼−10 mV for 8 min after which it increased to
∼+10 mV. According to Fig. 1(f), this time corresponded to com-
plete diOH-BDA removal and the peak maximum for intermediate
C . The zeta potential remained steady ∼+10 mV until tUV = 28 min,
1
with this time corresponding approximately to the time at which all
intermediates were consumed. Beyond this time the zeta potential
-
-
10
20
returned to the neat TiO value indicating most traces of organics
2
had been removed from the surface.
0
5
10
15
20
25
30
35
40
45
50
3.2. Discussion
Irradiation time (min)
Fig. 3. Change in TiO2 zeta potential during irradiation of (ꢀ) aqueous TiO2 suspen-
sion and TiO2 suspension in the presence of: (•) BDA, (ꢂ) OH-BDA and (ꢃ) diOH-BDA.
3.2.1. Degradation pathways
To highlight the similarities and differences in the degradation
pathways of the studied diacids, we first indicate the typical reac-
tions that are expected, and then apply these reactions to each
butanedioic acid.
−1
Experimental conditions: TiO2: 0.2 g L ; diacid initial amount: 42 ␮mol; initial pH:
3
± 0.05.
anionic form, favouring its adsorption through electrostatic attrac-
tion. For the other parent compounds, the amount of dissociated
species in solution was 26 and 6% for OH-BDA and BDA, respec-
tively.Thus, based on the pKa values alone [20], it would be expected
The degradation of carboxylate anions and carboxylic acids is
believed to occur through the photo-Kolbe reaction [22]:
RCOO + h+ → R + CO2
−
•
(1)
(2)
that diOH-BDA adsorbs to the greatest extent on TiO . In contrast
RCOOH + h+ → R + H + CO2
•
+
2
to this finding, Chen et al. [21] reported no significant difference in
the extent of adsorption of the same diacids at pH 2.9, on the same
•
The reaction pathways of an organic radical R are well docu-
mented [23–25]. They include addition of O , abstraction of H atoms
TiO , but at a much higher concentration so that the change in con-
2
2
and cleavages. In the presence of O , alkyl radicals are transformed
centration due to adsorption might have been still more difficult to
be measured.
2
into alkylperoxy (Eq. (3)) and alkoxy radicals (Eq. (4)), as is shown
by the main overall equations:
The surface coverages at the initial concentration used in this
−2
study were 0.23, 0.41, and 0.50 molecules nm for BDA, OH-BDA
and diOH-BDA, respectively. With these low surface coverages, the
degradation kinetics fitted the Langmuir–Hinshelwood model.
•
•
R + O → RO
(3)
2
2
•
•
•
2
RO2 → 2RO + O and RO + RH → ROOH + R•;
2
2
3
.1.3. TiO2 zeta potential during the reaction
TiO2 zeta potential was measured during the diacid photocat-
•
•
ROOH → RO + OH
(4)
alytic degradation (Fig. 3). This technique has been previously used
to monitor the presence of adsorbed organics on the photocata-
lyst surface in bulk phase under irradiation [18]. The zeta potential
of neat TiO2 remained constant at value of ∼+50 mV under irra-
diation. A comparison of the initial zeta potential values obtained
for the different diacids shows that the additional OH functional
groups led to a more negative zeta potential, which is indicative of
•
Depending on the chemical structure of the RO , namely in the
present case either R– CH O or R– CH(OH)O, the following reac-
tions are suggested to take place:
•
•
2
•
•
R– CH O → RCHO + H
(5)
(6)
2
•
•
R– CH(OH)O → RCOOH + H
greater adsorption of anions on the surface of TiO . The discrep-
•
•
2
2
R– CH O + O → RCHO + HO
(7)
2
2
ancy between initial zeta potential values (time = 0) in Figs. 3 and 2
is believed to be due to the carbon burn off step (i.e., removal of
adsorbed organic impurities) in the case of Fig. 3, which allowed a
greater initial adsorption of the parent ionised diacids leading to a
more negative initial TiO zeta potential.
Within the first 4 min of irradiation, the TiO2 zeta potential in
the BDA system increases substantially from approximately +20 to
•
•
•
R– CH(OH)O + O → RCOOH + HO
(8)
2
2
•
•
R– CH O + RH → RCH OH + R
(9)
2
2
•
R– CH(OH)O + RH → RCH(OH)2 + R
(10)
(11)
2
RCH(OH)2 ꢄ RCHO + H O
2
+
40 mV. This increase corresponds to the point where intermedi-
It is important to underline that these reactions involve either
no oxidant (Eqs. (4)–(6) and (9)–(11)) or only O (Eqs. (3), (7), (8)),
ate A reached its maximum concentration (Fig. 1(b)). Beyond this
1
2

W. Irawaty et al. / Catalysis Today 178 (2011) 51–57
55
and no h+ or OHs. Accordingly, their probability of occurring is
•
As in the case of the formation of ethanoic acid from BDA,
the observed formation of hydroxyethanoic acid from diOH-BDA
implies an additional C–C bond breaking via the decarboxyla-
tion of hydroxypropanedioic acid and reactions equivalent to
those corresponding to Eqs. (15) and (16) for the resulting rad-
•
high. Moreover, regarding the RO s, their decomposition (Eqs. (5)
and (6)) and their reactions with O2 (Eqs. (7) and (8)) are more
probable than their reactions with organic molecules (Eqs. (9) and
10)) whose concentration is much lower than that of O2.
In the following sections we successively indicate which prod-
(
•
ical HOOC– CHOH. Consequently, hydroxypropanedioic acid can
ucts can be formed from each of the diacids studied according to
the pathways corresponding to Eqs. (1)–(11).
lead to both dihydroxypropanedioic acid and hydroxyethanoic acid,
which might explain why it was not detected. In addition, given the
pKa1 and pKa2 values of hydroxypropanedioic acid of 2.37 and 4.74,
respectively [20], approximately 82.5% of the acid is dissociated
at pH 3 and will most likely be easily adsorbed on the positively
3.2.1.1. Case of butanedioic acid (BDA). Given the symmetric struc-
ture of BDA (inset in Fig. 1(a)), decarboxylation can take place at
−
either of the two COOH/COO groups in the carbon chain. Conse-
charged surface of TiO , hence avoiding detection in solution.
2
quently, CO is evolved and a carbon-centred radical is formed (Eqs.
2
•
•
(
1) and (2)). The R being HOOC–CH – CH , the expected product
3.2.1.3. Case of hydroxybutanedioic acid (OH-BDA). When OH-BDA
2
2
is HOOC–CH –CHO (3-oxopropanoic acid); Eqs. (3)–(5) and (7)),
was employed as the starting reactant, depending on whether the
2
which was tentatively identified to correspond to intermediate A1
initial decarboxylation concerns the carboxylic group adjacent to
•
(
Fig. 1(b)), whereas the formation of HOOC–CH –CH OH (Eqs. (3),
either the CHOH group or the CH group, the R (Eqs. (1) and (2)) is
2
2
2
•
•
(
4), (9)) is less probable and, in fact, this potential intermediate
suggested to be either HOOC–CH – CHOH or HOOC–CHOH– CH .
2
2
are
•
product was not observed, at least under our analytical condi-
The
expected
products
of
HOOC–CH – CHOH
2
•
tions. Oxidation of 3-oxopropanoic acid can easily occur upon OH
HOOC–CH –COOH (propanedioic acid; Eqs. (3), (4), (6), (8))
2
attack of the gem-diol resulting from hydration of the aldehyde
group (Eqs. (12)–(14)). This leads to propanedioic acid, which was
observed in the solution (Fig. 1(b)):
and HOOC–CH –CHO (3-oxopropanoic acid; Eqs. (3), (4), (10),
(11)). As presented in Fig. 1(d), propanedioic acid was identified,
while 3-oxopropanoic acid – reported to be the main intermediate
2
product of OH-BDA [26] – was believed to be intermediate A [19].
1
HOOC–CH –CHO + H O ꢅ HOOC–CH –CH(OH)
(12)
2
2
2
2
Alternatively, the former product can also result from the latter
•
•
•
HOOC–CH –CH(OH) + OH → HOOC–CH – C(OH) + H O (13)
product through oxidation involving OH (Eqs. (12)–(14)), which
2
2
2
2
2
is less probable. The same reactions as in the case of BDA would
lead to the formation of ethanoic acid.
•
•
2
HOOC–CH – C(OH) + O → HOOC–CH –COOH + HO
(14)
2
2
2
2
•
The formation of ethanoic acid (Fig. 1(b)) implies both the break-
ing of a C–C bond and the addition of a H atom to obtain the CH3
group. The C–C bond breaking can result from the decarboxyla-
In fact, although the R issued from Eqs. (1) and (2) was different
•
•
(HOOC–CH – CHOH in place of HOOC–CH – CH ), the identified
2
2
2
products from OH-BDA and BDA were the same. However, accord-
ing to the pathways proposed, propanedioic acid requires further
tion of propanedioic acid or its anions (Eqs. (1) and (2)) yielding
•
•
HOOC– CH , while the H atom can be abstracted from a BDA
oxidation by a OH, to be formed from BDA via 3-oxopropanoic acid
2
molecule (Eq. (15)) or stem from Eq. (5):
(Eqs. (12)–(14)), whereas it can also be formed from OH-BDA with-
out involving OH. However, that was not reflected by the temporal
variations in the low concentration of propanedioic acid, which
•
•
•
HOOC– CH + RH → HOOC–CH + R
(15)
(16)
2
3
•
•
HOOC– CH + H → HOOC–CH
were similar in both cases (Fig. 1(b and d)).
2
3
•
By contrast, the radical HOOC–CHOH– CH – resulting from
2
−
3.2.1.2. Case of dihydroxybutanedioic acid (diOH-BDA). In the case
the decarboxylation of the COOH/COO group adjacent to the CH2
•
of diOH-BDA, because of the symmetrical chemical structure of
group – would lead first to HOOC–CHOH–CH O (Eqs. (3) and
2
•
this diacid (inset in Fig. 1(e)), the R (Eqs. (1) and (2)) can only
(4)) and then to HOOC–CHOH–CHO (2-hydroxy-3-oxopropanoic
•
be HOOC–CHOH– CHOH. From this radical, the expected products
acid) by both Eq. (5) and (7) and to HOOC–CHOH–CH OH (2,3-
2
are HOOC–CHOH–COOH (hydroxypropanedioic acid; Eqs. (3), (4),
dihydroxypropanoic acid) via a less probable reaction (Eq. (9)).
Additionally, oxidation of HOOC–CHOH–CHO would have produced
HOOC–CHOH–COOH (Eqs. (12)–(14)). None of these three products
were detected.
(
6), (8)) and HOOC–CHOH–CHO (2-hydroxy-3-oxopropanoic acid;
Eqs. (3), (4), (10), (11)). None of the chromatographic peaks can
be attributed to hydroxypropanedioic acid whose retention time
(
tr) was 6.3 min under our analytical conditions. In the absence
of a commercial standard, the tr of HOOC–CHOH–CHO can only
be deduced by analogy. Given that tr of HOOC–CH –CHO was
These observations strongly suggest that decarboxylation pre-
−
dominantly occurs at the COOH/COO group adjacent to the CHOH
•
2
group thus leading to the radical HOOC–CH – CHOH. This is not
2
7
.7 min, it is tentatively suggested that the intermediate product
unexpected because the OH group of OH-BDA can easily form a
C1 (tr = 7.6 min; Fig. 1(f)) might be HOOC–CHOH–CHO.
hydrogen bond with the surface OH groups of TiO . Consequently,
2
−
It could be envisaged that HOOC–CHOH–COOH was oxidised to
the adjacent COOH/COO group would also be more easily bound to
+
a diol-diacid (HOOC–C(OH) –COOH (dihydroxypropanedioic acid))
TiO and hence more susceptible to react with h (Eqs. (1) and (2)).
2
2
•
−
by the removal, through a OH attack, of the relatively labile H atom
By contrast, the other COOH/COO group would be more distant
carried by the central C atom, followed by processes corresponding
to Eqs. (3) and (4) leading to the radical HOOC–C(OH) O–COOH and
then:
from the surface and hence not prone to direct electron trans-
fer. This type of reasoning is further developed in the following
sections.
•
To conclude this section, the use of a set of reactions has allowed
us to explain the formation of the three acids (3-oxopropanoic,
propanedioic and ethanoic acids) that were identified in the course
of the degradation of BDA and OH-BDA, and to propose that the
initial decarboxylation of OH-BDA occurred preferentially at the
position adjacent to the alcohol group. In the case of diOH-BDA,
hydroxypropanedioic acid was not detected although its forma-
tion was forecasted according to the same set of reactions, possibly
because it was rapidly transformed into dihydroxypropanedioic
•
•
HOOC– C(OH)O–COOH + RH → HOOC–C(OH) –COOH + R (17)
2
However, this product was not detected. Although the C2 inter-
mediate product possessed the same retention time (8 min) as
HOOC–C(OH) –COOH, its UV spectrum differed. On the other hand,
the relatively low concentration reached by this intermediate C2
may indeed indicate that it was a secondary oxidation product and
that its formation would compete with that of hydroxyethanoic
acid as indicated in the following paragraph.
2

5
6
W. Irawaty et al. / Catalysis Today 178 (2011) 51–57
acid and hydroxyethanoic acid and/or because of its high ionisation
percentage it remained predominantly adsorbed. Also, the forma-
tion of 2-hydroxy-3-oxopropanoic acid suggested from the set of
reactions was not certain in the absence of a standard.
molecules (Table 1), the fraction of diacid molecules attracted to
TiO2 by electrostatic forces was about 1/2 for diOH-BDA, 1/4 for
OH-BDA and only 1/16 for BDA, which meant much higher proba-
bilities of hitting the surface for the ionised hydroxylated diacids.
Second, it seems reasonable to propose that the residence time of
the ionised hydroxylated diacids at the surface would be increased
by the possibility of forming an additional hydrogen bond with
TiO2 via the alcohol group as mentioned in the previous sections.
For diOH-BDA, the increase in residence time should be even more
pronounced than for OH-BDA because this possibility exists what-
3.2.2. Mineralisation rates
For the three diacids, the amount of CO detected in the system
2
was almost the same over approximately the first 4 min of irradia-
tion (Fig. 1(a, c and e)). As decarboxylation is very likely the easiest
mineralisation pathway (Eqs. (1) and (2)), it should contribute the
−
most to CO generated during this initial period. To wit, no interme-
ever the COO anion hitting the surface; this reasoning may explain
2
−
diate products still containing the two initial COOH/COO groups
why the initial zeta potential caused by these hydroxylated diacids
differed more than expected from the respective ionised fractions
(Table 1).
were detected.
The higher mineralisation rate of BDA, observed between
4
< tUV < 16 min, can be explained by several reasons. One of the
After the initial increase in zeta potential, no significant change
occurred until about tUV = 22 min when the starting reactant was
BDA (Fig. 3). The corresponding zeta potential value was around
−
reasons could be easier decarboxylation of the second COOH/COO
group for BDA than the hydroxylated diacids. To account for this
easier decarboxylation of BDA, it is hypothesized that the forma-
+40 mV against +50 mV for an aqueous suspension of TiO . This
2
tion of a hydrogen bond between TiO and the OH group of OH-BDA
small difference is supposed to be due to the adsorption of ionised
remaining BDA (up to 18 min; Fig. 1(b)) and carboxylate anions of
the intermediate products. The zeta potential reached +50 mV at
tUV > 32 min showing the total elimination of negatively charged
intermediate products, which was consistent with the achieve-
ment of the maximum concentration in CO2 as a function of tUV
(Fig. 1(a)) because the near-to-complete-mineralisation products
are assumed to be carboxylic acids.
When the starting reactant was OH-BDA, a progressive increase
in zeta potential was observed beyond 8 min, this time correspond-
ing to the complete disappearance of OH-BDA in solution. However,
the zeta potential attained the same value as in the case of BDA
only after about 22 min. It is suggested that this difference may
have been caused by a higher concentration in ionised interme-
diate products, especially 3-oxopropanoate and ethanoate anions
(Fig. 1(d)). Also, note that the zeta potential reached +50 mV at
the same time (about 38 min; Fig. 3) as the achievement of the
2
maintains the COOH group not adjacent to the OH group away
from the surface and thus hinders its decarboxylation. In the case
of diOH-BDA, the two OH groups point in opposite directions, so
that only one of these groups can easily be hydrogen-bound to
TiO ; accordingly, the situation is equivalent to that of OH-BDA,
2
that is, one of the COOH group is not close to the surface. By con-
trast, the absence of a OH group in BDA allows both COOH groups
to interact with TiO2 indifferently. In other words, the suggested
difference in the adsorption mode of BDA and the hydroxylated
diacids might explain why the CO2 generated was higher for BDA
in region II despite a lower oxidation state of the non-carboxylic C
atoms and a lower extent of adsorption (Fig. 2). This is a case where
molecules that are less mobile at the solid surface reduce the pos-
sibilities of attack at several molecular positions by surface-located
active species and, consequently, can decrease the mineralisation
rate.
Additionally, up to tUV = 8 min – which corresponded to the dis-
maximum concentration in CO (Fig. 1(c)).
2
appearance of either OH-BDA or diOH-BDA – the TiO agglomerates
In the case of diOH-BDA as the starting reactant, an increase in
zeta potential from about −10 mV to +12 mV was observed once
diOH-BDA was no longer detected in the solution, that is, beyond
about tUV = 8 min (Fig. 1(f)). The value reached was approximately
the same as that found initially when OH-BDA was the starting
molecule (Fig. 3). It may mean that the corresponding density in
2
were surrounded by ionised molecules of these diacids as shown
by the zeta potential measurements discussed in the following sec-
tion. This phenomenon might have facilitated desorption of the
−
intermediate products – which all contained a COOH/COO group
–
and, accordingly, delayed their decarboxylation. The higher con-
−
centration reached by 3-oxopropanoic acid issued from OH-BDA
than from BDA (Fig. 1(d) and (b), respectively) and also the temporal
COO groups at the surface depended on the presence of an adja-
cent OH group in the intermediate products, which should be the
case given the formula of diOH-BDA. Indeed, hydroxyethanoic acid
was identified, dihydroxypropanedioic acid was presumed to be
formed and 2-hydroxy-3-oxopropanoic acid was tentatively sug-
gested to be generated. As aforementioned, this adjacent OH group
would increase the residence time of carboxylate anions at the sur-
face. The zeta potential value of about +10–12 mV (tUV = 12–26 min;
Fig. 3), attributed mainly to hydroxylated carboxylate anions,
was maintained until the intermediate products had disappeared
at tUV = 26–28 min (Fig. 1(f)). The zeta potential value of an
aqueous suspension of TiO2 was then reached at a time approx-
imately corresponding to total mineralisation (Fig. 1(e)), as was
expected.
It is clear here that the zeta potential variations not only
reflected the concentrations in the initial reactant and intermedi-
ate products – which were all carboxylic acids – but also the longer
residence time of these compounds on the TiO2 surface when an
alcohol group was adjacent to the carboxylate group. The conclu-
sions drawn from these measurements are consistent with both
the selectivity – deduced from the pathways we have proposed –
for the decarboxylation of OH-BDA and the adsorption modes sug-
gested to interpret the distinct mineralisation rates of the three
diacids studied.
variation of intermediate C (unfortunately, not quantified) formed
1
from diOH-BDA (Fig. 1(f)) may lend support to this interpretation.
3.2.3. Zeta potential variations
The initial zeta potential values (tUV = 0) showed that the TiO2
surface charge was more negative with a greater adsorption of
the more hydroxylated diacid (Fig. 3). The solution concentra-
−
−11
−1
mol L at pH 3, was negligible with
tion of OH , that is, 10
−
4
−1
respect to the initial diacid concentration (2.08 × 10 mol L ).
Consequently, the TiO surface charge depended on the mono- or
2
di-ionised diacid concentration at pH 3 (Table 1) and indeed the
zeta potential varied with this concentration, although not propor-
tionally (see the next paragraph for a tentative interpretation).
The difference in surface charge with respect to an aqueous
suspension of TiO2 considerably decreased over about the first
5
min of irradiation when the starting reactant was BDA, whereas
it remained the same when it was OH-BDA or diOH-BDA (Fig. 3).
The origin of this contrasting behaviour could be, in the case of the
hydroxylated diacids, a faster replacement at the TiO2 surface of
the ionised diacid that has been degraded by non-degraded ionised
diacid. Indeed, two arguments can be put forward in favour of a
faster replacement. First, on the basis of the percentages of ionised

W. Irawaty et al. / Catalysis Today 178 (2011) 51–57
57
4
. Conclusions
References
[
1] Y. Paz, Photocatalytic treatment of air: from basic aspects to reactors, in: H.I.
de Lasa, B.S. Rosales (Eds.), Adv. Chem. Eng., Academic Press, Amsterdam, 2009,
pp. 289–336.
In an effort to further illustrate the impact of adsorption modes
in photocatalysis of aliphatic acids, this study focused on deter-
mining to what extent the substitution of one hydroxyl group or
two hydroxyl groups at different positions in BDA – which should
change the binding to the TiO2 surface – affected the disappear-
ance rates, possible competitive adsorption with the intermediate
products and the mineralisation rates. To interpret the apparent
discrepancies found between the adsorbed amounts in the dark and
the photocatalytic disappearance rates on one side, and the min-
eralisation kinetics on the other side, it is hypothesised that the
hydroxyl substitution modifies the type of adsorption, the asso-
ciated geometry and the mobility in the adsorbed phase, so that
the average distance between the surface and one of the carboxyl
groups would increase. Accordingly, the possibility of surface-
located direct electron transfer, believed to effect decarboxylation,
would be hindered for this carboxyl group. This hypothesis is sup-
ported by zeta potential measurements. The initial zeta potential
values and those recorded in the course of the photocatalytic degra-
dation can indeed be explained by assuming that the residence
time of a carboxylate anion near the surface is increased by the
presence of a vicinal hydroxyl group. Although it does not pro-
vide direct information on the nature of the low adsorbed species
and on the adsorption modes, electrophoresis proves, neverthe-
less, to be instructive about the ability of the charged molecules
to bind to the surface. This investigation shows that, in practice,
higher removal rates of target pollutants, even those having com-
mon structural moieties, do not warrant higher mineralisation rates
of these pollutants.
[2] H. Zhang, G. Chen, D.W. Bahnemann, J. Mater. Chem. 19 (2009) 5089–5121.
[
[
[
3] D. Friedmann, C. Mendive, D. Bahnemann, Appl. Catal. B 99 (2010) 398–406.
4] Y. Paz, Appl. Catal. B 99 (2010) 448–460.
5] P. Pichat, Appl. Catal. B 99 (2010) 428–434.
[6] Y. Paz, Solid State Phenomena 162 (2010) 135–162.
[7] D. Mitoraj, H. Kisch, Solid State Phenomena 162 (2010) 49–75.
[8] P. Pichat, J. Adv. Oxid. Technol. 13 (2010) 238–246.
[9] H. Tran, J. Scott, K. Chiang, R. Amal, J. Photochem. Photobiol. A 183 (2006) 41–
52.
[10] F. Denny, J. Scott, K. Chiang, W.Y. Teoh, R. Amal, J. Mol. Catal. A 263 (2007)
3–102.
9
[
[
11] T. Colin, A. Bories, Y. Sire, R. Perrin, Water Sci. Tehnol. 51 (2005) 99–106.
12] N. Quichi, M.E. Morgada, G. Piperata, P. Babay, R.T. Gettar, M.I. Litter, Catal.
Today 101 (2005) 253–260.
13] A.P. Schwab, D.S. Zhu, M.K. Banks, Chemosphere 72 (2008) 986–994.
14] K. Gopal, S.S. Tripathy, J.L. Bersillon, S.P. Dubey, J. Hazard. Mater. 140 (2007)
[
[
1–6.
[15] U.S. Environmental Protection Agency, Basic information about disinfection
byproducts in drinking water: total trihalomethanes, haloacetic acids, bromate,
and chlorite. http://water.epa.gov/drink/contaminants/basicinformation/
disinfectionbyproducts.cfm (accessed 30.06.2011).
[16] C.M. Villanueva, M. Kogevinasa, J.O. Grimalt, Water Res. 37 (2003) 953–
9
58.
17] A. Versari, M. Castellari, U. Spinabelli, S. Galassi, J. Chem. Technol. Biotechnol.
6 (2001) 485–488.
[18] S.W. Lam, K. Chiang, T.M. Lim, R. Amal, G.K.-C. Low, J. Photochem. Photobiol. A
87 (2007) 127–132.
19] W. Irawaty, D. Friedmann, J. Scott, R. Amal, J. Mol. Catal. A 335 (2011) 151–
57.
[
7
1
[
1
[20] G.W. Gokel, Dean’s Handbook of Organic Chemistry, second ed., McGraw-Hill,
New York, 2004.
[
21] Q. Chen, J.M. Song, F. Pan, F.L. Xia, J.Y. Yuan, Environ. Technol. 30 (2009)
103–1109.
1
[
[
22] B. Kraeutler, A.J. Bard, J. Am. Chem. Soc. 100 (1978) 5985–5992.
23] C. von Sonntag, H.-P. Schuchmann, Angew. Chem. Int. Ed. 30 (1991) 1229–
1253.
Acknowledgements
[
24] K. Stemmler, U. von Gunten, Atm. Environ. 34 (2000) 4241–4252.
25] C. von Sonntag, Free-radical-induced DNA Damage and its Repair: A Chemical
Perspective, Springer, Berlin, 2006.
[
The authors gratefully acknowledge the financial support pro-
vided to this study by the Australian Research Council (ARC) Centre
of Excellence for Functional Nanomaterials.
[
26] J.-M. Herrmann, H. Tahiri, C. Guillard, P. Pichat, Catal. Today 54 (1999) 131–
141.