Titania surface chemistry and its influence on supported metal catalysts

Article abstract of DOI:10.1016/j.poly.2019.05.012

This work characterized the surface chemistry of a number of different titania samples including four commercial anatase samples, an anatase sample that we synthesized, the pyrogenic titania samples P25 and P90, and a commercial rutile sample. X-ray photoelectron spectroscopy (XPS), inductively coupled plasma-optical emission spectroscopy (ICP-OES), and Diffuse Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS), were used to identified surface species that might interfere with the acid/base properties of the surface hydroxyls. All commercial anatase samples were contaminated by sulfur, which diminished their effectiveness as metal oxide supports for heterogeneous catalysis and has implications for their utility as photocatalysts. Hydrogen-bonded surface hydroxyls remained after calcination up to 400 °C for all anatase samples, in contrast to rutile and the pyrogenic titania materials P25 and P90, in which they were eliminated. Ru(0) catalysts on titania without hydrogen-bonded surface hydroxyls showed enhanced C–O hydrogenolysis selectivity in the presence of water while Ru(0) catalysts on titania with hydrogen-bonded surface hydroxyls showed diminished selectivity in water, suggesting that surface hydrophilicity is important for this reaction. Heteroepitaxy between rutile RuO₂ and rutile TiO₂ is not essential for the creation of small evenly-spaced supported Ru(0) nanoparticles, which are important in many catalytic reactions.

Full text of DOI:10.1016/j.poly.2019.05.012

Accepted Manuscript
Titania surface chemistry and its influence on supported metal catalysts
Akbar Mahdavi-Shakib, Samra Husremovic, Sohee Ki, Jessica Glynn, Lauren
Babb, Janine Sempel, Ioannis Stavrinoudis, Juan Manuel Arce Ramos, Ryan
Nelson, Lars C. Grabow, Thomas J. Schwartz, Brian G. Frederick, Rachel
Narehood Austin
PII:
S0277-5387(19)30324-9
https://doi.org/10.1016/j.poly.2019.05.012
POLY 13937
DOI:
Reference:
To appear in:
Polyhedron
Received Date:
Accepted Date:
15 March 2019
7 May 2019
Please cite this article as: A. Mahdavi-Shakib, S. Husremovic, S. Ki, J. Glynn, L. Babb, J. Sempel, I. Stavrinoudis,
J.M.A. Ramos, R. Nelson, L.C. Grabow, T.J. Schwartz, B.G. Frederick, R.N. Austin, Titania surface chemistry and
its influence on supported metal catalysts, Polyhedron (2019), doi: https://doi.org/10.1016/j.poly.2019.05.012
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.

Titania surface chemistry and its influence on supported metal
catalysts.
2
1
1
1
1
Akbar Mahdavi-Shakib , Samra Husremovic , Sohee Ki, Jessica Glynn , Lauren Babb , Janine
1
1
3
4
3
Sempel , Ioannis Stavrinoudis , Juan Manuel Arce Ramos , Ryan Nelson , Lars C. Grabow ,
5
2
1
Thomas J. Schwartz , Brian G. Frederick , Rachel Narehood Austin
1. Department of Chemistry, Barnard College of Columbia University, 3009 Broadway, New
York New York 10027 USA rna2113@columbia.edu
2. Department of Chemistry, University of Maine, Orono ME 04469
3
. Department of Chemical and Biomolecular Engineering, University of Houston 4726
Calhoun Rd. Houston, TX 77204-4004
4
. Department of Chemistry, Bowdoin College, 255 Main Street Brunswick ME 04011
. Department of Chemical and Biomedical Engineering, University of Maine, Orono ME
5
04469
Abstract: This work characterized the surface chemistry of a number of different titania samples
including four commercial anatase samples, an anatase sample that we synthesized, the pyrogenic
titania samples P25 and P90, and a commercial rutile sample. X-ray photoelectron spectroscopy
(XPS), inductively coupled plasma-optical emission spectroscopy (ICP-OES), and Diffuse
Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS), were used to identify surface
species and surface contaminants that might interfere with the acid/base properties of the surface
hydroxyls. All commercial anatase samples were contaminated by sulfur, which diminished their
effectiveness as metal oxide supports for heterogeneous catalysis and has implications for their
utility as photocatalysts. Hydrogen-bonded surface hydroxyls remained after calcination up to 400
oC for all anatase samples, in contrast to rutile and the pyrogenic titania materials P25 and P90, in
1

which they were eliminated. Ru(0) catalysts on titania without hydrogen-bonded surface hydroxyls
showed enhanced C-O hydrogenolysis selectivity in the presence of water while Ru(0) catalysts on
titania with hydrogen-bonded surface hydroxyls showed diminished selectivity in water, suggesting
that surface hydrophilicity is important for this reaction. Heteroepitaxy between rutile RuO and
2
rutile TiO is not essential for the creation of small evenly-spaced supported Ru(0) nanoparticles,
2
which are important in many catalytic reactions.
Introduction:
Earth-abundant metal oxides play important roles in a wide variety of technologies.1-2 They
can serve as photocatalysts for organic chemical reactions, water splitting,4-5 or solar energy
3
6
conversion . They can be used for oxidation, dehydration, dehydrogenation, desulfurization,
oxychlorination, C-O bond activation, and isomerization reactions,7-8 as well as for sensor9-10 and
optoelectronic applications.11 These oxides are also commonly used as a support for reduced
metal catalysts.7, 12 Transition metal oxides doped into metal hydrides can store hydrogen,13 while
other metal oxides act as key interfaces for biocompatible materials.14 Mesoporous metal oxides
can provide shape selectivity to the reactions they catalyze.15 The surface chemistry of metal
oxides affects how they function in many of these roles.1, 7-8, 14 Increasingly, attention has focused
on proton-coupled electron transfer to and from metal oxides, requiring a chemical link to the
surface as redox chemistry occurs.16 Passivation of metals by metal oxides is often critical to their
function.17-18
TiO is one of the most studied metal oxides because of its wide range of applications in
2
19
20
21
22
photocatalysis , catalysis , solar cells , and biocompatible implants . The counter ion of the
titanium precursor, solvent, surfactant, and temperature all significantly impact the crystal structure,
2

electronic structure, and the surface chemistry of TiO .23 Therefore, detailed physico-chemical
2
characterization of these materials is necessary. Here, we have performed detailed physico-
chemical characterization of the most commonly used commercial TiO samples and explored
2
some of the effects of these characterized properties.
We have shown that metal oxide-supported noble metal catalysts can be coupled with fast
pyrolysis for the conversion of biomass to fuels,24-32 which helps address the some of the problems
associated with our reliance on fossil fuels.33-34 Thermal conversion of lignocellulosic biomass into
second generation biofuels using fast pyrolysis has been proposed24-25, 29, 35-36 and evaluated37-39 as
a potentially economically viable route for the sustainable production of liquid fuels.40-42 Our group
and others have found that Ru supported on TiO is a particularly effective catalyst for the
2
deoxygenation of phenol, a model compound which serves to simplify the complexity of bio-oils.43-
45
DFT studies of the reaction mechanism show that an amphoteric support is required for the
direct deoxygenation (DDO) of phenol to benzene, without hydrogenation of the ring.43 In the
proposed mechanism, the reaction is initiated by heterolytic dissociation of H at the metal support
2
interface. Heteroepitaxy between RuO , a common precursor in the synthesis of Ru/TiO catalysts,
2
2
and rutile TiO is often invoked as a reason for the high activity of Ru/TiO catalysts, although this
2
2
effect has not been fully explored.46-50 Our DFT calculations were performed on rutile (110) and
experiments were performed on the Ru/P25 TiO , which contains anatase and rutile nanoparticles,
2
leaving open questions about the role of crystal structure in catalysis.43
Given the importance of titania in a variety of catalytic applications, in this work we have
characterized the surface chemistry of a number of different titania samples including four
commercial anatase samples, an anatase sample that we synthesized, the pyrogenic titania
samples P25 and P90, and a commercial rutile sample. We used X-ray photoelectron
spectroscopy (XPS), inductively coupled plasma-optical emission spectroscopy (ICP-OES), and
3

Diffuse Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS) to thoroughly characterize
the metal oxide surfaces. DRIFTS measurements were also used to probe questions of surface
hydrophobicity/hydrophilicity. Transmission electron microscopy (TEM) was used to show that all
surfaces of titania can serve as nucleating agents for nanoparticulate Ru(0), thus ruling out an
essential role for heteroepitaxy in creating these hybrid materials. These studies test hypotheses
about why pyrogenic titania materials behave more like rutile than anatase despite being
predominantly anatase, and our observations point to implications for catalytic reactions that utilize
these materials.
Methods and Materials
Materials. Titanium dioxide P-25 (Evonik), Titanium dioxide P-90 (Evonik), TiO rutile (US
2
Research Nanomaterials), TiO2 anatase (Sigma Aldrich, denoted “SA”, US Research
Nanomaterials, denoted “USR”, Alfa Aesar, denoted “AA”, Nanografi, denoted “NG”) were all
purchased from their respective manufacturers. Composition was confirmed by XRD (Fig S1).
RuCl ·3H O, liquefied phenol, solid phenol, and diethyl ether were purchased from Sigma-Aldrich.
3
2
200 mesh Cu TEM grids and silicon nitride TEM grids were purchased from Ted Pella Inc.
(Redding, CA).
Support treatment. To decrease the sulfur content of commercial anatase samples, selected
anatase samples were calcined (5 h, 400 °C, O ), reduced (5 h, 400 °C, H ), and calcined (5 h, 400
2
2
°
C, O ) under flow conditions. These materials are indicated as “anatase_CRC” (anatase_calcined,
2
reduced, calcined).
Anatase synthesis. Anatase was synthesized following a published procedure.51 The resulting
material, denoted “BC” (indicating the material was synthesized at Barnard College), has a surface
2
area of 159 m /g, and a bimodal distribution of pores with maxima around 4 and 12 nm.
4

Catalyst Preparation. Catalysts were prepared on a 5 g scale with a metal loading of both 1 wt%
and 2 wt% Ru. The metal precursor (RuCl3 . 3H O, 130 or 260 mg) was dissolved in Milli Q grade
2
water (18 M) and added to the support until a paste-like consistency was obtained. The mixture
o
was dried overnight at 90 C and crushed with a mortar and pestle to uniform consistency.
Catalyst Reduction. Catalysts were reduced immediately prior to each reaction in a 25 mL Parr
4590 microreactor sealed with a flat graphite gasket controlled by a Parr 4848 controller operated
by SpecView. After sealing the catalyst in the reactor, the system was purged three times with Ar
gas and gradually heated to 573 K under flowing H gas. The reactor was kept under flowing H for
2
2
another 0.5 h at 573 K. Then, an additional 550 psig of H was added and the system was
2
maintained at that pressure and temperature for 1 h. The reactor was subsequently cooled to room
temperature prior to venting the system.
Metal Determination. The wt% metal loading of the catalysts was determined using a Thermo
Scientific iCAP 600 ICP-OES spectrometer with microwave-assisted digestion. Approximately 25 to
30 mg of catalyst was weighed and placed in a CEM EasyPrep microwave digestion vessel with 8
ml of a 3-to-1 mixture of HCl and HNO (trace metal grade). The vessels were sealed and placed in
3
a CEM MARS 6 microwave. The temperature was raised to 220 °C over two hours and then held
for 30 min. The samples were cooled to room temperature before being diluted to a total volume of
50 ml in MilliQ water. The digested samples were analyzed by ICP and the ruthenium
concentration was determined by comparing the emission to a calibration curve made using
ruthenium standards with concentrations of 0 ppm, 4 ppm, and 12 ppm. Independent analysis of
selected samples by Galbraith Laboratories confirmed these results.
Transmission Electron Microscopy (TEM). Conventional and high-resolution TEM was
performed at Columbia University Nanoinitiative (CNI) electron microscopy facilities using an FEI
Talos F200X transmission/scanning transmission microscope. The analyzed catalysts were mixed
5

with a minimal amount of 2-propanal and the mixture was suspended on either a lacey carbon film
on a Cu grid or a silicon nitride membrane grid with a membrane thickness of 50 nm, frame
thickness of 50 m and nine 0.1 mm x 0.1 mm windows. The operating voltage of the FEI Talos
was 200 kV. It was seen that prolonged exposure to the electron beam resulted in amorphization of
the particles, and, therefore, in the results presented here, care was taken to minimize the imaging
time. Additionally, it was found that the carbon layer in the holey carbon grids was not fully
amorphous, making it difficult to unambiguously distinguish the particles from the supporting
carbon layer. As a result, the silicon nitride membrane grids were used when there was a need for
obtaining the clearest images of the particles.
XPS and BET analysis. TiO samples were placed on carbon tape for XPS analysis. Samples
2
-8
were outgassed at room temperature and pressures < 10 torr for 3 h before the analyses. XPS
analyses were performed using a SPECS PHOIBUS 100 MCD plus analyzer at a pass energy of
1
00 eV with an Al K훼 X-ray source working at 240 W and 20 mA.
Nitrogen adsorption isotherms were measured at 77 K using a Micromeritics ASAP-2020
instrument. Samples were degassed at 350 °C for 4 h prior to the measurements. The pore size
distributions were calculated using the BJH method,52-53 and surface areas were calculated using
the BET method.54
FTIR, DRIFTS analysis. DRIFT spectra were recorded using a Nicolet 6700 Fourier Transform
Infrared (FTIR) spectrometer equipped with a Praying MantisTM accessory (Harrick Scientific
Production, IBC). KBr powder was placed in the sample cup inside a Harrick Scientific high
temperature reaction chamber (HVC). The sample cup in the HVC was placed on a temperature-
controlled sample stage equipped with a cartridge heater and a thermocouple. The sample cup
temperature was controlled by a Harrick Scientific Automatic Temperature controller (ATC-024-1).
6

The TiO samples were spread on top of the KBr powder in the sample cup for DRIFTS
2
studies. After the material was loaded in the chamber, the environmental chamber was heated
externally overnight with the sample maintained at 70 °C under an O flow (Grade 2, Matheson).
2
The O was dried with a 13X molecular sieve trap (4 to 8 mesh, Acros Organics) immersed in a dry
2
ice-acetone bath. Step-wise temperature-programmed calcination was performed in the
environmental chamber while several DRIFTS spectra were recorded at each temperature. All the
DRIFTS spectra were recorded under dry O flow by accumulation of 512 scans at 8 cm-1
2
resolution using a DTGS detector. Single beam spectra measured after drying KBr at elevated
temperatures were used for background correction of sample spectra at the corresponding
temperatures.
Catalytic Reactions with Phenol. Following procedures used in prior published work43, 55, 4 ml of
liquefied phenol (~10 wt% water, Sigma Aldrich) was introduced into a 25 mL Parr reactor with 120
mg of catalyst reduced immediately prior to the reaction. The system was purged three times with
Ar and pressurized to 5 psi under an Ar atmosphere. The temperature was then gradually
increased to 573 K while stirring at 650 RPM. The autogenic pressure in the reactor was 50 psi
after the set temperature was reached, at which point an additional 650 psi of H was added to the
2
system. Vapor-liquid equilibrium calculations indicate that most of the liquid in the reactor remains
at this condition (see Supporting Information). The reaction time after final pressurization was 15
min. The system was repressurized if a 50 psi drop in pressure was observed. After the reaction,
the system was rapidly cooled using an ice bath to prevent extended reaction times. Samples were
immediately analyzed using GC/MS. The Weisz-Prater criterion was estimated to be 0.035 for the
P25-supported catalyst (see Supporting Information), suggesting the reactions carried out here
were performed in the absence of mass transfer limitations. Reactions without water were done as
described above except 4.5 grams of solid phenol and 150 mg of catalyst were used. Control
7

experiments showed no reactivity of phenol in the absence of H under these conditions. As a
2
reference, reactions done with Ru/P25 with and without water yielded results identical to those
previously reported.43
Gas Chromatography Mass Spectrometry (GC/MS). Two L of the reaction mixture were diluted
in 1.5 mL of diethyl ether. Samples were analyzed using an Agilent 7820GC System with a 7963
series autoinjector and 5977E network mass selective detector with a HP-5MS cross-linked 5% PH
ME siloxane capillary column (dimensions of 30 m × 0.25 mm × 0.25 μm) using UHP He. The
method consisted of a 2.2 min solvent delay, and a 5 min hold at 318 K, followed by a ramp at 10
K/min to 498 K. Product concentrations were obtained with reference to an external calibration
curve. Selectivity was defined based on the relative amounts of products, according to Equation 1.
Equation 1 Definition of selectivity
CProduct,i
Selectivity 
100%
i CProduct,i
Results:
3.1 Surface chemistry determined by XPS
Pure anatase is commercially produced via the “sulfate method” while pyrogenic TiO is
2
produced from the gas-phase reaction of TiCl .23, 56. The presence of sulfate in commercial anatase
4
samples is mentioned by several authors,23, 57-58 but is most often neglected in the literature. In
addition, sulfate is a common additive to TiO to create Brønsted acidity59 and to increase its
2
photo-oxidation activity.60-61 Therefore, it is necessary to elucidate the influence of sulfate on the
surface chemistry of commercial anatase materials.
8

All of the TiO samples studied in this work have been analyzed using XPS and ICP. SA,
2
6+
2-
USR, AA, and NG anatase samples showed a S2p peak at 168.9 eV, characteristic of S , not S
62
(
Figure 1). Consistent with the XPS results, ICP analysis also showed the presence of sulfate for
SA, USR, AA, and NG anatase samples (Table 1). Since ICP is a bulk analysis method and XPS is
a surface analysis method, the agreement between the sulfur contents measured by XPS and ICP
indicates that the majority of the sulfate resides on the surface of the materials (see Section S.3 of
the supplemental material).
Anatase (SA)
(
A)
(B) 1.20x10
4
S"2p
Si"2s
As received
Pretreated
Anatase (SA)
1
1
1
.16
.12
4
1.5x10
Anatase (AA)
Anatase (USR)
.08
4
1
.20x10 Anatase (AA)
1.0
0.5
0.0
1
1
1
.18
.16
x 10
Anatase NG
Anatase (BC)
Rutile
.14
4
Anatase (USR)
1
.36x10
1
1
.34
.32
P90
1.30
1.28
170 165 160 155 150
1
70 165 160 155 150
BE / e.v.
BE / e.V.
9

Figure 1. XPS analysis of the TiO samples. A) XPS spectra of the as-received TiO samples in the
2
2
sulfur 2p and Si 2s region. The second (red) Anatase NG spectrum was multiplied by 10 due to the
low intensity of the S signal, which was only visible with additional scans. Spectra are offset for
clarity. Survey XPS scans are provided in the supplemental material and reveal Ti, O, and C peaks
(Figure S2). B) XPS analysis of commercial anatase samples before (black spectra) and after
400°C calcination-reduction-calcination pretreatment (red spectra). These results showed that
sulfate is eliminated after pretreatment at 400 °C.
The commercial anatase samples were subjected to a calcination/reduction/calcination
cycle (each at 400 °C for 5 hr) in an attempt to decrease their sulfur content. The sulfur content
decreased below the XPS (Figure 1B) and ICP-OES (Table 1) detection limits following this
treatment. Additionally, to compare data to a sulfate-free anatase sample, anatase was
synthesized at Barnard College from TiCl as the titanium source. This anatase, labeled “anatase
4
BC”, was free from sulfur as characterized by XPS (Figure 1A).
Table 1. Chemical and BET analysis of the TiO samples
2
Material
S-content by
S-content by
Surface area
a
2
-1
XPS (S
ICP-OES (ppm)
(m g )
-2
atoms nm )
Pyrogenic P90
P25
NDb
<500
<500
3660
<500
1361
100
55
NDb
Anatase
SA
1.5
97
SA_CRC
AA
NDb
NAc
136
0.5
1
0

AA_CRC
USR
NDb
0.5
<500
782
NAc
97
USR_CRC
NG
NDb
0.15
NDb
NDb
<500
1310
NAc
NAc
280
159
33
BC
Rutile
US research
<500
a
Information on how the surface sulfur coverage was determined is provided in the supplemental
information.
b
None Detected
c
Not Available
3
.2 DRIFTS characterization of TiO samples
2
FTIR spectroscopy provides a method for observing both surface sulfate species and
surface hydroxyl groups and detecting perturbations to these key surface moieties that originate
from the different TiO sources. In addition, spectra recorded at different temperatures can guide
2
the design of the pretreatment methods that remove sulfate.
DRIFTS spectra of P90 recorded during step-wise temperature-programmed calcination are
shown in Figure 2A. The DRIFTS spectrum of the sample recorded at 30 °C shows several bands,
indicating many chemical species are present on the surface at this temperature. The 1357, 1444,
-1
and 1547 cm bands are assigned to the carbonate/bicarbonate species that are formed upon
exposure of the surface to atmospheric CO .63 The 1624 cm band is assigned to the bending
-1
2
mode of physisorbed water. The broad absorbance feature in the 2500-3500 cm-1 region is
assigned to hydrogen-bonded water and surface hydroxyls.64-65 The bands in the 3600-3800 cm-1
region are from isolated surface hydroxyls and water molecules coordinated to surface Lewis acid
1
1

66
sites. DRIFTS analysis of P90 shows that water is removed at 200 °C, carbonates at 300 °C, and
isolated surface hydroxyls at 400 °C.
DRIFTS spectra of anatase AA are shown in Figure 2B. Similar to P90, water is eliminated
at 200 °C and carbonates are eliminated at 300 °C. There is an additional band in this spectrum at
1
311 cm-1 that continuously blue-shifts to 1358 cm-1 at 300 °C. This blue shift is indicative of
67
surface sulfate species. The sulfate band is eliminated after calcination at 400 °C. Similar spectral
behavior was observed for other commercial anatase samples (see Figure 2C for anatase NG and
Figures S3 and S4 for anatase USR and SA). Observation of the sulfate band in the spectra of
these samples is consistent with the XPS and ICP data. Interestingly, the broad hydrogen-bonded
-1
hydroxyl absorbance feature in the 2500-3600 cm range is not eliminated even after calcination at
400 °C. Anatase BC was synthesized following a sulfate-free method, and DRIFTS spectra of this
sample show that water and carbonate are removed upon calcination at 300 °C; however, isolated
and hydrogen-bonded surface hydroxyls are retained even after calcination at 400 °C (See Figure
S5).
1
2

1
3

Figure 2. DRIFTS spectra of TiO samples recorded during the calcination process from 30 °C to
2
400 °C. The band at 1180 cm^-1 in Figure 2C which gradually shifts to 1356 cm-1 upon
heating is assigned S=O stretching of the sulfate groups.67 The bands at 1357, 1444, and
-1
63, 68
-1
1
547 cm are assigned to either surface carbonates or bicarbonate.
The 1624 cm band is
assigned to the bending mode of water. The broad absorbance feature in the 2500-3600 cm-1
region contains the vibrational bands of water and H-bonded surface hydroxyls.64-65 The bands in
-1
the 2800-3000 cm region correspond to C-H stretching vibrations of hydrocarbon impurities. The
-1
complex absorbance features in the 3600-3800 cm region are due to the vibrational modes of
66
isolated hydroxyls. Spectra are offset for clarity.
Surface hydroxyls are postulated as active sites for the DDO of phenol over Ru/TiO , and
2
these species can be clearly observed in the spectra collected at 300 °C. Figure 3 shows three
-1
-1
-1
distinct bands at 3708 cm , 3673 cm , and 3630 cm for the P90 and anatase BC. Observation of
identical bands between anatase BC and P90 is consistent with P90 being 90% (w/w) anatase. In
parallel work, we have assigned the FTIR absorbances that correspond to isolated surface
hydroxyls on anatase nanopowders and used ab initio atomistic thermodynamics to define their
thermodynamic stability under reaction conditions.69 The 3708 cm band is assigned to a terminal
-1
-1
hydroxyl species on the anatase (001) surface and the 3673 cm band is assigned to a bridging
hydroxyl species on the anatase (101) surface.
We observed that the presence of surface sulfate species perturbs the surface hydroxyl
bands. Anatase SA, which has the highest sulfate content, shows a featureless absorbance in this
region. However, distinct features are observed for the other commercial anatase samples that all
have lower sulfur contents (Table 1). The vibrational frequencies of the surface hydroxyl groups on
the sulfated anatase are slightly different from those on the sulfate-free anatase; the 3708 cm-1
1
4

-1
-1
-1
band was red-shifted by approximately 2 cm while the 3673 cm band was red-shifted by 11 cm .
In addition to the frequency shift, the relative intensities also change in the presence of sulfate. The
exact interpretation of these observations is outside the scope of this paper; however, they suggest
that sulfate species may influence the acidity, basicity, and relative population of either the
individual surface hydroxyls or the exposed crystal facets.
OH A,(001)
tr
(
A)
(B)
USR
NG
AA
SA
Sulfate(band
P90
BC
0
.5
Anatase (SA)
Anatase (USR)
Anatase (NG)
0
.05
OH A,(101)
br
Anatase (AA)
Anatase (BC)
Rutile
P90
3
800 3750 3700 3650 3600
4
000 3500 3000 2500 2000 1500 1000
-1
-
1
Wavenumber / cm
Wavenumber / cm
Figure 3. DRIFTS spectra of TiO samples after overnight calcination at 300 °C. A) DRIFTS
2
spectra in the 1000-4000 cm-1 range, spectra are offset for clarity B) isolated surface hydroxyl
-1
bands and the corresponding assignment of the 3708 cm band to the hydrogen-bond receiving
-1
terminal hydroxyls of anatase (001) surface and the 3673 cm band to the bridging hydroxyls of
the anatase (101) surface.69
1
5

Another notable difference among the anatase samples was the broad absorbance feature
observed in the 2500-3500 cm-1 region. Figure 4 shows the DRIFTS spectra for all the TiO2
samples following overnight calcination at 400 °C. No bands were observed for the P90 sample,
indicating the surface is completely dehydroxylated. In contrast, a broad absorbance from 2500-
-1
3
500 cm was observed for all the anatase samples, indicating that these samples are still partially
hydroxylated and that these hydroxyl groups are in close enough proximity to participate in
hydrogen bonding, consistent with the observations of Hadjiivanov et al.23 Isotope exchange
reactions with D O demonstrated that these hydroxyls are accessible and therefore excluded
2
assignment to the OH stretch of trapped water molecules.70
1
6

400°C
Anatase (SA)
0.05
Anatase (USR)
Anatase (NG)
Anatase (AA)
Anatase (BC)
Rutile
P90
3
600
3200
2800
-1
Wavenumber / cm
-1
Figure 4. DRIFTS spectra of the P90, rutile, and anatase samples in the 2500-3800 cm region
after overnight calcination at 400°C. Spectra are offset for clarity.
1
7

3.3 TEM
Upon reduction in flowing hydrogen, Ru/TiO catalysts form small evenly spaced nanoparticles of
2
ruthenium, which are an essential component of selective HDO catalysts.47, 55, 71-72 There are a
number of questions about how nanoparticles form and what aspects of their structure are critical
for reactivity. RuO has a rutile crystal structure with lattice parameters well matched to those of
2
rutile TiO , which has spawned speculation that heteroepitaxy between RuO and rutile TiO is
2
2
2
important in nucleating Ru(0) nanoparticles.49-50, 73-75 In our prior work, catalysts that had been
calcined prior to reduction formed much larger, irregular ruthenium particles that preferentially
catalyzed HYD reactions, which we attributed to the favorability of HYD on ruthenium.55 In this
work, TEM images of reduced Ru/rutile, Ru/anatase, Ru/P25, and Ru/P90 all show the presence of
small Ru nanoparticles with an average particle diameter of 2 nm that did not vary significantly
among supports (Figure 5 and Table 2). Prior work using CO chemisorption55 to determine
ruthenium particle size led to results consistent with particle diameters determined by TEM.
1
8

Figure 5 TEM images of active catalysts. Ru(0) nanoparticles are visible as darker spots against
the lighter titania background. The white arrow points to one such Ru(0) nanoparticle. (A)
Pretreated Ru on P90. (B) Pretreated Ru on P25
Ru on anatase (Nanographi)
(C) Pretreated Ru on rutile (D) Pretreated
1
9

Support
Mean diameter (nm) and standard deviation
P90
1.7 +/- 0.5
1.8 +/- 0.5
1.7 +/- 0.6
2.0 +/- 0.5
P25
Anatase (SA and NG)
Rutile
Table 2: Mean diameter of ruthenium particles (nm) on TiO supports as obtained from an analysis
2
of TEM images. 200 particles from each material were measured.
3.4 Phenol deoxygenation reactions
We used the hydrodeoxygenation of phenol as a probe reaction to assess the catalytic
activity of these materials (see Figure 6 for the reaction network). Ruthenium supported on Sigma
Aldrich and Alfa Aesar anatase were not active for this reaction (phenol conversions less than 1%).
The as-received anatase material from US Research achieved 6.5% phenol conversion. When this
material was treated by calcining, reducing, and calcining to remove the sulfur, the phenol
conversion increased to 22% at the same contact time. Nanographi anatase, which has very low
2
sulfur content per surface area (0.14 S atoms per nm ), was active as received.
2
0

Figure 6 Reaction pathways for phenol in HDO catalysis. The top pathway, in which phenol is
converted directly to benzene, is the direct deoxygenation pathway (DDO). The bottom pathway, in
which phenol is sequentially hydrogenated, dehydrated, and hydrogenated, is the hydrogenation
pathway (HYD).
We have previously shown that the reactivity of Ru/TiO catalysts relies on a critical acid-
2
base interaction at the interface between small nanoparticles of Ru and the amphoteric TiO2
support (Figure 7).43 Hydrogen adsorbs on the small Ru(0) particles, after which a basic surface
hydroxyl facilitates H-H bond heterolysis. In a “rebound step”, the protonated surface hydroxyl (now
an acid), donates a proton to the phenolic oxygen, which weakens the C-O bond. C-O bond
cleavage and hydride transfer complete the reaction cycle. Water favors the DDO pathway by
increasing the surface concentration of protonated hydroxyls. Interestingly, the rutile-, P25-, and
P90-supported catalysts all show an increase in DDO in the presence of water, whereas no
increase was observed for the anatase-supported catalysts (Figure 8 and Table S1). This
observation correlates with the greater stability of the hydrogen bonded surface hydroxyls in
anatase as described above.
2
1

H
H
H
H
H
O
H
H
H
O
H
H
H
H
O
O
O
H
O
H
+
H O
2
H
H
H
Ru
Support
Ru
Support
Ru
Ru
O
O
Ti
Ti
Ti
Ti
Support
Support
Ti
Ti
Ti
Ti
Figure 7 Reaction mechanism for selective DDO catalyst (adapted from 43). Surface hydroxyls
facilitate heterolytic H-H bond cleavage and after being protonated serve as acids that donate a
proton to the phenolic oxygen, weakening the C-O bond. Water suppresses desorption of the
protonated hydroxyl, keeping the concentration of protonated surface hydroxyls relatively high.
Figure 8. The role of water in influencing catalyst selectivity for the formation of benzene. The
selectivity enhancement represents the percent change in the relative amount of benzene formed
through the direct deoxygenation pathway to the relative amounts of all other products formed
2
2

through the hydrogenation pathway with and without the presence of 10 wt% water in the reaction
for the same catalyst under comparable phenol conversions.
Discussion
In our earlier work, we demonstrated that small particles of Ru are required to achieve high
selectivity for DDO, suggesting the active site for direct deoxygenation is at the interface between
the particle and the support.55 This model is consistent with recent work by Crossley et al.
showing the rate of gas-phase DDO of cresol is proportional to the number of interfacial sites and
47, 71
specifically to the total perimeter length of the Ru-TiO interface.
We have also shown that
2
water is a co-catalyst and provided a mechanistic explanation, based on DFT calculations, for its
43
role. Our earlier studies were performed using TiO P25 as the support.
2
In this work, we observed that reactivity of the catalyst strongly depends on the support
used. P90 and rutile are analogous to P25 in terms of DDO selectivity and response to water. The
observation that all the commercial anatase materials we studied were contaminated by surface
sulfate species explains the diminished catalytic activity we saw for untreated materials. We
showed that sulfate impurities of commercial anatase samples could be removed after
pretreatment at 400 °C. However, catalysts made from all non-pyrogenic anatase samples
behave very differently from all other catalysts in terms of their DDO selectivity and response to
water. It is surprising that P90 and P25 behave more like rutile than pure anatase although they
are mostly anatase (P90 is 90% anatase and 10% rutile and P25 is 85% anatase and 15% rutile).
Understanding the difference between pyrogenic TiO and pure (non-pyrogenic) anatase TiO
2
2
was central to this study.
The rutile structure of RuO , which is a precursor in most Ru(0) nanoparticle syntheses, has
2
lead to the hypothesis that rutile TiO should be particularly good in supporting Ru(0) catalysts.49-50,
2
73-75
The results from Crossley et al., in which deoxygenation of m-cresol is insensitive to TiO2
2
3

crystal structure71 while deoxygenation of guaiacol is not, provide no clear-cut answer to whether
72
the structure of titania plays a role in DDO catalysts. In this work we show that these small Ru(0)
particles can form on any anatase crystal type, ruling out the hypothesis that there is some sort of
epitaxial matching between rutile RuO {110} and rutile TiO {110} that either stabilizes the
2
2
RuO phase or facilitates high dispersion in the activated catalyst.
2
The FTIR characterization of these materials points to potential differences in the
hydrophilicity of the titania surface, a phenomenon that is known to be important in certain
applications. The DRIFT spectra of the low-temperature-synthesized and commercial anatase
materials all show a broad hydrogen-bonded hydroxyl absorption that is not present in the spectra
o
-1
for pyrogenic TiO after calcination above 200 C. The absence of the 1620 cm water-bending
2
mode indicates that this band is not associated with chemisorbed or physisorbed water but is
instead due to hydroxyls that are present in heterogeneous bonding conformations that hydrogen
bond to nearby oxide oxygens. We have shown that these hydroxyls can be exchanged with D O,
2
demonstrating that they are accessible surface hydroxyls.70 Their persistence after calcination at
400 C indicates that their thermal stability is comparable or greater than the isolated hydroxyls,
69
66
which we and others predict to be thermodynamically stable under reaction conditions.
Hadjiivanov in his 1996 review paper mentions that anatase samples undergo irreversible
dehydroxylation upon when heated at up to 600 C.23 Pyrogenic TiO is prepared at higher
o
2
o
temperatures (between 1000 – 2400 C), and under dry conditions, which explains why this
material is less hydroxylated. Consequently, the low-temperature anatase surfaces should be
much more hydrophilic than the pyrogenic anatase surface.
Water has been implicated in the catalytic activity of a number of titania-based
heterogeneous catalysts. In the gas phase, Crossley and coworkers have shown that the DDO
reaction rate increases in the presence of water, although the effect is negated when the amount of
2
4

water increases substantially relative to the number of Ru perimeter sites. They propose that water
blocks active sites when present in quantities significantly in excess of the number of Ru perimeter
71
sites, thereby leading to this decrease. We suggest that this observation can also be rationalized
based on density functional theory results presented in our previous work.43 We showed that the
presence of a water molecule increases the activation barrier for proton transfer from the Ru
+
nanoparticle to the TiO surface by 0.58 eV because H O -like transition states are less stable than
2
3
+
H transition states. Correspondingly, the DDO pathway requires heterolytic hydrogen cleavage
across the Ru-TiO interface to create the protonated hydroxyl site, which can proceed through an
2
+
+
H -like or H O -like transition state. Similar conclusions regarding the detrimental effect of water on
3
heterolytic H activation across the Au/TiO interface have been drawn from kinetic measurements,
2
2
DFT calculations, and H-D exchange experiments.76
In the reactions carried out in the present study, the hydrophilic nature of the low-
temperature anatase surfaces may lead to locally-enhanced water concentrations in the vicinity of
the Ru perimeter sites, which in turn would lead to a higher barrier for regenerating the active site
for DDO and the correspondingly-observed reduction in DDO rates. Kinetic water uptake
measurements by Faria and coworkers indicate that the rate of water uptake on anatase is much
more rapid than that on rutile, confirming the more hydrophilic nature of anatase surfaces.74
Similarly, Stone and coworkers performed microcalorimetry measurements of water adsorption on
anatase and rutile powders that indicate water adsorbs more strongly on anatase surfaces (heat of
-1
reversible adsorption = -60 kJ mol ) than on rutile surfaces (heat of reversible adsorption = -44 kJ
-1 77
mol ) . Thus, it is reasonable to assume that the surface concentration of water will be higher for
anatase-supported catalysts than for rutile-supported catalysts, leading to depressed rates of
heterolytic hydrogen cleavage. Conversely, the stronger heat of adsorption of water on anatase
surfaces could lead to additional site-blocking by water for these systems. Both explanations are
2
5

consistent with our reactivity observations, and additional work is needed to discriminate between
them.
The unique reactivity of Ru supported on pyrogenic TiO could similarly be due to locally-
2
manipulated water concentrations. While pyrogenic TiO is mostly composed of anatase, it is
2
mixed at the nanoscale with rutile crystallites. The hydrophobic nature of these crystallites may be
sufficient to disrupt hydrogen-bonding networks of water distributed around the Ru nanoparticles,
thereby decreasing the local concentration of water in P25- and P90-supported materials and
leading to reactivity that is closer to rutile-supported catalysts than anatase-supported catalysts.
Notably, the kinetic water uptake measurements from Faria’s group show that the rates of water
uptake on P25 lie between those for anatase and rutile,74 supporting the notion that the nanoscale
incorporation of rutile domains should lead to different amounts of water present near the Ru
nanoparticles in our catalysts. While this hypothesis is consistent with the reactivity and
characterization data presented here, additional work is needed to test it. Ultimately water could
activate and/or deactivate either of the two competing pathways (HYD and DDO) that occur in
these catalyst reactions.78 Further studies to explore, in detail, the mechanistic role of water as a
function of surface chemistry and morphology are underway.
Conclusion
This work focused on characterizing the surfaces of titania and three important ways in
which they do and do not support catalytically-active nanoparticulate Ru(0)-based materials. We
examined the presence of surface sulfur on a number of different titania materials and the
influence this sulfur has on surface hydroxyls. We examined the presence of hydrogen bonded
surface hydroxyls on titania, which is indicative of surface hydrophobicity/hydrophilicity. We
studied whether the crystal structure impacts the ability of titania to support Ru(0) nanoparticles.
The results of these investigations have implications for our catalytic work, but also for the work of
2
6

many other research efforts that use titania for the range of uses outlined in the introduction.
In this work, we show that commercial anatase samples contain varying amounts of sulfur
on the surface and that this sulfur shifts hydroxyl stretching frequencies and diminishes catalytic
activity in the phenol conversion reaction studied. We also show that all low-temperature
synthesized anatase samples, regardless of sulfur content, contain thermally stable hydrogen-
bonded surface hydroxyls that appear to foster a more hydrophilic surface than pyrogenic titania or
rutile surfaces, in which these hydrogen-bonded surface hydroxyls do not form. Finally, all titania
materials studied can nucleate the small Ru(0) nanoparticles that are important in a number of
Ru(0)-based heterogeneous catalysts, ruling out an essential role for heteroepitaxy matching
between rutile RuO and rutile TiO in their formation. Additional questions, however, about the role
2
2
of electron transfer between the reducible metal oxide support and metal nanoparticle and the
78
influence of water on the kinetic paramters of DDO v. HYD reactions , are not addressed by the
characterizations provided here and will be the subject of further study.
Acknowledgements:
This work was supported by the U.S. National Science Foundation under award number NSF
1565843. SH was the recipient of two Hoffman grants from Bates College. We thank Simon
Billinge, Katayun Barmak, Soham Banerjee, Daniela Stuck, and Amirali Zangiabadi for helpful
discussions and Amiralia Zangiabadi for assistance with TEM imaging. We thank George
Bernhardt for assistance in measuring XPS data.
Declaration of Interest: none
2
7

1
.
Fierro, J. L. G., Preface. In Metal oxides: Chemistry and Applications, Fierro, J. L. G., Ed. Taylor and
Francis: 2006; p xi.
Wu, Y.; van Ree, T., 1 - Introduction: Energy technologies and their role in our life. In Metal Oxides
in Energy Technologies, Wu, Y., Ed. Elsevier: 2018; pp 1-16.
Stone, V. F. J.; Davis, R. J., Syntheis, characterization, and photocatalytic activity of titania and
niobia mesoporous molecular sieves. Chem. Mater. 1998, 10, 1468-1474.
Zong, X.; Li, C., 7 - Photocatalytic water splitting on metal oxide-based semiconductor
photocatalysts. In Metal Oxides in Heterogeneous Catalysis, Védrine, J. C., Ed. Elsevier: 2018; pp 355-399.
Wang, J.; van Ree, T.; Wu, Y.; Zhang, P.; Gao, L., 8 - Metal oxide semiconductors for solar water
splitting. In Metal Oxides in Energy Technologies, Wu, Y., Ed. Elsevier: 2018; pp 205-249.
Yuan, X.; Wu, X.; Wang, J.; Wu, Y.; van Ree, T., 12 - Solar-driven fuel production by metal-oxide
thermochemical cycles. In Metal Oxides in Energy Technologies, Wu, Y., Ed. Elsevier: 2018; pp 321-340.
Védrine, J. C., 1 - Fundamentals of heterogeneous catalysis. In Metal Oxides in Heterogeneous
Catalysis, Védrine, J. C., Ed. Elsevier: 2018; pp 1-41.
Goulas, K. A.; Mironenko, A. V.; Jenness, G. R.; Mazal, T.; Vlachos, D. G., Fundamentals of C–O bond
activation on metal oxide catalysts. Nature Catalysis 2019.
Wetchakun, K.; Samerjai, T.; Tamaekong, N.; Liewhiran, C.; Siriwong, C.; Kruefu, V.; Wisitsoraat, A.;
2
.
3
.
4
.
5
.
6
.
7
.
8
.
9
.
Tuantranont, A.; Phanichphant, S., Semiconducting metal oxides as sensors for environmentally hazardous
gases. Sensors and Actuators B: Chemical 2011, 160 (1), 580-591.
1
8
1
2
1
0.
Tiemann, M., Porous Metal Oxides as Gas Sensors. Chemistry – A European Journal 2007, 13 (30),
376-8388.
1.
016, 15, 383-396.
2.
Yu, X.; Marks, T. J.; Facchetti, A., Metal oxides for optoelectronic applications. Nature Materials
8 - Main industrial processes using metal oxides as catalysts or support and future trends in
heterogeneous catalysis. In Metal Oxides in Heterogeneous Catalysis, Védrine, J. C., Ed. Elsevier: 2018; pp
4
1
01-549.
3. Chen, B.; Liu, Z.; Li, C.; Zhu, Y.; Fu, L.; Wu, Y.; van Ree, T., 9 - Metal oxides for hydrogen storage. In
Metal Oxides in Energy Technologies, Wu, Y., Ed. Elsevier: 2018; pp 251-274.
4. Silva-Bermudez, P. r. n. R. M. r. n. R. S. E., The atomic ordering and its correlation to protein
adsorption on biocompatible metal oxide coatings. Frontiers in Bioengineering and Biotechnology.
5. Ren, Y.; Ma, Z.; Bruce, P. G., Ordered mesoporous metal oxides: synthesis and applications.
Chemical Society Reviews 2012, 41 (14), 4909-4927.
6. Damatov, D.; Laga, S. M.; Mader, E. A.; Peng, J.; Agarwal, R. G.; Mayer, J. M., Redox reactivity of
1
1
1
collodal nanoceria and use of optical spectra as an in situ monitor of Ce oxidation states. Inorganic
Chemistry 2018, 57, 14401-14408.
2
8

1
7.
Saji, V. S.; Choe, H. C.; Brantley, W. A., An electrochemical study on self-ordered nanoporous and
nanotubular oxide on Ti–35Nb–5Ta–7Zr alloy for biomedical applications. Acta Biomaterialia 2009, 5 (6),
2
1
303-2310.
8. Asserghine, A.; Filotás, D.; Nagy, L.; Nagy, G., Scanning electrochemical microscopy investigation of
the rate of formation of a passivating TiO2 layer on a Ti G4 dental implant. Electrochemistry
Communications 2017, 83, 33-35.
1
9.
Ohtani, B., design and Development of Active Titania and Related Photocatalysts. In Photocatalysis
and Water Purification: From Fundamentals to Recent Applications, Pichat, P., Ed. Wiley-VCH: 2013; pp 74-
1
2
02.
0.
Centi, G.; Perathoner, S., Nano-architecture and reactivity of Titania catalytic materials. Quasi-1D
nanostructures. Catalysis 2007, 20, 367-402.
1. Singh, R.; Dutta, S., A review on H2 production through photocatalytic reactions using TiO2/TiO2-
assisted catalysts. Fuel 2018, 220, 607-620.
2. Liu, K.; Cao, M.; Fujishima, A.; Jiang, L., Bio-Inspired Titanium Dioxide Materials with Special
Wettability and Their Applications. Chemical Reviews 2014, 114 (19), 10044-10094.
3. Hadjiivanov, K.; Klissurski, D. G., Surface Chemistry of Titania (Anatase) and Titania-supported
Catalysts. Chem. Soc. Rev. 1996, 107, 534-545.
4. Bridgwater, A. V., Renewable fuels and chemicals by thermal processing of biomass. Chemical
Engineering Journal 2003, 91, 87-102.
2
2
2
2
2
1
2
5.
815.
6.
Elliott, D. C., Historical Developments in Hydroprocessing Bio-oils. Energy & Fuels 2007, 21, 1792-
Bridgwater, A. V.; Cottam, M.-L., Opportunities for Biomass Pyrolysis Liquids Production and
Upgrading. Energy & Fuels 1992, 6 (2), 113-120.
7. Bridgwater, A. V.; Double, J. M., A strategic assessment of liquid fuels from biomass. In Research in
2
Thermochemical Biomass Conversion, Bridgwater, A. V.; Kuester, J. L., Eds. Elsevier Applied Science: New
York, 1988; pp 98-110.
2
8.
chemical families. Biomass Bioenergy 2007, 31, 222-242.
9. Huber, G. W.; Iborra, S.; Corma, A., Synthesis of Transportation Fuels from Biomass: Chemistry,
Catalysts, and Engineering. Chem. Rev. 2006, 106, 4044-4098.
0. Ingram, L.; Mohan, D.; Bricka, M.; Steele, P.; Strobel, D.; Crocker, D.; Mitchell, B.; Mohammad, J.;
Garcia-Perez, M.; Chaala, A.; Pakdel, H.; Kretschmer, D.; Roy, C., Characterizaton of bio-oils in
2
3
Cantrell, K.; Pittman, C. U., Pyrolysis of wood and bark in an auger reactor: Physical properties and chemical
analysis of the produced bio-oils. Energy & Fuels 2008, 22 (1), 614-625.
3
1.
NMR Spectroscopy. Energy & Fuels 2009, 23 (5), 2707-2718.
2. Oasmaa, A.; Meier, D., Characterisation, Analysis, Norms & Standards. In Fast Pyrolysis of Biomass:
A Handbook Bridgwater, A. V., Ed. CPL Press: Newbury, UK, 2005; Vol. 3, pp 19-59.
3. Nashawi, I. S.; Malallah, A.; Al-Bisharah, M., Forecasting World Crude Oil Production Using
Multicyclic Hubbert Model. Energy Fuels 2010, 24, 1788-1800.
4. Hemminger, J. C., Science for Energy Technology: Strengthening the link between Basic Research
and Industry. A report of a subcommittee to the Basic Energy Sciences Advisory Committee 2010.
5. Huber, G. W., Breaking the Chemical and Engineering Barriers to Lignocellulosic Biofuels: Next
Mullen, C. A.; Strahan, G. D.; Boateng, A. A., Characterization of Various Fast-Pyrolysis Bio-Oils by
3
3
3
3
Generation Hydrocarbon Biorefineries. National Science Foundation. Chemical, B., Environmental, and
Transport Systems Division, Ed. NSF: Washington, D.C., 2008.
2
9