An unexplored O2-involved pathway for the decarboxylation of saturated carboxylic acids by TiO2 photocatalysis: An isotopic probe study

Article abstract of DOI:10.1002/chem.201001704

The aerobic decarboxylation of saturated carboxylic acids (from C ₂ to C₅) in water by TiO₂ photocatalysis was systematically investigated in this work. It was found that the split of C ¹-C² bond of the acids to release CO₂ proceeds sequentially (that is, a C₅ acid sequentially forms C₄ products, then C₃ and so forth). As a model reaction, the decarboxylation of propionic acid to produce acetic acid was tracked by using isotopic-labeled H₂¹⁸O. As much as ≈42% of oxygen atoms of the produced acetic acids were from dioxygen (¹⁶O₂). Through diffuse reflectance FTIR measurements (DRIFTS), we confirmed that an intermediate pyruvic acid was generated prior to the cut-off of the initial carboxyl group; this intermediate was evidenced by the appearance of an absorption peak at 1772 cm^-1 (attributed to C=O stretch of α-keto group of pyruvic acid) and the shift of this peak to 1726 cm ^-1 when H₂¹⁶O was replaced by H ₂¹⁸O. Consequently, pyruvic acid was chosen as another model molecule to observe how its decarboxylation occurs in H₂ ¹⁶O under an atmosphere of ¹⁸O₂. With the α-keto oxygen of pyruvic acid preserved in the carboxyl group of acetic acid, ≈24% new oxygen atoms of the produced acetic acid were from molecular oxygen at near 100% conversion of pyruvic acid. The other ≈76% oxygen atoms were provided by H₂O through hole/OH radical oxidation. In the presence of conduction band electrons, O₂ can independently accomplish such C¹-C² bond cleavage of pyruvic acid to generate acetic acid with ≈100% selectivity, as confirmed by an electrochemical experiment carried out in the dark. More importantly, the ratio of O₂ participation in decarboxylation increased along with the increase of pyruvic acid conversion, indicating the differences between non-substituted acids and α-keto acids. This also suggests that the O ₂-dependent decarboxylation competes with hole/OH-radical-promoted decarboxylation and depends on TiO₂ surface defects at which Ti _4c sites are available for the simultaneous coordination of substrates and O₂. Acid aerobics! An O₂-involved pathway for the decarboxylation of saturated carboxylic acids by TiO₂ photocatalysis is clarified, and has been found to be composed of two major tandem steps. An oxygen atom of O₂ is incorporated into the product acid in the second step (see scheme).

Full text of DOI:10.1002/chem.201001704

FULL PAPER
DOI: 10.1002/chem.201001704
An Unexplored O2-Involved Pathway for the Decarboxylation of Saturated
Carboxylic Acids by TiO2 Photocatalysis: An Isotopic Probe Study
Bo Wen, Yue Li, Chuncheng Chen, Wanhong Ma,* and Jincai Zhao*[a]
Abstract: The aerobic decarboxylation
of saturated carboxylic acids (from C2
to C5) in water by TiO2 photocatalysis
1772 cmꢀ1 (attributed to C=O stretch
of a-keto group of pyruvic acid) and
the shift of this peak to 1726 cmꢀ1
when H216O was replaced by H218O.
Consequently, pyruvic acid was chosen
as another model molecule to observe
how its decarboxylation occurs in
H216O under an atmosphere of 18O2.
With the a-keto oxygen of pyruvic acid
preserved in the carboxyl group of
acetic acid, ꢁ24% new oxygen atoms
of the produced acetic acid were from
molecular oxygen at near 100% con-
version of pyruvic acid. The other
ꢁ76% oxygen atoms were provided by
H2O through hole/OH radical oxida-
tion. In the presence of conduction
band electrons, O2 can independently
1
2
ꢀ
accomplish such C
C bond cleavage
was systematically investigated in this
of pyruvic acid to generate acetic acid
with ꢁ100% selectivity, as confirmed
by an electrochemical experiment car-
ried out in the dark. More importantly,
the ratio of O2 participation in decar-
boxylation increased along with the in-
crease of pyruvic acid conversion, indi-
cating the differences between non-
substituted acids and a-keto acids. This
also suggests that the O2-dependent de-
carboxylation competes with hole/OH-
radical-promoted decarboxylation and
depends on TiO2 surface defects at
which Ti4c sites are available for the
simultaneous coordination of substrates
and O2.
1
ꢀ
work. It was found that the split of C
C2 bond of the acids to release CO2
proceeds sequentially (that is, a C5 acid
sequentially forms C4 products, then C3
and so forth). As a model reaction, the
decarboxylation of propionic acid to
produce acetic acid was tracked by
using isotopic-labeled H218O. As much
as ꢁ42% of oxygen atoms of the pro-
duced acetic acids were from dioxygen
(16O2). Through diffuse reflectance
FTIR measurements (DRIFTS), we
confirmed that an intermediate pyruvic
acid was generated prior to the cut-off
of the initial carboxyl group; this inter-
mediate was evidenced by the appear-
ance of an absorption peak at
Keywords: carboxylic acids · decar-
boxylation · isotopic labeling · pho-
tocatalysis · titanium dioxide
Introduction
bonds and formation of new bonds for hydrocarbon moiet-
ies is tuned by holes, electrons, and subsequent underlying
reactants such as O2 and H2O in this intricate heterogeneous
system. In spite of the complexity of initial organic pollu-
tants in TiO2-mediated photodegradation processes, carbox-
ylic acids are always observed to accumulate and account
for the majority of converted initial substrates besides
CO2.[1,2] Apparently, the formation of these fragment acids
and their further decomposition to achieve complete miner-
As one of the most important methods for the elimination
of pollutants, TiO2 photocatalysis has been extensively stud-
ied and widely used in this field for decades.[1–3] Upon UV ir-
radiation, most organic pollutants can be effectively degrad-
ed through surface-mediated redox processes performed by
photon-excited hole–electron pairs. However, to date, little
is known about how the mechanism of the cleavages of old
ꢀ
alization constitute the most fundamental steps of C C
bond destruction, which has always been believed to directly
affect or even sometimes decide the whole fateful efficiency
of photocatalytic transformations.[1,2] However, the details of
decarboxylation that affords the decomposition of the acids
as well as the details of subsequent carboxylation are not
entirely clear yet, even in the cases of most simple saturated
straight-chain carboxylic acids with less than five carbon
atoms.
[a] Dr. B. Wen, Dr. Y. Li, Dr. C. Chen, Dr. W. Ma, Dr. J. Zhao
Beijing National Laboratory for Molecular Sciences
Key Laboratory of Photochemistry, Institute of Chemistry
Chinese Academy of Sciences, Beijing 100190 (China)
Fax : (+86)108-261-6495
E-mail: whma@iccas.ac.cn
jczhao@iccas.ac.cn
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201001704.
Chem. Eur. J. 2010, 16, 11859 – 11866
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
11859

Serpone et al. have systematically examined the photoca-
talytic decarboxylation of saturated straight-chain carboxylic
acids in aerated aqueous TiO2 dispersions, which showed
that the degradation intermediates were mainly composed
of shorter-chain monocarboxylic acids.[4] Roughly, the decar-
ever, the established mechanism is not completely consistent
with the experimental observations: 1) although -ols and
-ones are relatively stable under most oxidative condi-
tions,[12] it is always shorter-chain acids rather than -ols/ones
that are detected as the main intermediates with bare TiO2
as photocatalyst;[4] 2) even if sometimes -ols and -ones are
too unstable to attain observable concentration in the reac-
tions, holes/OH radicals cannot oxidize them to produce
acids without O2, for example, it is acetone rather than pyr-
uvic acid or acetic acid as the final product in the oxidation
of isopropanol promoted only by holes/OH radicals in the
absence of O2.[13] This means that there are other possible
unexplored O2-involved pathways for the decarboxylation of
carboxylic acids by TiO2 photocatalysis.
1
2
ꢀ
boxylation can be considered as the split of C C bond in
the substrate acid and simultaneous release of CO2
[Eq. (1)].
O2=H2O
2
1
2
1
ꢀðCH Þ ꢀ CH ꢀ COOH
-ðCH Þnꢀ1ꢀ CH2ꢀ COOH þ CO2
ƒƒƒƒ!
2
n
2
2
TiO2=UV
ð1Þ
The prevalent mechanism for the decarboxylation has
1
2
ꢀ
been proved that the split of C C bond exclusively origi-
nates from hole oxidation [Eq. (2)],[4–6] in which h+ is a pho-
togenerated hole in the valence band. Subsequent inter-
mediate products are alkanes formed under deaerated con-
ditions by means of so-called photo-Kolbe reactions
[Eqs. (2) and (3)][5] (ecꢀb is a photogenerated electron in the
conduction band). Such phenomena have also been ob-
served in the thermal oxidative decarboxylation of these
acids by high-valence metal ions, such as CoIII,[7] PbIV,[8] and
MnIII,[9] although the main decarboxylation products were al-
kenes rather than alkanes since no reductive species [like
the electrons shown in Eq. (3)] were involved. When O2 is
present, it will react with the generated carbon-centered
radicals to finally form hydroxylated and carbonylated inter-
mediates (i.e., -ols/ones)[6] obeying the Russell mechanism
for the decomposition of peroxyl radicals via quaternion-
peroxide intermediates[10] [Eqs. (4)–(7)].
In this work, detailed 18O-isotopic-labeling experiments
(with H218O and 18O2) were carried out to investigate the de-
carboxylation of saturated straight-chain carboxylic acids by
TiO2 photocatalysis (unless specially indicated, an anatase
TiO2 film supported on a glass plate was used). Through an-
alyzing the distribution of isotopic-labeled oxygen atoms
from O2 or H2O in product acids along with the conversion
of substrate acids, we will accurately show the relative con-
tribution of O2 to the decarboxylation process. We also em-
ployed diffuse reflectance FTIR measurements (DRIFTS)[14]
combined with 18O-isotopic-labeling method to identify im-
portant intermediates in the decarboxylation process. Mean-
while, well-defined electrochemical experiments[15] were per-
formed for the observation of the influence of electrons and
O2 on focused decarboxylation reactions without the in-
volvement of holes/OH radicals. Through performing these
detailed investigations, we expect that some results of this
research can be useful in understanding the intrinsic role of
O2 as well as clarifying the mechanisms of decarboxylation
in TiO2 photocatalytic reactions.
þhþ
ꢀCO2
RCH2COOꢀ
RCH
ð2Þ
ð3Þ
C
ƒƒƒ!
2
ꢀ
RCH2 þ ecb þ Hþ ! RCH3
C
C
C
RCH2 þ RCH2 ! RCH2ꢀCH2R
Results and Discussion
C
C
RCH2 þ O2 ! RCH2OO
ð4Þ
ð5Þ
ð6Þ
O2 þ eꢀcb þ Hþ ! OOH
C
To gain more understanding of the decarboxylation process
in the presence of O2, first of all we re-examined the degra-
dation of some saturated straight-chain carboxylic acids
(from acetic acid to valeric acid) in aerated aqueous solu-
tions by TiO2 photocatalysis. The temporal concentration
changes of substrate acids and generated intermediates in
the reactions were measured by ion chromatography (IC;
Figure S1 in the Supporting Information). Table 1 lists the
main generated intermediates and their highest yields
during the degradation of the acids. Note that formic acid
was detected as an intermediate in the degradation of every
acid, but the yield of it was only determined in the degrada-
tion of acetic acid due to its very low concentration and the
peak overlap between it and propionic acid in IC when in-
vestigating the degradation of other acids. Moreover, we cal-
culated the total organic carbon (TOC) values of the re-
maining substrates and generated intermediates at different
time intervals during the degradation of the acids according
to the IC results and compared them with those of the
C
C
RCH2OO þ OOH ! RCH2OOOOH
RCH2OOOOH þ RCH2COOH !RCH2OH þ RCHOþ
O2 þ CO2
ð7Þ
OH radicals are also regarded to be important oxidative
species that act as “diffusive holes” in this decarboxylation
process.[11] Moreover, O2 may be reduced by photogenerated
electrons to form H2O2, which could be further transformed
into OH radicals to promote decarboxylation, although this
conversion, which is restricted by several integral steps,
should only provide a minor contribution to the decarboxyl-
ACHTUNGTRENNUNG
ation process.[2] In the alleged universal mechanism argu-
ment, the direct release of CO2 arbitrarily attributed to
hole/OH radical oxidation as well as the involvement of O2
through free-radical reactions has been the key point. How-
11860
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 11859 – 11866

Photocatalysis
FULL PAPER
Table 1. The main generated intermediates and their highest yields
during the degradation of C2–C5 acids by TiO2 photocatalysis as well as
the proportion of calculated TOC to measured TOC at the time point of
2 h UV irradiation in the degradation of the acids.
over, there must be some intermediates prior to the cleav-
1
2
ꢀ
age of C C bond and release of CO2, otherwise, we can
only observe the direct generation of CO2 rather than the
accumulation of every shorter-chain acid.[16] Following the
above notions, we chose propionic acid as the model sub-
strate, because its main decarboxylation product, acetic acid,
was stable enough to achieve very high accumulation con-
centration for ready analysis, to specifically search for the
most suitable precursor for the CO2 departure and clarify
the role of O2 in designed TiO2 photocatalytic experiments.
Since the oxidative decarboxylation and subsequent for-
mation of new carboxyl group promoted by holes and OH
radicals requires H2O participation regardless of hydrolysis
or addition reactions, we performed an isotopic labeling ex-
periment using H218O (96%) to observe the 18O profile of
the produced acetic acid in the aerobic decarboxylation of
propionic acid by TiO2 under UV irradiation (Table 2). This
Substrate[a]
Intermediates
Yields[b] [%]
P[c] [%]
CH3COOH
C2H5COOH
n-C3H7COOH
HCOOH
1.3
31.5
17.3
21.0
13.6
7.0
99.1
97.6
98.5
CH3COOH
CH3COOH
C2H5COOH
CH3COOH
C2H5COOH
n-C3H7COOH
n-C4H9COOH
95.1
1.1
[a] Experimental conditions: solution of acid in water (30 mL, 2 mm) in a
50 mL vessel. [b] The highest yield of this intermediate during the degra-
dation of the substrate acid, which was calculated according to yield=
(highest intermediate concentration)/(initial substrate concentration).
[c] P=calculated TOC/measured TOC, which was based on the IC and
TOC data obtained at the time point of 2 h UV irradiation.
Table 2. The decarboxylation of C2H5COOH by TiO2 photocatalysis in
H218O and the isotopic distribution of produced CH3COOH.[a]
whole solutions measured by a TOC analyzer (Figure S2 in
the Supporting Information). The proportions of calculated
TOC to measured TOC at the time point of 2 h UV irradia-
tion are also listed in Table 1.
The less than 5% differences between the calculated
TOC values and the measured ones confirmed that the de-
tected, linear, non-substituted, shorter-chain monocarboxylic
acids constituted the main degradation intermediates. More-
over, regardless of the initial substrate acids (from C2 to C5),
1) these shorter-chain acid intermediates remained in the so-
lution and almost accounted for the total conversion of ini-
tial substrate besides CO2; 2) nearly all the presumable infe-
rior acids (nꢀ1, nꢀ2, …) originating from the parent acid,
even C1 formic acid, steadily appeared without exception
and could be quantified, although their features of accumu-
lation to attain the highest concentrations differed; 3) acetic
acid was always generated and significantly accumulated
during the degradation of all longer-chain acid substrates.
These results, highly similar to those of Serponeꢁs previous
work,[4] draw the outline for the degradation and
Conver-
Yield[c]
[%]
Percentage of various
sion[b] [%]
CH3COOH isotopomers[d] [%]
16O16O
16O18O
18O18O
16O[e]
18O[e]
1
2
25.1
55.6
20.0
16.2
23.0
14.5
38.9
33.9
38.1
51.6
42.5
31.5
57.5
68.5
[a] Experimental conditions: solution of C2H5COOH in H218O (1.5 mL,
20 mm) in a 10 mL vessel. [b] The conversion of propionic acid. [c] The
fractional yield of acetic acid, which was calculated according to yield=c
(generated acetic acid)/c (converted propionic acid). [d] Calculated by
the integration of corresponding peak area obtained from mass spectrom-
etry. [e] The total proportion of 16O or 18O in the carboxyl group of
CH3COOH, which was calculated according to 16O=16O16O+16O18O/2,
18O=18O18O+16O18O/2.
way, the O2 entrance into product acids can be also re-
vealed. Since the formed acetic acid as intermediate could
also be further degraded, its steady concentration just ac-
counted for about 0.2 equiv of the converted propionic acid.
Blank experiments were performed to examine the oxygen
exchange between the carboxyl groups of these acids (acetic
and propionic acid) and H218O and showed that such ex-
change was negligible (Table S1 in the Supporting Informa-
tion), which ensured the credibility of our isotopic labeling
experiments. Table 2 shows the results of the experiments at
25.1 and 55.6% conversion of propionic acid. Although the
concentration of dissolved 16O2 was much lower than that of
H218O, quite impressive 16O abundance was detected in the
acetic acid product. As high as ꢁ42% of oxygen atoms in
the acetic acid were exclusively from 16O2 at 25.1% conver-
sion of propionic acid (Table 2, entry 1). Besides the genera-
tion of anticipated CH3C18O18OH, CH3C16O16OH and
CH3C16O18OH were also generated, distinctly indicating that
O2 can accomplish the formation of the new carboxyl group
in product acid either by itself or via a possible concerted
decarboxylACHTUNGTRENNUNGation of a carboxylic acid by TiO2 photocatalysis
1
2
ꢀ
(Scheme 1): the C C bond split of the acids is invariably
sequential (i.e., a C5 acid sequentially forms C4 products,
then C3 and so forth) until formic acid was generated; no di-
acids or -ols/ones as obligatory intermediates were detected
as described in the mechanism given in Equations (2)–(7).
This strongly suggests the fact that decarboxylation is to
1
2
ꢀ
simul
ACHTUNGTRENNUNGtaneously break the old C C bond of substrate acid
and establish new carboxyl group of the inferior acid. More-
1
2
ꢀ
Scheme 1. One-by-one cleavages of C C bonds in carboxylic acids.
Chem. Eur. J. 2010, 16, 11859 – 11866
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11861

J. Zhao, W. Ma et al.
reaction. At higher conversion of 55.6%, the ratio of O2 par-
ticipation to H2O participation decreased, indicating that de-
carboxylation of the saturated acid is a quite complex pro-
cess for the involvement of both O2 and H2O. To clarify in
which steps O2 is incorporated into the product acids, we
further examined the possible intermediates prior to the
cut-off of the initial carboxyl group in the decarboxylation
of propionic acid by TiO2 photocatalysis.
Considering that the possible intermediates are not as
stable as substrate acids, we employed the in situ analysis
technique diffuse reflectance FTIR measurements
(DRIFTS) to observe the decarboxylation of propionic acid.
In a typical process, propionic acid (100 mmol per g of cata-
lyst, corresponding to less than one monolayer of propionic
acid adsorbed on TiO2 surface) was adsorbed on TiO2 (De-
gussa P25 powder) and exposed to UV irradiation in the
presence of H2O and O2 in an in situ reactor (see details in
Experimental Section). The spectrum of propionic acid ad-
sorbed on TiO2 before irradiation is shown at the top of
Figure 1. The positive and negative absorption bands in the
propionate. Furthermore, a broad band centered at around
3250 cmꢀ1 also appeared and can be assigned to surface ad-
sorbed H2O.[20] A band at about 1625 cmꢀ1, which belongs to
H2O, should also be visible but was overlapped by the
strong carbonyl absorption at 1690 cmꢀ1.[20] Other spectra in
Figure 1 were collected at different time intervals upon UV
ꢀ
irradiation. The intensity of C H and carboxylate bands all
decreased as the reaction proceeded, indicating that pro-
pionic acid was continuously decarboxylated and degraded.
At the same time, the intensity of the band at 3250 cmꢀ1 in-
creased, which can be due to the generation of H2O as a re-
action product.
To obtain the information about the generated intermedi-
ates in the decarboxylation process, we further subtracted
the DRIFTS spectrum collected before UV irradiation from
the one after irradiation to get the difference spectrum. The
upper spectrum in Figure 2 is the difference spectrum ob-
Figure 2. Difference DRIFTS spectra obtained for the decarboxylation of
propionic acid by TiO2 photocatalysis after 1 min of UV irradiation using
H216O (upper line) and H218O (lower line). The generation of a-keto acid
intermediate in the decarboxylation process is indicated by the appear-
ance of positive absorption peaks at 1772 cmꢀ1 (upper line) and
1726 cmꢀ1 (lower line) in the difference spectra.
Figure 1. Diffuse reflectance FTIR (DRIFTS) spectra recorded during
decarboxylation of propionic acid by TiO2 photocatalysis. These spectra
were collected at different time intervals (from top to bottom: 0, 1, 5, 10,
20 min) upon UV irradiation. The starting concentration of propionic
acid was 100 mmol per g of catalyst.
tained after 1 min of UV irradiation. The appearances of all
the negative bands in the alkyl and carboxylate regions of
the difference spectrum were consistent with the
spectrum indicate the formation and disappearance of spe-
cies adsorbed on TiO2, respectively. The negative hydroxyl
bands at 3680 and 3732 cmꢀ1 indicate that surface hydroxyl
groups of TiO2 were consumed upon propionic acid adsorp-
tion.[17] Meanwhile, as propionic acid was adsorbed on TiO2
surface, the positive bands at 2890, 2948, and 2982 cmꢀ1,
decarboxylACTHNGUTERNNUaG tion and degradation of substrate propionic
acid. Moreover, an authentic positive absorption peak at
1772 cmꢀ1 was observed and indicated the generation of an
intermediate in the decarboxylation process. This positive
peak at 1772 cmꢀ1 could be assigned to the vibration of C=
O stretch of a keto group in the intermediate, which should
be the a-keto group of pyruvic acid rather than a b-carbonyl
group of malonaldehydic acid. Such an assignment is based
on two aspects: 1) from intermediate profile, acetic acid
could be readily generated from pyruvic acid (Figure S3 in
the Supporting Information), while no reasonable shorter-
chain acid could be produced from b-carbonyl precursor
[Eq. (8)]; 2) the DRIFTS spectrum of pyruvic acid adsorbed
on TiO2 was taken as the standard for comparison and
showed a similar absorption peak (Figure S4 in the Support-
ing Information).
ꢀ
which characterize C H stretching of the acid, were ob-
served.[18] In the carboxylate region, five bands at 1300,
1430, 1472, 1532, and 1690 cmꢀ1 were also identified. Based
on previous studies of the identification of absorption bands
in the DRIFTS spectrum of acetic acid adsorbed on TiO2
and considering the structure similarity between acetic acid
and propionic acid,[19] these bands can be individually as-
signed. That is, the bands at 1300 and 1690 cmꢀ1 represent
molecularly adsorbed propionic acid, while the bands at
1430, 1472, and 1532 cmꢀ1 can be attributed to dissociated
11862
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 11859 – 11866

Photocatalysis
FULL PAPER
electrons (or other active free radicals) to accomplish the
transformation, since inactivated triplet O2 cannot directly
react with pyruvic acid. Therefore, we refer to the well-de-
fined electrochemical experiment in which electrons either
accumulated in the conduction band of TiO2 film electrode
or were readily taken away through tuning external cur-
rent.[15] As this is a dark experiment, the net reaction of pyr-
uvic acid with conduction band electrons (eꢀcb) and O2 can be
directly examined and the influence of holes/OH radicals is
avoided. In a typical process, a cathodic bias potential
(1.6 V vs. NHE) was continuously applied to a TiO2 film
electrode in an electrochemical cell that was filled with aer-
ated pyruvic acid solution. In this way, we can simulate the
photogeneration of electrons in the conduction band of TiO2
without UV illumination so that holes and/or OH radicals
are no longer generated. Consequently, we could directly
observe whether this pyruvic acid precursor reacts with ecꢀb
and dioxygen to implement decarboxylation (Figure 3).
Subsequently, we tried to identify which oxidant, OH rad-
ical (generated from H2O) or O2, delivers the new oxygen
atom in this a-keto group through utilizing H218O to carry
out the same experiment. We found that the peak clearly
shifted from 1772 to 1726 cmꢀ1 when we replaced ordinary
H2O with H218O (the lower spectrum of Figure 2). The shift
of 46(ꢂ4) cmꢀ1 induced by isotopic effect fits fairly well with
that of 43 cmꢀ1 predicted by Hookeꢁs law for this kind of C=
O stretch mode, indicating that such an a-keto group was
delivered from H2O through oxidative attack of holes and/
or OH radicals rather than dioxygen. In addition, the
oxygen exchange between a-keto group of pyruvic acid and
H218O is insignificant at the short timescale of 1 min
(Table S2 in the Supporting Information). Thus we conclude
that the pristine acids are indeed transformed into a-keto
acids prior to decarboxylation. This carbonylation mainly
originates from H2O participation by means of either an
electron transfer coupled with hydrolysis process [Eq. (9)]
or an a-H abstraction followed by an OH radical-rebound
process [Eqs. (10) and (11)]. Alternatively, a process of
“one-hole oxidation” followed by the release of another
electron from the substrate to the conduction band of TiO2
could also possibly accomplish this carbonylation [Eq. (12)],
similar to the “current doubling” effect observed in the pho-
tocatalytic oxidation of alcohols by TiO2.[13] Such an a-keto
acid intermediate should be easily destroyed and thus
always exhibited rather weak signal in the difference
DRIFTS spectra.
Figure 3. Temporal concentration changes of various acids in the electro-
&
chemical reactions using pyruvic acid ( ) and propionic acid (N) as sub-
*
strates. The concentrations of decomposed pyruvic acid ( ) and produced
~
acetic acid ( ) in the pyruvic acid reaction are also shown here. In the re-
actions, a cathodic bias potential (1.6 V vs. NHE) was continuously ap-
plied to the TiO2 film electrode without UV irradiation.
Within this dark electrochemical reaction, pyruvic acid
was indeed decomposed and converted to acetic acid slowly
but quantitatively (with almost 100% selectivity). That is,
only acetic acid, rather than any less than C2 acid (i.e.,
formic acid) or keto-, aldehyde-, dicarboxylic acids, was gen-
erated in this process. For comparison, we also took pro-
pionic acid as the initial substrate in the electrochemical re-
action to observe its behavior under the same conditions
and found no evident decarboxylation processes. These re-
sults adequately illustrate that the intermediate a-keto acid
is crucial for eꢀcb and O2 to give rise to the decarboxylation
of a carboxylic acid. In contrast to the situation of general
TiO2 photocatalytic reactions, H2O2 was not detected during
all the electrochemical reactions, which indicates that O2
may play a different role in the transformation of a-keto
acid to shorter-chain acid from that of capturing eꢀcb to be
converted into H2O2 or other active oxygen derivates.[2,3]
Accordingly, O2 definitely participates and reliably achieves
the decarboxylation of propionic acid through being activat-
Now that an intermediate in the decarboxylation process
has been successfully identified, and this intermediate was
determined to mainly originate from H2O participation, O2
would reasonably take part in the subsequent transforma-
tion; that is, the decarboxylation of this pyruvic acid inter-
mediate to an acetic acid product. Theoretically, the partici-
pation of O2 should be accompanied by photogenerated
Chem. Eur. J. 2010, 16, 11859 – 11866
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11863

J. Zhao, W. Ma et al.
ed by eꢀcb without the aid of holes and/or OH radicals as
soon as propionic acid is transformed into a-keto acid. Note
that the a-keto acid could also be transformed to shorter-
chain acid by the oxidation of holes and/or OH radicals;[21]
the transformation of the a-keto acid intermediate to prod-
uct acid can be realized by both hole/OH radical pathway
and electron/O2 pathway.
Based on the above discovered characteristics of the de-
carboxylation process, we can draw the outline of one
tandem reaction in which O2 is involved in the integrated
decarboxylation process: the substrate acid is transformed
to a-keto acid by holes or OH radicals, then the a-keto acid
further releases CO2 to afford a shorter-chain monoacid by
the synergistic effect of eꢀcb and O2. For the last step, the
clarification of the contribution of this reaction pathway to
practical TiO2 photocatalytic reactions is needed as this
transformation could also be realized by the oxidation of
holes and/or OH radicals. Thus, we designed an 18O2 (95%)
isotopic-labeling experiment to examine the transformation
of pyruvic acid to acetic acid by TiO2 photocatalysis
(Table 3), which enables us to directly observe the distribu-
substrates in TiO2 photocatalytic reactions has not been re-
ported before. A possible explanation for the origination of
this phenomenon is that a-keto acids may take the priority
of adsorption compared to non-substituted acids at TiO2 sur-
face defects,[22] at which four-coordinate Ti (Ti4c) sites are
available for the simultaneous coordination of substrates
and O2.[23] While at five-coordinate Ti (Ti5c) sites on the ter-
race of TiO2 surface, the a-keto acids and non-substituted
acids should be chemisorbed in identical fashion and affords
no significant differences in absorption priority.[24] Hender-
son et al. have similarly illustrated that substrate and O2 ad-
sorption play determining roles in the control of selectivity
between isobutene and isobutane as the product of gas-
phase photodecomposition of trimethyl acetate (TMA) on
TiO2ACHTNUGRTNEUNG
(110).[25] At a fixed initial TMA coverage and in the
presence of O2, the photodecomposition selectivity switches
from initially isobutene to the mixture of isobutene and iso-
butane and then back to predominantly isobutene within
this reaction, in which the latter selectivity change originates
from the enhanced impact of O2 at decreased substrate con-
centration.[25] However, further studies are evidently needed
to get the accurate conclusion about this process.
Table 3. The transformation of CH3COCOOH by TiO2 photocatalysis
using 18O2 and the isotopic distribution of produced CH3COOH.[a]
Conclusion
Through isotopic-labeling examinations combined with
DRIFTS and electrochemical experiments, we have illustrat-
ed an unexplored O2-involved pathway for the decarboxyla-
tion of saturated carboxylic acids by TiO2 photocatalysis. An
alternative pathway, involving eꢀcb and O2, exists in addition
to the direct hole/OH radical oxidation pathway and affords
the decarboxylation of carboxylic acids by TiO2 photocataly-
Conver-
Yield[c]
[%]
Percentage of various
sion[b] [%]
CH3COOH isotopomers[d] [%]
16O16O
16O18O
18O18O
1
2
3
4
5
44.3
69.0
84.9
96.1
100.0
29.5
44.9
52.9
56.3
55.9
99.8
91.5
84.4
77.3
75.9
0.2
8.5
15.6
22.7
24.1
0
0
0
0
0
1
2
ꢀ
sis. Through this way, the C C bond cleavage of the initial
saturated straight-chain carboxylic acid is accomplished in
two steps: at first, the pristine acid is oxidized to a-keto acid
by holes or OH radicals without loss of carbon atoms; and
then, the intermediate a-keto acid is decarboxylated to
shorter-chain acid through the eꢀcb/O2 process, in which an
oxygen atom of O2 is selectively incorporated into product
acid.
[a] Experimental conditions: solution of pyruvic acid in H216O (10 mL,
10 mm) in a 50 mL vessel. [b] The conversion of pyruvic acid. [c] The
fractional yield of acetic acid, which was calculated according to yield=c
(generated acetic acid)/c (converted pyruvic acid). [d] Calculated by the
integration of corresponding peak area obtained from mass spectrometry.
tion of oxygen atoms from O2 or H2O in newly established
carboxyl groups. The incorporation of an oxygen atom from
18O2 was clearly indicated by the generation of
CH3C16O18OH. Moreover, no single 18O18O profile in pro-
duced acetic acid illustrates that at least one oxygen atom of
the substrate a-keto acid remains in the new carboxyl group
Experimental Section
Materials and general instrumentation: Formic acid, acetic acid, propion-
ic acid, butanoic acid, and valeric acid were obtained from Acros. Pyruvic
acid was purchased from Sigma–Aldrich. All the acids were of analytical
grade purity and used as received. H218O was purchased from Jiangsu
Changshu Chemical Limited. The isotopic enrichment was 96% as deter-
mined by mass spectrometry. 18O2 was purchased from Cambridge Iso-
tope Laboratories, Inc. The isotopic enrichment was 95%. Conducting
glass sheets (<20 W per square, Laibao Technologies Limited, transpar-
ent conductive oxide coated glass plates of fluorine-doped SnO2) were
employed as substrates for TiO2 film coating. Deionized and doubly dis-
tilled water was used throughout this study.
1
2
ꢀ
as the old C C bond is cleaved, supporting our protocol
for the specific a-scission of CO2.
Similar to the trend of dioxygen incorporation in the cor-
responding H218O isotopic-labeling experiment (Table S3 in
the Supporting Information), a gradual increase in the pro-
portion of CH3C16O18OH (from 0.2 to 24.1%) was observed
along with the increase of conversion yield, corresponding
to a gradually enhanced participation of O2. To our knowl-
edge, such an unexpected phenomenon that the proportion
of O2 entrance into products depends on the conversion of
Kinetic runs were analyzed by ion chromatography (IC). Samples of ap-
proximately 0.5 mL were taken out at regular time intervals during the ir-
radiation. The concentrations of carboxylic acids and their degradation
11864
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 11859 – 11866

Photocatalysis
FULL PAPER
intermediates were measured by a DX-120 ion chromatograph with an
IonPac AS11 column (Dionex). The eluent was 0.8 mm NaHCO3 solu-
tion. The flow rate was 0.5 mLminꢀ1. The identification of the intermedi-
ates by IC was preformed by comparing the retention times with those of
standards. Total organic carbon (TOC) was measured by a Tekmar Dohr-
mann Apollo 9000 TOC analyzer. The detection of H2O2 was performed
using standard DPD-POD method.[26]
two-compartment reactor and standard two-electrode mode. The electro-
des consisted of a TiO2 film working electrode and a Pt foil counter elec-
trode. The two electrodes were separated in different compartments of
the reactor. The electrolyte used was 0.01m NaClO4. The active area of
the TiO2 electrode was 6 cm2. A potentiostat model DJS-292 was used to
bias the TiO2 electrode and to measure the electrochemical parameters.
A cathodic bias potential (1.6 V vs. NHE) was applied to the TiO2 elec-
trode. The initial concentration of all the substrate acids was 0.5 mm.
GC-MS analyses were carried out on a Finnigan Trace GC ultra gas chro-
matograph with a 25 m DB-5 column, coupled with a Finnigan Trace
DSQ mass spectrometer. The injector port was set for split operation at
2508C. The temperature program of the column was as follows: at 808C,
hold time=5 min. The isotopic enrichment of various acids was exam-
ined by EI-MS. Response factors for the isotopomers of these acids were
within a small experimental error and thus it was assumed that all iso-
topomers would also have identical response factors.
Acknowledgements
This work was financially supported by 973 projects (Nos. 2010CB933503
and 2007CB613306) and the NSFC (Nos. 20920102034, 20877076, and
20772129) and CAS.
Photocatalytic tests for kinetic and isotopic studies: The photocatalytic
degradation reactions of carboxylic acids for kinetic studies were all per-
formed with a TiO2 film (anatase), which was prepared and characterized
according to the method described in our earlier study.[15] The photoac-
tive area of the TiO2 film was 6 cm2. The reactions were carried out in
Pyrex vessels (50 mL) with a solution volume of 30 mL. The initial con-
centration of all the substrate acids was 2 mm. Irradiations were carried
out using a 100 W mercury lamp (Toshiba SHL-100UVQ-2).
[1] D. F. Ollis, E. Pelizzetti, N. Serpone, Environ. Sci. Technol. 1991, 25,
1522–1529.
[2] M. R. Hoffmann, S. T. Martin, W. Choi, D. W. Bahnemann, Chem.
Rev. 1995, 95, 69–96.
[3] A. Fujishima, X. Zhang, D. A. Tryk, Surf. Sci. Rep. 2008, 63, 515–
582.
In all the isotopic labeling experiments, a TiO2 film with a photoactive
area of 3 cm2 was used. The photocatalytic reactions of propionic acid
and pyruvic acid in H218O were carried out in Pyrex vessels (10 mL). A
solvent volume of 1.5 mL was used each time. The initial concentration
was 20 mm for propionic acid and 10 mm for pyruvic acid, respectively. Ir-
radiations were carried out using a 500 W Xe lamp (Beijing Trusttech
Co. Ltd.). The photocatalytic reactions of pyruvic acid using 18O2 were
performed in Pyrex vessels (50 mL). A solution volume of 10 mL at ini-
tial acid concentration of 10 mm was used each time. Before the 18O2 re-
action, the solution in the vessel was placed under vacuum with a pump
and then purged with argon three times to remove the air, followed by
saturation with 18O2. Irradiations were carried out with a 100 W mercury
lamp (Toshiba SHL-100UVQ-2). After the isotopic-labeling reaction was
stopped, the solution was extracted by ethyl ether. Then the ethyl ether
was evaporated and the resulting liquid was analyzed by GC-MS.
[4] N. Serpone, J. Martin, S. Horikoshi, H. Hidaka, J. Photochem. Pho-
tobiol. A 2005, 169, 235–251.
[5] B. Kraeutler, A. J. Bard, J. Am. Chem. Soc. 1978, 100, 5985–5992.
[6] J. Schwitzgebel, J. G. Ekerdt, H. Gerischer, A. Heller, J. Phys.
Chem. 1995, 99, 5633–5638.
[7] S. S. Lande, J. K. Kochi, J. Am. Chem. Soc. 1968, 90, 5196–5207.
[8] J. D. Bacha, J. K. Kochi, J. Org. Chem. 1968, 33, 83–93.
[9] J. M. Anderson, J. K. Kochi, J. Am. Chem. Soc. 1970, 92, 2450–2460.
[10] G. A. Russell, J. Am. Chem. Soc. 1957, 79, 3871–3877.
[11] a) T. Sakata, T. Kawai, K. Hashimoto, J. Phys. Chem. 1984, 88,
2344–2350; b) C. Guillard, J. Photochem. Photobiol. A 2000, 135,
65–75.
[12] a) S. Yurdakal, G. Palmisano, V. Loddo, V. Augugliaro, L. Palmisa-
no, J. Am. Chem. Soc. 2008, 130, 1568–1569; b) M. Zhang, C. Chen,
W. Ma, J. Zhao, Angew. Chem. 2008, 120, 9876–9879; Angew.
Chem. Int. Ed. 2008, 47, 9730–9733; c) A. M. Khenkin, R. Neu-
mann, J. Am. Chem. Soc. 2008, 130, 14474–14476; d) M. Zhang, Q.
Wang, C. Chen, L. Zang, W. Ma, J. Zhao, Angew. Chem. 2009, 121,
6197–6200; Angew. Chem. Int. Ed. 2009, 48, 6081–6084.
[13] T. Ohno, S. Izumi, K. Fujihara, Y. Masaki, M. Matsumura, J. Phys.
Chem. B 2000, 104, 6801–6803.
[14] a) F. Guzman, S. S. C. Chuang, J. Am. Chem. Soc. 2010, 132, 1502–
1503; b) M. A. Newton, M. D. Michiel, A. Kubacka, M. Fernandez-
Garcia, J. Am. Chem. Soc. 2010, 132, 4540–4541.
[15] J. Yang, C. Chen, H. Ji, W. Ma, J. Zhao, J. Phys. Chem. B 2005, 109,
21900–21907.
[16] Generally, the formation of short-lived intermediates from reactants
is necessary for the following generation and accumulation of rela-
tively stable intermediates or products in stepwise chemical reac-
tions. For details, see: F. A. Carey, R. J. Sundberg, Advanced Organ-
ic Chemistry: Part A: Structure and Mechanisms, 2nd ed., Plenum
Press, New York, 1984.
DRIFTS measurements: A Thermo Nicolet 6700 FT-IR spectrometer
equipped with a mercury cadmium telluride (MCT) detector was used
for the diffuse reflectance FTIR measurements (DRIFTS). A praying
mantis diffuse reflectance accessory and a reaction cell equipped with a
heater (Harrick Scientific) form the reaction system. Degussa P25
powder was housed at a sample cup inside the reaction cell. A dome cov-
ered the cell containing three windows. Two of the windows were made
of ZnSe and permitted entry and exit of the infrared beam and the dif-
fusely scattered infrared light while the third one was used for UV irradi-
ation (quartz). A constant flow of 100 sccm of 20% O2 and 80% N2 was
maintained throughout the reaction. UV irradiations were performed
with a 500 W Xe lamp (Beijing Trusttech Co. Ltd.).
The TiO2 powder was heated to 3608C and kept at this temperature for
20 min to provide a clean catalyst surface. Then the catalyst was cooled
to 308C and the spectrum of the TiO2 surface was collected. Interfero-
grams over the range of 650 to 4000 cmꢀ1 were recorded by averaging
100 scans with a resolution of 4 cmꢀ1. A processing time of 0.6 min was
needed for one spectrum. Before every experiment the background spec-
trum was collected and then subtracted from subsequent spectra after the
introduction of reaction substrates. Afterwards a pulse of the substrate
acid (100 mmol per g catalyst) was injected into the reaction cell (as the
[17] M. Minella, M. G. Faga, V. Maurino, C. Minero, E. Pelizzetti, S. Co-
luccia, G. Martra, Langmuir 2010, 26, 2521–2527.
[18] J. B. Lambert, H. F. Shurvell, D. A. Lightner, R. G. Cooks, Organic
Structural Spectroscopy, Prentice-Hall, London, 1998.
[19] a) L. F. Liao, C. F. Lien, J. L. Lin, Phys. Chem. Chem. Phys. 2001, 3,
3831–3837; b) W. Rachmady, M. A. Vannice, J. Catal. 2002, 207,
317–330; c) M. J. Backes, A. C. Lukaski, D. S. Muggli, Appl. Catal.
B 2005, 61, 21–35.
[20] J. Fan, J. T. Yates, J. Phys. Chem. 1994, 98, 10621–10627.
[21] E. R. Carraway, A. J. Hoffman, M. R. Hoffmann, Environ. Sci. Tech-
nol. 1994, 28, 786–793.
real density of five-coordinate Ti surface sites on the ideal TiO CHUTNGTRENN(NUG 110) sur-
2A
[27]
face is about 10ꢀ5 molmꢀ2
,
less than one monolayer of acid molecules
could be obtained in the experiment considering the specific BET surface
area of about 50 m2 gꢀ1 for P25). Spectra were continuously collected
with gas flushing excess acids away from the TiO2 surface until they get
stable. For the reaction with H216O or H218O, a pulse of H216O or H218O
was respectively injected into the reaction cell.
Electrochemical measurements: The electrochemical reactions of the
acids with the conduction band electrons and O2 were performed in a
[22] J. A. Dean, Langeꢀs Handbook of Chemistry, 15th ed., Science Press,
New York, 1998.
Chem. Eur. J. 2010, 16, 11859 – 11866
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11865

J. Zhao, W. Ma et al.
[23] a) G. Lu, A. Linsebigler, J. T. Yates, J. Phys. Chem. 1995, 99, 7626–
7631; b) A. Linsebigler, G. Lu, J. T. Yates, J. Phys. Chem. 1996, 100,
6631–6636; c) M. A. Henderson, J. Phys. Chem. B 2005, 109, 12062–
12070.
[26] T. Wu, G. Liu, J. Zhao, H. Hidaka, N. Serpone, J. Phys. Chem. B
1999, 103, 4862–4867.
[27] M. A. Henderson, J. Phys. Chem. B 2004, 108, 18932–18941.
[24] U. Diebold, Surf. Sci. Rep. 2003, 48, 53–229.
[25] M. A. Henderson, J. M. White, H. Uetsuka, H. Onishi, J. Catal.
2006, 238, 153–164.
Received: June 16, 2010
Published online: September 20, 2010
11866
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 11859 – 11866