Enhanced activity and stability of Ru-TiO2rutile for liquid phase ketonization

Article abstract of DOI:10.1016/j.apcata.2016.10.025

Stabilization of oxygen vacancies on metal oxides (e.g. TiO₂) in liquid phase is an important challenge for the utilization of these materials in artificial photosynthesis, environmental remediation and biomass conversion. To create materials with low-energy barriers for vacancies formation and high stability in aqueous environments, we have developed partially hydrophobic (contact angle ≥90°) TiO₂rutile decorated with Ru nanoparticles. Negligible catalytic activity was observed when hydrophilic (contact angle 51°) 5 wt.% Ru/TiO₂anatase was utilized in hot liquid water, while amphiphilic 5 wt.% Ru/TiO₂rutile (contact angle ～90°) retained its catalytic activity. Fine-control of crystalline structure (lattice matching) of TiO₂and Ru allowed us to accelerate the rate of reaction, while the high surface hydrophobicity of the support enabled the stabilization of Ti³⁺cations in aqueous and organic environments.

Full text of DOI:10.1016/j.apcata.2016.10.025

Accepted Manuscript
Title: Enhanced Activity and Stability of Ru-TiO2 Rutile for
Liquid Phase Ketonization
´
´
Author: Nicolas Aranda-Perez M. Pilar Ruiz Javier Echave
Jimmy Faria
PII:
S0926-860X(16)30520-8
DOI:
Reference:
http://dx.doi.org/doi:10.1016/j.apcata.2016.10.025
APCATA 16041
To appear in:
Applied Catalysis A: General
Received date:
Revised date:
Accepted date:
9-8-2016
5-10-2016
23-10-2016
´
´
Please cite this article as: Nicolas Aranda-Perez, M.Pilar Ruiz, Javier Echave, Jimmy
Faria, EnhancedActivityand StabilityofRu-TiO2Rutile forLiquidPhase Ketonization,
Applied Catalysis A, General http://dx.doi.org/10.1016/j.apcata.2016.10.025
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

Enhanced Activity and Stability of Ru-TiO2 Rutile for Liquid Phase Ketonization
Nicolás Aranda-Pérez, M. Pilar Ruiz, Javier Echave, and Jimmy Faria*
Abengoa Research, Campus Palmas Altas, C/ Energía Solar 1, 41014 Seville, Spain
*Corresponding author: Jimmy Faria, ing.jimmyfaria@gmail.com

Graphical abstract
θ < 90°
Hydrophillic-Anatase
Activity Ratio (H2O:n-
hexane)
1,00
0,75
0,50
0,25
0,00
100
80
60
40
20
0
θ
ContactAngle (degrees)
θ > 90°
Hydrophobic-Rutile
5% Ru/TiO2 5% Ru/TiO2 5% Ru/TiO2
5% Ru/TiO2
Anatase
5% Ru/TiO2
Rutile
5% Ru/TiO2
P25
P25
Rutile
θ
Ruthenium and TiO2 rutile stabilize Ti3+ Lewis acid sites for decarboxylative C-C coupling of
acids with remarkably stability in hot condensed water.
Highlights:
1) Liquid-phase ketonization of acetic acid over three catalysts of Ru on different titania
supports (anatase, rutile and P25) demonstrates the outstanding effect of the support
hydrophobicity on the catalyst stability;
2) The catalyst of Ru supported on rutile (the most hydrophobic support studied) presents
the highest ketonization activity and stability;
3) Partially hydrophobic materials can facilitate the high temperature reactions in aqueous
environments for biomass conversion, environmental remediation, artificial photosynthesis and
green chemistry.
Abstract
2

Stabilization of oxygen vacancies on metal oxides (e.g. TiO2) in liquid phase is an important
challenge for the utilization of these materials in artificial photosynthesis, environmental
remediation and biomass conversion. To create materials with low-energy barriers for vacancies
formation and high stability in aqueous environments, we have developed partially hydrophobic
(contact angle ≥ 90°) TiO2 rutile decorated with Ru nanoparticles. Negligible catalytic activity
was observed when hydrophilic (contact angle 51°) 5 wt.% Ru/TiO2 anatase was utilized in hot
liquid water, while amphiphilic 5 wt.% Ru/TiO2 rutile (contact angle ~ 90°) retained its catalytic
activity. Fine-control of crystalline structure (lattice matching) of TiO2 and Ru allowed us to
accelerate the rate of reaction, while the high surface hydrophobicity of the support enabled the
stabilization of Ti3+ cations in aqueous and organic environments.
Keywords: ketonization, titanium dioxide, hydrothermal stability, hydrophobicity
1. Introduction
Biomass is the largest reservoir of renewable carbon on the planet reaching annual
production rates of 170 billion metric tons,[1] and the potential of replacing 27% of the annual
world total transport fuel demand (i.e. diesel, kerosene and jet fuel), which could avoid around
2.1 gigatonnes (Gt) of CO2 emissions.[2] However, harvesting and transforming the biomass into
liquid transportation fuels and chemicals is not simple.[3] It requires complex chemical
transformation steps that can reduce the oxygen content of the biomass, while preserving the
carbon and hydrogen.[4–6] The liquefaction of biomass via fast pyrolysis is an interesting
alternative as the liquid product and energy density is maximized while the oxygen-to-carbon
ratio is reduced. The resulting pyrolysis oil (bio-oil) is a complex mixture of water and
oxygenated compounds (e.g. short acids, aldehydes, anhydrosugars, furans, and phenolics).
This bio-oil can be used directly or processed for heat, power, biofuels, and chemicals.[7–9] In
order to make a fungible fuel the pyrolysis oil must undergo a catalytic upgrading process to
reduce the excess oxygen and increase C-C length of the small molecules, while preserving the
energy content.[10–14]
To accomplish this goal, the small acids, which can account up to 30 % of the aqueous
phase, must be transformed into larger hydrocarbons before undergoing hydrogenation to avoid
3

the formation of low-value light gases (e.g. methane, ethane) during hydrotreating. Ketonization
of the acids can be employed to reduce the oxygen content by elimination of CO2 and H2O,
while making a ketone that can serve as an alkylation agent of aromatics downstream.[15]
Additionally, the presence of water in high concentrations makes unavoidable the development
of stable catalytic materials that can operate under hydrothermal stress.[16,17]
On transition metal oxides surfaces, decarboxylative C-C coupling or ketonization reaction
takes place on coordinatively unsaturated metal cations or oxygen vacancies where the
carboxylates are adsorbed to form intermediate species that lead to the formation of
ketones.[18,19] The catalytic activity of the metal oxides is strongly dependent on the presence of
surface metal cations/oxygen anions pairs that act as Lewis acids and bases towards the
carboxylates. On metal oxides with reducible properties (i.e. TiO2, CeO2, ZrO2), it is possible to
generate oxygen vacancies at temperatures above 300 °C for ketonization reactions in vapor
phase.[15] At lower temperatures, however, the ketonization reaction does not occur on bare
amphoteric oxides as the unsaturated metal cations are not formed.[20] One alternative to
promote this process is to add doping agents (e.g. Na, K, Cs)[21]21 and metals with high
hydrogen-dissociation activity.[20,22,23] On metal oxides doped with reducible metals, e.g.
Ru/TiO2, it is possible to increase significantly the ketonization rate by creating additional active
sites upon reduction in H2.[20,22,24] For instance, in bio-oil upgrading, where the ketonization of
short-chain acids allows oxygen removal and C-C chain growth with zero consumption of
hydrogen, it is particularly important to perform the reaction at mild temperatures (<250° C) to
avoid secondary reactions.[25,26] Operating in aqueous environments at moderate temperatures
(>150 °C) is remarkably difficult due to the structural collapse of the metal oxide[27] and
competitive adsorption of water onto unsaturated metal cations, like partially unsaturated Ti
cations (i.e. Lewis acid sites).[28–30]
A detailed experimental study of the mechanisms controlling the stability of zeolite catalysts
in liquid water revealed that defects on the crystalline structure are the controlling factor.[31] The
defective sites, typically hydroxylated metal-oxygen bonds, serve as nucleation sites for the
initiation of the degradation process.[32] The stability of the zeolite is dramatically improved by
either functionalization of these OH-functionalities by silylation with organosilanes[16] or
synthesis in fluorinated media to create near-perfect crystalline structures. [32–35] The surface
4

wettability of amphoteric metal oxides like TiO2 is far more complex than conventional metal
oxides and zeolites, as these materials can reversibly switch from hydrophobic to
superhydrophillic upon exposure to UV-light.[36] The formation of surface hydroxyl groups on Ti3+
cations, generated by the UV-Light, increases the surface hydrophilicity. More recently, X. Feng
et al.[37] reported that it is possible to achieve switchable superhydrophobicity on TiO2 rutile
nanorods films, as a result of the combination of micro- and nanoscale hierarchical surface
structures, the orientation of the crystal planes and the surface photosensitivity.
Inspired on this previous research, we have developed a material that combines the high
activity of TiO2 rutile-ruthenium interface with the hydrophobic modification of the surface by
thermal dehydroxylation. To investigate the interaction of the crystalline phases of TiO2 and the
effect of hydrophobicity, three catalysts composed of Ru supported on TiO2 anatase, TiO2 P25
(mixture of anatase:rutile 80:20 in weight), and TiO2 rutile were synthesized. A detailed
characterization of the different materials and catalysts stability studies in aqueous phase
enabled the identification of properties-structure-performance relationships. As a showcase we
have selected the ketonization reaction of acetic acid, which is present in high-concentrations in
bio-oil mixtures and is one of the main contributors to the corrosiveness and instability of
pyrolysis oil.[7,9,38]
The simplified ketonization reaction of acetic acid is the following:
2CH3COOH → CH3COCH3 + CO2 + H2O
Eq. 1
Experimental studies have shown that ketonization rate of reaction is strongly dependent on
the number of α-hydrogen of the molecule.[39–41] That is on carboxylic acids with higher number
of α-hydrogen the faster the reaction rate. In addition, transition state analysis of kinetic data
obtained for the ketonization of carboxylic acids with different chain-length suggested that this
reaction proceeds via a β-ketoacid intermediate. The reaction rate significantly decreases when
long-chain carboxylic acids were reacted as the C-C coupling reaction was hindered by bulkier
alkyl-groups.[22] Density Functional Theory study on ZrO2 slab showed that the formation of
ketones through a concerted mechanism is kinetically unfavorable. In contrast, the pathway
involving a α-hydrogen abstraction step had much lower activation energy, due to the formation
of a highly reactive nucleophile that reacts with a nearby carboxylate molecule to produce a β-
5

ketoacid intermediate that produces the ketone upon decarboxylation.[42] Based on this we have
decided to use as a probe molecule acetic acid, which is a molecule containing three α-
hydrogen.
2. Materials and methods
2.1.Catalyst synthesis
First, three different commercial TiO2 were selected attending its phase composition:
anatase (TiO2 G5), rutile (TiO2 Rutile), and anatase/rutile in a proportion of 80/20 (TiO2 P25).
TiO2 P25 and TiO2 G5 were kindly provided by Evonik Industries and rutile was purchased to
Sigma Aldrich. The three catalysts of the study were synthesized by excess volume
impregnation of the Ru precursor on the different TiO2 supports. The Ru loading was 5 wt.%
using Ru (III) Chloride precursor (Sigma Aldrich, 99.98%). The required amount of the metal
precursor was dissolved in deionized water and left for solution for 4h; subsequently TiO2 was
added and kept under stirring (500 rpm) overnight. Afterwards, evaporation of water was
performed by heating at 150°C for 3h. Once the impregnation is accomplished, the catalyst was
dried at 100°C overnight and finally calcined at 400°C for 4 h with a heating ramp of 2°C/min.
The same preparation procedure was followed for all the samples. In this work the catalysts are
labeled as Ru/TiO2 anatase, Ru/TiO2 rutile and Ru/TiO2 P25.
2.2.Catalyst characterization
The catalysts were characterized by micro-fluoresce of X-Rays (MXRF), transmission
electron microscopy (TEM), X-ray diffraction (XRD), N2-physisorption, temperature-programmed
reduction (TPR), and temperature programmed desorption of NH3 (NH3-TPD), and powder
contact angle.
2.2.1. X-rays microfluorescence (XRMF)
Compositional analysis of the catalysts was performed employing a microfluorescence X-
Ray spectrometer Eagle-III from EDAX employing a 40 W Rh X-Ray tube. The detection system
employed an 80 mm2 Lithium-drifted Silicon crystal liquid N2 cooled with a Be window that can
6

allow Na and U simultaneous detection with a resolution of <150 eV for MnKa at 5000 counts
per second (CPS) and <185 eV for MnKa at 15,000 CPS. The optics of the Eagle-III consisted
on a mono-capillary system operating at 300 m and open aperture of 2 mm spot. The powder
samples were placed in the sample holder and introduced ¨as synthesized¨ inside the analysis
chamber. The analysis was performed using a total data acquisition time of approximately 15
seconds on a 300 m scanning area.
2.2.2. High angle annular dark-field scanning transmission electron microscopy
(HAADF), High-resolution transmission electron microscopy and energy
dispersive X-ray spectroscopy (HR-TEM-EDS)
High angle annular dark-field scanning transmission electron microscopy (HAADF-STEM)
and high-resolution transmission electron microscopy (HRTEM) was carried out using a FEI
Tecnai F30, operated at 300 kV equipped with a Gatan CCD camera, an EDS (EDAX) detector
and a Gatan Tridiem Energy Filter. Each sample was prepared by sonicating in n-butanol to
improve the dispersion of the particles. The solution was dropped onto a holey carbon coated
300 mesh copper grid.
2.2.3. X-ray diffraction (XRD)
XRD analysis was carried out on a D8I Bruker for dust samples with incident slits with Cu
anode. The data were collected in an angle range 0-45º. A semi-quantitative method was used
to determine the crystalline structure of TiO2 comparing to Corindon data-base.
2.2.4. N2-Physisorption
BET surface area, pore size and volume were measured using N2 physisorption at liquid
nitrogen temperature on a Micromeritics ASAP 2020C. Degasification was performed at a
temperature of 200ºC for 4 h.
2.2.5. Temperature programmed reduction (TPR)
TPR analysis of 100 mg of the catalysts were conducted in Autochem 2920 using a gas
mixture of 10%H2 in Ar at a flow rate of 30 sccm with a linear heating ramp of 5ºC/min up to
900ºC and holding time of 1 min. H2 consumption was determined by a thermal conductivity
detector.
7

2.2.6. Ammonia temperature programmed desorption (NH3-TPD)
NH3-TPD was carried out on Autochem 2920 using a gas mixture of 15% NH3 in He. The
surface was firstly clean with He, then it was reduced in 50 sscm of H2 at 230ºC during 3 h to
replicate reduction conditions, then it was cleaned again with He at 230ºC to evaporate any H2O
formed. Once 35ºC was reached, 15%NH3/He was passed through the sample for 60 min in
which it was chemisorbed. Finally, a temperature ramp of 10ºC/min up to 900ºC was
programmed to study the desorption profile, which is related to the concentration of exposed
cations on the catalyst. Total acidity was measured by integrating the area under the curve of
NH3-TPD.
2.2.7. Powder contact angle (PCA)
Powder contact angle was measured using DCAT21 from Dataphysics. The samples were
measured first with n-hexane for calibration and then with deionized water.
2.3. Ketonization experiments
Ketonization reactions were carried out in a 150 mL batch reactor (Berghof BR-100) with a
Teflon liner of 100 mL to avoid corrosion of the reactor. A scheme of the experimental setup is
shown in Figure S1. Firstly, 200 mg of catalyst and 65 mL of n-hexane (VWR, 99%) or
deionized water (depending on the experiment), were loaded on the liner. In a first stage, the
catalyst was reduced by placing the insert on the reactor and pressurizing, after purge with N2,
with 30 bar of H2 stirring at 750 rpm. The reduction was conducted at 230°C and the pressure
increased autogenously (60 bar) for 3h. After cooling the reactor to room temperature, the H2
was purged and N2 at 5 bar was introduced into the reactor. Then after a heating ramp of 2h,
the target temperature of the reaction was reached (200°C, 210°C, 220°C depending on the
experiment), at this moment, which indicates the starting time of the reaction, 7 mL of acetic
acid were injected through an injection tank, and the pressure was increased to 50 bar.
Magnetic stirring was kept at 750 rpm to maintain the catalyst suspended to avoid mass transfer
limitations. Reactions were run for 15 min, 30 min, 1h and 2h to establish the kinetics for each
of the temperatures and time. Once the time of the reaction was completed, the reactor was
quickly cool down to 10°C. Then, it was depressurized to ambient pressure and two liquid
8

aliquots were taken for gas chromatography (GC) analysis and quantification. An Agilent
Technologies 7890C GC equipped with autosampler and autoinjector was used for
quantification of the products and reactants. The column used for separating the constituents
was HP-5 (30mx0.32 mm, 0.25m). The mass balance for these experiments was typically
close to about 95 %.
Recycling experiments were conducted to study the catalyst stability in the ketonization
reaction, following the same procedure aforementioned. After each reaction cycle the reactor
effluent was filtered in a Nylon filter of 0.2 µm and dried overnight at 80°C. The dried catalyst
was weighted and placed back inside the reactor for the next reaction experiment. To make the
experiments comparable, the catalyst to feed ratio (m/v) was maintained constant (acetic
acid:catalyst 1:0.61).
3. Results and Discussion
3.1.Characterization results
3.1.1. X-rays microfluorescence (XRMF)
The catalyst samples were characterized by X-rays microflurescence (XRMF) to determine
the exact loading of ruthenium achieved after impregnation. The powder samples were placed
in the sample holder and introduced ¨as synthesized¨ inside the analysis chamber. In Table 1
are summarized the compositional results obtained for the three catalysts samples. The results
indicate that the loading of ruthenium varied from 3.9 to 6.9 wt. %. This value was close to the
calculated loading of 5 wt. % of Ru on TiO2. This information was employed to determine the
degree of exact degree of reduction of the metal and support in the temperature programmed
reduction experiments in section 3.1.5. While the real loadings, ranging from 3.9 to 6.9 wt. % of
Ru on TiO2, were employed in the calculation of the extent of reduction of the catalyst the
theoretical value of 5 wt. % metal loading was employed in the data presentation and
discussions for clarity and homogeneity.
9

3.1.2. High angle annular dark-field scanning transmission electron microscopy
(HAADF), High-resolution transmission electron microscopy and energy
dispersive X-ray spectroscopy (HR-TEM-EDS)
High-resolution transmission electron microscopy (TEM) characterization of the samples
revealed significant differences in geometry of TiO2 and size of Ru nanoparticles (see Figures 1
and 2 a-c). On 5 wt. % Ru/TiO2 rutile catalyst, highly uniform hexagonal particles of TiO2 were
observed with average particle size of 44 ± 11 nm, while anatase and P25 catalysts were
composed of spherical particles of 30 ± 8 nm and 33 ± 9 nm, respectively (Figure 1). Notably,
the particle size distribution of anatase and P25 were significantly narrower than the one
obtained for anatase.
Energy dispersive X-ray (EDX) of the samples allowed the identification of the Ru
nanoparticles (Figure 2 a-ii, b-ii, and c-ii). To properly determine the particle size the three
catalysts a significant number of images at low and high magnification ( 60 images) were
processed. First, the identification of the ruthenium particles was performed using EDX, then the
particles were sized (see Figure 2). On anatase, Ru nanoparticles had an average particle size
of 6.9 ± 3.7 nm with a wide distribution of particles (Table 2); while on P25 the distribution of Ru
particles size was narrower with an average particle size of 4.2 ± 1.6 nm. In sharp contrast, TiO2
rutile was able to stabilize smaller Ru nanoparticles of 3.1 ± 1.1 nm with a narrower distribution.
This enhancement of Ru dispersion on rutile can be explained in terms of metal-support
interactions.[43] The differences in crystallographic and electronic structure between rutile and
anatase are key to rationalize the particle size distribution. Initially, ruthenium clusters were
deposited using a wetness impregnation method under the same operating conditions. After the
10

impregnation the RuCl3 deposited on the samples is decomposed at high temperatures in the
presence of air. This process leads to the formation of RuO2 clusters with rutile structure on the
surface of the TiO2. Here, the interfacial lattice matching of RuO2 and TiO2 favors the formation
of Ru-O-Ti bonds reducing the mobility of Ru clusters during calcination at 400 °C.[43–48] While it
is possible that under reducing conditions at high temperature ruthenium particles undergo
particle sintering,[49] previously reported data has shown that small metal clusters, such as Pt,
can be effectively stabilized on surface defects or vacant sites of TiO2.[50] The authors showed
that on TiO2 P25 the Pt clusters preferentially grow on the rutile surface, creating an uneven
distribution of particles. Detailed ab-initio pseudopotential calculations for Pt and Pt2 on the
stoichiometric and reduced TiO2 surfaces and oxygen vacancies showed that in rutile the lower
formation energies of oxygen vacancies (Vo) and stronger interaction of Pt with Vo sites favored
the stabilization of a larger fraction of metal cluster with a smaller size on the rutile surface,
compared to anatase.[51] Hence, it is possible that on Ru/TiO2 rutile surface the stabilization of
smaller particles is due to a combination of the lattice matching of RuO2 and TiO2 rutile during
calcination and the metal-vacancy interaction during the reduction process. This will be in line
with the trend of dispersion observed, in which Ru cluster sizes were increasing monotonically
from rutile to P25 to anatase. Although Ru/TiO2 rutile is the catalyst with the smallest surface
area (23 m2/g), compared to Ru/TiO2 anatase (114 m2/g), its ruthenium dispersion is the highest
among the three catalyst samples.
The metal dispersion was estimated using a semi-empirical correlation of the particle size of
metal clusters obtain my HRTEM to dispersion published by Larson.[52] Previous work has
demonstrated of the validity of this approach.[53–56] Dalla and Taylor pioneering work in
correlating metal dispersion and H2 chemisorption showed that it is possible to determine the
metal particle size from the hydrogen chemisorption on reduced metals. For instance, X. Shen
et al.[55] reported good agreement between the measurements of metal dispersion of ruthenium
on TiO2 by hydrogen chemisorption and two indirect characterization techniques like X-ray
diffraction line broadening and Transmission Electron Microscopy. Similarly, J. Okal et al.[56]
observed that dispersion of ruthenium nanoparticles on alumina determined by hydrogen
chemisorption and TEM characterization methods were in good agreement, except in those
11

cases where impurities were present. Other methods, like the one developed by G. Bergeret
and P. Gallezot allow the estimation of metal dispersion based on particle size.[57] This method,
however, employs the atomic area of the topmost surface atoms, which is a parameter rather
difficult to determine. The technique herein employed is an improved version of the previously
reported alternative as it takes into consideration the possible deviations related to the
estimation of the atomic area of the surface atoms. Table 2 summarizes the results of the metal
dispersion. Notably, the catalyst supported on TiO2 rutile presented the highest Ru dispersion,
with a value of 31 %, while the anatase-supported catalyst presented the lowest value (14 %)
followed by P25 with 24 %.
3.1.3. X-ray diffraction (XRD)
X-ray diffraction characterization of the samples (Figure 3-a) showed high-degree of
crystallinity, evidenced by sharp intense diffraction peaks, corresponding to anatase (◊) and
rutile (•). In Ru/P25 catalyst, the fingerprints of both anatase and rutile were observed as this
material is a mixture of anatase:rutile 80:20 in weight. On Ru/TiO2 anatase the larger particle
size of RuO2 allowed the identification of RuO2 (*) fingerprints. The peaks observed at 28, 35
and 55 degrees can be indexed to the planes (110), (101) and (211) of tetragonal RuO2,
respectively (JCPDS 21-1172).[58] The absence of RuO2 on rutile and P25 based catalysts can
be attributed to the high dispersion of Ru clusters and overlaying of diffraction peaks from TiO2
rutile. Notably, the size of Ru clusters was independent of the support surface area.
3.1.4. N2-Physisorption
Textural characterization of the catalysts by N2-physisorption (Table 3) revealed that on Ru
supported TiO2 anatase the specific surface area was significantly larger (93 m2/g) than the one
observed on Ru supported on P25 (57 m2/g) and rutile (25 m2/g). The loss in surface upon
deposition of Ru clusters can be attributed to the structural rearrangement of the metal oxide
surface, which leads to the pore collapse.[59]
The Barrett-Joyner-Halenda (BJH) analysis of desorption isotherms of N2 (Figure 4 a-b)
indicated that all of the materials were primarily mesoporous. On anatase and P25 based
12

catalysts a narrow pore size distribution was observed with average pore size, determined by
the Brunauer, Emmett and Teller method (BET), of 143 and 159 Å, respectively (Table 3). On
rutile catalyst the pore size distribution was significantly wider with pores ranging from 20 to
1000 Å with an average pore size of 247 Å, which could be indicative of fractal-like macro- and
meso- porosity.[60–62]
3.1.5. Temperature programmed reduction (TPR)
Temperature programmed reduction was employed to determine the reducibility of the
different catalysts (Figure 5). The peaks were fitted using up to six Gaussian curves with the
idea of facilitate the analysis of the results. Notably, it is possible to identify a low-temperature
reduction peak (1) of Ru at 80-90 °C for the 5 % wt. Ru supported on TiO2 rutile. A series of
convoluted peaks were observed between 100°C and 250 °C for all the catalysts. On
Ru/anatase catalyst, a peak at 350 ºC appeared. Finally, a broad peak (2) was observed
starting at 450 °C and extending up to 650°C on Ru/rutile catalyst. It is interesting to note that
on 5 % wt. Ru supported on TiO2 P25 peak 1 and 2 were not observed. Also, on anatase
supported catalyst, peak 1 was not observed and peak 2 was slightly shifted to higher
temperatures.
The profiles obtained by temperature programmed reduction of the Ru/TiO2 catalysts
showed the presence of multiple reducible species. Previously reported data have shown
similar TPR profiles for Ru/TiO2/C.[20] In that case, however, the presence of high-surface area
provided by the activated carbon enabled the stabilization of Ru clusters that were not in contact
with TiO2. The reducibility of these clusters was strongly affected by the particle size of the
catalyst. For this reason, they attributed the reduction peaks at 121 and 183 °C to Ruthenium
with different particle sizes. The rest of the peaks between 200 and 500 °C were tentatively
assigned to reduction of titanium and ruthenium oxides in close interaction, which delayed the
13

reduction of Ru, but enhanced the reduction of TiO2. This could explain the presence of several
reduction peaks at temperatures below 500 °C.
Isothermal reduction experiments performed by J.E. Rekoske and M. A. Barteau[63]
demonstrated that the reduction rates and density of surface defects on TiO2 strongly depend
on the crystalline phase. That is on anatase the concentration unsaturated Ti cations is smaller
and the rate of reduction are slower than in rutile. These observations are in agreement with
Density Functional Theory (DFT) calculations that showed that on rutile the formation energies
of oxygen vacancies (Vo) is lower than in anatase.[64] In the case of Ru/TiO2 anatase, herein
reported, the TEM-EDX characterization revealed that in this support the particle size
distribution of Ru was significantly larger and wider than the one observed on polymorph P25
and rutile with values of 6.9 ± 3.7, 4.2 ± 1.6, and 3.1 ± 1.1 nm, respectively. Therefore, in the
particular case of Ru/TiO2 anatase it is possible that the TPR peak centered at 350 °C was
caused by a combination of different factors, including; larger particle size of ruthenium, lower
reducibility of anatase support, and weaker interaction between Ru-TiO2 surfaces. In the starting
material, i.e. TiO2 P25 (see Figure S3), the reduction profile is quite different from that observed
on Ru/TiO2 P25. At low temperatures the hydrogen consumption was negligible. A broad peak
between 550 and 800 °C was observed, which is in line with previously reported data for
TiO2.[65] This peak corresponded to reduction of the support at high-temperatures.
On Ru/TiO2 P25 the reduction peak was shifted to temperatures ranging from 100 to 300 °C
that are related to reduction of ruthenium and titanium oxides. The absence of low temperature
reduction peaks could be related to the strong influence that the reducibility of the support could
exert on the ruthenium nanoparticles as TiO2 anatase and polymorph P25 (80:20 anatase:rutile)
are less reducible than rutile.[63]
The extent of reduction of RuO2 and TiO2 oxides was determined using the moles of
hydrogen consumed on each experiment (see Table 4). The results indicate that in Ru/TiO2
anatase, P25, and rutile the amount of hydrogen consumed vary significantly with the support
with values of 527.1, 196.7, and 457.7 mol of H2/gof catalyst, respectively. While the hydrogen
consumption per mass for anatase was higher than rutile, the H2 uptakes in terms of surface
area followed the opposite trend. For instance, on Ru/TiO2 rutile the H2 consumed was 3.2
times larger than in anatase. Notably, the extent of reduction for ruthenium was complete for all
14

the catalysts, while in the case of TiO2 the degree of reduction increased from anatase, P25 to
rutile, with values of 20 %, 24 %, and 35 % reduction.
Similar observations have been reported by J.E. Rekoske and M. A. Barteau when studying
the rates of reduction of polymorph TiO2 anatase and rutile under isothermal conditions.[63] In
that case a difference of 2.5 time larger rates of surface reduction were obtained on rutile
compared to anatase. The differences were attributed to higher concentration of Ti cation on
low index planes of TiO2 rutile. DFT studies on the reducibility of TiO2 rutile and anatase crystals
showed that the less dense structure of anatase distortions the octahedron inducing a lift of
degeneracy of the lowest Ti levels favouring the formation of a low spin state. As a result, TiO2
anatase is less reducible than rutile.[64] Furthermore, the addition of H-H dissociation catalyst
enhances the reducibility of TiO2. J.E. Rekoske and M. A. Barteau demonstrated through kinetic
fitting that addition of ruthenium changed the rate limiting step that controlled the reduction
process. On bare TiO2 surface, i.e. without ruthenium, the rate limiting step was the interaction
between the surface O atom and the H atom adsorbed on the surface. In contrast, when
ruthenium was present the rate controlling step was the replacement of the surface O by
oxygen from the bulk. They proposed that the role of ruthenium was the homolytic dissociation
H-H that quickly spill-over the TiO2 surface. These observations were in line with DFT
calculations of H-H dissociation on Ru-TiO2 anatase. The results showed that in the presence of
sufficient hydrogen it is thermodynamically favourable to transfer hydrogen to the metal oxide
(TiO2 anatase) with the corresponding electron transfer to the empty states of the support
reducing the oxide.[66]
Furthermore, the strong interaction of Ru and TiO2 changes the electronics of the surface.
A recent X-Ray photoelectron spectroscopy (XPS) study of TiO2/C has shown the electronic
properties of TiO2 are significantly modified in the presence of Ru nanoparticles.[20] Upon
addition of Ru a shift to lower binding energies of approximately 1 eV was observed on the Ti
2p3/2 peak, suggesting that Ti cations have become electron enriched. This ‘‘partial reduction’’
was tentatively attributed to an electronic effect from Ru. Ex-situ reduction of the catalyst prior
XPS characterization further shifted the Ti 2p3/2 to even lower biding energies, which indicated
the reduction of Ti4+ to Ti3+.[49,67] These electronic effects are strongly affected by the catalyst
dispersion. G. Haller and D. E. Resasco demonstrated that it is possible to achieve a higher
15

degree of metal-support interaction by increasing metal dispersion.[68] They observed that on
reduced Rh/TiO2 and Ir/TiO2 catalysts prepared by ion-exchange technique, with metal
dispersions of nearly 100 %, the catalytic activity towards isomerization and dehydrogenation of
alkanes were unusually higher than those observed on catalysts supported on inert metal
oxides (e.g. SiO2) or catalysts prepared using conventional impregnation technique on TiO2. It is
possible to associate the higher concentration of partially unsaturated Ti3+ to a higher degree of
TiO2 reduction caused by the greater metal dispersion. The same observation has been
reported by electron paramagnetic resonance (EPR) and X-ray photoelectron spectroscopy
(XPS) when Pt and Ru supported on CeO2 and TiO2 were pre-reduced in H2.[22,63,69]
Therefore, the differences in H2 uptake per surface area and extent of reduction observed
between Ru/TiO2 rutile and anatase could be due to a combination of factors. One hand, the
higher reducibility of rutile TiO2 caused by the larger surface density of metal cations and anions
on ruitle low index planes that accelerates the rate of reduction and on the other the higher
dispersion of Ru clusters that increases the hydrogen spill-over and the degree of metal-support
interaction.
3.1.6. Ammonia temperature programmed desorption (NH3-TPD)
Temperature programmed desorption experiments of ammonia (NH3-TPD) was employed to
determine the concentration of acid sites in the catalyst. Figure 6 shows the results of TPD for
the three Ru/TiO2 catalysts in stacked format for clarity. The desorption of NH3 starts in the
three catalysts at temperatures ranging 80 to 100 ºC. While it is possible to use the desorption
temperatures of NH3 to infer the strength active sites the possible interference of other
phenomena hinders the utilization of this technique to obtain energetic parameters. The studies
reported by R. J. Gorte have shown that in fact when characterizing catalytic materials such as
high-surface area zeolites and metal oxides it is possible that mass transport limitations and
adsorbate-adsorbate interactions can lead to shifts in the temperature programmed desorption
profiles of probe molecules.[70,71] His studies demonstrated that it is not possible to obtain
energetic information form TPD techniques. Instead, he proposed that the only information that
can be gathered from TPD are the total concentration of active sites, surface reaction
16

mechanisms, and gas phase products. Similar conclusions were reported by S. Kouva et al.[72]
for TPD-NH3 performed on aluminosilicates. Here, it is important to keep in mind that these
conclusions were drawn from the analysis of TPD-NH3 data obtained from high-surface area
zeolites, where significant intra-particle diffusion limitations and adsorbate-adsorbate repulsion
interactions can lead to significant shifts in the desorption temperature.[72] However, on metal
oxides with macro- and meso- porosity, like the ones employed in the present study, the intra-
particle diffusional limitations are significantly reduced.[73] Therefore, the only possible cause of
any deviation in the desorption temperatures can be due to adsorbate-adsorbate interactions.
For this reason, it is rather difficult to obtain valid conclusions regarding the strength of the acid
sites present on the materials herein studied. Instead, the total acid site density has been
calculated (see Table 5).
As shown in Table 5 the total acid site concentration per mass of catalyst was significantly
larger on Ru/TiO2 P25 (551 mol NH3/g) and anatase (503 mol NH3/g), compared to rutile (312
mol NH3/g). This is due to the larger specific surface areas of TiO2 P25 and anatase that
provide a higher concentration of active sites in terms of catalyst mass. However, when the acid
site density was expressed in terms of surface area rutile and P25 showed the highest values,
respectively. These results are in line with the temperature programmed reduction results,
where rutile had the largest H2 uptake per surface area compared to anatase. This result will be
in line with temperature programmed reduction data. The higher metal dispersion of Ru clusters
and the more reducible nature of TiO2 rutile are responsible for the larger concentration of acid
sites and H2-uptake per surface area.
3.2. Ketonization results
3.2.1. Effect of reaction time and temperature
Acetic acid ketonization reactions were carried out over the three catalysts prepared, at
different reaction times and temperatures, in aqueous or organic reaction medium. As a
benchmark, we used n-hexane to measure the reaction kinetics at different temperatures to
17

avoid competitive adsorption between acetic acid and the solvent. Previous work published by
C. K. Lee et al.[74] showed that heat of adsorption of organic molecules on TiO2 nanotubes
materials strongly depends on the polarity of the probe molecule. The results indicated that the
heat of adsorption progressively increased with the polarity of the adsorbate. In a recent report
T. Pham et al.[22] showed that on Ru/TiO2/C water, CO2 and acetone compete with acetic acid
for the active sites. The heat of adsorption obtained by fitting a Langmuir-Hinshelwood kinetic
model for water, CO2 and acetone were -81.80, 109.73, and -87.54 KJ/mol, respectively. While
these values were significantly smaller than the one of acetic acid (-133.97 KJ/mol) these
products bound sufficiently stronger with the surface to compete with acetic acid for the active
sites.
The results of acetic acid conversion at different reaction times (up to 8 h) and different
temperatures (200, 210 and 220 ºC) in n-hexane are presented in Figure 7. At 200 ºC, for all
the catalysts the conversion initially increased rapidly up to 10 % for rutile catalyst (Figure 7a),
and 20 % for P25 and anatase catalysts (Figures 7b and c). However, after 2 h of reaction the
conversion reached a plateau for all the catalysts. It has been reported that in the ketonization
of acetic acid in vapor phase over Ru/TiO2, acetone and water competitively adsorb on the
catalyst surface,[24] which could explain the decrease in the reaction rate as the concentration of
water and acetone increases with reaction time. It is important to mentioned that competitive
adsorption effects could be accentuated when polar solvents are employed. These effects have
been reported previously by S. Mukherjee and M. A. Vannice,[75,76] in liquid phase hydrogenation
of citral on palladium supported on silica catalyst on solvent with different physical and
electronic properties.
Intrinsic catalytic activities of the different catalysts were attained employing the ketonization
data and acid site density. Assuming that each ammonium adsorbate binds to two acid sites
and assuming that the majority of the acidity of the catalyst between is caused by exposed Ti
cations on the surface that can act as active sites in the decarboxylative ketonization, it is
possible to determine the TOF of each catalyst.[77] Table 6 summarizes the values of TOF
18

calculated at short reaction times at 200, 210, and 220 ºC, for the three catalysts. Notably, the
TOF increased for anatase, P25, and rutile as function of reaction temperature.
This trend indicates that in the case of rutile the intrinsic kinetic rates for the acetic acid
ketonization are nearly two times faster compared to anatase. Although, there is a lower
concentration of acid sites per mass of catalyst on 5 wt. % Ru/TiO2 rutile the reaction rate per
site is always larger compared to the other two catalysts. To rationalize these differences in
intrinsic activity one should consider the possible electronic and structural effects of Ru on TiO2
rutile and anatase. Metal support interactions between Ru clusters an TiO2 could lead to
electronic modification on Ti and Ru.[20] For instance, combined experimental and theoretical
calculation studies of density of states (DOS), Hirshfeld charge analysis, and electron
deformation density (EDD) demonstrated that during the oxidation of CO the higher activity
observed on Ru-TiO2 was largely due to the lower activation energies. The results were
attributed to the electron charge transfer from the Ru to TiO2. This metal-support interaction was
mainly localized at the interface of Ru-TiO2 via the Ru-O bond. Similar results have been
reported by Bell et al.[78] were the specific activity and selectivity of the catalyst for the Fischer-
Tropsch reaction was dominated by the interface between Ru and TiO2. In those examples Ru
was directly involved in the reaction mechanism stabilizing surface reaction intermediates that
were kinetically relevant for the CO oxidation and Fischer-Tropsch reactions. However, in the
case of decarboxylative ketonization the partially unsaturated cations are believed to be the
active sites for the stabilization of the carboxylates intermediates that subsequently decompose
to form ketones, water and carbon dioxide. While the exact surface reaction mechanism is still a
matter of debate, it has been shown that the most plausible mechanism involves a bi-molecular
coupling of two carboxylate species to form a -ketoacid intermediate.[15,22] Thus, it is not
entirely clear if the interface Ru-TiO2 is the controlling parameter for the ketonization. Instead,
H-H spillover on the Ru-TiO2 is a more plausible explanation for such marked changes in
surface reducibility, acidity, and catalytic activity observed when Ru is added to TiO2.[15,20,24,63] In
a recent publication, G. Pacchioni[23] presented a comprehensive revision with a DFT
perspective of the ketonization of carboxylic acids over TiO2 and ZrO2 surfaces. In this revision it
19

was shown that the role of reducible metals supported on TiO2 and ZrO2 is still not well
understood. The simple H-spill-over is just one of the possible roles of the metal clusters on
TiO2, but many other possible interactions could be also present in the system that can be
controlling the surface reaction, including; defect migration and stabilization by metal clusters,
formation of oxide nanoclusters, or encapsulated core-shell structures, and interstitial
stabilization of metal atoms inside the metal oxide framework.[79,80] In this scenario, structural
effects can play a more decisive role. For instance, M. A. Barteau demonstrated employing
simply electrostatic theory analysis that on TiO2 rutile (110) and anatase (001) coordinatively
unsaturated Ti4+ cations are more reducible on TiO2 rutile than anatase. The results indicate that
on TiO2 rutile the intrinsic Lewis acidity of Ti+4 cations at the (110) is higher than that of the
metal cations in anatase (001) planes.[81] Also, the distances between Ti cations varies
substantially with the crystalline structure. For instance, the number of next-nearest Ti atoms on
anatase is 2, compared to rutile with 10 nearest atoms at Ti-Ti distances of 2.96 (2) and 3.57 Å
(10).[64] As a result, the distance of two surface Ti cations is reduced substantially. As a result, it
is possible to reduce the intermolecular distance between surface carboxylates. This is
particular important as recent kinetic data of acetic acid ketonization has shown that the rate
limiting step in this reaction is the surface coupling of two carboxylate species.[82] Thereby, one
can imagine that reducing the interatomic distances between surface cations could ease the
formation of the coupling product.
3.2.2. Aqueous vs. organic phase ketonization
In order to study the effect of the solvent on the reaction, experiments on aqueous phase
were perform at 220 ºC for 1 h to compare with the results obtained under the same conditions
with n-hexane as solvent.
To determine the stability of these catalysts in aqueous environment, the ratio of catalytic
activity per gram of catalyst for the ketonization of acetic acid was determined for the three
catalysts in n-hexane and water (Figure 8). Here, it was observed that on 5 wt. % Ru TiO2
anatase and P25 the ratio of activity in water: n-hexane was zero, while in the case of rutile this
value was 0.87. This indicates that on 5 wt. % Ru/TiO2 rutile the catalytic activity in water was
20

nearly the same as in n-hexane, while in anatase and P25 the activity was negligible. It has
been proposed that in the presence of condensed water at high temperatures fast deactivation
of metal oxides takes place due to the structural collapse.[27] On zeolites the presence of
defectives sites, typically OH-metal moieties, determines the stability of the crystalline structure
in hot liquid water.[34,35] In addition, the water molecules formed in the reaction compete with
acetic acid for the active sites of the catalyst, what could be more acute on more hydrophilic
surfaces.[83]
It has been reported that the absence of structural defects or hydroxyl functional groups
changes the surface wettability, making it more hydrophobic.[84,85] Therefore, one could use
the contact angle as an indirect measurement of the surface hydroxyl groups or defects on
metal oxides. To determine the contact angle of the different powders we employed the
technique developed by Teipel and Mikonsaari[85] based on Washburn´s equation.[86] The
contact angles resulting from the water-n-hexane penetration experiments allow the
determination of the average surface wettability. All the TiO2 samples were previously calcined
to 400ºC. Figures S2 and S3 present the mass ratio of water adsorbed per mass of sample as
function of time for TiO2 anatase, P25, and rutile. The results indicated that on TiO2 anatase it is
possible to adsorb a large amount of water quickly (~1.5 gH2O/gsample in 400 seconds), while on
TiO2 rutile the mass of water and rate of adsorption were significantly smaller (~0.4 gH2O/gsample
in 1766 seconds). As a result, the liquid-solid-gas contact angle of anatase and rutile were 51.0
and 89.5 degrees, respectively (see Figure 8). These values are in line with previous reports on
TiO2 freshly cleaved rutile (110).[36] Interestingly, on P25 the behavior was intermediate between
anatase and rutile. The mass absorbed was slightly higher than the one observed on anatase
(~1.7 gH2O/gsample in 1766 seconds), but the profile was significantly different. In this sample, it
was possible to identify two different regions of adsorption before saturation.
Considering that the contact angle is a function of the kinetic rate of water adsorption, which
in the case of P25 changes significantly with time, it is difficult to calculate a single contact
angle. As an alternative, we calculated the contact angle using the two sections of the curve. As
21

a result, we obtained a contact angle of 89.5 and 43 degrees for section 1 and 2 of the curve,
respectively. These two contact angles resembled that of anatase and rutile, indicating that on
P25 there are domains with high hydrophilicity and hydrophobicity. If one assumes that the
contact angle of the mixed surface behaves like a physical mixture of rutile and anatase one
can calculate the average contact angle as a linear combination of the two crystalline oxides at
a ratio 80:20 anatase:rutile. This results in a contact angle of 52.29 degrees. One can
immediately recognize that the hydrophilicity of anatase dominates explaining the similarities in
performance of 5 %wt. Ru on TiO2 supported on P25 and anatase in aqueous phase.
The low activity of the more hydrophilic catalysts (5 wt. % Ru TiO2 anatase and P25)
compared to rutile could be due to either strong competition of water molecules on more
hydrophilic TiO2 anatase and P25 or the structural collapse of the porous structure of the
catalyst. Previous studies from D. E. Resasco have shown that in zeolites the stability in
condensed water at high temperatures (>150 C) is controlled by the concentration of structural
defects on the lattice of the crystal.[31,32] These defects are typically OH- terminated, which
confers a hydrophilic character to these materials. Functionalization of these hydroxyl groups
with organosilanes stabilizes the defects preventing the hydrolytic cleavage of the Si-O-Si
bonds, while making the catalyst hydrophobic.[16] The hydrophobicity per se is not what prevent
the deactivation, but rather the neutralization of highly reactive defects on the zeolite structure.
While it is possible that the higher hydrophobicity of Ru/TiO2 rutile could indicate a lower
concentration of OH-groups or defects, at present there is no direct evidence to support it.
Furthermore, if one considers that in terms of surface area Ru/TiO2 rutile contains nearly 2
times more surface acid sites and 3 times higher surface hydrogen uptake, which are closely
related to partially unsaturated Ti cations or defects, then it is more difficult to rationalize the
stabilization solely based on the concept of density of surface defects or OH- groups.
Alternatively, it is possible that strongly bound water molecules to highly hydrophilic TiO2
anatase and P25, blocking the adsorption sites for the formation of surface acetates, which are
the precursors of ketonization products.[19,21,67,87,88] It has been showed that on TiO2
anatase water molecules dissociatively adsorb on partially unsaturated Ti cations. The results
showed that on surface oxygen vacancies H2O dissociates through the transfer of one proton to
a nearby oxygen atom, forming two hydroxyl groups for every vacancy.[28] Therefore, one could
22

expected that on a partially reduced surface, like the ones herein reported, with a high
concentration of unsaturated Ti cations, quick saturation of the surface by OH- groups should
take place upon exposure to condensed water. While having a higher density of surface
unsaturated Ti cations will most likely result in a higher hydrophilicity upon water exposure, it is
important to mention that this is a reversible process. For instance, detailed single-crystal
studies reported by R. Wang et al.[36] revealed that TiO2 wettability can reversibly switch from
hydrophobic to superhydrophillic upon exposure to UV-light. Initially, TiO2 freshly cleaved (rutile)
(110) showed a contact angle of 74 degrees and upon 30 min exposure to 40 mW/cm2 the
contact angle decreased to zero degrees. More importantly, on polycrystalline anatase this
switch was more easily achieved with light intensities as low as 0.1 mW/cm2 and a much shorter
time of UV illumination. The change in surface wettability was attributed to the formation of
surface hydroxyl groups on Ti3+ cations, generated by the UV-Light, creating hydrophilic
moieties that increased water-wettability. More recently, X. Feng et al.[37] reported that on TiO2
rutile nanorods films it is possible to achieve switchable superhydrophobicity as a result of the
combination of micro- and nanoscale hierarchical surface structures, the orientation of the
crystal planes, and the surface photosensitivity. In a recent publication K.K. Varanasi
demonstrated the electronic configuration of the metal cations at the outer layer could also
influence the Hydrophilic-Lipophilic balance (HLB) of the surface.[89] Here, it was reported that
in rare earth metal oxides, the electronic orbitals that can serve as H-bonding sites for water
molecules, i.e. the unfilled 4f orbitals are shielded from interactions with the surrounding
environment by the full octet of electrons in the 5s2p6 outer shell. In the alumina, however,
aluminium atoms are electron-deficient, with six electrons in their three sp2-hybrid orbitals,
which favour the formation of H-bonding with interfacial water molecules. If one compares the
electronic structure of Al3+ and Ti4+ one can immediately recognize that in both cases the sp2-
hybrid orbitals contain six electrons. Therefore, one should expect TiO2 to be inherently
hydrophilic. However, that is not the case for TiO2 rutile.
The exact nature of the stability of TiO2 rutile supported catalysts is still not entirely clear.
The experimental evidence indicates that surface hydrophobicity is critical for the stability in hot
liquid water, but the true nature of the phenomena requires additional studies. Elucidating the
23

exact mechanism of stabilization will be critical for the development of metal oxides with high-
stability for reactions in hot liquid water.
3.2.3. Catalyst recyclability
Finally, to demonstrate the enhanced stability of the 5 wt. % Ru supported on TiO2 rutile a
series of reaction with recycled catalyst was performed. Figure 9 presents the catalytic activity
after several reaction cycles employing water as reaction solvent. Notably, the catalytic activity
and conversion remained unaltered after three reaction cycles with values of 483, 463, and 450
µmol/g*s-1. The higher hydrothermal stability of the rutile supported catalyst could be tentatively
attributed to surface hydrophobic character.[32] This can be inferred from fact that rutile formation
requires the thermal treatment of TiO2 to temperatures above 600 °C,[90] where significant
surface dehydroxylation occurs.[91] Also, one could attribute the higher stability in hot
condensed water to the thermodynamic stability of TiO2 rutile compared to anatase.[92] In many
synthesis methods the initial crystalline TiO2 formed in generally anatase. This can be attributed
to the lower energetic barriers for the arrangement of short-range ordered TiO6 octahedra into
long-range anatase during synthesis.[93] Thus, it is possible that under the proper reaction
conditions the reversed synthesis process, i.e. dissolution and amorphization, could take place
more easily on anatase than rutile. However, at this point additional studies are required to
identify the controlling parameters of the enhanced stability of 5 wt. % Ru/TiO2 rutile.
4. Conclusions
Liquid phase acetic acid ketonization reactions have been studied on Ru catalysts
supported on three different TiO2 (anatase, rutile and a mixture of rutile: anatase). The
combination of the lattice matching effect between the rutile phase of ruthenium oxide and TiO2,
followed by thermal treatment, enhanced both catalytic activity and hydrothermal stability
24

without the need of any surface functionalization by organosilanes. This resulted in faster
reaction rates compared to the same catalysts based on P25 and anatase. The rutile
hydrophobicity improved the stability of the Ru-TiO2 rutile catalyst, resulting in negligible losses
in catalytic activity after several reaction cycles in aqueous phase. While the true nature of the
stability in hot liquid water remains elusive the impact of the surface Hydrophilic-Lipophilic
balance (HLB) is clearly. To properly describe the surface wettability of TiO2 and, more
importantly, the enhanced stability of Ru/TiO2 rutile in aqueous phase at high temperatures
further studies are required. Achieving this goal will require the study of all the parameters
influencing the surface wettability of TiO2 and its stability in hot condensed water, including;
porosity, crystalline structure, nano- and micro- scale architecture, electronic structure, surface
chemistry, as well as the possible blocking towards the active sites caused water competitive
adsorption. We envision that the concept of combining lattice matching effects and surface
hydrophobization on Ru-TiO2 could be extrapolated to other materials for applications as
diverse as high-temperature photocatalysis, thermochemical water splitting, commodity
chemical synthesis (e.g. HCl oxidation), and biomass conversion, where hydrothermal stability
of the catalyst is critical for the process economic viability.
Supporting Information
Detailed information of the experimental setup and the calculations of the contact angle is
available on the Supporting information file.
Acknowledgement
This work was funded by Abengoa Research and the European Union Seventh Framework
Programme (FP7/2007-2013) under grant agreement no. 604307. The authors would like to
thank Dr. J.C. Serrano, Dr. M. A. Díaz, and Dr. J. Arauzo their helpful discussions, Dr. G. Rincón
for the help with XRD and Contact Angle Measurements. We thank Dr. D. E. Resasco for helpful
discussions on liquid phase ketonization and hydrothermal stability.
25

References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
H. Röper, Starch‐Stärke 54 (2002) 89–99.
N. Tanaka, Technology Roadmap: Biofuels for Transport, Paris, 2011.
G.W. Huber, S. Iborra, A. Corma, Chem. Rev. 106 (2006) 4044–4098.
A. Corma Canos, S. Iborra, A. Velty, Chem. Rev. 107 (2007) 2411–2502.
D.E. Resasco, S. Crossley, AIChE J. 55 (2009) 1082–1089.
D.E. Resasco, S.P. Crossley, Catal. Today 257 (2015) 185–199.
J.C. Serrano-Ruiz, J. a. Dumesic, Energy Environ. Sci. 4 (2011) 83–99.
A. V. Bridgwater, Biomass and Bioenergy 38 (2012) 68–94.
M. Ringer, V. Putsche, J. Scahill, Large-Scale Pyrolysis Oil Production: A Technology
Assessment and Economic Analysis, Golden, 2006.
[10]
[11]
[12]
[13]
S. Xiu, A. Shahbazi, Renew. Sustain. Energy Rev. 16 (2012) 4406–4414.
D. Mohan, C.U. Pittman, P.H. Steele, Energy and Fuels 20 (2006) 848–889.
J. Piskorz, D.S. Scott, D. Radlein, ACS Symp. Ser. Am. Chem. Soc. (1988) 165–178.
J.H. Marsman, J. Wildschut, P. Evers, S. de Koning, H.J. Heeres, J. Chromatogr. A 1188
(2008) 17–25.
[14]
[15]
[16]
S. Czernik, A. V. Bridgwater, Energy & Fuels 18 (2004) 590–598.
T.N. Pham, T. Sooknoi, S.P. Crossley, D.E. Resasco, ACS Catal. 3 (2013) 2456–2473.
P.A. Zapata, J. Faria, M.P. Ruiz, R.E. Jentoft, D.E. Resasco, J. Am. Chem. Soc. 134
(2012) 8570–8578.
[17]
[18]
[19]
[20]
[21]
T.N. Pham, D. Shi, D.E. Resasco, Appl. Catal. B Environ. 145 (2014) 10–23.
K.S. Kim, M.A. Barteau, J. Catal. 125 (1990) 353–375.
K.S. Kim, M. a Barteau, Langmuir 6 (1990) 1485–1488.
T.N. Pham, D. Shi, T. Sooknoi, D.E. Resasco, J. Catal. 295 (2012) 169–178.
O. Nagashima, S. Sato, R. Takahashi, T. Sodesawa, J. Mol. Catal. A Chem. 227 (2005)
231–239.
[22]
[23]
[24]
[25]
T.N. Pham, D. Shi, D.E. Resasco, J. Catal. 314 (2014) 149–158.
G. Pacchioni, ACS Catal. 4 (2014) 2874.
T.N. Pham, D. Shi, D.E. Resasco, Top. Catal. 57 (2013) 706–714.
C.A. Gaertner, J.C. Serrano-Ruiz, D.J. Braden, J.A. Dumesic, J. Catal. 266 (2009) 71–
78.
[26]
[27]
R.W. Snell, B.H. Shanks, ACS Catal. 3 (2013) 783–789.
R.M. Ravenelle, F. Schübler, A. Damico, N. Danilina, J.A. Van Bokhoven, J.A. Lercher,
C.W. Jones, C. Sievers, J. Phys. Chem. C 114 (2010) 19582–19595.
[28]
[29]
R. Schaub, N. Lopez, I. Stensgaard, J.K. Norskov, F. Besenbacher, Phys. Rev. Lett. 87
(2001) 266104-1-266104–4.
Z. Wang, X. Hao, S. Gerhold, Z. Novotny, C. Franchini, E. McDermott, K. Schulte, M.
Schmid, U. Diebold, J. Phys. Chem. C 117 (2013) 26060–26069.
[30]
[31]
[32]
U. Diebold, Surf. Sci. Rep. 48 (2003) 53–229.
P.A. Zapata, Y. Huang, M.A. Gonzalez-Borja, D.E. Resasco, J. Catal. 308 (2013) 82–97.
L. Zhang, K. Chen, B. Chen, J.L. White, D.E. Resasco, J. Am. Chem. Soc. 137 (2015)
11810–11819.
[33]
Y. Román-Leshkov, M. Moliner, J.A. Labinger, M.E. Davis, Angew. Chemie - Int. Ed. 49
26

(2010) 8954–8957.
[34]
[35]
[36]
A. Corma, L.T. Nemeth, M. Renz, S. Valencia, Nature 412 (2001) 423–425.
R. Gounder, M.E. Davis, J. Catal. 308 (2013) 176–188.
R. Wang, N. Sakai, A. Fujishima, T. Watanabe, K. Hashimoto, J. Phys. Chem. B 103
(1999) 2188–2194.
[37]
[38]
X. Feng, J. Zhai, L. Jiang, Angew. Chemie - Int. Ed. 44 (2005) 5115–5118.
J.. Dielbold, A Review of the Chemical and Physical Mechanisms of the Storage Stability
of Fast Pyrolysis Bio-Oils, Golden, 1999.
[39]
H. Idriss, C. Diagne, J.P. Hindermann, A. Kiennemann, M.A. Barteau, J. Catal. 155
(1995) 219–237.
[40]
[41]
[42]
J.E. Rekoske, M.A. Barteau, Ind. Eng. Chem. Res. 50 (2011) 41–51.
M.A. Barteau, Chem. Rev. 96 (1996) 1413–1430.
A. Pulido, B. Oliver-Tomas, M. Renz, M. Boronat, A. Corma, ChemSusChem 6 (2013)
141–151.
[43]
E. V. Kondratenko, A.P. Amrute, M.-M. Pohl, N. Steinfeldt, C. Mondelli, J. Pérez-
Ramírez, Catal. Sci. Technol. 3 (2013) 2555–2558.
[44]
[45]
J. Zhang, P. Zhou, J. Liu, J. Yu, Phys. Chem. Chem. Phys. 16 (2014) 20382–20386.
J. Xu, X. Su, H. Duan, B. Hou, Q. Lin, X. Liu, X. Pan, G. Pei, H. Geng, Y. Huang, T.
Zhang, J. Catal. 333 (2016) 227–237.
[46]
[47]
C. Hernandez-Mejia, E.S. Gnanakumar, A. Olivos-Suarez, J. Gascon, H.F. Greer, W.
Zhou, G. Rothenberg, N. Raveendran Shiju, Catal. Sci. Technol. 6 (2015) 577–582.
D. Teschner, G. Novell-Leruth, R. Farra, A. Knop-Gericke, R. Schlögl, L. Szentmiklósi,
M. González Hevia, H. Soerijanto, R. Schomäcker, J. Pérez-Ramírez, N. López, Nat.
Chem. 4 (2012) 739–45.
[48]
[49]
[50]
[51]
T. Omotoso, S. Boonyasuwat, S.P. Crossley, Green Chem. 16 (2014) 645.
G.L. Haller, D.E. Resasco, Adv. Catal. 36 (1989) 173–235.
H. Iddir, M.M. Disko, S. Ogut, N.D. Browning, Micron 36 (2005) 233–241.
H. Iddir, V. Skavysh, Ö. Serdar, N.D. Browning, M.M. Disko, Phys. Rev. B 73 (2006) 1–
4.
[52]
[53]
[54]
[55]
M.I. Larsson, (12) United States Patent, 7,813,523 B1, 2010.
R.A.D. Betta, J. Catal. 34 (1974) 57–60.
K.C. Taylor, J. Catal. 38 (1975) 299–306.
X. Shen, L.J. Garces, Y. Ding, K. Laubernds, R.P. Zerger, M. Aindow, E.J. Neth, S.L.
Suib, Appl. Catal. A Gen. 335 (2008) 187–195.
[56]
[57]
J. Okal, M. Zawadzki, L. Kepiński, L. Krajczyk, W. Tylus, Appl. Catal. A Gen. 319 (2007)
202–209.
G. Bergeret, P. Gallezot, Handbook of Heterogeneous Catalysis, 2nd ed., Weinheim:
Wiley-VCH, 1997.
[58]
[59]
[60]
[61]
[62]
[63]
[64]
[65]
G. Borah, P. Sharma, Indian J. Chem. 50A (2011) 41–45.
B.C. Gates, Chem. Rev. 95 (1995) 511–522.
D. Avnir, D. Farin, P. Pfeifer, Nature 308 (1984) 261–263.
S. Gheorghiu, M.O. Coppens, AIChE J. 50 (2004) 812–820.
M.O. Coppens, G.F. Froment, Chem. Eng. Sci. 49 (1994) 4897–4907.
J.E. Rekoske, M.A. Barteau, J. Phys. Chem. B 101 (1997) 1113–1124.
A. Bouzoubaa, A. Markovits, M. Calatayud, C. Minot, Surf. Sci. 583 (2005) 107–117.
L.E. Briand, L. Gambaro, H. Thomas, J. Therm. Anal. 44 (1995) 803–821.
27

[66]
[67]
[68]
[69]
H.-Y.T. Chen, S. Tosoni, G. Pacchioni, ACS Catal. (2015) 5486–5495.
K.S. Kim, M.A. Barteau, J. Catal. 125 (1990) 353–375.
D.E. Resasco, G.L. Hailer, J. Phys. Chem. 288 (1984) 4552–4556.
S. Ricote, G. Jacobs, M. Milling, Y. Ji, P.M. Patterson, B.H. Davis, Appl. Catal. A Gen.
303 (2006) 35–47.
[70]
[71]
[72]
R.A. Demmin, R.J. Gorte, J. Catal. 90 (1984) 32–39.
R.J. Gorte, J. Catal. 28 (1996) 405–414.
S. Kouva, J. Kanervo, F. Schüßler, R. Olindo, J.A. Lercher, O. Krause, Chem. Eng. Sci.
89 (2013) 40–48.
[73]
[74]
M.O. Coppens, Colloids Surfaces A Physicochem. Eng. Asp. 187–188 (2001) 257–265.
C.K. Lee, S.K. Fen, H.P. Chao, S.S. Liu, F.C. Huang, Adsorpt. Int. Adsorpt. Soc. 18
(2012) 349–357.
[75]
[76]
[77]
S. Mukherjee, M.A. Vannice, J. Catal. 243 (2006) 108–130.
S. Mukherjee, M.A. Vannice, J. Catal. 243 (2006) 131–148.
J.M.G. Amores, V. Sanchez Escribano, G. Ramis, G. Busca, Appl. Catal. B Environ. 13
(1997) 45–58.
[78]
T. Komaya, A.T. Bell, Z. Wengsieh, R. Gronsky, F. Engelke, T.S. King, M. Pruski, J.
Catal. 150 (1994) 400–406.
[79]
[80]
E.W. McFarland, H. Metiu, Chem. Rev. 113 (2013) 4391–4427.
X. Shao, S. Prada, L. Giordano, G. Pacchioni, N. Nilius, H.J. Freund, Angew. Chemie -
Int. Ed. 50 (2011) 11525–11527.
[81]
[82]
[83]
J. Woning, R.A. Van Santen, Chem. Phys. Lett. 101 (1983) 541–547.
J.M. Vohs, Chem. Rev. 113 (2013) 4136–4163.
J. Struijk, M. D’Angremond, W.J.M.L. de Regt, J.J.F. Scholten, Appl. Catal. A, Gen. 83
(1992) 263–295.
[84]
[85]
[86]
[87]
[88]
J. Drzymala, Adv. Colloid Interface Sci. 50 (1994) 143–185.
U. Teipel, I. Mikonsaari, Part. Part. Syst. Charact. 21 (2004) 255–260.
E.W. Washburn, Phys. Rev. 17 (1921) 273–283.
R. Martinez, M.C. Huff, M.A. Barteau, J. Catal. 222 (2004) 404–409.
R. Pestman, R.M. Koster, J. a. Z. Pieterse, V. Ponec, a. van Duijne, J. Catal. 168 (1997)
255–264.
[89]
G. Azimi, R. Dhiman, H.-M. Kwon, A.T. Paxson, K.K. Varanasi, Nat. Mater. 12 (2013)
315–20.
[90]
[91]
[92]
[93]
P.S.O.F. Anatase, Am. Mineral. 53 (1969) 1929–1930.
D.N. Misra, Nat. Phys. Sci. 240 (1972) 14–15.
D.A.H. Hanaor, C.C. Sorrell, J. Mater. Sci. 46 (2011) 855–874.
A. Matthews, Am. Mineral. 61 (1976) 419–424.
28

a-i)
a-ii)
b-ii)
c-ii)
a-iii)
Dsitribution (%)
Cumulative (%)
100%
80%
60%
40%
20%
0%
10
20
30
40
50
60
>60
100 nm
10 nm
100 nm
50 nm
Particle size (nm)
b-i)
Dsitribution (%)
Cumulative (%)
b-iii)
100%
80%
60%
40%
20%
0%
10
20
30
40
50
60
>60
200 nm
Particle size (nm)
Dsitribution (%)
Cumulative (%)
c-i)
100%
80%
60%
40%
20%
0%
c-iii)
10
20
30
40
50
60
>60
500 nm
Particle size (nm)
Figure 1: Structural of 5 wt. % Ru supported on TiO2 rutile (a), P25 (b), and anatase (c)
obtained by scanning transmission electron microscopy (STEM) at two magnifications (i-ii).
Particle size distribution obtained from HRTEM images of TiO2 rutile (a-iii green), P25 (red-iii b),
and anatase (blue-ciii).
29