H2 photo-production from methanol, ethanol and 2-propanol: Pt-(Nb)TiO2 performance under UV and visible light

Article abstract of DOI:10.1016/j.mcat.2017.12.023

In this work we analyzed the photo-production of hydrogen using titania-based systems able to profit from UV and visible light photons. For this purpose, we prepared Niobium-doped titania and a titania reference by a microemulsion method, subjected these oxide precursors to calcination and subsequently introduced Pt as co-catalyst by a chemical reduction method. These materials were characterized in terms of the structural and morphological properties of the oxide and metal phases. Using these materials, we measured the reaction rate and quantum efficiency of the hydrogen photo-production using methanol, ethanol, and 2-propanol as sacrificial agents. Significant activity enhancement was observed in the Niobium-doped material with respect to the titania reference material. The study focuses on interpreting the differences presented (between the two samples) among the three alcohols in the hydrogen yield and provides a physico-chemical study to understand the roots of the activity. Such study was mainly based on the analysis of the reaction mechanism using in-situ infrared spectroscopy together with the analysis of the energetics of the reaction taking into account the fate of the sacrificial alcohol during reaction.

Full text of DOI:10.1016/j.mcat.2017.12.023

Molecular Catalysis 446 (2018) 88–97
Contents lists available at ScienceDirect
Molecular Catalysis
journal homepage: www.elsevier.com/locate/mcat
H2 photo-production from methanol, ethanol and 2-propanol:
Pt-(Nb)TiO2 performance under UV and visible light
Olga Fontelles-Carceller, Mario J. Mun˜oz-Batista, José Carlos Conesa, Anna Kubacka∗,
Marcos Fernández-García∗
Instituto de Catálisis y Petroleoquímica, CSIC, C/Marie Curie, 2, 28049 Madrid, Spain
a r t i c l e i n f o
a b s t r a c t
Article history:
In this work we analyzed the photo-production of hydrogen using titania-based systems able to profit
from UV and visible light photons. For this purpose, we prepared Niobium-doped titania and a titania
reference by a microemulsion method, subjected these oxide precursors to calcination and subsequently
introduced Pt as co-catalyst by a chemical reduction method. These materials were characterized in
terms of the structural and morphological properties of the oxide and metal phases. Using these mate-
rials, we measured the reaction rate and quantum efficiency of the hydrogen photo-production using
methanol, ethanol, and 2-propanol as sacrificial agents. Significant activity enhancement was observed
in the Niobium-doped material with respect to the titania reference material. The study focuses on inter-
preting the differences presented (between the two samples) among the three alcohols in the hydrogen
yield and provides a physico-chemical study to understand the roots of the activity. Such study was
mainly based on the analysis of the reaction mechanism using in-situ infrared spectroscopy together
with the analysis of the energetics of the reaction taking into account the fate of the sacrificial alcohol
during reaction.
Received 13 November 2017
Received in revised form
15 December 2017
Accepted 16 December 2017
Keywords:
Anatase
Doping
Quantum efficiency and yield
Selectivity
Rate
© 2017 Elsevier B.V. All rights reserved.
1. Introduction
sponding to the visible light region [1–8]. In the context of the
photocatalytic production of hydrogen, the use of visible photons
Hydrogen appears as an energy vector potentially important for
a future where greener and more sustainable chemical processes
are required. This is due to the fact that the molecule is ideal to
store energy (ca. 3 times more than conventional natural gas per
unit volume) and does not generate toxic or dangerous molecules
during its chemical transformation to release energy [1]. Produc-
tion of hydrogen trough photocatalysis using light and particularly
sunlight and bio-derived molecules such as alcohols can provide
a sustainable and attractive path with neutral carbon emissions,
leading to a greener future [2–9].
Titania corresponds to the most utilized material in photocatal-
ysis due to its inherent properties related to a modest cost, limited
toxicity and outstanding chemical properties under illumination.
However, its use is limited to the UV electromagnetic region due to
its wide band gap, always above 3.0 eV. The use of sunlight requires,
in a first approximation, the efficient handling of photons corre-
without compromising UV performance has been achieved by a
number of methods. A successful one is the doping of the anatase
material with specific cations such as B, Fe, Nb, Ce and others.
Using Pt or noble metals as co-catalyst, the above mentioned doped
titania samples displayed significant activity in the reforming of
alcohols under UV and/or visible illumination [10–16].
Several bio-alcohols have been tested in the literature as sacri-
ficial agents in the photo-production of hydrogen highlighting the
importance of several experimental variables in order to maximize
the production of hydrogen. Among them we can highlight the co-
catalyst chemical nature and physico-chemical properties, charge
carrier handling by the metal-support system, alcohol:water ratio,
or the light wavelength and intensity [17–33]. Methanol is the sim-
plest molecule from the structural and chemical points of view
and probably the molecule receiving most attention due to the
relatively high hydrogen production rates commonly observed
[14,15,34–37]. Irrespective of the alcohol chemical nature, the
studies show that hydrogen molecules are essentially produced
stoichiometrically from the alcohol, although the hydrogen atoms
are exchanged through scrambling with the corresponding atoms
of water molecules of the media during the production process of
∗
Corresponding authors.
E-mail addresses: ak@icp.csic.es (A. Kubacka), m.fernandez@icp.csic.es,
mfg@icp.csic.es (M. Fernández-García).
https://doi.org/10.1016/j.mcat.2017.12.023
2468-8231/© 2017 Elsevier B.V. All rights reserved.

O. Fontelles-Carceller et al. / Molecular Catalysis 446 (2018) 88–97
89
the hydrogen molecule [36,38]. A schematic representation of the
reactions that summarizes the most important chemical steps is as
follows:
cation content of the aqueous solution is 0.5 M. Water/(Ti+Nb) and
water/surfactant molar ratios were, respectively, 111 and 18 for all
samples. The resulting mixture was stirred for 24 h, centrifuged,
decanted, rinsed with methanol and dried at 300 K for 6 h. Fol-
lowing the microemulsion preparation method, the amorphous
powders were calcined under air for 2 h at 723 K. The co-catalyst
was introduced subsequently by a chemical deposition method
using a H2PtCl6 (Aldrich) solution. First, the calcined powder was
suspended by stirring in a deionized water solution for 30 min.
After that, the proper quantity of H2PtCl6 was added to the solu-
tion (to get a 0.5 wt.% of Pt on metal basis) and kept under stirring
5 min more. The reduction was carried out using a NaBH4 (Aldrich)
aqueous solution (Pt/NaBH4 molar ratio 1/5). The final solid was
rinsed with deionized water, collected by centrifugation and dried
at 353 K. Samples are named Ti (Pt-TiO2 reference system) and Nb
(Pt-Nb doped-TiO2). Niobium (when present) and Platinum content
of the solids were measured with total reflection x-ray fluorescence
(Bruker − S2 PicoFox TXRF Spectrometer) rendering values equal
to 0.5 wt% for Pt in both catalysts and 2.5 mol% (cationic basis as
Nb0.025Ti0.0975Ox) for the Nb sample within an error below 2.1and
1.5% for, respectively, Niobium and Platinum components.
XRD profiles of the samples were obtained using a poly-
crystal Xı´Pert Pro PANalytical diffractometer using Ni-filtered
Cu K␣ radiation with a 0.02◦ step. Crystallite sizes reported
were calculated from XRD patterns using the Williamson-Hall
method which takes into account the strain and particle size
contributions to the XRD peak broadening [42].The BET surface
areas and average pore sizes and pore volumes were measured
by nitrogen physisorption (Micromeritics ASAP 2010). UV–vis
diffuse-reflectance spectroscopy experiments were performed on
a Shimadzu UV2100 apparatus using BaSO4 or Teflon as a refer-
ence, and the results presented as the Kubelka-Munk transform
[43]. Band gap analysis for an indirect/direct semiconductor was
done following standard procedures; e.g. plotting (hva)n (n = ½
or 2 for indirect or direct semiconductor; hv = excitation energy,
a = absorption coefficient, assumed to be proportional to the
Kubelka-Munk transform in the relevant wavelength range) vs.
energy and obtaining the corresponding intersection of the linear
fit with the baseline [44].
Cat + hꢀ → Cat + h+ + e−
(1)
(2)
(3)
(4)
h
+ + H2Oads → OH◦ + H+
RCH2OHads + h+ OH◦ → Products + H+ H O
(
)
(
)
2
2H+ + 2e− → H2
In brief, light absorption triggers the creation of electron and
holes (Eq. (1)). These charge carriers reach the surface and pro-
duce several chemical species (Eqs. (2)–(4)). In the case of the
alcohol reforming, holes or hole-related species such as hydroxyls
are responsible of the attack (or oxidation) of the alcohol (Eq. (3))
while electrons are consumed in the production of hydrogen (Eq.
(4)) [2–4]. A point usually dismissed in the literature is connected
with the carbon-containing molecules produced in step 3. Hydro-
gen production depends critically on this issue or more specifically
on the hydrogen remaining on the carbon-containing products due
to the already mentioned stoichiometric relationship essentially
holding between hydrogen production and hydrogen content of
the sacrificial molecules [36,38].
Analysis of catalytic performance concerning the study of
several alcohols as sacrificial agents has been presented for noble-
metal promoted titania materials [25,39–41]. Alcohols having at
least one hydrogen atom bonded to the (carbon) alpha position
render higher reaction rates. Progress in understanding the differ-
ences in hydrogen production rates as a function of the alcohol
nature showed the importance of alcohol polarity and the expo-
nential of the alcohol potential oxidation in controlling the rate
of hydrogen production [40,41]. However the understating of the
physico-chemical interpretation of photo-activity in terms of alco-
hol chemical nature has not considered the chemical variable
concerning the critical role that the different chemical products
generated as a function of the alcohol chemical nature must have
in hydrogen production. This point is obviously related to the
chemical properties of the active centers transforming the alcohol
sacrificial molecules.
In this contribution we study the hydrogen photo-production
using several alcohols as sacrificial molecules and taking into
account the use of UV and visible light illumination in order to
progress in creating systems able to profit from sunlight. For this
task we carried out a study using methanol, ethanol and 2-propanol
and measured the reaction rates as well as the true quantum
efficiency. Although all these molecules have hydrogen in alpha
positions (to the hydroxyl moiety), significant differences in hydro-
gen photo-production rate are expected. We utilized titania and,
more importantly, Niobium-doped titania as supports of the Pt-
based catalysts in order to achieve high activity under both UV and
visible illumination [14,15]. This work provides an interpretation
of catalytic output concerning activity and selectivity of the alco-
hol photo-reforming reaction through an in-situ infrared analysis
of the alcohol activation together with a thermodynamic analysis
of the energy of the process and its connection with the hydrogen
production rate.
Diffuse Reflectance Infrared Fourier Transform Spectra (DRIFTS)
were taken in a Bruker Vertex 80 FTIR spectrometer using a MCT
detector and running under OPUS/IR software. The set-up consists
of a praying mantis DRIFTS accessory (Harrick Scientific) and a
reaction cell (HVC, Harrick Scientific). The reaction mixture was
prepared by injecting in a nitrogen carrier (50 mL min−1) a 3:7
alcohol:water mixture (0.08 mL min−1) with a syringe pump before
entering the DRIFTS cell. The DRIFTS spectra were collected in the
range of 4000–600 cm−1with a resolution of 2 cm−1, by averaging
10 scans over a total of 1.2 s. In DRIFTS experiments in-situ light
excitation was carried out using 365 nm (488, 550 nm) radiation. A
Hg-Xe 500 W lamp with a dicroic filter 280–400 nm coupled with
a 365 nm (25 nm half-width) filter (LOT-Oriel) were used to select
the light excitation. Each sample, without any previous treatment
(expect flowing nitrogen for 10 min), was subjected in a continu-
ous mode (without modifying gas mixture) to a single, multi-step
experiment which aims to test the: i) adsorption of the reactive
mixture under dark conditions, ii) reaction mixture evolution under
illumination conditions, and iii) subsequent stay at dark conditions.
Spectra were taken after different exposure times to verify any
evolution behavior.
2. Experimental
2.1. Preparation and characterization of catalysts
Materials were prepared using a microemulsion method by
addition of Titanium tetraisopropoxide to an inverse emulsion con-
taining either an aqueous solution of hydrated Niobium nitrate
(Sigma) or just water dispersed in n-heptane, using Triton X-
100 (Aldrich) as surfactant and hexanol as cosurfactant. Total
2.2. Description of reactor and catalytic outputs
Photocatalytic measurements at liquid medium were carried
out using a cylindrical-type, batch pyrex (cutting absorption edge
at ca. 300 nm) reactor of 4 cm depth (z coordinate in the numerical

90
O. Fontelles-Carceller et al. / Molecular Catalysis 446 (2018) 88–97
Fig. 1. Local volumetric rate of photon absorption (Einstein cm−3 s−1) obtained under UV illumination for the Ti (upper panels) and Nb (lower panels) samples and concerning
reaction with methanol (left column), ethanol (middle column) and 2-propanol (right column).
Fig. 2. Local volumetric rate of photon absorption (Einstein cm−3 s−1) obtained under Visible illumination for the Ti (upper panels) and Nb (lower panels) samples and
concerning reaction with methanol (left column), ethanol (middle column) and 2-propanol (right column).
analysis) and circular section with a 2 cm radius (r coordinate; for
details see Ref. [14]). The reactor is filled with ca. 50 mL of a sus-
pension of the catalyst in 3:7 (v/v) RCH2COH/H2O mixture medium
maintained at a constant temperature (293 1 K). The ratio is cho-
sen to provide maximum activity according to previous reports
[14]. The catalyst suspension (0.5 g L−1) was first degassed with an
Ar stream for around 20 min. Subsequently, the Ar flow was settled
down to 10 mL min−1 and stabilized before reaction. Ar is used as
carrier to displace reaction gases from the reactor to the detection
system. The solution inside the reactor was irradiated from above
using a Hg-Xe lamp (500 W) and dichroic filters (LOT Quantum
Design) allowing exposure of the catalysis to the UV (280–400 nm)
or Visible (420–680 nm) wavelength range. The reaction rates for
hydrogen production were evaluated at 3 h from the start of the
irradiation, where a pseudo-stationary situation is reached. The
hydrogen production rate was analyzed using an on-line Mass
spectrometry (Onmistart 300), gas (TCD/FID detection using HP-
PLOT-Q/HP-Innowax columns and an Agilent 6890 apparatus) and
liquid chromatography (xD8-C18/5 microm, 4.6 × 150 mm Agilent
HPLC Column Eclipse and a Varian Pro Star 230 apparatus).
Quantum efficiency was calculated, according to the IUPAC rec-
ommendation [45], as the ratio of the number of molecules reacting
by the number of photon interacting with the sample (Eq. (5)).
ꢀ
ꢁ
r
molm−3s−1
ꢀ
ꢁ
ꢁq % = 100 ×
( )
(5)
S < ea,v
>
Einsteinm−3s−1

O. Fontelles-Carceller et al. / Molecular Catalysis 446 (2018) 88–97
91
The reaction rate of hydrogen production (r in Eq. (5)) is divided
by the selectivity factor (S) and the averaged local volumetric
photon absorption rate (<ea,v>). The so-called selectivity factor
(dimensionless) is calculated as a summation over all chemical
reactions generating protons (Eq. (3)) and takes the value:
ꢂ
S =
niSi
(6)
i
Where Si is the fractional selectivity to product i, and ni is the
inverse of number of charge carrier species required to obtain one
mol of hydrogen [46].
To determine the denominator, we also need to obtain the
solution of the radiative transfer equation (RTE; Eq. (7)) in the
heterogeneous reactor(s). The RTE for liquid phase measured the
variation of radiation intensity (associated to a beam of rays at
wavelength ␭ in the direction of a solid angle vector, ˝) through a
direction of the space (s) [47].
ꢀ ꢁ
ꢃ
dIꢂ,˝
x
ꢀ ꢁ
ꢀ ꢁ
ꢀ
ꢁ
ꢀ ꢁ
ꢄꢂ
= −ꢃꢂIꢂ,˝
x
− ꢄꢂIꢂ,˝
x
+
p
˝’ → ˝ Iꢂ,˝
x
d˝ꢀ (7)
’
4ꢅ
ds
ꢀ
˝ =4ꢅ
Where ꢃꢂ is the absorption coefficient; ꢄꢂ is the scattering coef-
ꢀ
ꢁ
ficient; and p ˝’ → ˝ is the scattering phase measured with the
Henyey and Greenstein phase function, as usually done for titania
samples [47]. ꢃꢂ and ꢄꢂ coefficients are obtained from transmit-
tance and reflectance measurements in model suspensions of the
catalysts [46,47]. Eq. (7) assumes that (i) the emission radiation
is negligible (at room temperature), and (ii) steady state condi-
tion during the photocatalytic processes. Eq. (7) is solved using
the discrete ordinate method and considering the corresponding
boundary equations (known inlet radiation from the upper win-
dow and null reflection in the inner wall of the reactor surfaces, see
Ref. [46] for details). Once the spatial and angular distribution of the
intensity is obtained, the local volumetric rate of photon absorption
(ea,v) is calculated at each r − z point of the reactor according to [14]:
ꢃ
ꢃ
ea,v x =
ꢃꢂ
·
Iꢂ,˝ x d˝dꢂ
( )
(8)
( )
ꢂ
˝=4ꢅ
Fig. 3. XRD patterns (A) and UV–vis spectra (B) of the samples.
The ea,v
x
( )
parameter is presented for the Ti and Nb samples
in Figs. 1 and 2 under, respectively, UV and visible illumination. In
these figures a “r-z” section (0 < r ≤ 2; 0≤ z ≤ 4 cm) of the reactor
is shown. The reactor volume is constructed from this section by
the azimuthal symmetry around the central axis of the cylindrical
reactor, having all sections equal values of the observable.
for titania-based materials [25,39–41], Ti and Nb samples present
maximum rates for methanol, followed by ethanol and 2-propanol,
irrespective of the illumination source. Larger differences using
the three alcohols are observed for the Nb sample with respect
to the Ti reference but the trend of both samples is rather similar.
The selectivity of carbon containing products generated is summa-
rized in Table 2. For methanol, both samples generate formic acid
and metyl formate but the Nb sample also shows the production
of formaldehyde under UV illumination. For ethanol in all illumi-
nation conditions both samples generate acetaldehyde and ethyl
acetate. We did not detect carbon containing molecules originated
3. Results and discussion
The basic characterization of the two materials (Ti and Nb) is
presented in Fig. 3 using XRD and UV–vis spectroscopy. The analy-
sis of the corresponding XRD patterns and UV–vis spectra renders
physico-chemical information summarized in Table 1. From the
point of view of the titania support, Table 1 collects rather sim-
ilar values for all observables for the two samples. The anatase
structure (PDF 21-1272; space group I41/amd) of titania therefore
displays equal morphological properties as well as band gap values.
The physico-chemical analysis of the samples was completed using
chemical analysis, XPS, microscopy and other techniques, described
in a previous report [14]. For both samples, Pt has a similar primary
crystallite size of ca. 1.3 0.1 Å and, for the Nb sample, the Nb/Ti
surface ratio measured by XPS took a value of 0.023:0.977 (rather
close to the nominal one of 0.025:0.975).
+
from the fragmentation of the C2 molecules detected. Finally, for
2-propanol larger differences are observed among our samples. The
Nb sample generates acetone and propanone-diisopropyl-acetal
while the Ti sample only renders acetone. Again no fragmentation
+
is detected and only the mentioned C3 products are detected. So,
the presence of Niobium at the surface of the material makes small
changes in the active centers as to trigger the generation of a few
products not observed in the case of Ti. In any case and according
to Table 2, these are minor products accounting only for no more
than 20% of the alcohol molecules transformed. Note that CO2 is
not detected through all tests carried out (Table 2).
The rate of hydrogen production using the two samples and illu-
mination sources and the three alcohols tested here are presented
in panel A of Fig. 4. As may be expected considering previous results
The balance equations required to generate the different carbon
containing products are presented in Table 3. The different prod-

92
O. Fontelles-Carceller et al. / Molecular Catalysis 446 (2018) 88–97
Table 1
Main physico-chemical properties of the catalysts.
Sample
Bandgap (eV)
Size (nm)
BET Area (m2/g)
Pore volume (cm3/g)
Pore size (nm)
Nb
Ti
3.15
3.18
11.9
12.3
80.5
82.1
0.131
0.129
5.6
5.5
Standard error: 0.04 eV; 0.5 nm; 3.0 m2g−1
.
Table 2
Carbon-containing products detected in the alcohols photo-transformation under UV and visible illumination.
Alcohol
Sample
Nb
Illumination
UV
Product
Selectivity (%)a
Formic acid
Formaldehyde
85
8
Methanol
Methyl formate
Formic acid
Methyl formate
Formic acid
Methyl formate
Formic acid
Methyl formate
Acetaldehyde
Ethyl acetate
Acetaldehyde
Ethyl acetate
Acetaldehyde
Ethyl acetate
Acetaldehyde
Ethyl acetate
Acetone
7
91
9
Vis
UV
Vis
UV
Vis
UV
Vis
UV
Vis
90
10
98
2
89
11
97
3
81
19
71
29
74
26
80
20
100
100
Ti
Ethanol
Nb
Ti
2-propanol
Nb
Ti
Propanone diisopropyl acetal
Acetone
Propanone diisopropyl acetal
Acetone
UV
Vis
Acetone
a
Average Standard error; 4.6%.
Table 3
Standard Gibbs energy of formation of the alcohols and of the alcohol oxidation reactions.
Alcohol
ꢆfG◦(J mol−1
−166.6·103
)
Balanced equation
(−)ꢆG◦(J mol−1
)
CH4O + 2 h+ → CH2O + 2H+
64.1·103
42.3·103
−53.2·103
47.2·103
16.9·103
27.6·103
155.4·103
Methanol
CH4O + 4 h+ + H2O → CH2O2 + 4H+
2CH4O + 4 h+ → C2H4O2 + 4H+
C2H6O + 2 h+ → C2H4O + 2H+
2C2H6O + 4 h+ → C4H8O2 + 4H+
C3H8O + 2 h+ → C3H6O + 2H+
3C3H8O + 2 h+ → C9H20O2 + H2O + 2H+
−174.8·103
−180.3·103
Ethanol
2-propanol
ucts require a different number of (hole-related) charge carriers but
as the number of charge carries consumed matches the number of
protons generated, this simplifies significantly the calculation of
the quantum efficiency as the S (selectivity) factor always takes a
factor of 1/2. So, the reaction rate has to be multiplied by a factor
of 2 as commonly carried out. This can be also seen easily consid-
ering Eq. (4). Using the S value and the average volumetric rate of
photon absorption, we calculate the true quantum efficiency (Eq.
(5)). Panel B of Fig. 4 summarizes the corresponding results. Under
UV/visible illumination, maximum quantum efficiency values of
5.2/2.4 are obtained using methanol, 4.2/1.1 for ethanol, and 1.9/0.9
for 2-propanol. All these values correspond to the Nb sample, which
displays quantum efficiency values between 1.8–2.8 times higher
than those of the Ti reference sample. The trends observed for the
reaction rate and quantum efficiency as a function of the sample
(Nb vs. Ti) and illumination source are essentially the same, indi-
cating that the reaction rate is in this case a good indicator of the
“true” photo-activity of sample. Therefore, we can use the reaction
rate in order to investigate the origin of activity differences among
the alcohols tested.
Only UV illumination results are presented due to the practical
absence of significant changes with respect to visible illumination,
although signals generated under reaction conditions were signifi-
cantly weaker in the later case, likely due to the lower reaction rates.
Note that considering the selectivity values collected in Table 2, an
essentially identical mechanism is expected under all illumination
conditions here tested. So, UV-DRIFT results are representative of
all illumination conditions. In Figs. 5 and 6 an in-situ infrared study
of the two catalysts was carried out by obtaining difference spec-
tra taking as background the first spectrum recorded during each
step (from i to iii) of the treatment of the samples consisting in a
sequential: i) adsorption of the alcohol in presence of water vapour,
ii) reaction mixture (alcohol:water) under illumination conditions,
and iii) subsequent stay at dark conditions under the same reactive
mixture used in previous steps of the experiment. Samples were
not pre-treated prior IR measurements.
Adsorption of methanol prior to illumination provides evi-
dence of the existence of methanol and methoxy species (upper
row in Figs. 5 and 6). For Ti and Nb samples, methanol (C
H
contributions at 2940/2831 and 2945/2836 cm−1, respectively)
seems to have larger intensity than methoxy (2920/2816 and
2920/2817 cm−1, respectively) species [48–50]. This is concomi-
To understand such issue, we first carried out an in-situ analysis
of the alcohol photo-transformation using infrared spectroscopy.

O. Fontelles-Carceller et al. / Molecular Catalysis 446 (2018) 88–97
93
yls and/or water. Note that the same type of isolated hydroxyl
radical (3693, 3700 cm−1) interact with the alcohol in the case of
methanol and ethanol. Some differences are, however, observed
in the 3600–3000 cm−1 region concerning water and/or interact-
ing hydroxyls. Differences between ethanol and methanol in this
region would be related to the different alcohol coverage at the
surface and the competition with water [46,51–54]. In the case of
2-propanol and for both samples, bands at 2972, 2932-2930, and
2868–2864 cm−1 (ꢀCH), 1468 cm−1 (␦asCH3), 1380 cm−1 (␦sCH3)
and ␦CH mode at 1282 cm−1 were observed with several band
assigned to the CO stretch vibration mode in the 1175–1130 cm−1
region [55,56]. The latter (C O) bands describe the variety of inter-
actions of the alcohol and the surface, generating several (more
than 2) 2-propanol and propoxide species. It is relatively difficult
to observe other products at the initial contact of the 2-propanol
and the catalysts.
Under reaction conditions (illumination) the consumption of
the alcohols (in the three cases here tested) is rather visible (nega-
tive bands) in the region of the C H and C O stretching vibration
modes. All adsorbed species (alcohol and alcoxy species) are con-
sumed and/or desorbed from the surface of the Ti and Nb samples.
In the case of methanol, Ti displays appearance (bands with pos-
itive intensity) of new bands at 1191, 1388, 1515 and 2868 cm−1
while Nb shows new contributions at 1122, 2865, 2963 and a broad
band at ca. 1400 cm−1. The ca. 1120–1190 cm−1 peaks would indi-
cate the formation of a carbonyl species while the 1515–1400 cm−1
would suggest the formation of formate and, less likely, other car-
boxylate species [48–53]. This indicates that the surface species
detected by infrared agrees with detection of formic acid as main
product of the methanol transformation. As mentioned, distinc-
tive of the methanol case with respect to other alcohols is the
detection of carbonyl species at the surface of both catalysts. Alde-
hydes may evolve from the (activated) carbonyl species detected
(1190–1120 cm) although formaldehyde is only observed at the
gas phase in the case of the Nb sample under UV illumination.
The detection of (adsorbed) methyl formate is on the other hand
not obvious as the most characteristic bands at ca. 1200 and
1750 cm−1 [57] were not observed. In the case of ethanol, and
for both samples we observe bands at ca. 2950, 1500-1402, 1367-
1260 and ca. 1294 cm−1 strongly indicative of carboxylate (and
carbonate) moieties including formates [48–53]. Finally, in the case
of 2-propanol we encountered a relatively similar situation with
bands at ca. 2955-2954, 1444 and 1351 cm−1. The absence of bands
around/above 1500 cm−1 would indicate the lower importance of
carboxylates (particularly formate species) [55,56]. Note that we
cannot detect acetone (no hint around or above 1650 cm−1), but
this is not an uncommon situation due to the strong interaction
that 2-propanol has with titania surfaces. Such a strong interaction
(with respect to reaction products) drives to a rather small coverage
for other chemical compounds and particularly acetone [55,56,58].
Overall, under reaction we see the production of carboxy-
late/carbonate species as the main surface product in the case of
methanol and ethanol, indicating that the hole attack to the alco-
hol would end up in the formation of oxidized species related to
the corresponding alcohol/alcoxy adsorbed species. In the case of
methanol, carboxylate species generate formic acid and methyl for-
mate. The latter is produced by interaction of the carboxylate and
one methanol molecule [57]. For ethanol the carboxylate or carbon-
ate species only end up in the formation of ethyl acetate (Table 2).
For methanol, surface carbonyl species were additionally detected
but only in one case (Nb under UV illumination) the formation of
the aldehyde was observed as a product (Table 2). In the case of 2-
propanol the carboxylate-type species are apparently less present
at the surface of both Ti and Nb materials, likely due to the low cov-
erage of products and the relatively easy desorption of the main
reaction product, acetone. In any case, we also observed evolu-
Fig. 4. Reaction rate (A) and quantum efficiency (B) of the hydrogen photo-
production using different alcohols. Me, Et and Is stand for methanol, ethanol and
2-propanol (isopropanol).
tantly observed with the decrease of isolated hydroxyl of titania
at ca. 3693 cm−1 as well as the decreasing of the water (and inter-
acting hydroxyl radicals) signals at ca. 3650-3200 and 1640 cm−1
.
C
O stretching bands for these two (methanol/methoxy) adsor-
bate species appear around 1051 cm−1 but the bands are so
broad that cannot be easily resolved. Additional peaks in the ca.
1500–1100 cm−1 region are indicative of the formation of other
species. Although the region around 1450 cm−1 has contribution
from methanol/methoxy species, the presence of peaks at ca.
1370-1350, 1205, and 1165–1145 cm−1 indicate the formation of
carboxylate or carbonate species [50–53]. In view of the relative
similitude observed in these frequencies after the saturation with
the ethanol:water mixture, the dominance of carbonate species
may be inferred if such results are jointly considered with the
selectivity data collected for methanol and ethanol in Table 2. In
the case of ethanol (central row in Figs. 5 and 6), the C H region
showed peaks at ca. 2977-2975, 2930, 2900 and 2875 cm−1 for
both (Ti and Nb) samples. Additional bands related to the alco-
hol are presented at ca. 1382-1279 (␦asCH3), 1140 (C C stretch)
and 1092-1089 and 1055-1048 (C O stretch) cm−1 [25,54]. The
C
O stretch doublet also provides evidence of ethanol and ethoxy
species in both Ti and Nb samples. As mentioned, other bands (par-
ticularly around 1450 and 1390–1380 cm−1) indicate the presence
of carboxylate species but mainly carbonate species formed dur-
ing the contact of the alcohol and the surface. Additional (negative)
contributions are, as occurring with methanol, related to hydrox-

94
O. Fontelles-Carceller et al. / Molecular Catalysis 446 (2018) 88–97
Fig. 5. DRIFTS spectra for the in situ analysis of the methanol (upper panel), ethanol (middle panel) and 2-propanol (lower panel) photo-transformation occurring under
reaction conditions for the Ti sample. Spectra obtained during saturation (dark conditions) with the alcohol:water mixture, under reaction-illumination conditions and
subsequently at dark (post-reaction) conditions are presented.
tion of the acetone. The propanone-diisopropyl-acetal molecule is
formed by reaction of acetone and two molecules of 2-propanol
[59].
To further progress in the analysis of the effect of the alcohol
chemical nature we discuss the influence of their physico-chemical
properties on hydrogen photo-production. The importance of the
interaction of lone pairs of the OH groups with unoccupied Ti

O. Fontelles-Carceller et al. / Molecular Catalysis 446 (2018) 88–97
95
Fig. 6. DRIFTS spectra for the in situ analysis of the methanol (upper panel), ethanol (middle panel) and 2-propanol (lower panel) photo-transformation occurring under
reaction conditions for the Nb sample. Spectra obtained during saturation (dark conditions) with the alcohol:water mixture, under reaction-illumination conditions and
subsequently at dark (post-reaction) conditions are presented.
3d states to activate the alcohol molecule suggests that alcohol
polarity may influence activity, as suggested previously by several
authors [40,41,60]. However, methanol markedly differs from the
expected tendency if polarity drives photo-activity for hydrogen
production. This point has been argued to be based in chemi-
cal differences concerning, in first place, the CO production [41].

96
O. Fontelles-Carceller et al. / Molecular Catalysis 446 (2018) 88–97
Table 4
CO can play a series of roles depending of its quantity as well
as its handling at the metal-support interface, which could be
positive for hydrogen production through the water gas shift reac-
tion [48], but can be a poison at the surface of the metal with
detrimental effects in photo-activity [61]. Whatever the role of
the CO in the reaction is, here we do not detect such molecule
using infrared while for other catalysts it has been detected in
similar experimental conditions [50]. Also, a different alcohol
oxidation mechanism was proposed for methanol. This is based
in the easy formation and surface stability (or no easy evolu-
tion) of formate species created from methanol with respect to
other carboxylates or (hydrogen)-carbonates formed from higher
alcohols [41]. However, at least when comparing methanol with
ethanol using infrared spectroscopy, we show essentially the same
behavior of the titania surfaces against the alcohol and evolving
molecules coming from the hole-attack. In these two cases, the
alcohol evolution seems to proceed initially through a stepwise
mechanism with no change in the chain length with consecutive
(hole-attack) steps such as: alcohol (alcoxy) → aldehyde (carbonyl-
related species) → acid (carboxylates). As mentioned, carboxylate
species also evolve forming methyl formate and ethyl acetate. Such
a mechanism has been previously proposed [35]. Thus, we can
dismiss the formate vs. other carboxyaltes hypothesis as being of
primary significance in our case.
Average oxidation potential (equation Eox = -ꢀG◦/(n·F)) for the different alcohol
reactions taking place in the reactor for the different catalysts and illumination
conditions.
Catalyst
Illumination
Alcohol
E◦ox. (V) versus
NHEa
UV
Vis
UV
Vis
UV
Vis
UV
Vis
UV
Vis
UV
Vis
0.037
0.044
0.042
0.052
0.112
0.119
0.101
0.092
0.241
0.213
0.143
0.143
Nb
Ti
Methanol
Nb
Ti
Ethanol
2-
Nb
Ti
propanol
a
Average standard error; 8.7%.
Alternatively or in parallel, some authors highlight the signif-
icance of thermodynamic issues of the alcohol oxidation step(s)
in driving hydrogen photo-production. According to current theo-
ries of electron transfer reactions between an electron donor and
valence band holes in a semiconductor [41,62], the experimental
rate constant (kexp) for such reactions can follow the relation:
ꢄ
ꢅ
0
0
−
E
−E
)
/RT
ox
VB TiO
(
2
Kexp ∝ e
(9)
Where EV0B TiO is the valence band potential of titania and Eo0x is
(
)
the oxidation2potential of the donor, which according to the dis-
cussion presented in the introduction (Eqs. (1)–(4)) is the alcohol.
If this charge transfer controls the reaction rate we can expect a
linear relationship between the kexp expression in Eq. (9) and the
hydrogen photo-production rate (which, by the way, takes place at
constant temperature in our experiments). On the other hand, note
that, in this study and using our samples, we showed that the rate
is also a real measurement of the reaction (quantum) efficiency.
Of course, a real calculation should take into account the different
products formed during the reaction and the corresponding ther-
modynamic values. This is a consequence of the fact that hydrogen
is extracted from the alcohol molecule in each step of the equations
considered in Table 3, as mentioned in the introduction [36,38].
Concretely, Eo0x is calculated (equal to (−)ꢆG◦/nF, being the Fara-
day constant F = 96.485 kC mol−1.) from the standard Gibbs energy
of alcohol oxidation (ꢆG◦) and the number of charge species (n)
involved in the reaction. These parameters are presented in Table 3
and were calculated using tabulated values of the Gibbs energies
and the balance equations also presented in such table [63]. As men-
tioned, for each alcohol we consider the selectivity to the different
(alcohol) oxidation products and thus for each sacrificial molecule
we obtained an average value for its Eo0x parameter (Table 4). It is
important to note that, this average allows to obtain different val-
ues for different illumination sources, a fact required by our (and
in general for all) catalytic experiments. The Eo0x parameter values
are plotted against the hydrogen production rate in Fig. 7 for the
different alcohols and illumination conditions here tested.
Fig. 7. Rates of H2 production versus exp(Eox.) of the alcohol.
tions in terms of the alcohol nature within our series. It should be
pointed out that the three alcohols have hydrogen in alpha position
of the hydroxyl moiety but show marked differences in the hydro-
gen photo-production rate. In particular, we note that methanol fits
in the linear trends, similarly to the other alcohols. This strongly
suggests the similar role and evolution of the alcohol in the photo-
production of hydrogen, irrespective of the sample and excitation
wavelength.
In Fig. 7 we observed a quasi-linear trend for the two samples
and illumination conditions. This suggests that, for all cases, the
rate is dominated by thermodynamic reasons and that the active
center and mechanism do not suffer strong differences or varia-

O. Fontelles-Carceller et al. / Molecular Catalysis 446 (2018) 88–97
97
4. Conclusions
[15] U. Caudillo-Flores, M.J. Mun˜oz-Batista, J. Cortés, M. Fernández-García, A.
Kubacka, Mol. Catal. 437 (2017) 1–10.
[16] J.C. Colmenares, A. Magdziarz, M.A. Aramendia, A. Marinas, J.M. Marinas, F.J.
Urbano, J.A. Navío, Catal. Commun. 16 (2011) 1–6.
[17] T. Kawan, T. Sakata, Nature 286 (1980) 474–476.
[18] A. Galinska, J. Walendziewski, Energy Fuels 11 (2005) 1143–1147.
[19] Y.X. Li, Y.Z. Me, S.Q. Peng, G.X. Lu, S.B. Li, Chemosphere 63 (2006) 1312–1318.
[20] A. Pastora, D.I. Kondarides, X.E. Verykios, Catal. Today 124 (2007) 94–102.
[21] X. Fu, J. Long, X. Wang, D.C.Y. Leung, Z. Ding, L. Wu, Z. Zhang, Z. Li, X. Fu, Int. J.
Hydrogen Energy 33 (2008) 6484–6491.
[22] W. Sun, S. Zhang, Z. Liu, W. Wang, Z. Mao, Int. J. Hydrogen Energy 33 (2008)
1112–1117.
[23] M. Bouker, Catal. Lett. 142 (2012) 923–929.
[24] Y. Ma, Q. Xu, X. Zong, D. Wang, G. Wu, X. Wang, C. Li, Energy Environ. Sci. 5
(2012) 6345–6351.
[25] E. Pulido Melián, J.A. Ortega Méndez, C. Rodríguez López, M. Nereida Suárez,
J.M. Don˜a Rodríguez, J.A. Navío, D. Fernández Hevia, Int. J. Hydrogen Energy
38 (2013) 11737–11748.
In this contribution we examine the photo-production of
hydrogen from bio-alcohol molecules using bare anatase and
Niobium-doped anatase materials having a 0.5 wt% Pt as co-
catalyst. These materials have similar morphological properties
(particularly BET surface area and primary particle size of the
noble metal and anatase components). Three alcohol molecules,
methanol, ethanol, and 2-propanol, and two illumination sources
corresponding to UV and visible ranges were utilized. Catalytic out-
put was measured through the reaction rate and the true quantum
efficiency observables. The parallel behavior of both observables
along our sample series indicates that the reaction rate renders a
realistic measurement of the catalytic properties of the solids. The
Niobium-doped sample presented an improved performance under
all experimental conditions tested, proving quantum efficiency val-
ues going from ca. 1–5%. Enhancement ratios from 1.8 to 2.8 were
observed with respect to the titania reference system.
For all illumination conditions and samples we interpret cat-
alytic activity in hydrogen production with the combined use of an
in-situ infrared study as well as the thermodynamic foundation of
the reaction rate. From a mechanistic point of view we observed
differences between alcohols but those are of minor relevance to
justify quantitatively the differences appearing in hydrogen pro-
duction rates using the three alcohols here studied. Differences in
activity were correlated quantitatively with the average oxidation
potential of the alcohol if the selectivity in the bio-molecule oxida-
tion is taken into account to interpret hydrogen photo-production.
For all samples and illumination conditions here tested, the work
showed that the thermodynamic of the interaction between the
solid(s) and the sacrificial molecule is likely driving the catalytic
output and interprets the different rates and quantum efficiency
values obtained.
[26] D. Slament, V. Tristanidi, M. Ibadurrbhran, Energy Res. 37 (2013) 1372–1381.
[27] G.N. Nomikos, P. Panagiotopoulous, D.I. Kondarides, X.E. Verykios, Appl. Catal.
B 146 (2014) 249–257.
[28] A. Kubacka, M.J. Mun˜oz-Batista, M. Fernández-García, S. Obregón, G. Colón,
Appl. Catal. B 163 (2015) 214–222.
[29] S. Obregón, M.J. Mun˜oz-Batista, M. Fernández-García, A. Kubacka, G. Colón,
Appl. Catal. B 179 (2015) 468–478.
[30] Y.-J. Yuan, J.-R. Tu, Z.-J. Ye, Z.-T. Yu, Z.-G. Zou, Appl. Catal. B 188 (2016) 13–22.
[31] Z. Chen, X. Jian, C. Zhu, C. Shi, Appl. Catal. B 199 (2016) 241–251.
[32] Q. Hu, J. Huan, G. Li, J. Chen, Z. Zhang, Z. Deng, Y. Jiang, W. Guo, Y. Cao, Appl.
Surf. Sci. 369 (2016) 201–206.
[33] M. Zhang, R. Sun, Y. Li, Q. Shi, L. Xie, J. Chen, X. Xu, H. Shi, W. Zhao, J. Phys.
Chem. C 120 (2016) 10746–10756.
[34] J.G. Highfield, M.H. Chen, P.T. Ngyen, Z. Chen, Energy Environ. Sci. 2 (2009)
991–1002.
[35] G.L. Chiarello, M.H. Aguirre, E. Selli, J. Catal. 273 (2010) 182–190.
[36] T.A. Kandiel, I. Ivanova, D.W. Bahnemann, Energy Environ. Sci. 7 (2014)
1420–1426.
[37] A. Naldom, M.D. Ángelo, M. Altomare, M. Morelli, R. Scott, F. Morazzoni, E.
Selli, V. Del Santo, Appl. Catal. B 130–131 (2013) 239–248.
[38] Galinska, J. Walendziewski, Energy Fuels 19 (2005) 1143–1147.
[39] C.R. López, E. Pulido Melián, J.A. Ortega Méndez, J.M. Don˜a Rodríguez, O.
González Díaz, J. Photochem. Photobiol. A 312 (2015) 45–54.
[40] W.-T.- Chen, A. Chan, Z.H.N. Al-Azri, A.G. Dosado, M.A. Nadeem, D.
Sun-Waterhouse, H. Idriss, G.I.N. Waterhouse, J. Catal. 329 (2015) 499–513.
[41] 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.
Acknowledgements
[42] G.K. Willianson, W.H. May, Acta Metall. 1 (1953) 22–29.
[43] P. Kubelka, J. Opt. Soc. Am. 38 (1948) 448–457.
[44] M. Fernández-García, A. Martínez-Arias, J.C. Hanson, J.A. Rodriguez, Chem.
Rev. 104 (2004) 4063–4105.
[45] S.E. Braslavsky, A.M. Braun, A.E. Cassano, A.V. Emeline, M.I. Litter, L.
Palmisano, V.N. Parmon, N. Serpone, Pure Appl. Chem. 83 (2011) 931–1014.
[46] M.J. Mun˜oz-Batista, A. Kubacka, A.B. Hungría, M. Fernández-García, J. Catal.
330 (2015) 154–166.
We are thankful to MINECO (Spain) for supporting the work
carried out through the ENE2016-77798-C4-1-R grant. We thanks
Dr. Rodriguez-Ramos group for performing HPLC analysis of liquid
samples.
[47] M.L. Satuf, R.J. Brandi, A.E. Cassano, O.M. Alfano, Ind. Eng. Chem. Res. 44
(2005) 6643–6649.
[48] A. Yamakata, T.-A. Ishibashi, H. Onishi, J. Phys. Chem. B 106 (2002) 9122–9125.
[49] P.A. Panatoyov, S.T. Burrows, J.P. Morris, J. Phys. Chem. C 116 (2012)
6623–6635.
[50] O. Fontelles-Carceller, M.J. Mun˜oz-Batista, E. Rodríguez-Castellón, J.C. Conesa,
M. Fernández-García, A. Kubacka, J. Catal. 347 (2017) 157–169.
[51] A. Matson, L. Osterlund, J. Phys. Chem. C 114 (2010) 14121–14132.
[52] A.C. Sola, D. Garzón Sousa, J. Aran˜a, O. González Díaz, J.M. Don˜a Rodríguez, P.
Ramírez de la Piscina, N. Homs, Catal. Today 266 (2016) 53–61.
[53] C.E. Naroyakkara, J.K. Dillon, V.H. Grassian, J. Phys. Chem. C 118 (2014)
25487–25495.
[54] D.L. Carvalho, L.E.P. Borges, L.G. Appel, P. Ramírez de la Piscina, N. Homs, Catal.
Today 213 (2013) 115–121.
[55] F. Arsac, D. Bianchi, J.M. Chovelon, A.C. Ferronato, J.M. Herrmann, J. Phys.
Chem. A 110 (2006) 4202–4212.
[56] F. Arsac, D. Bianchi, J.M. Chovelon, C. Ferronato, J.M. Herrmann, J. Phys. Chem.
A 110 (2006) 4213–4222.
[57] M. El-Roz, Ph. Bazin, Phys. Chem. Chem. Phys 17 (2015) 11277–11285.
[58] M.J. Mun˜oz-Batista, U. Caudillo-Flores, F. Ung-Medina, M.C. Chávez-Parga, J.A.
Cortés, A. Kubacka, M. Fernández-García, Appl. Catal. B 201 (2016) 400–410.
[59] H. Maarse, Volatile Compounds in Foods and Beverages, 1991, New York.
[60] Y.Z. Yang, C.H. Chang, H. Idriss, Appl. Catal. B Environ. 67 (2006) 217–222.
[61] G.H. Haselman, D. Elder, ACS Catal. 7 (2017), 4688–4675.
[62] V. Balzani, F. Scandola, in: Michael Gratzel (Ed.), Energy Resources Through
Photochemistry and Catalysis, Academic Press Inc, New York, 1983, pp. 2–48.
[63] J.A. Dean, (Ed.) Langeı´s Handbook of Chemistry, 15th ed. 1999, New York.
References
[1] N. Vecirolu, F. Barbir, Int. J. Hydrogen Energy 17 (1992) 391–404.
[2] X. Chen, S.S. Chen, L. Guo, S.S. Mao, Chem. Rev. 110 (2010) 6503–6570.
[3] K. Maeda, K. Domen, J. Phys. Chem. Lett. 1 (2010) 2655–2661.
[4] A. Kubacka, M. Fernández-García, G. Colón, Chem. Rev. 112 (2012) 1555–1614.
[5] K. Villa, X. Domenech, S. Malato, M.I. Maldonado, J. Peral, Int. J. Hydrogen
Energy 38 (2013) 12718–12724.
[6] Y. Ma, X. Wang, Y. Jia, X. Chen, H. Han, C. Li, Chem. Rev. 114 (2014)
9987–10043.
[7] J.C. Colmenares, R. Luque, Chem. Soc. Rev. 43 (2014) 765–778.
[8] X. Li, J. Yu, J. Low, Y. Fang, J. Xiao, X. Chen, J. Mater. Chem. A 3 (2015)
2485–2534.
[9] J.C. Colmenares, Nanophotocatalysis in selective transformations of
lignocellulose-derived molecules: a green approach for the synthesis of fuels,
fine chemicals, and pharmaceuticals, in: Nurxat Nuraje, Ramazan Asmatulu,
Guido Mul (Eds.), Green Photo-active Nanomaterials: Sustainable Energy and
Environmental Remediation, The Royal Society of Chemistry, Warsaw, 2016,
pp. 168–201 (chap. 8).
[10] Y. Li, G. Ma, S. Peng, G. Lu, S. Li, Appl. Surf. Sci. 254 (2008) 6831–6836.
[11] M.A. Khan, S.I. woo, O.B. Yang, Int. Hydrogen Energy 33 (2008) 5345–5351.
[12] X. Sun, H. Liu, J. Dong, J. Wei, Y. Zhang, Catal. Lett. 135 (2010) 219–225.
[13] S. Piskuniv, O. Lisovski, J. Begens, M. Wessel, E. Spohr, J. Phys. Chem. C 119
(2015) 18686–18696.
[14] O. Fontelles-Carceller, M.J. Mun˜oz-Batista, J.C. Conesa, M. Fernández-García,
A. Kubacka, Appl. Catal. B 216 (2017) 133–145.