Catalytic routes and oxidation mechanisms in photoreforming of polyols

Article abstract of DOI:10.1016/j.jcat.2016.08.009

Photocatalytic reforming of biomass-derived oxygenates leads to H₂ generation and evolution of CO₂ via parallel formation of organic intermediates through anodic oxidations on a Rh/TiO₂ photocatalyst. The reaction pathways and kinetics in the photoreforming of C₃–C₆ polyols were explored. Polyols are converted via direct and indirect hole transfer pathways resulting in (i) oxidative rupture of C–C bonds, (ii) oxidation to α-oxygen functionalized aldoses and ketoses (carbonyl group formation) and (iii) light-driven dehydration. Direct hole transfer to chemisorbed oxygenates on terminal Ti(IV)-OH groups, generating alkoxy-radicals that undergo ?-C–C-cleavage, is proposed for the oxidative C–C rupture. Carbonyl group formation and dehydration are attributed to indirect hole transfer at surface lattice oxygen sites [Ti?O?Ti] followed by the generation of carbon-centered radicals. Polyol chain length impacts the contribution of the oxidation mechanisms favoring the C–C bond cleavage (internal preferred over terminal) as the dominant pathway with higher polyol carbon number.

Full text of DOI:10.1016/j.jcat.2016.08.009

Journal of Catalysis xxx (2016) xxx–xxx
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Catalytic routes and oxidation mechanisms in photoreforming of polyols
a,b,
Kai E. Sanwald a, Tobias F. Berto a, Wolfgang Eisenreich a, Oliver Y. Gutiérrez a, , Johannes A. Lercher
⇑
⇑
a TU München, Department of Chemistry and Catalysis Research Center, Lichtenbergstrasse 4, D-85747 Garching, Germany
b Pacific Northwest National Laboratory, Institute for Integrated Catalysis, P.O. Box 999, Richland, WA, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Photocatalytic reforming of biomass-derived oxygenates leads to H2 generation and evolution of CO2 via
parallel formation of organic intermediates through anodic oxidations on a Rh/TiO2 photocatalyst. The
reaction pathways and kinetics in the photoreforming of C3–C6 polyols were explored. Polyols are con-
verted via direct and indirect hole transfer pathways resulting in (i) oxidative rupture of C–C bonds,
Received 20 July 2016
Revised 19 August 2016
Accepted 21 August 2016
Available online xxxx
(ii) oxidation to
a-oxygen functionalized aldoses and ketoses (carbonyl group formation) and (iii)
light-driven dehydration. Direct hole transfer to chemisorbed oxygenates on terminal Ti(IV)-OH groups,
generating alkoxy-radicals that undergo ß-C–C-cleavage, is proposed for the oxidative C–C rupture.
Carbonyl group formation and dehydration are attributed to indirect hole transfer at surface lattice oxy-
gen sites [Tiꢀ ꢀ ꢀOꢀ ꢀ ꢀTi] followed by the generation of carbon-centered radicals. Polyol chain length impacts
the contribution of the oxidation mechanisms favoring the C–C bond cleavage (internal preferred over
terminal) as the dominant pathway with higher polyol carbon number.
Keywords:
Photocatalysis
TiO2
H2 production
Photoreforming
Glycerol
Polyols
Oxidation mechanism
Ó 2016 Elsevier Inc. All rights reserved.
1. Introduction
depth [9–12], the role of the chemically coupled reactions have
been hardly explored mechanistically and kinetically. In such reac-
Photocatalytic H2 generation from bio-derived oxygenates
(‘photoreforming’) is a desired pathway for the production of a
chemical energy carrier utilizing solar energy as it lowers the
energy requirements compared to water cleavage [1–5]. Aqueous
glycerol, an abundant by-product from triglyceride transesterifica-
tion, and polyol-containing wastewaters, e.g. from industry or
catalytic upgrading conceptually, could be suitable feedstocks for
photoreforming [1–4,6]. Thereby valorization and/or purification
of those resources are coupled to the production of H2. Photore-
forming benefits from a narrow energetic separation of the two
redox half-reactions (E0(H+/H2) = 0 V; e.g. for glycerol photore-
forming E0(CO2/C3H8O3) = ꢁ0.004 V vs. NHE [7]), which provides
a large overpotential at the anode facilitating cathodic H2 evolu-
tion. Moreover, compared to overall water splitting, substitution
of the oxygen evolution reaction for anodic oxygenate oxidation
to CO2 eliminates the need of separation and the back-reaction of
H2 and O2 [8].
tions, the co-catalyst decorated semiconductor acts as coupled
micro-electrochemical cell [13–15]. Anodic half-reactions are
thought to occur on the semiconductor surface as a consequence
of interfacial transfer of photogenerated holes through either direct
transfer to the oxygenate or via an indirect mechanism, e.g. medi-
ated by ÅO(H)-radicals [6,16,17]. The co-catalyst serves as cathode,
electron trap, and, thus, H2 evolution site and does not participate
in the anodic half-reactions [18,19]. Yet, H2-evolution and thus
oxygenate degradation rates (due to charge balance) are influenced
by co-catalyst nature [7], loading and particle size [7,20] as well as
its composition and morphology [21,22]. The co-catalyst could
even aid the suppression of surface back-reactions at the cathode
[8,23]. These factors provide essential means for optimization
of the efficiency of electron transfer at the semiconductor/
co-catalyst interface.
It is established that H2-evolution rates over TiO2-based photo-
catalysts depend on oxygenate nature and coverage [3,7,24–28].
However, there is ambiguity in the anodic pathways and mecha-
nisms toward full oxidation to CO2. Initial anodic transformations
of polyols over metal/metal oxide loaded TiO2 were proposed to
result in the formation of the corresponding aldehydes, which
undergo either decarbonylation followed by water-gas shift [17]
or cleavage of formic acid [19]. Oxidation to carboxylic acids that
decarboxylate [6,16] or initial polyol dehydration to compounds
with non-functionalized carbon atoms [17] was also suggested
While the electron-hole recombination and charge carrier
transport to the surface and their relation to physicochemical
properties of the photocatalyst have been explored with great
⇑
Corresponding authors at: TU München, Department of Chemistry and Catalysis
Research Center, Lichtenbergstrasse 4, D-85747 Garching, Germany (J.A. Lercher).
Fax: +49 89 28913544.
E-mail addresses: oliver.gutierrez@mytum.de (O.Y. Gutiérrez), johannes.
lercher@ch.tum.de (J.A. Lercher).
http://dx.doi.org/10.1016/j.jcat.2016.08.009
0021-9517/Ó 2016 Elsevier Inc. All rights reserved.
Please cite this article in press as: K.E. Sanwald et al., Catalytic routes and oxidation mechanisms in photoreforming of polyols, J. Catal. (2016), http://dx.
doi.org/10.1016/j.jcat.2016.08.009

2
K.E. Sanwald et al. / Journal of Catalysis xxx (2016) xxx–xxx
prior to C–C-cleavage. On the other hand, photoreforming of linear
aldehydes was reported to involve sequential cleavage of formic
acid to form the C1-deficient aldehyde in a single reaction pathway
[19,29].
reactor at water level is 8.08 ꢀ 1017 sꢁ1 (k < 390 nm). Typically,
75 mg of photocatalyst were ultrasonically dispersed in 100 mL
of a 20 mM aqueous reactant solution. The system was deaerated
by four consecutive evacuations and Ar filling cycles. All reactions
were carried out at 288 K and an Ar pressure of 1 bar.
Evolved gases during photocatalytic reactions were analyzed
online via gas chromatography (Shimadzu, GC 2010 Plus) on a
Chromosorb 101 and a Molsieve 5 Å column with a TCD, FID and
a methanizer catalyst unit. Ar was used as a carrier gas.
During aerobic photooxidation of oxygenates, multiple reaction
pathways were accounted for by a two-site model for direct and
indirect mechanisms [30–32]. Holes trapped at surface lattice oxy-
gen sites [Tiꢀ ꢀ ꢀOꢀ ꢀ ꢀTi] abstract a H-atom from a C–H bond whereas
direct hole transfer yields alkoxy radicals chemisorbed on terminal
OH-groups. We hypothesize that under photoreforming conditions
(anaerobic environment and hydrogen evolving) similar mecha-
nisms operate, which determine the routes toward complete oxi-
dation. We have shown in a preceding contribution that the
anodic transformations of ethylene glycol during photoreforming
may be rationalized on this basis [18]. We further propose that
identical anodic transformations are followed across families of
compounds. However, the impact of molecular polyol structure
and associated surface adsorption complexes on the contribution
of the different reaction pathways to the complete oxidation is lar-
gely unknown.
Aliquots of the liquid-phase were manually taken at regular
times from the photocatalyst suspension via a sample valve. The
solid was separated by filtering through 0.2 lm nylon filters.
Liquid-phase species were quantitatively analyzed by 1H NMR
spectroscopy. Additionally, high-performance liquid chromatogra-
phy (HPLC) analysis was performed for photoreforming experi-
ments with C4–C6 oxygenates.
Reactant conversions XA were calculated according to XAð%Þ ¼
NA
ð1 ꢁ ðtÞÞ ꢀ 100, where NA(t) is the time-dependent molar amount
NA;0
of reactant A and NA,0 is the initial molar amount of reactant A,
i.e. present at the beginning of the reaction.
Here, we explore the mechanisms and kinetics of photoreform-
ing of C3–C6 polyols on benchmark TiO2 P 25 with Rh as co-catalyst.
On the basis of quantitative analysis of gas- and liquid-phase spe-
cies we establish general relationships between structural reactant
functional groups, anodic reaction pathways and oxidation mecha-
nisms. We provide rationalization of the impact of anodic surface
chemistry on the photoreforming kinetics in terms of oxygenate
conversion and associated H2-evolution rates.
2.2.2. 1H NMR analysis
For quantitative 1H NMR measurements reactor aliquots were
mixed in a 1:1 volume ratio with an external standard that con-
tained 20 mM 1,3,5-trihydroxybenzene in D2O adjusted to pH 3
with DCl. 1H NMR spectra with water signal suppression were
recorded at 305 K using an Avance III HD 500 System (Bruker Bios-
pin, Rheinstetten, Germany) with an UltraShield 500 MHz magnet
(11.75 T) and a BBI 500 S2 probe head (5 mm, inverse 1H/X with Z-
gradient). The resonance frequency of 1H was 500.13 MHz. The
spectra were acquired using the one-dimensional NOESY sequence
from the Bruker library ‘‘noesygppr1d” with presaturation of the
residual water signal during the relaxation delay and the mixing
time using spoil gradients. Longitudinal relaxation times (T1) were
determined by the inversion recovery pulse method. Relaxation
delay and acquisition time were set to 26 s and 4.1 s, respectively.
The sum of the latter corresponds to at least three times T1 of the
slowest relaxing 1H-nucleus (formic acid) and ensures quantitative
analysis. Typically, 64 or 128 scans, with 64 k data points were col-
lected. An exponential window function with a line broadening of
0.2 Hz was applied prior to Fourier transformation and the spectra
were manually phased, baseline corrected and integrated using
Mestre-C 8.1.1 software package. Liquid-phase species were iden-
tified according to their chemical shifts (referenced to the internal
standard, see SI-Table 1) and in comparison with spectra of com-
mercial references recorded under identical conditions. Overlap-
ping signals were deconvoluted using Lorentzian-Gaussian shape
type fitting functions. Quantification was done on the basis of the
integrated signal intensities which were further calibrated against
prepared solutions of known concentration in order to account for
signal damping due to the water-suppression in close proximity to
the water resonance frequency.
2. Experimental
A comprehensive list of compounds used during the study, and
detailed experimental procedures regarding the characterization of
the photocatalyst are compiled in the Supporting Information.
Briefly, experiments are conducted over AEROXIDEÒ TiO2 P 25
(referred to as TiO2 hereafter), commonly employed as benchmark
semiconductor, decorated with nanoparticulate Rh as a co-catalyst
(1 wt.% loading, particle size 1.9 nm ( 0.7 nm), dispersion 57%).
Further physicochemical properties of the materials are described
in the Supporting Information.
2.1. Photocatalyst preparation
The 1 wt.% Rh-decorated P 25 photocatalyst, hereafter referred
to as Rh/TiO2, was prepared via incipient wetness impregnation.
TiO2 P 25 was dried under a static air atmosphere at 473 K for
2 h prior to impregnation. Subsequently, the semiconductor was
treated with appropriate amounts of an aqueous solution of RhCl3-
ꢀxH2O in ultrapure water. The resulting precursor was kept in syn-
thetic air at 383 K for 1 h (100 mL minꢁ1, 5 K minꢁ1) and heated to
623 K (100 mL minꢁ1, 5 K minꢁ1) for 1 h. The material was allowed
to cool to room temperature before treatment in H2 at 623 K
(100 mL minꢁ1, 5 K minꢁ1) for 1 h. A TiO2 reference was prepared
subjecting as-received TiO2 P 25 to the same temperature and
gas treatments described before.
2.2.3. 13C NMR analysis
For 13C NMR measurements, reactor aliquots were again mixed
in a 1:1 volume ratio with D2O adjusted to pH 3 with DCl. 13C NMR
spectra were recorded at 300 K using an Avance III 500 System
(Bruker Biospin, Rheinstetten, Germany) with an UltraShield
500 MHz magnet (11.75 T) and a Cryo-QNP probe head (5 mm,
direct 13C/31P/29Si/19F with Z-gradient). The resonance frequency
of 13C was 125.07 MHz. The spectra were acquired using the one-
dimensional ‘‘zgpg” sequence from the Bruker library using
waltz16 1H-decoupling and a 30° 13C-pulse. Relaxation delay and
acquisition time were set to 4 s and 1.2 s, respectively. Typically,
1024 scans with 64 k data points were collected. An exponential
2.2. Photocatalytic test
2.2.1. Photoreforming experiments
Kinetic experiments were performed in a Pyrex top-irradiation
photo-reactor connected to a closed gas-circulation system [18].
The setup is equipped with facilities for online gas-analysis and
liquid-phase sampling. Irradiation is provided by a 300 W Xe lamp
with a cold mirror 1 (CM 1). A water filter with quartz windows
closes the top of the photo-reactor. The photon flux within the
Please cite this article in press as: K.E. Sanwald et al., Catalytic routes and oxidation mechanisms in photoreforming of polyols, J. Catal. (2016), http://dx.
doi.org/10.1016/j.jcat.2016.08.009

K.E. Sanwald et al. / Journal of Catalysis xxx (2016) xxx–xxx
3
window function with a line broadening of 1 Hz was applied prior
to Fourier transformation and the spectra were manually phased,
baseline corrected and integrated using Mestre-C 8.1.1 software
package. Signals were assigned according to their chemical shifts
in comparison with spectra of commercial references recorded
under identical conditions.
H2/CO2-ratios (Fig. 1B), which decreased throughout the reac-
tion period, were above the theoretical ratio of 2.3 expected from
Scheme 1 (ratio of 3.8 after 12 h). This observation indicates that
full oxidation of glycerol to CO2 occurs with the formation of
organic intermediates in the liquid-phase. A list of molecular struc-
tures and overall reaction equations are furthermore given in SI-
Table 3. Analysis of the aqueous-phase (Fig. 2) revealed that
formaldehyde (613 lmol after 12 h) was generated as the major
2.2.4. HPLC analysis
species linearly accumulating in solution with reaction time. The
oxidation of formaldehyde is slow in the presence of other oxy-
genates during photoreforming due to its comparatively low
apparent adsorption constant on TiO2 [18].
Complementary HPLC analysis on an Agilent HP series 100
equipped with a refractive index (RI, 313 K) and a diode array
ultraviolet (UV) detector was performed for quantification of ery-
thritol, arabitol and sorbitol. Liquid-phase species were separated
on an ion-exclusion column (Rezex ROA, Phenomenex) held at
343 K. 0.005 M H2SO4 was used as the mobile phase flowing at
0.5 mL minꢁ1. Reactor aliquots were mixed in a 4:1 volume ratio
with an aqueous solution containing 1,3-propylene glycol as inter-
nal standard before injection. Integrated signal areas were cali-
brated against prepared solutions of known concentration.
Combined determination of gas- and liquid-phase concentra-
tions of reaction products and intermediates allowed to quantita-
tively follow photoreforming kinetics with closed mass-balances.
Three C3-species are formed from glycerol in low amounts:
glyceraldehyde (GAD, 24
(DHA, 45 mol after 12 h) generated via two electron oxidation
of a primary or secondary carbon atom of glycerol, respectively,
as well as hydroxyacetone (HA, 5 mol after 12 h) formally result-
lmol after 12 h) and dihydroxyacetone
l
l
ing from a light-driven dehydration. A steady-state concentration
of GAD in solution is attained after 4 h indicating consumption of
GAD by consecutive reactions. Similarly, the amount of DHA
tended to level off with increasing reaction time while the concen-
tration of HA increased gradually with reaction time. In addition to
small amounts of formic acid (25
species, i.e. glycolaldehyde (275 mol after 12 h), acetaldehyde
(30 mol after 12 h) (whose temporal profile is indicative of a sec-
ondary reaction product), and traces of acetic acid (2 mol after
lmol after 12 h), three C2-
3. Results and discussion
l
l
3.1. Kinetics of photoreforming of glycerol and C3 intermediates
l
12 h) were detected. Interconversion and/or equilibration of
GAD and DHA via keto-enol tautomerism was not observed under
reaction conditions. We concluded that at least three parallel reac-
tion pathways for glycerol oxidation exist, because three non-
interconvertible C3-products were observed (Scheme 2). Further
analysis indicated that a fourth pathway exists yielding glycolalde-
hyde and formaldehyde directly from glycerol (Path 4, vide infra).
3.1.1. Glycerol photoreforming
Photoreforming of glycerol (Scheme 1) proceeded with contin-
uous evolution of H2 and CO2 (1405 lmol and 366 lmol, respec-
tively after 12 h reaction time) reaching a conversion of 39% after
12 h. The H2-evolution rates declined over time from a maximum
value of 143 l l
mol hꢁ1 to about 95 mol hꢁ1 after 12 h (Fig. 1A).
This decline is related to a first-order dependence on concentration
(SI-Fig. 3, see kinetic model in Section 3.2 and Supporting Informa-
tion). Chemical transformations were not observed in absence of
illumination.
3.1.2. Path 1: Glyceraldehyde (GAD) photoreforming
Photoreforming of GAD yielded H2 and CO2 (1264
lmol and
387 mol, respectively after 12 h) in the gas-phase. In contrast to
l
Scheme 1. Cathodic, anodic half-reactions and overall photoreforming reaction of glycerol.
Fig. 1. (A) Course of H2-evolution rates during photoreforming of C3-oxygenates. (B) Course of H2/CO2-ratios. Curved lines are drawn as a guide to the eye. Dashed horizontal
lines in (B) represent the stoichiometric ratios expected from the overall reaction equations. Reaction conditions: 75 mg photocatalyst, 100 mL aqueous oxygenate solution
(20 mM), 288 K, 1 bar Ar, 300 W Xe-lamp (CM1).
Please cite this article in press as: K.E. Sanwald et al., Catalytic routes and oxidation mechanisms in photoreforming of polyols, J. Catal. (2016), http://dx.
doi.org/10.1016/j.jcat.2016.08.009

4
K.E. Sanwald et al. / Journal of Catalysis xxx (2016) xxx–xxx
Fig. 2. Analysis of species generated in anodic half-reactions during glycerol photoreforming. (A) Course of the amount of glycerol. Quantities of different (B) C3-, (C) C2- and
(D) C1-species generated. Reaction conditions: 75 mg photocatalyst, 100 mL aqueous glycerol solution (20 mM), 288 K, 1 bar Ar, 300 W Xe-lamp (CM1).
glycerol photoreforming, H2-evolution from GAD occurred with
constant rate over time (Fig. 1A). Analysis of the liquid-phase spe-
cies (Fig. 3) showed glycolaldehyde (614
dominant species accompanied by formic acid (steady-state
amount of 421 mol after 12 h), small amounts of formaldehyde
(158 mol after 12 h) and traces of acetaldehyde and acetic acid
(18 mol and 7 mol after 12 h, respectively). Oxidation of GAD
lmol after 12 h) as the
l
l
l
l
did not yield glyceric acid as intermediate prior to C–C-scission
as proposed in previous studies on photocatalytic conversion of
glycerol [6,16,33] based on observations from electrocatalytic oxi-
dation of glycerol and GAD [34–38].
The temporal profiles indicate that anodic conversion of GAD
involves C–C-cleavage to form glycolaldehyde and a C1-species.
Scheme 2. Initial conceptual reaction pathways for glycerol photoreforming over
Rh/TiO2 on the basis of identification of three C3-species generated in the liquid-
phase.
Fig. 3. Course of anodic reaction products formed during glyceraldehyde photoreforming. (A) Course of amount of glyceraldehyde. (B) Temporal profiles of C2- and C1-species
generated. Reaction conditions: 75 mg photocatalyst, 100 mL aqueous glyceraldehyde solution (20 mM), 288 K, 1 bar Ar, 300 W Xe-lamp (CM1).
Please cite this article in press as: K.E. Sanwald et al., Catalytic routes and oxidation mechanisms in photoreforming of polyols, J. Catal. (2016), http://dx.
doi.org/10.1016/j.jcat.2016.08.009

K.E. Sanwald et al. / Journal of Catalysis xxx (2016) xxx–xxx
5
As the amounts of CO2 and trace CO (4.4
duced do not account for the amount of GAD converted (711
l
mol CO after 12 h) pro-
centrations similar to those observed during the reforming of
DHA alone.
lmol
after 12 h), the C1 moiety cleaved from GAD must be contained in
formic acid, which forms at equimolar concentrations than glyco-
ladehyde at low reaction time. Subsequent photoreforming of gly-
coladehyde proceeds via a selective C–C bond cleavage yielding
equimolar amounts of formic acid and formaldehyde [18,19]. As
the apparent adsorption constants of both species differ by two
orders of magnitude (see Chapter 3.2) consecutive oxidation of for-
mic acid to CO2 occurs more rapidly than conversion of formalde-
hyde to formic acid. Indeed, formaldehyde conversion is negligible
under reaction conditions while CO2 is generated from formic acid
oxidation [18]. This supports our conclusion of the C–C bond cleav-
age of GAD proceeding via formation of formic acid and glycolalde-
hyde (see the Supporting Information, see also SI-Fig. 4). The
reaction pathway for full anodic conversion of glycerol to CO2 via
GAD (Path 1), glycolaldehyde and the respective C1-species is
depicted in Scheme 3.
Thus, we conclude that the photoreforming of DHA proceeds via
glycolic acid as C2-intermediate in the first reaction step simulta-
neously generating one H2-equivalent at the cathode (Scheme 4).
Consecutively, glycolic acid undergoes fast oxidative decarboxyla-
tion to form second equivalents of formaldehyde and H2
(Scheme 4).
Our reaction pathway studies are guided by the principle that
stable organic species may be formed after two electron oxidation.
In contrast, the formation of glycolaldehyde and formaldehyde
from DHA requires the formation of a new C-H bond at the car-
bonyl carbon atom, which corresponds to a redox-neutral transfor-
mation, i.e., one electron oxidation of a primary carbon of DHA to
form formaldehyde along with one electron reduction of the sec-
ondary carbon. Under photoreforming conditions, a redox-neutral
transformation of this kind must be slow because the availability
of electrons at the semiconductor surface is expected to be greatly
limited in the presence of the co-catalyst acting as an electron trap.
An alternative redox-neutral process would involve diffusion of a
radical oxidant species to the metal, where the reduction step
would occur competing with H2-evolution. Thus, generation of gly-
colaldehyde as relevant C2-intermediate in the reforming of DHA is
considered a minor side reaction.
3.1.3. Path 2: Dihydroxyacetone (DHA) photoreforming
The ratio of the amounts of H2 and CO2 (1247
lmol and
686 mol, respectively after 12 h) evolved determined for DHA
l
photoreforming was close to two during the entire reaction
(Fig. 1B), i.e., the ratio corresponding to the overall reaction equa-
During photoreforming of the next higher homologous ketose of
DHA (erythrulose, refer to SI-Figs. 6 and 7), also the characteristic
H2/CO2-ratio of two was observed, while the corresponding
C3-intermediate acid could not be detected. Thus, in the photore-
forming of linear ketoses intermediate Cnꢁ1 acid, if formed, rapidly
reacts further yielding a Cnꢁ2 aldehyde as apparent primary
product.
tion (SI-Table 3). Along with formic acid (11
liquid-phase contained formaldehyde in quantities, which exceed
the amount of DHA converted (1038 mol and 656 mol, respec-
tively after 12 h) (Fig. 4). This indicates that more than one equiv-
alent of formaldehyde was formed from DHA and consecutive
oxidations. The H2/CO2 ratio and quantities of liquid-phase prod-
ucts indicate that DHA photoreforming proceeds via formation of
two equivalents of formaldehyde and one equivalent of CO2. Only
small amounts of C2-species, i.e. glycolaldehyde and glycolic acid
lmol after 12 h), the
l
l
3.1.4. Path 3: Hydroxyacetone (HA) photoreforming
The amounts of H2 and CO2 (222 lmol and 57 lmol, respec-
(24 lmol and 2 lmol, respectively after 12 h), were present. Dur-
tively after 12 h) evolved during photoreforming of HA are about
80% lower than those from DHA, GAD, and glycerol. In general,
H2-evolution rates are lower for compounds carrying non-
oxygen-functionalized carbons compared to their counterparts
with maximum number of oxygen-functionalities. In particular,
ing reforming of an equimolar mixture of DHA and glycolic acid,
the latter was converted at much higher rates than the former
(SI-Fig. 5), whereas formaldehyde and CO2 were observed at con-
a
-oxygen functionalization is required for efficient hole transfer
[18,24,25,39]. Note that the amounts of H2 obtained from photore-
forming of 20 mM solutions of 1,2- and 1,3-propanediol (997 mol
and 447 mol after 12 h, respectively) were lower than those from
l
l
glycerol (1405
lmol after 12 h) under identical reaction
conditions.
Liquid-phase analysis (Fig. 5) during HA photoreforming dis-
closed acetic acid as a main primary product and formaldehyde
(99
lmol and 56
l
mol, respectively, after 12 h). Formic acid is vir-
tually absent (1
lmol after 12 h). Acetaldehyde, which is known to
undergo slow consecutive oxidation to acetic acid, and hence to
accumulate in liquid-phase [40,41], was not observed. This sug-
gests that acetic acid was indeed formed as a primary product from
HA. Consecutive oxidation of acetic acid may progress either to
CO2, or to CH4 and CO2 (Photo-Kolbe reaction) [18,42–44] sug-
gested by the detection of CH4 in the gas-phase (5 lmol CH4 after
12 h). In line with the concluded decomposition of HA via acetic
acid (associated to low H2-evolution rates), considerably lower
H2-evolution rates from acetic acid/acetate solutions in compar-
ison with C1–C3 alcohols and polyols have been obtained over
noble metal decorated TiO2 [25,44].
In analogy to DHA photoreforming, we conclude that formalde-
hyde is generated with acetic acid upon C–C scission of HA. The
reaction pathway for glycerol photoreforming via intermediate
formation of HA is represented in Scheme 5. HA contents during
glycerol photoreforming were the smallest among the C3-
intermediates generated while the consecutive reforming reactions
Scheme 3. Reaction pathway for photocatalytic reforming of glycerol (Path 1):
Glycerol is oxidized to glyceraldehyde. Subsequent reaction steps involve sequen-
tial cleavage of formic acid through intermediate glycolaldehyde (Cnꢁ1 aldehyde)
and formaldehyde (Cnꢁ2 aldehyde) toward full oxidation to CO2.
Please cite this article in press as: K.E. Sanwald et al., Catalytic routes and oxidation mechanisms in photoreforming of polyols, J. Catal. (2016), http://dx.
doi.org/10.1016/j.jcat.2016.08.009

6
K.E. Sanwald et al. / Journal of Catalysis xxx (2016) xxx–xxx
Fig. 4. Course of anodic reaction products formed during dihydroxyacetone photoreforming. (A) Course of amount of dihydroxyacetone. (B) Temporal profiles of C2- and C1-
species generated. (C) Expanded view of (B) to show the reaction products generated in low amounts. Reaction conditions: 75 mg photocatalyst, 100 mL aqueous
dihydroxyacetone solution (20 mM), 288 K, 1 bar Ar, 300 W Xe-lamp (CM1).
3.2. Path 4: Direct oxidative C–C-cleavage of glycerol
Photoreforming of glycerol DHA, GAD, and HA showed different
initial H2-evolution rates under identical reaction conditions. How-
ever, the rates converge towards
a
maximum value of
200
l
mol hꢁ1 with increasing initial concentration in agreement
with a Langmuir-type kinetics (SI-Fig. 8). This agrees also well with
photocatalytic conversion of C2-oxygenates over Rh/TiO2 [18].
Assuming the intrinsic rate constants to be substrate independent,
this indicates that photoreforming rates of oxygenates primarily
depend on their apparent adsorption constants and associated sur-
face coverages (under constant illumination conditions) irrespec-
tive of the nature of the charge carrier transfer mechanism.
Hence, a model based on Langmuir adsorption applies to describe
the kinetics of photoreforming of the studied oxygenates as quan-
titatively outlined in the Supporting Information.
Direct oxidative C–C-cleavage of glycerol to yield glycolalde-
hyde and formaldehyde, which had been proposed during aerobic
conversion of glycerol over non-metallized TiO2 P 25 [31], had to
be included in the model (besides the parallel reaction pathways
from glycerol that yield C3 oxygenates) in order to achieve reason-
able agreement with the experimental data. Evidence for this
transformation is provided from the product distributions obtained
for photoreforming of higher polyols and is discussed in
Section 3.3.
Kinetic modeling confirmed a negligible contribution of HA for-
mation from glycerol and successive oxidations (Path 3). In conclu-
sion, we propose the reaction network depicted in Scheme 6 for
glycerol photoreforming from gas-, liquid-phase analysis and
kinetic modeling. Anodic glycerol conversion predominantly
involves oxidative C–C-scission (61%) to yield glycolaldehyde and
formaldehyde. Oxidations of primary and secondary carbons,
Scheme 4. Reaction pathway for photocatalytic reforming of glycerol (Path 2): Two
electron oxidation of glycerol to form dihydroxyacetone. Consecutive dihydroxy-
acetone conversion via glycolic acid yields CO2 and two equivalents of
formaldehyde.
of HA are significantly slower compared to GAD and DHA. Thus, the
light induced dehydration of glycerol to HA is a side-reaction.
Please cite this article in press as: K.E. Sanwald et al., Catalytic routes and oxidation mechanisms in photoreforming of polyols, J. Catal. (2016), http://dx.
doi.org/10.1016/j.jcat.2016.08.009

K.E. Sanwald et al. / Journal of Catalysis xxx (2016) xxx–xxx
7
Fig. 5. Course of anodic reaction products formed during photoreforming of hydroxyacetone. (A) Course of amount of hydroxyacetone. (B) Temporal profiles of C2- and C1-
species generated. Reaction conditions: 75 mg photocatalyst, 100 mL aqueous hydroxyacetone solution (20 mM), 288 K, 1 bar Ar, 300 W Xe-lamp (CM1).
3.3. Photoreforming of C4–C6 polyols
The courses of the H2-evolution rates during photoreforming of
erythritol, arabitol and sorbitol (C4, C5, and C6 polyols, respectively)
are presented in Fig. 6. The 49% decline from initial H2-evolution
rates over 12 h of reaction time, particularly steep during the first
4 h for all three polyols, is more pronounced than for glycerol pho-
toreforming (33%). The course of the evolution of intermediates
and reaction products in the anodic half-reactions during sorbitol
photoreforming is exemplarily depicted in Fig. 7 (corresponding
data for erythritol and arabitol are found in SI-Figs. 12 and 13).
The product distributions in the photoreforming of erythritol,
arabitol and sorbitol show common features. All shorter chain
(Cnꢁ1, Cnꢁ2, etc.) aldoses with decreasing carbon number down to
C2 and formaldehyde were encountered during photoreforming
of the corresponding Cn polyol. However, higher initial selectivities
Scheme 5. Reaction pathway for photocatalytic reforming of glycerol (Path 3):
were obtained to Cnꢁ2 aldoses (being constant over time) com-
Light-driven dehydration of glycerol to form hydroxyacetone. Consecutive oxida-
pared to Cnꢁ1 aldoses and formaldeyde. Furthermore, glycolalde-
tion of hydroxyacetone leads to the formation of acetic acid and formaldehyde
hyde is the major species in each case accumulating over time in
liquid-phase, with the highest and almost constant selectivity with
increasing reaction time (SI-Figs. 14–16). Larger amounts of glyco-
laldehyde were produced at similar amounts of H2 generated with
increasing polyol carbon number (SI-Table 5). These observations
indicate that glycolaldehyde is not formed as a result of a reaction
cascade, but as a primary product from the polyols. Thus, initial
intermediates toward full oxidation to CO2.
which lead to the formation of GAD (26%) and DHA (13%), repre-
sent minor reaction channels. A comparison of experimental and
fitted photoreforming kinetics of glycerol, GAD and DHA and the
corresponding apparent adsorption constants of C1–C3 oxygenates
are given in SI-Figs. 9–11 and SI-Table 4.
Scheme 6. Reaction network for photocatalytic reforming of glycerol. Initial anodic glycerol transformation involves four reaction pathways. C–C-cleavage to form
glycolaldehyde and formaldehyde occurs primarily over formation of carbonyl groups to produce glyceraldehyde or dihydroxyacetone. Light-driven dehydration to
hydroxyacetone constitutes a side reaction.
Please cite this article in press as: K.E. Sanwald et al., Catalytic routes and oxidation mechanisms in photoreforming of polyols, J. Catal. (2016), http://dx.
doi.org/10.1016/j.jcat.2016.08.009

8
K.E. Sanwald et al. / Journal of Catalysis xxx (2016) xxx–xxx
with an initial selectivity of 3%) are absent in the cases of arabitol
and sorbitol photoreforming. These low selectivities to Cn ketoses
and Cn aldoses from polyols corroborate the minor contribution
of pathways yielding Cn species with primary or secondary sp2-
carbon.
Compounds bearing non-functionalized carbon as a conse-
quence of dehydration reactions were not observed except for
small quantities of acetaldehyde (e.g., of 30 lmol after 12 h during
photoreforming of erythritol) that were present with all polyols
and decreasing with increasing polyol chain length. The origin,
properties and impact of multiple formic acid esters (summed to
a total amount of ‘‘formates”) encountered with minor abundance
as a consequence of the presence of cyclic hemiacetal intermedi-
ates (erythrose, arabinose, xylose, lyxose and glucose) are currently
under investigation.
Thus, initial anodic transformations of erythritol, arabitol and
sorbitol involve multiple oxidative C–C-scission pathways. The
reaction pathways for sorbitol photoreforming are exemplarily dis-
played in Scheme 7 (reaction pathways for erythritol and arabitol
conversion are found in SI-Schemes 1 and 2). Generally, oxidative
cleavage of each individual C–C bond is feasible in the first step.
However, cleavage of the internal C–C bonds, i.e. C2–C3 for erythri-
tol and arabitol and C2–C3 preferred over C3–C4 for sorbitol, con-
stitutes the dominant reaction pathways.
Fig. 6. Course of H2-evolution rates during photoreforming of C4–C6 polyols.
Reaction conditions: 75 mg photocatalyst, 100 mL aqueous oxygenate solution
(20 mM), 288 K, 1 bar Ar, 300 W Xe-lamp (CM1).
polyol conversion must predominantly proceed through oxidative
cleavage of an internal C–C bond, i.e. the C2–C3 bond in erythritol,
arabitol, and sorbitol (Scheme 7, and SI-Schemes 1 and 2).
Constant selectivity profiles for C2+ aldoses over time indicate
that every C–C bond may also undergo oxidative cleavage in the
first step although the internal cleavage is preferred. Similar distri-
butions of these aldoses were encountered in studies conducted
over noble metal decorated TiO2 [29,45]. However, we did not find
evidence for the generation of gluconic acid as reported earlier
[29,45].
3.4. Mechanistic aspects of anodic transformations of polyols
Cn aldoses generated from oxidation of a terminal polyol car-
bon, i.e. erythrose from erythritol, arabinose and lyxose from ara-
bitol, and glucose from sorbitol are present in small quantities
(initial selectivities of 1–8% decreasing with increasing polyol car-
bon number). Cn ketoses resulting from oxidation of a secondary
carbon in 2-position (i.e., formation of erythrulose from erythritol
A unified description involving two different reaction sites is
proposed to rationalize the elementary steps in the anodic trans-
formations of polyols, i.e. oxidative C–C cleavage (direct charge
carrier transfer), oxidation to the corresponding aldehydes and
ketones (indirect charge carrier transfer) and dehydration (side
reaction within indirect mechanism).
Fig. 7. Temporal profiles of anodic reaction products formed during sorbitol photoreforming. (A) Course of amount of sorbitol and C6-intermediates. Course of (B) C5/C4-, (C)
C3-/C2- and (D) C1-species generated. Reaction conditions: 75 mg photocatalyst, 100 mL aqueous sorbitol solution (20 mM), 288 K, 1 bar Ar, 300 W Xe-lamp (CM1).
Please cite this article in press as: K.E. Sanwald et al., Catalytic routes and oxidation mechanisms in photoreforming of polyols, J. Catal. (2016), http://dx.
doi.org/10.1016/j.jcat.2016.08.009

K.E. Sanwald et al. / Journal of Catalysis xxx (2016) xxx–xxx
9
Scheme 7. Reaction network for photocatalytic reforming of sorbitol. Initial anodic conversion may proceed through four pathways. C–C-scission (C2–C3) to form
glycolaldehyde and erythrose constitutes the dominant reaction pathway. Cleavage of all other C–C bonds and formation of glucose represent minor reaction channels.
Oxidative C–C cleavage is initialized by direct electron transfer
from ‘deep surface hole traps’, which are constituted by chemi-
sorbed oxygenates on terminal OH-groups at oxygen coordina-
tively unsaturated Ti-sites [18,31,46] as illustrated in Scheme 8,
Sequence (1). As a consequence of hole transfer to the adsorbate,
alkoxy radicals are formed which undergo b-C–C-scission [47] to
form one aldehyde equivalent and a carbon-centered radical. Oxi-
dation of the latter radicals to a second aldehyde moiety under
anaerobic conditions is hypothesized to proceed preferentially
through injection of an electron into the conduction band of
TiO2, i.e. a ‘‘current doubling” process. Formally, the reaction prod-
ucts from C–C-cleavage reflect the outcome of a Malaprade peri-
odic acid oxidation chemistry [48].
Beyond that, holes may be trapped at surface lattice oxygen sites
[Tiꢀ ꢀ ꢀOꢀ ꢀ ꢀTi] [18,31,46]. These sites act as shallow hole traps that
catalyze ÅOH-radical/Fenton’s reagent like chemical transforma-
tions (Scheme 8, Sequence (2)), i.e., the formation of the correspond-
ing aldehydes and ketones from polyols [18,31]. Carbon-centered
radicals from H-abstraction may form stable organic species upon
current doubling (Scheme 8, sequence (3)). In a side reaction during
photoreforming, the radicals may undergo an acid-catalyzed ß-shift
(Scheme 8, sequence (3a)) which subsequently results in the
associated dehydration products [18,49–51], e.g., production of
HA from glycerol. Acid-catalyzed elimination of water upon
ß-shift of a carbon-centered radical to form a carbon radical (or
intermediate radical cation upon proton association) stabilized by
a vicinal carbonyl group (Scheme 8) is well-known to occur during
free OH-radical mediated oxidations [49–51].
with a minor contribution of a direct hole transfer mechanism
[53]. Polyols possess a high tendency toward coordination to Ti
(IV) cations. Evidence for a bidentate mode of binding was pro-
vided from the changes in the pre-edge structure in Ti K-edge
XANES [55]. Those were interpreted in terms of an almost quanti-
tative change in coordination environment around Ti(IV) sites from
square pyramidal toward octahedral upon adsorption of polyols. In
case of glycerol, IR-spectroscopic studies provide evidence that
glycerol dissociatively establishes a bridging alkoxy bond via a pri-
mary OH-functionality to a coordinatively unsaturated Ti(IV) site
[56,57]. An increasing tendency for polyol chemisorption with
increasing number of OH-functionalities was shown for competi-
tive adsorption of water and polyols on hydrated Al2O3 [56,58].
Our investigations into the anodic reaction networks of ethy-
lene glycol [18] and C3–C6 polyols show a change from oxidation
of a primary carbon (reaction of ethylene glycol to glycolaldehyde
via an indirect mechanism) to C–C-cleavage (direct mechanism) for
glycerol and higher polyols as the dominant reaction pathway. The
relative contributions of direct and indirect mechanisms to the
anodic conversion of Cn polyols are presented in Fig. 8 (Data for
C2-polyol was taken from Ref. [18]). The contributions were
derived from the rate constants from kinetic modeling in case of
ethylene glycol and glycerol and the initial selectivities of the pri-
mary products for C4–C6 polyol photoreforming. The direct mech-
anism becomes progressively dominant with increasing polyol
chain length, i.e. the relative contribution increases from 15% for
ethylene glycol to 61% for glycerol and 98% for sorbitol photore-
forming, respectively. This is in line with the complexation of coor-
dinatively unsaturated Ti(IV)-sites expected to be more effective,
i.e. a larger polyol fraction in a chemisorbed state at constant con-
centration, with increasing number of anchoring OH-groups. In
parallel, the participation of indirect reaction channels diminishes
accordingly. The decreasing selectivity for dehydration with
Aerobic photooxidation studies demonstrated that the contri-
butions of direct and indirect oxidation mechanisms critically
depend on molecular reactant structure [46,52–54]. For instance,
photocatalytic degradation of phenol over TiO2 P 25 proceeded
predominantly through interaction with ‘shallowly trapped holes’
Please cite this article in press as: K.E. Sanwald et al., Catalytic routes and oxidation mechanisms in photoreforming of polyols, J. Catal. (2016), http://dx.
doi.org/10.1016/j.jcat.2016.08.009

10
K.E. Sanwald et al. / Journal of Catalysis xxx (2016) xxx–xxx
OH
OH
OH
OH
Ti
R
R'
Ti
R
R'
Ti
R
R'
Ti
R
R'
O
OH
Ti
O
O
O
O
O
O
O
H
H
HO
+h+
(1)
Ti
Ti
Ti
O
O
O
O
OH
R'
OH
R'
HO
HO
H
OH
Ti
OH
Ti
OH
Ti
OH
Ti
H
+h+
(2)
(3)
Ti
Ti
Ti
Ti
O
O
O
O
OH
O
R'
R'
R'
_e- to CB
_H+
OH
OH
OH
O
O
ß-shift
R'
R'
[H+]
(3a)
+R''CH2OH
_R''CH·OH
_H2O
OH
Scheme 8. Illustration of proposed anodic transformations via direct and indirect oxidation mechanisms. (1) Direct hole transfer to a chemisorbed oxygenate on
coordinatively unsaturated Ti(IV)-OH site to generate an alkoxide radical followed by ß-C–C-scission. Nucleophilic attack of water is proposed for active site recovery and
product release. (2) Generation of carbon centered H-loss radicals via interaction with shallowly trapped holes at surface lattice oxygen sites. (3) Formation of stable organic
species through either current doubling of carbon-centered radicals or dehydration via acid-catalyzed ß-shift of the radicals.
Fig. 8. Relative contributions of direct (C–C-cleavage) and indirect (oxidation to Cn aldoses and ketoses and dehydration) charge transfer mechanisms to the anodic
transformations of C2–C6 polyols during photoreforming over Rh/TiO2 derived from reaction pathway analysis. Data for C2-polyol was taken from Ref. [18].
increasing polyol carbon number that leads to the absence of the
respective products during photoreforming of C4–C6 polyols cor-
roborates a side-reaction within the indirect mechanism.
Accordingly, the anodic reaction network of glycerol photore-
forming follows four reaction pathways. Oxidative C–C-cleavage
leads to glycolaldehyde and formaldehyde as the dominant path-
way (61%). Oxidation of the primary or secondary carbon leads to
glyceraldehyde (26%) or dihydroxyacetone (13%), respectively.
Light-driven dehydration of glycerol to hydroxyacetone proceeds
in a minor side-reaction. With increasing polyol chain length
(number of carbons and OH-groups) the selectivities shift to favor
anodic conversion through C–C-cleavage even further with internal
C–C bonds preferentially cleaved over terminal ones. This is attrib-
uted to the increasing number of anchoring OH-functionalities
favoring conversion through a direct hole transfer mechanism.
4. Conclusions
Photocatalytic rates of linear C1–C3 oxygenates are primarily
dependent on the substrate specific apparent adsorption constants
and follow a Langmuir adsorption model. In the anodic half-
reactions of photoreforming linear polyols are converted via (i)
oxidative rupture of C–C bonds, (ii) oxidation to the corresponding
aldoses or ketoses or (iii) light-driven dehydration, while evolving
H2 at the cathode. The first pathway is proposed to result from
direct hole transfer to the chemisorbed oxygenate on terminal Ti
(IV)-OH groups to form alkoxy-radicals which undergo ß-C–C-
cleavage. The latter pathways are attributed to an indirect mecha-
nism initiated by hole trapping at surface lattice oxygen sites
[Tiꢀ ꢀ ꢀOꢀ ꢀ ꢀTi].
Acknowledgments
We would like to thank Clariant for fruitful discussions within
the framework of MuniCat and the iC4 PhotoCOO project. The
authors are grateful to Donald M. Camaioni for critical reading of
Please cite this article in press as: K.E. Sanwald et al., Catalytic routes and oxidation mechanisms in photoreforming of polyols, J. Catal. (2016), http://dx.
doi.org/10.1016/j.jcat.2016.08.009

K.E. Sanwald et al. / Journal of Catalysis xxx (2016) xxx–xxx
11
[23] Y. Hang Li, J. Xing, Z. Jia Chen, Z. Li, F. Tian, L. Rong Zheng, H. Feng Wang, P. Hu,
H. Jun Zhao, H. Gui Yang, Nat. Commun. 4 (2013) 2500.
[24] H. Bahruji, M. Bowker, P.R. Davies, L.S. Al-Mazroai, A. Dickinson, J. Greaves, D.
James, L. Millard, F. Pedrono, J. Photoch. Photobio. A 216 (2010) 115–118.
[25] H. Bahruji, M. Bowker, P.R. Davies, F. Pedrono, Appl. Catal. B 107 (2011) 205–
209.
[26] G.N. Nomikos, P. Panagiotopoulou, D.I. Kondarides, X.E. Verykios, Appl. Catal. B
146 (2014) 249–257.
[27] X. Fu, J. Long, X. Wang, D.Y.C. Leung, Z. Ding, L. Wu, Z. Zhang, Z. Li, X. Fu, Int. J.
Hydrogen Energy 33 (2008) 6484–6491.
the manuscript. Additionally, we would like to thank the German
Federal Ministry of Education and Research (BMBF) for financial
support (project no. 01RC1106A). K.E.S. gratefully acknowledges
financial support by the Fond der Chemischen Industrie (FCI).
The authors thank Xaver Hecht for BET and H2-chemisorption mea-
surements and Martin Neukamm for AAS measurements. Christine
Schwarz and Martina Haack are acknowledged for technical assis-
tance concerning NMR and HPLC analysis, respectively.
[28] V.M. Daskalaki, D.I. Kondarides, Catal. Today 144 (2009) 75–80.
[29] R. Chong, J. Li, Y. Ma, B. Zhang, H. Han, C. Li, J. Catal. 314 (2014) 101–108.
[30] C. Minero, E. Pelizzetti, P. Pichat, M. Sega, M. Vincenti, Environ. Sci. Technol. 29
(1995) 2226–2234.
Appendix A. Supplementary material
[31] C. Minero, A. Bedini, V. Maurino, Appl. Catal. B 128 (2012) 135–143.
[32] V. Maurino, A. Bedini, M. Minella, F. Rubertelli, E. Pelizzetti, C. Minero, J. Adv.
Oxid. Technol. 11 (2008) 184–192.
[33] T. Montini, V. Gombac, L. Sordelli, J.J. Delgado, X. Chen, G. Adami, P. Fornasiero,
ChemCatChem 3 (2011) 574–577.
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jcat.2016.08.009.
References
[34] J. Schnaidt, M. Heinen, D. Denot, Z. Jusys, R.J. Behm, J. Electroanal. Chem. 661
(2011) 250–264.
[1] K. Shimura, H. Yoshida, Energy Environ. Sci. 4 (2011) 2467–2481.
[2] R.M. Navarro, M.C. Sanchez-Sanchez, M.C. Alvarez-Galvan, F.d. Valle, J.L.G.
Fierro, Energy Environ. Sci. 2 (2009) 35–54.
[3] D. Kondarides, V. Daskalaki, A. Patsoura, X. Verykios, Catal. Lett. 122 (2008)
26–32.
[4] Y. Ma, X. Wang, Y. Jia, X. Chen, H. Han, C. Li, Chem. Rev. 114 (2014) 9987–
10043.
[5] X. Chen, S. Shen, L. Guo, S.S. Mao, Chem. Rev. 110 (2010) 6503–6570.
[6] M. Cargnello, A. Gasparotto, V. Gombac, T. Montini, D. Barreca, P. Fornasiero,
Eur. J. Inorg. Chem. 2011 (2011) 4309–4323.
[35] C.A. Martins, M.J. Giz, G.A. Camara, Electrochim. Acta 56 (2011) 4549–4553.
[36] Y. Kwon, K.J.P. Schouten, M.T.M. Koper, ChemCatChem 3 (2011) 1176–1185.
[37] H.J. Kim, J. Lee, S.K. Green, G.W. Huber, W.B. Kim, ChemSusChem 7 (2014)
1051–1056.
[38] L. Roquet, E.M. Belgsir, J.M. Léger, C. Lamy, Electrochim. Acta 39 (1994) 2387–
2394.
[39] X. Fu, X. Wang, D.Y.C. Leung, Q. Gu, S. Chen, H. Huang, Appl. Catal. B 106 (2011)
681–688.
[40] A. Gallo, M. Marelli, R. Psaro, V. Gombac, T. Montini, P. Fornasiero, R. Pievo, V.D.
Santo, Green Chem. 14 (2012) 330–333.
[7] Z.H.N. Al-Azri, W.-T. Chen, A. Chan, V. Jovic, T. Ina, H. Idriss, G.I.N. Waterhouse,
J. Catal. 329 (2015) 355–367.
[8] F. Dionigi, P.C.K. Vesborg, T. Pedersen, O. Hansen, S. Dahl, A. Xiong, K. Maeda, K.
Domen, I. Chorkendorff, J. Catal. 292 (2012) 26–31.
[9] W. Jiao, L. Wang, G. Liu, G.Q. Lu, H.-M. Cheng, ACS Catal. 2 (2012) 1854–1859.
[10] J. Zhang, Q. Xu, Z. Feng, M. Li, C. Li, Angew. Chem. Int. Ed. 120 (2008) 1790–
1793.
[41] T. Sakata, T. Kawai, Chem. Phys. Lett. 80 (1981) 341–344.
[42] T. Sakata, T. Kawai, K. Hashimoto, J. Phys. Chem.-US 88 (1984) 2344–2350.
[43] B. Kraeutler, A.J. Bard, J. Am. Chem. Soc. 100 (1978) 5985–5992.
[44] X.-J. Zheng, L.-F. Wei, Z.-H. Zhang, Q.-J. Jiang, Y.-J. Wei, B. Xie, M.-B. Wei, Int. J.
Hydrogen Energy 34 (2009) 9033–9041.
[45] M. Bellardita, E.I. García-López, G. Marcì, L. Palmisano, Int. J. Hydrogen Energy
41 (2016) 5934–5947.
[11] Y.K. Kho, A. Iwase, W.Y. Teoh, L. Mädler, A. Kudo, R. Amal, J. Phys. Chem. C 114
(2010) 2821–2829.
[46] J.F. Montoya, M.F. Atitar, D.W. Bahnemann, J. Peral, P. Salvador, J. Phys. Chem. C
118 (2014) 14276–14290.
[12] A. Tanaka, S. Sakaguchi, K. Hashimoto, H. Kominami, ACS Catal. 3 (2013) 79–
85.
[47] A. Boto, D. Hernández, R. Hernández, E. Suárez, J. Org. Chem. 68 (2003) 5310–
5319.
[13] A.J. Bard, M.A. Fox, Acc. Chem. Res. 28 (1995) 141–145.
[14] A.J. Bard, J. Photochem. 10 (1979) 59–75.
[48] G. Dryhurst, Periodate Oxidation of Diol and Other Functional Groups:
Analytical and Structural Applications, vol. 2, Pergamon Press, Oxford, 1970
(Chapter 2 and 3).
[49] A.L. Buley, R.O.C. Norman, R.J. Pritchett, J. Chem. Soc. B (1966) 849–852.
[50] C. Walling, R.A. Johnson, J. Am. Chem. Soc. 97 (1975) 2405–2407.
[51] D. Jiang, S. Barata-Vallejo, B.T. Golding, C. Ferreri, C. Chatgilialoglu, Org.
Biomol. Chem. 10 (2012) 1102–1107.
[15] J. Chen, D.F. Ollis, W.H. Rulkens, H. Bruning, Water Res. 33 (1999) 1173–1180.
[16] K. Lalitha, G. Sadanandam, V.D. Kumari, M. Subrahmanyam, B. Sreedhar, N.Y.
Hebalkar, J. Phys. Chem. C 114 (2010) 22181–22189.
[17] P. Panagiotopoulou, E.E. Karamerou, D.I. Kondarides, Catal. Today 209 (2013)
91–98.
[18] T.F. Berto, K.E. Sanwald, W. Eisenreich, O.Y. Gutiérrez, J.A. Lercher, J. Catal. 338
(2016) 68–81.
[52] C. Minero, G. Mariella, V. Maurino, E. Pelizzetti, Langmuir 16 (2000) 2632–
2641.
[19] R. Chong, J. Li, X. Zhou, Y. Ma, J. Yang, L. Huang, H. Han, F. Zhang, C. Li, Chem.
Commun. 50 (2014) 165–167.
[53] C. Minero, G. Mariella, V. Maurino, D. Vione, E. Pelizzetti, Langmuir 16 (2000)
8964–8972.
[20] M. Murdoch, G.I.N. Waterhouse, M.A. Nadeem, J.B. Metson, M.A. Keane, R.F.
Howe, J. Llorca, H. Idriss, Nat. Chem. 3 (2011) 489–492.
[21] R. Su, R. Tiruvalam, A.J. Logsdail, Q. He, C.A. Downing, M.T. Jensen, N.
Dimitratos, L. Kesavan, P.P. Wells, R. Bechstein, ACS Nano 8 (2014) 3490–3497.
[22] W. Jones, R. Su, P.P. Wells, Y. Shen, N. Dimitratos, M. Bowker, D. Morgan, B.B.
Iversen, A. Chutia, F. Besenbacher, G.J. Hutchings, PCCP 16 (2014) 26638–
26644.
[54] J.M. Kesselman, O. Weres, N.S. Lewis, M.R. Hoffmann, J. Phys. Chem. B 101
(1997) 2637–2643.
[55] I.A. Shkrob, M.C. Sauer, D. Gosztola, J. Phys. Chem. B 108 (2004) 12512–12517.
[56] J.R. Copeland, X.-R. Shi, D.S. Sholl, C. Sievers, Langmuir 29 (2013) 581–593.
[57] J.R. Copeland, I.A. Santillan, S.M. Schimming, J.L. Ewbank, C. Sievers, J. Phys.
Chem. C 117 (2013) 21413–21425.
[58] W. van Bronswijk, H.R. Watling, Z. Yu, Colloid. Surface A 157 (1999) 85–94.
Please cite this article in press as: K.E. Sanwald et al., Catalytic routes and oxidation mechanisms in photoreforming of polyols, J. Catal. (2016), http://dx.
doi.org/10.1016/j.jcat.2016.08.009