Exploration of the preparation of Cu/TiO2 catalysts by deposition-precipitation with urea for selective hydrogenation of unsaturated hydrocarbons

Article abstract of DOI:10.1016/j.jcat.2016.05.011

Deposition-precipitation with urea (DPu) was used to deposit copper on TiO₂ and prepare metal Cu/TiO₂ catalysts with different Cu loadings from 0.6 to 4.2 wt%. Cu²⁺ complexes adsorb on TiO₂ at pH_TiO2(around 6), i.e., when the titania surface is supposed to be positively charged. To attempt to explain these results, a study on the deposition behavior of Cu²⁺ was carried out as a function of the pH of the preparation solution and the Cu loading. The samples were characterized by XRD, UV-visible, TPR and electron microscopy leading to a chemical mechanism of copper deposition: the presence of some negatively charged Ti-O sites on the TiO₂ surface even at low pH could be responsible for the Cu²⁺ complex adsorption, but it has not been possible to discard the occurrence of a surface reaction with OHs of titania. In fine, these adsorbed Cu²⁺ species formed Ti-O-Cu bonding, and were proposed to act as nucleation sites for the formation of Cu₂(OH)₃NO₃ particles onto the support. As DPu proceeded, these particles grew according to Ostwald ripening mechanism, and became unstable and transformed into CuO depending on the final Cu loading and DPu duration. As a result, the size of Cu⁰ nanoparticles obtained after reduction at 350 °C strongly depends on the Cu loading and the DPu time: very small Cu⁰ particles (<1 nm) in the low loaded samples (<1 wt% Cu) resulting from the presence of small particles of Cu₂(OH)₃NO₃ after preparation, moderate Cu⁰ particle size (～3 nm) in samples initially containing larger particles of Cu₂(OH)₃NO₃ and additional large particles (>10 nm) in samples containing CuO. The Cu catalysts exhibit interesting potential for selective hydrogenation of butadiene at relatively low temperature with good activity and high selectivity. The highest activity and stability were obtained with the samples containing small metal Cu particles, i.e., the samples exempt of CuO after preparation.

Full text of DOI:10.1016/j.jcat.2016.05.011

Journal of Catalysis 340 (2016) 95–106
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Exploration of the preparation of Cu/TiO2 catalysts
by deposition–precipitation with urea for selective
hydrogenation of unsaturated hydrocarbons
⇑
Zhao Wang, Dalil Brouri, Sandra Casale, Laurent Delannoy, Catherine Louis
Sorbonne Universités, UPMC Univ Paris 06, UMR CNRS 7197, Laboratoire de Réactivité de Surface, Tour 43-44, 3ème étage, Case 178, 4 Place Jussieu, F-75252 Paris, France
a r t i c l e i n f o
a b s t r a c t
Article history:
Deposition–precipitation with urea (DPu) was used to deposit copper on TiO2 and prepare metal Cu/TiO2
catalysts with different Cu loadings from 0.6 to 4.2 wt%. Cu2+ complexes adsorb on TiO2 at pH lower than
Received 24 February 2016
Revised 13 May 2016
Accepted 16 May 2016
PZCTiO (around 6), i.e., when the titania surface is supposed to be positively charged. To attempt to
2
explain these results, a study on the deposition behavior of Cu2+ was carried out as a function of the
pH of the preparation solution and the Cu loading. The samples were characterized by XRD, UV–visible,
TPR and electron microscopy leading to a chemical mechanism of copper deposition: the presence of
some negatively charged Ti–O sites on the TiO2 surface even at low pH could be responsible for the
Cu2+ complex adsorption, but it has not been possible to discard the occurrence of a surface reaction with
OHs of titania. In fine, these adsorbed Cu2+ species formed Ti–O–Cu bonding, and were proposed to act as
nucleation sites for the formation of Cu2(OH)3NO3 particles onto the support. As DPu proceeded, these
particles grew according to Ostwald ripening mechanism, and became unstable and transformed into
CuO depending on the final Cu loading and DPu duration. As a result, the size of Cu0 nanoparticles
obtained after reduction at 350 °C strongly depends on the Cu loading and the DPu time: very small
Cu0 particles (<1 nm) in the low loaded samples (<1 wt% Cu) resulting from the presence of small parti-
cles of Cu2(OH)3NO3 after preparation, moderate Cu0 particle size (ꢀ3 nm) in samples initially containing
larger particles of Cu2(OH)3NO3 and additional large particles (>10 nm) in samples containing CuO. The
Cu catalysts exhibit interesting potential for selective hydrogenation of butadiene at relatively low tem-
perature with good activity and high selectivity. The highest activity and stability were obtained with the
samples containing small metal Cu particles, i.e., the samples exempt of CuO after preparation.
Ó 2016 Elsevier Inc. All rights reserved.
Keywords:
Deposition–precipitation with urea
Cu/TiO2
Selective butadiene hydrogenation
>60 ꢁC
CO þ 2NHþ þ 2OHꢂ
ð1Þ
1. Introduction
COðNH Þ þ 2H O
ꢀꢀꢀꢀꢀꢀ!
2
2
2
2
4
A detailed study on the mechanism of deposition–precipitation
with urea (DPu) was carried out by Louis et al. [4,5] in the case of
the preparation of Ni/SiO2 catalysts. They found that electrostatic
interaction between positively charged Ni complexes and nega-
tively charged SiO2 surface was the initial driving force for the
adsorption of NiII complexes, followed by Si–O–Ni bond formation
that acted as nucleation sites for the precipitation of nickel hydrox-
ide and/or nickel phyllosilicate due to silica dissolution [5]. As a
result, the samples prepared by the DPu method showed smaller
metal particle size with narrower size distribution after reduction
than samples prepared by impregnation, and much higher metal
loadings than samples prepared by ion adsorption.
In 1977, a new method of preparation of supported catalysts
appeared in the academic literature, with the articles of Geus
et al. [1–3]. This method called deposition–precipitation is based
on the gradual and continuous increase of pH of the suspension
of metal precursor and oxide support resulting from controlled
addition of a base (e.g., sodium hydroxide, ammonia or urea). This
allows the control of the precipitation of the metal ions onto the
surface of the support and not in solution. Compared to NaOH
and ammonia, urea is more suitable to control precipitation
because the addition of urea before thermal decomposition above
60 °C allows homogeneous basification all over the solution
(Eq. (1)):
Copper deposition–precipitation with urea has also been stud-
ied on silica support by Geus at al. [1,2,6] and on alumina by
Kaushik et al. [7]. It also leads to the deposition of copper phyllosil-
icate (chrysocolla) and copper aluminate respectively, due to a
⇑
Corresponding author.
E-mail address: Catherine.louis@upmc.fr (C. Louis).
http://dx.doi.org/10.1016/j.jcat.2016.05.011
0021-9517/Ó 2016 Elsevier Inc. All rights reserved.

96
Z. Wang et al. / Journal of Catalysis 340 (2016) 95–106
mechanism of support dissolution–reprecipitation with the conse-
quence that reduction into metal particles occurs only at high tem-
perature. Copper DPu has also been recently performed on titania
in association with gold to prepare bimetallic Au–Cu particles
[8,9]. It was found that Cu (around 1 wt%) could be easily and
quantitatively deposited on titania in spite of the fact that as alu-
mina, titania exhibits a higher PZC (ꢀ6) than silica (ꢀ2) and that
the first step of deposition is probably also electrostatic interaction
between copper complex and titania surface. Moreover, deposited
copper could be easily reduced into small metal particles at mod-
erate temperature. However, no detailed study was performed to
investigate the mechanism of DPu of copper on this oxide support.
This is the goal of the present study.
solution during DPu, aliquots of solution containing the precipitate
arising from the solution of urea and Cu heated at 80 °C, were
withdrawn after 50 and 200 min of DPu. After centrifugation, it
was dried under vacuum but without washing because of the too
small amount gathered.
As a complement of the experiment above (Experiment I), a sec-
ond series of Cu/TiO2 samples were prepared with the same nom-
inal Cu loading (2.5 wt%) but with different DPu durations, 3, 5, 6, 7
and 20 h (Experiment II). The procedure of sample preparation was
the same as above, except the DPu time. Here, aliquots of 40–50 mL
of suspension were withdrawn after the different DPu durations
followed by centrifugation, washing and drying under vacuum at
room temperature.
From a catalytic point of view, copper is an interesting catalyst
for various types of reactions such as water–gas-shift reaction
[10,11], methanol synthesis [12–14], propene epoxidation [15–
17], alcohol oxidation [18,19] and in particular selective hydro-
genation of CO2 [20] and of furfural [21]. Some explorations have
been performed on the catalytic property of metallic Cu for selec-
tive hydrogenation of unsaturated hydrocarbons [22–25] but more
than 20 years ago.
The goal of this paper was therefore to investigate the genesis
of Cu/TiO2 samples with different Cu loadings prepared by
deposition–precipitation with urea, i.e., the evolution of both the
pH, the Cu loading and the nature of supported species, then the
consequence on the metal particle size after reduction and on
the catalytic properties in selective hydrogenation of butadiene
performed in excess of propene. This reaction can be considered
as a model reaction that mimics the challenging industrial process
of catalytic purification of alkene cuts from diene impurities, a
reaction already studied in our laboratory over other catalytic
systems.
Sample reduction was performed as under hydrogen
(100 mL minꢂ1
) from room temperature to 350 °C with a
2 °C minꢂ1 heating rate, and maintained at 350 °C for 2 h.
2.2. Sample characterization
The pH evolution of the mixture solution was recorded using a
pH probe (PHM201 Portable pH Meter). It was calibrated at pH = 4
and pH = 7 at 80 °C and the electrode sensitivity of the pH probe
lines in the range 97–100%.
The metal loadings of the samples were measured by X-ray flu-
orescence (XRF) with a spectrometer XEPOS HE (AMETEK). The
method used for the XRF analyses was based on calibration curves
obtained from standard samples for each element.
X-ray Diffraction (XRD) of as-prepared Cu/TiO2 samples was
performed on a diffractometer (D8 Bruker Company) operating
with Cu Ka radiation (1.5418 Å; 40 kV and 30 mA) with a Ni filter.
The 2 theta angle range was from 40° to 52°. The crystallite size
was calculated by Scherrer equation [26] taking into account
instrumental broadening correction. All diffractograms were stan-
dardized relatively to the (200) peak of anatase TiO2 at 2h = 48.1
since it does not interfere with any of the diffraction peaks of cop-
per compounds.
2. Experimental section
UV–visible spectroscopy in liquid (transmission mode) and
solid phase (diffuse reflectance mode) was used for the complete
detection of the Cu2+ species present in the supernatant and in
wet and as-prepared Cu/TiO2 samples, respectively. The spectra
were recorded on a Cary 5000 spectrometer in the range of
1200–200 nm at room temperature. The as-prepared sample pow-
ders were finely crushed and lightly pressed into the sample holder
of the diffuse reflectance cell. Titanium oxide was used as a refer-
ence in order to display the band specifically belonging to Cu
species.
2.1. Catalyst preparation
A series of Cu/TiO2 catalysts were prepared with different Cu
loadings, 1, 2, 2.5, 3 and 5 wt% Cu, by deposition–precipitation with
urea during the same duration of 20 h according to the following
procedure: 100 mL of distilled water and 1 g of support (TiO2 P25
Evonik, 80% anatase and 20% rutile) were introduced into a
double-wall reactor. The mixture was heated at 80 °C. Given
amounts of Cu(NO3)2ꢃ3H2O (Sigma–Aldrich) (0.077, 0.0968,
0.1168 and 0.1986 g) were dissolved into the TiO2 suspension solu-
tion (1.6–8.2 ꢄ 10ꢂ3 molCu Lꢂ1) in order to achieve the desired
nominal Cu loadings. After addition of a large excess of urea CO
(NH2)2 (Sigma–Aldrich) (5 g, i.e., [urea]/[Cu] P 100), the mixture
was stirred at 80 °C in the closed reactor for 20 h. The solid was
separated from the liquid phase by centrifugation and subse-
quently repeatedly washed with distilled water and centrifuged
(three times). Then the samples were dried under vacuum at room
temperature for 24 h (so-called as-prepared sample).
The evolution of the Cu loading on TiO2 as a function of time
and pH was studied during the preparation of the 2.5 wt%
Cu/TiO2 sample (using 3 g of TiO2 in 300 mL). Aliquots of
40–50 mL of suspension were withdrawn at different pHs, and
treated as above, i.e., centrifuged, washed and dried. The pH
evolution during Cu/TiO2 preparation was measured for four
samples (2, 2.5, 3 and 5 wt% Cu) every 5 min during the first 5 h
of DPu. As blank experiments, pH evolution of aqueous solutions
(100 mL) containing only urea (8.2 ꢄ 10ꢂ1 mol Lꢂ1) or urea and
copper (8.2 ꢄ 10ꢂ3 molCu Lꢂ1, i.e., equivalent to 5 wt% Cu on TiO2)
or urea and TiO2 (1 g) was also recorded during heating at 80 °C.
Moreover, to study the nature of the Cu precipitate formed in
Temperature programmed reduction (TPR) was performed in a
Micromeritics Autochem
P using 100 mg of as-prepared Cu/TiO2
samples. Ar purge (25 mL minꢂ1) was performed for 30 min to
remove the air in the reactor at room temperature (RT) before
increasing temperature from room temperature to 550 °C at a lin-
ear rate (7.5 °C minꢂ1
) in a 5% H2/Ar gas mixture flow
(25 mL minꢂ1).
Since metallic Cu particles are easily oxidized in contact with
air, and known to redisperse under oxygen [9], TEM measurements
were performed using a Gatan environmental cell holder enabling
to reduce Cu/TiO2 onto a Mo-grid at 350 °C and to transfer it into a
JEOL 2010 microscope transmission electron microscopy (TEM)
without contact with air. The reduction in the cell was performed
under hydrogen (100 mL minꢂ1), and the temperature program
was as follows: the samples were heated from RT to 350 °C in
30 min, maintained at 350 °C for 15 min, then cooled down to RT.
Statistical analysis of the Cu particle size distributions (PSD) in
reduced samples was obtained by counting more than 200 parti-
cles, using the software Image J. The average particle diameter
P
P
was deduced from the equation dm
¼
nidi= ni where ni is the

Z. Wang et al. / Journal of Catalysis 340 (2016) 95–106
97
Table 1
Final Cu loadings in the Cu/TiO2 samples after 20 h DPu and pH at which the color of the suspension changed in the first 5 h of DPu.
Nominal Cu loading (wt%)
Concentration of Cu2+
Actual Cu Loading
after DPu 20 h (wt%)
pH at which the color changes
White to green
(10ꢂ3 mol Lꢂ1
)
Green to gray
2
2.5
3
5
3.2
4.0
4.9
8.2
8.2
ꢀ1.6
2.1
2.5
4.2
–
4.1
3.9
3.8
3.7
–
–
–
5.5
5 (B2)
5.2 (green-muddy)
7.2 (green black)
number of the particles of diameter di. The detection limit of the
particles is ꢀ1 nm. Aberration-corrected electron microscopy was
used to identify small Cu particles by different techniques, HRTEM,
STEM high-angle annular dark-field (STEM-HAADF) and energy-
dispersive X-ray spectroscopy (EDS). The microscope was a JEOL
ARM 200F microscope, equipped with a cold field emission gun
and a CEOS aberration corrector of the objective lens, operated at
200 kV.
0.04
0.03
0.02
0.01
0.00
800
c
650
b
2.3. Butadiene hydrogenation reaction
The performance of Cu/TiO2 catalysts in selective butadiene
hydrogenation was tested on 100 mg of Cu/TiO2 catalyst (sieve
fraction, 125–200 lm) in a plug flow microreactor in Pyrex
a
(4 mm of internal diameter). Experiments were performed in tem-
perature programmed reaction from RT to 300 °C. The catalysts
were in situ activated under pure hydrogen with flow rate of
100 mL minꢂ1, from room temperature to 350 °C with a heating
rate of 2 °C minꢂ1 followed by a plateau at 350 °C for 2 h. Then,
the catalysts were cooled down in hydrogen flow to RT. The reac-
tion mixture (0.3% butadiene, 30% propane, 20% hydrogen and
49.7% helium) was introduced into the microreactor with a total
flow rate of 50 cm3 minꢂ1, which corresponded to a space velocity
of 30 L gꢂ1 hꢂ1 (GHSV = 20,000 hꢂ1). For the temperature pro-
grammed reactions, the catalysts were heated under this gas mix-
ture at a rate of 1 °C minꢂ1, from RT up to 300 °C. Gas analysis at
the outlet of the reactor was performed every 15 min, which
means that the gas composition was recorded every 15 °C. The
starting test temperature was 30 °C. A gas chromatograph (Peri-
chrom PR 2100, FID detector) with a 7.5 m column (1/8 in.), which
was filled with sebaconitrile 25% Chromosorb PAW 80/100 mesh),
was applied to analyze the gas at the outlet of the reactor. Reten-
tion times and sensitivity factors for the reactants and products
were calibrated using several gas mixtures. Software Winilab 4.0
was used for the data acquisition and extraction of the concentra-
tion of each gas composition. As mentioned in [27], preliminary
tests were made to attest that under the reaction conditions
described and at low conversion, internal and external diffusion
limitations could be neglected. Catalytic activities, expressed in
450 500 550 600 650 700 750 800 850
Wavelength / nm
Fig. 1. UV–visible spectra of the supernatant collected after the DPu preparation for
20 h of 2.5 wt% Cu/TiO2 (a), 5 wt% Cu/TiO2 (b), of a Cu(NO3)2 aqueous solution (c).
3. Results
The XRF analyses of the as-prepared Cu/TiO2 samples after 20 h
DPu with different Cu loadings show that around 80% of Cu was
deposited on the TiO2 support during the DPu process (Table 1).
The Cu2+ ions left in the very light blue supernatants of the 2.5
and 5 wt% Cu/TiO2 preparation were characterized by UV–visible
(Fig. 1). The spectra show a Cu2+ d–d band around 650 nm
(Fig. 1a and b), strongly shifted with respect to the one of the Cu
(H2O)26+ complex of a Cu(NO3)2 solution around 800 nm (Fig. 1c).
The band can be attributed to the d–d transition of a [Cu(NH3)4]2+
species according to [28]. The formation of this complex results
from the decomposition of urea into ammonia (Eq. (1)) and from
the fact that NH3 is a stronger ligand than H2O. Fig. 2 shows that
the as-prepared samples exhibit different colorations depending
on the Cu loadings: the solid samples with Cu loading lower than
1.6 wt% are green whereas those containing more Cu are gray.
l
mol of bꢂu1tadiene converted per second and per
g of Cu
(l
mol sꢂ1 gCu) were compared at 60 °C, i.e., at butadiene conver-
sion ꢀ20–30%.
Catalytic stability tests were also carried out in isothermal con-
dition at 105 °C using 100 mg of mechanical mixture of Cu/TiO2
(0.88, 1.55 and 2.5 wt%) and plain TiO2 support with the same grain
3.1. Cu deposition–precipitation process
size range (sieved fraction: 125–200
with the same amount of Cu (13.9
l
m) to operate experiments
mol): 100 mg of 0.88 wt%
3.1.1. Cu loading at different final pHs
1
l
To determine the evolution of the Cu loading during the DPu
preparation of the 2.5 wt% Cu/TiO2 sample, aliquots of the suspen-
sion were withdrawn at different pHs and times (see Section 2),
and the Cu loadings on TiO2 surface were measured by XRF after
sample washing and drying (Fig. 3). From Fig. 3, it appears that
most of the Cu is deposited on TiO2 below pH 6, which is the PZC
of TiO2, i.e., before 2.2 h of DPu. This was a surprising result since
Cu/TiO2, 56.8 mg of 1.55 wt% Cu/TiO2 and 43.2 mg of TiO2,
35.2 mg of 2.5 wt% Cu/TiO2 and 64.8 mg of TiO2. After activation
under H2 at 350 °C as above, the sample was cooled to 105 °C
under H2, the reaction mixture was introduced and the reaction
was conducted for 20 h. As above, gas analysis was performed
every 15 min.

98
Z. Wang et al. / Journal of Catalysis 340 (2016) 95–106
Fig. 2. Color of as-prepared Cu/TiO2 with different Cu loadings after 20 h of DPu.
change from green to gray after 7 h of DPu occurs while the Cu
loading had already reached a maximum.
DPu Time (h)
1.8
0.9 1.1
1.3 1.6
2.2
3
20
5 6 7
2.5
2.0
1.5
1.0
0.5
0.0
3.1.2. pH evolution of the mixture solution during deposition–
precipitation process
Since the key point in the DPu process is to gradually increase
the concentration of precipitating agent in the solution [30], the
pH evolution of the suspension was monitored as a function of
DPu time. Fig. 4 displays the pH evolution at 80 °C during the first
5 h of DPu for four different Cu concentrations corresponding to
final samples containing 2, 2.5, 3 and 5 wt% Cu.
Before analyzing the pH evolution during DPu, three blank
experiments also reported in Fig. 4 must be commented: the pH
evolution of a solution of urea/water (gray line, named B1), of a
solution of Cu2+/urea/water (violet line, named B2) and of a sus-
pension of urea/TiO2/water (black line, named B3). The concentra-
tions of Cu2+ and TiO2 are the same as those used for the
preparation of 5 wt% Cu/TiO2. Urea decomposition at 80 °C
(Eq. (1)) is attested by the increase of pH of the solution of urea
in water (curve B1) up to a final value close to 7.2. The delay in
the increase of pH between curves B3 and B1 results from the
buffering effect due to the presence of TiO2 [31]. The delay in
the increase of pH between curves B2 and B1 in the 4.5–6.0 pH
range attests that OHꢂ ions resulting from urea decomposition
are consumed and involved in a change of copper speciation. This
is also attested by the fact that in Experiment B2, the solution is
3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5
pH of DPu
Fig. 3. Evolution of the Cu loadings in the 2.5 wt% Cu/TiO2 sample as a function of
pH and DPu time. H: Experiment I, I: Experiment II.
at pH below the PZC, both the Cu complexes and the titania surface
are supposed to be positively charged and not supposed to interact,
and it is known at least for the Ni/SiO2 system also prepared by
DPu [5,29] that cationic Ni complex deposition starts at pH above
the PZC of silica when silica surface is negatively charged. It may
be also noted that the color of the suspension changed during
DPu; it was initially bluish at pH = 3.0, green from pH ꢀ 4.0 to 6,
i.e., between 1 and 2.2 h of DPu and finally gray-green at pH ꢀ 7
after 7 h of DPu (Table 2). Fig. 3 and Table 2 also show that all
the copper is already deposited within 3 h since the loading does
not vary any more for longer time, but that the sample color
8
80oC
7
80oC
6
4.5
4.0
5
3.5
3.0
Table 2
4
Evolution of the Cu loading and sample color as a function of the duration of DPu
preparation of a Cu/TiO2 sample with a nominal Cu content of 2.5 wt%. This table
gathers two sets of experiments (Experiments I and II; see Section 2).
1.0
1.1
1.2
Time / h
3
DPu time
(h)
Cu loading
(wt%)
Sample
color
Experiment
B1
B2
B3
2 wt% DPu
2.5wt% DPu
3 wt% DPu
5 wt% DPu
0.9
1.6
2.2
0.1
1.4
2.0
Bluish
Green
Green
Experiment I: no color change
Cu loading versus time/pH
2
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Time / h
3
5
6
7
2.1
2.2
2.1
2.1
2.1
Green
Green
Green
Gray
Experiment II: no Cu loading change
Sample color versus time
Fig. 4. Evolution of the pH as a function of time of the suspension during DPu (at
80 °C) (Cu/urea/TiO2 mixture) for different nominal Cu loadings: 2, 2.5, 3 and 5 wt%
Cu and of three aqueous solutions or suspension containing, 5 g urea (B1); 5 wt% Cu
and 5 g urea (B2) and 1 g TiO2 and 5 g urea (B3).
20
Gray

Z. Wang et al. / Journal of Catalysis 340 (2016) 95–106
99
initially clear blue turned opaque muddy green above pH = 5.2 (at
CuO
(a)
the inflexion point around 50 min), indicating copper precipitation,
then black-green at pH = 7.2, indicating a change in the composi-
tion of the precipitate. The UV–visible spectrum of the supernatant
gathered after precipitate centrifugation is completely flat (not
shown), which indicates that all the Cu ions initially present in
the solution have precipitated.
CuO+TiO2 mixture
4.24wt%
2.50wt%
2.08wt%
If one examines the pH evolution during DPu of the four differ-
ent Cu/urea/TiO2/water mixtures (i.e., 2, 2.5, 3 and 5 wt% Cu), one
can see in Fig. 4 that the pH curves initially behave the same
way as curve B3 (urea/TiO2/water), then exhibit differences in
the pH range 3.5–6. The differences of shape and time shift of
the pH curves increase as the copper concentration increases. Dur-
ing these experiments, a change in the color of the suspension from
whitish to green was observed at pH close to 4 as reported in
Table 1. These pHs slightly vary with the Cu concentration, and
coincide with the pH values of the inflexion points of the pH
curves. For the suspension corresponding to 5 wt% Cu DPu
(8.2 ꢄ 10ꢂ3 mol Lꢂ1) (pink curve), the inflexion point is much more
visible, and induces a marked delay in the further increase of pH.
To some extent, it compares with the shape of curve B2 (5 wt%
Cu2+/urea), but the presence of TiO2 induces a time delay before
pH starts increasing, and lowers the pH of the inflexion point from
ꢀ5.2 to ꢀ4.5; this will be discussed in Section 4. In addition, the
color of the 5 wt% Cu DPu suspension which was deep green from
pH 3.7, turned gray when pH reached 5.5; this later color change
was not observed with the other three samples, as they remained
green in the first 5 h of the DPu process. Note however that for
longer DP time (20 h), they also tend to turn grayish except for
the lower Cu loadings (Fig. 2).
1.62wt%
1.55wt%
0.88wt%
TiO2
32
34
36
38
40
42
44
46
48
50
52
2 theta
(b)
DCu / nm
18
16
14
--
Cu
: JCPDS: 01-070-3038
DPu time
20h
7h
6h
3h
40 41 42 43 44 45 46 47 48 49 50 51 52
2 theta
DPu 50min
3.2. Characterization of the Cu/TiO2 samples prepared by DPu
(c)
DPu 200min
As mentioned above, the color of the as-prepared samples
changes from green to gray depending on the Cu loadings
(Fig. 2); this indicates the presence of at least two different Cu spe-
cies on the TiO2 surface. Attempts to identify the nature of the Cu
species on TiO2 were performed through their characterization by
different techniques.
3.2.1. XRD
XRD was performed first on the as-prepared samples (Fig. 5a)
prepared for 20 h and containing different Cu loadings. In addition
to the peaks of anatase-TiO2 (JCPDS #: 01-084-1286) and rutile-
TiO2 (JCPDS #: 01-086-0147), a weak peak at 2h = 35.6° is visible
for the Cu loadings higher than 1.62 wt%, i.e., for the gray samples,
and its intensity increases with the Cu loading. This XRD peak was
also observed at the same angle (35.6°) with the mechanical mix-
ture of CuO in TiO2 (1.5 wt% Cu, gray color) (Fig. 5a). This indicates
that the gray color of the Cu/TiO2 samples containing more than
1.6 wt% Cu results from the presence of CuO (confirmed by com-
parison with the JCPDS file of CuO, 03-065-2309), which is known
to be black. These results also indicate that the Cu species respon-
sible for the green color of the sample in the low loaded samples is
not detected by XRD because it is either present in too low amount
or is amorphous. The average CuO particle sizes deduced from the
Scherrer equation in the samples containing 1.6, 2.1, 2.4 and 4.2 wt
% Cu are 10, 12, 13 and 15 nm, respectively.
10 15 20 25 30 35 40 45 50 55 60 65 70 75 80
2 theta
Fig. 5. XRD patterns of (a) as-prepared Cu/TiO2 with different Cu loadings after 20 h
of DPu. The blue dot line corresponds to the XRD pattern of a physical mixture of
CuO and TiO2 with ꢀ1.5 wt% Cu; . stands for rutile TiO2 (JCPDS: 01-084-1286); O
stands for anatase TiO2 (JCPDS: 01-086-0147); CuO peak position from JCPDS: 03-
065-2309; (b) XRD of the reduced samples with the same Cu loading (2.1 wt%) but
different DPu preparation times; (c) unwashed precipitate collected from a solution
containing urea and 5 wt% Cu after 50 min and 200 min at 80 °C; ꢄ: urea (JCPDS #
01-089-2835); H: Cu(OH)3NO3 (JCPDS # 01-075-0127); I: Cu(OH)3NO3 (JCPDS #
00-014-0687); r: Cu(CO3)(OH)3 (JCPDS # 00-004-0309); h: CuO (JCPDS # 00-048-
1545).
increases with DPu time although the Cu loading has reached its
maximum (ꢀ2.1 wt% Cu) in these samples.
The series of Cu/TiO2 samples prepared with a Cu loading of
2.1 wt% also showed color changes as a function of the DPu dura-
tion (Table 2). After sample reduction at 350 °C, the particle size
of metallic Cu in these samples was determined by XRD (Fig. 5b).
They show that the Cu0 particle size increases with the DPu time,
no matter the sample color. This means that the size of the parti-
cles of the Cu compound before reduction also evolves and
In order to identify the green Cu species, two precipitates were
collected from the B2 solution (Cu2+/urea) after precipitation for
50 min (green color) and after aging up to 200 min (black-green
color). They were analyzed by XRD after drying but without wash-
ing because of the too small amount (see Section 2). Apart from the
peaks of residual urea (noted ꢄ), all the peaks visible in the diffrac-
tion pattern of the precipitate prepared for 200 min (Fig. 5c, bot-

100
Z. Wang et al. / Journal of Catalysis 340 (2016) 95–106
1.0
0.8
0.6
0.4
0.2
0.0
0.60wt%
(a)
0.88wt%
1.55wt%
1.62wt%
2.08wt%
2.50wt%
4.24wt%
Cu2(OH)3NO3 + TiO2
CuO + TiO2
400 500 600 700 800 900 1000 1100 1200
Wavelength / nm
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0.6
(b)
(c)
0.5
DPu 7h
0.4
DPu 7h
DPu 6h
0.3
DPu 6h
DPu 3h
0.2
DPu 3h
0.1
0.0
wet TiO2 reference
dry TiO2 reference
400 500 600 700 800 900 1000 1100 1200
Wavelength / nm
400 500 600 700 800 900 1000 1100 1200
Wavelength / nm
Fig. 6. UV–visible spectra of (a) the as-prepared Cu/TiO2 samples with different Cu loadings after 20 h of DPu and physical mixtures of CuO and TiO2 and of Cu2(OH)3NO3 and
TiO2, both containing ꢀ1.5 wt% Cu; (b) the wet Cu/TiO2 samples containing (2.1 wt% Cu) but prepared during different durations; and (c) the same samples after drying at RT.
tom) can be attributed to CuO. The diffractogram of the precipitate
prepared for 50 min is more complex (Fig. 5c top). Most of the
peaks other than those of urea can be attributed to Cu2(OH)3(NO3)
with a monoclinic structure and with an orthorhombic structure.
One single peak at 2 theta ꢀ20° remains unassigned (noted r in
Fig. 5c top), and could correspond to the most intense peak of
malachite-type hydroxycarbonate, Cu2(CO3)(OH)2. As a comple-
ment of characterization of the green copper compound, the DRIFT
spectra of DPU Cu/TiO2 (50 min, ꢀ0.3 wt% Cu) and the green pre-
cipitate (50 min) were recorded. The two spectra (Fig. S1 in supple-
mentary material) show the same bands at 1430 and 1340 cmꢂ1
in the precipitate is copper hydroxynitrate, and it is unstable and
decomposes into CuO when pH increases at 80 °C; the low stability
of copper hydroxynitrate was also mentioned by Pereira et al. [32].
Hence, it can be proposed that the green color of the Cu/TiO2
samples prepared by DPu for 20 h with Cu loadings lower than
1.6 wt% Cu (Fig. 2) or prepared for 3–6 h for Cu/TiO2 (2.1 wt% Cu)
(Table 2) originates from the formation of Cu2(OH)3NO3 species
on TiO2. The interaction with the titania support is attested by
the fact that the Cu particles on titania whether they are either
Cu2(OH)3NO3 or CuO (620 nm) are much smaller than those of
the green precipitate (0.5–1 lm). It is also interesting to note that
that can be attributed to the
m4 and
m
2 ONOꢂ2 stretching vibrations
the interaction with titania stabilizes Cu2(OH)3NO3 and prevents
its decomposition into CuO after 20 h of DPu at 80 °C, i.e., at pH
around 7, since the samples with Cu loadings lower than 1.6 wt%
remain green. In contrast, the gray color of the samples with Cu
loadings higher than 1.6 wt% results from the presence of CuO aris-
ing from the partial or total decomposition of supported
Cu2(OH)3NO3.
of copper hydroxynitrate according to Pereira et al. [32]. These
bands cannot arise from malachite-type hydroxycarbonate,
Cu2(CO3)(OH)2 since COꢂ3 stretching vibrations must appear at 1506
and 1388 cmꢂ1 according to Kondrat et al. [33]. The green precip-
itate was analyzed by TEM and shows large particles between 0.5
and 1 lm (Fig. S2). Our conclusion is that the green Cu compound

Z. Wang et al. / Journal of Catalysis 340 (2016) 95–106
101
loaded Cu/TiO2, and suggest that these samples contain both cop-
per oxide and copper hydroxynitrate.
3.2.3. TPR
The reduction behavior of the as-prepared Cu/TiO2 samples
(20 h of DPu) was studied by TPR (Fig. 7). TiO2 does not show
any reduction peaks in the temperature range from room temper-
ature to 550 °C; thus, the two reduction peaks at ꢀ150 and ꢀ175 °C
correspond to the reduction of Cu species on TiO2. The intensity of
the first reduction peak (ꢀ150 °C) seems to be independent of the
Cu loading whereas the second one, initially around 175 °C, shifts
to higher temperature and increases in intensity with the Cu load-
ing. It becomes higher than the first peak when the Cu loading
exceeds 1.6 wt%. Bulk CuO is known to reduce directly from Cu2+
to Cu0 because of the lower energy requirement for metallic Cu for-
mation from Cu2+ than from Cu+ [43] (Eqs. (2) and (3)):
4,24wt%
2,50wt%
2,08wt%
1,62wt%
1,55wt%
0,88wt%
0,60wt%
CuOðSÞ þ H2 ! Cu0ðSÞ þ H2O
Cu2OðSÞ þ H2 ! 2Cu0ðSÞ þ H2O
D
H ¼ 14:5 kcal=mol
H ¼ 27:4 kcal=mol
ð2Þ
ð3Þ
100
200
300
400
500
D
Temperature / oC
The two TPR peaks should thus correspond to the reduction of
two different CuII species. Several groups [44–47] also found more
than one reduction peak in the TPR profile of oxide-supported Cu
samples. There is a general agreement to propose that large and
crystallized CuO particles reduce at higher temperature than smal-
ler CuO particles or dispersed CuII species in strong interaction
with the oxide support. Their reduction temperature, ꢀ175 °C
(Fig. 7), which is far lower than the one of bulk CuO (280 °C)
[43], could be interpreted as an effect of interaction between
CuO and TiO2 surface [46,48] and/or smaller CuO particle size
[49]. Accordingly, the two reduction peaks in Fig. 7 can be attribu-
ted to CuII ions that are interacting with TiO2 and with CuO. The
second reduction peak at ꢀ175 °C can be easily attributed to CuO
as its intensity increases as the Cu loading increases (Fig. 7); more-
over, it slightly shifts to higher temperature in agreement with the
fact the CuO particle size increases from 10 nm in 1.62 wt% Cu/TiO2
to 15 nm in 4.24 wt% Cu/TiO2 according to XRD. The reason for
which CuO was evidenced by TPR but not by XRD in the sample
with Cu loading lower than 1.55 wt%, may be the too low CuO con-
tent and/or too small particle size. According to the literature, the
first reduction peak might belong to amorphous Cu species or to
highly dispersed CuII species in strong interaction with TiO2 surface
[48].
Fig. 7. TPR of as-prepared Cu/TiO2 prepared with different Cu loadings after 20 h of
DPu.
3.2.2. UV–visible
Fig. 6a displays the UV–visible spectra of the as-prepared
Cu/TiO2 samples with different Cu loadings after 20 h of DPu. The
contribution of TiO2 was removed from the spectra using TiO2 as
a reference for the baseline. One can distinguish two types of
spectra, those corresponding to low Cu loadings (<1.6 wt%) and
those corresponding to high Cu loadings (>1.6 wt%).
The spectra of the low loaded samples are composed of an
absorption band around 420 nm, and a broad one in the 600–
1200 nm region that can be assigned to Cu(II) d–d transition
[34,35]. A band around 400 nm has been observed in numerous
systems consisting of transition metal ions doping titania or at
the surface of titania, and attributed to a band of interfacial charge
transfer (IFCT) from the O2p valence band of TiO2 to Mx+ ions [36–
41]. This band assignment has been very recently confirmed by
Platanitis et al. [42] in a study on Mo/TiO2 catalysts combining
experimental and theoretical study at the molecular level. This
study convincingly rationalizes the relationship between the pres-
ence of the band of charge transfer from titania surface oxygen to
metal (Mo) in UV–visible spectra and the establishment of a Ti–O–
Mo bridge, i.e., the formation of an inner sphere complex between
the Cu2+ complexes and the surface oxygen of titania. One may
wonder whether this Ti–O–Cu bonding occurs in solution during
DPu preparation or during the drying step. To attempt to answer
this question, the UV–visible spectra of a series of wet DPu Cu/
TiO2 samples (those prepared for different durations, Table 2) were
recorded. The spectra displayed in Fig. 6b show that the band at
420 nm is already present in the wet samples, and seems to
increase in intensity with the DPu duration, and to be stronger
after drying at RT (Fig. 6c). In addition, Fig. 6a shows that the spec-
tra of the low loaded samples are similar to the spectrum of the
mechanical mixture of Cu2(OH)3NO3 and TiO2 corresponding to
1.5 wt% Cu. This observation is again a good indication of the pres-
ence of Cu2(OH)3NO3 in the low loaded green samples.
3.2.4. TEM study of the metal particle size in Cu/TiO2 samples
The sizes of metallic Cu particles in the Cu/TiO2 samples pre-
pared for 20 h were estimated by TEM (Fig. 8) after reduction
under H2 at 350 °C, and transfer into the microscope without air
contact (see Section 2). It must be mentioned that TiO2 support
has wide particle size distribution with largest size around
50 nm and smallest ones around 10 nm (not shown). The average
particle size was deduced from the particle size distributions, also
shown in Fig. 8. In the 0.88 wt% Cu/TiO2 sample, only very small
metallic Cu particles (<1 nm) are present (Fig. 8a); they are visible
on the edges of the TiO2 particles in the TEM images. In the 1.55 wt
% Cu/TiO2 sample, in addition to small particles, particles with lar-
ger size (around 3 nm) are also observed on the TEM images
(Fig. 8b). For the 4.24 wt% Cu/TiO2 sample (Fig. 8c), the Cu particle
size distribution is more complex. Firstly, only few particles with
size ꢀ6 nm can be found, which is surprising for a sample with
such high Cu loading (ꢀ4.24 wt%). To deepen the analysis, dark
field imaging (HAADF) was performed to observe more easily the
metallic Cu particles (white dots) in the images (Fig. 8d). Few large
(ꢀ15 nm) and very small particles (<1 nm) were observed in the
sample. Moreover, EDS carried out on these two types of particles
confirmed that the large (Fig. 8e) and the small particles (Fig. 8f)
For the highly loaded samples (>1.6 wt% Cu), an additional
broadband in the range 450–800 nm in the spectra appears and
grows up with the Cu loading, with roughly the same shape and
position as the one of the CuO–TiO2 mechanical mixture (1.5 wt%
Cu) in Fig. 6a. The threshold around 850–900 nm corresponds to
the energy of the band gap of copper oxide (Eg ꢀ 1.4 eV). The bands
of Cu2(OH)3NO3 seem to be still present in the spectra of the highly

102
Z. Wang et al. / Journal of Catalysis 340 (2016) 95–106
40
35
0.88wt%
(g)
(a)
30
25
20
15
10
5
0.55nm
0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Particle size / nm
1.55wt%
(h)
20
(b)
15
10
5
4.5nm
0
1
2 3 4 5 6 7 8 9 10111213141516
Particle size /nm
(c)
30
25
20
15
4.24 wt%
(i)
8.5nm
(e)
10
(d)
d =14nm
Cu
5
0
3
4
6 7 8 9
5Particle si1z0e1/1n1m2 13 14 15
(f)
d =0.8 nm
Cu
Fig. 8. TEM bright field micrographs of Cu/TiO2 (20 h of DPu) after reduction at 350 °C with (a) 0.88; (b) 1.55; (c) 4.24 wt% Cu; (d), STEM-HAADF micrographs of 4.24 wt% Cu/
TiO2; (e) and (f), EDS analysis of two different zones of image (d) and the corresponding particle size distributions (g–i).

Z. Wang et al. / Journal of Catalysis 340 (2016) 95–106
103
100
80
60
40
20
100
80
(a)
a
60
0.88 wt%
1.55 wt%
1.62 wt%
2.08 wt%
2.50 wt%
4.24 wt%
40
20
0
b
c
0
0
50
100
150
Time / min
200
250
300
350
100
(b)
99
98
97
96
95
Fig. 11. Catalytic stability under reaction at 105 °C of reduced Cu/TiO2 samples
with different Cu loadings performed with 100 mg of catalytic bed and the same
amount of Cu (13.9
1.55 wt% Cu/TiO2 + 43.2 mg TiO2; and (c) 35.2 mg 2.5 wt% Cu/TiO2 + 64.8 mg TiO2.
1 lmol): (a) 100 mg of 0.88 wt% Cu/TiO2; (b) 56.8 mg of
0.88 wt%
1.55 wt%
1.62 wt%
2.08 wt%
2.50 wt%
4.24 wt%
particles, Cu0 and CuO, respectively. To summarize, the sample
with high Cu loading contains particles smaller than 1 nm, parti-
cles around 4 nm and large particles around 15 nm, resulting in a
wide particle size distribution.
3.3. Selective hydrogenation of 1,3-butadiene
(c)
8
6
4
2
0
The evolution of the 1,3-butadiene conversion with tempera-
ture over the series of Cu/TiO2 samples prepared for 20 h is pre-
sented in Fig. 9a. One can see in Fig. 9b that butadiene is
transformed only into butenes up to T100% (minimum temperature
corresponding to 100% butadiene conversion; note that this is an
approximate temperature, and that it is possible that the real one
is at a temperature between it and the one of the former measure-
ment). A low amount of butane starts to be formed above these
temperatures while propene also starts hydrogenating with a very
low conversion (<2%, Fig. 9c). Intuitively, one could have expected
that the conversion curve and T100% will gradually shift to lower
temperature as the Cu loading increases (Fig. 9a). Although the
sample containing the lowest Cu loading (0.88 wt%) is the least
active and the one containing the highest loading (4.24 wt%) is
the most active, the evolution of T100% of the samples with interme-
diate Cu loadings seems erratic. However, the calculation of the
activity of the catalysts at 60 °C expressed per mass unit of copper
shows a decrease as the Cu loading increases (Fig. 10). This is con-
sistent with the fact that the Cu particle size also increases when
the Cu loading increases.
0.88 wt%
1.55 wt%
1.62 wt%
2.08 wt%
2.50 wt%
4.24 wt%
30
45
60
75
90
105
120
135
150
Temperature / oC
Fig. 9. Evolution of the butadiene conversion (a), the selectivity to butenes (1-
butene + cis-2-butene + trans-2-butene) (b) and the conversion to alkanes (butane
and propane) (c) as a function of the reaction temperature for reduced Cu/TiO2 with
different Cu loadings (100 mg of catalyst).
30
T = 60oC
25
Finally, an experiment was performed in isotherm at 105 °C to
test the stability of three Cu/TiO2 catalysts (0.88, 1.55 and 2.5 wt
20
15
10
5
%), using the same mass of Cu (13.9
1 lmol) in 100 mg of cat-
alytic bed thanks to physical mixing of Cu/TiO2 and TiO2, to operate
with the same contact time. Fig. 11 shows that the butadiene con-
version of the 0.88 wt% Cu/TiO2 catalyst was 100% during the first
1 h of reaction, then decreased to around 10% in the following 3 h
with a rate 19 ppm minꢂ1. For the 1.55 wt% Cu/TiO2 sample, buta-
diene conversion was also 100% at the beginning of the reaction,
but it quickly decreased to around 5% after 2.5 h with a rate of
28 ppm minꢂ1. Initial butadiene conversion of the 2.5 wt% Cu/
TiO2 sample was slightly lower than 100% and declined still more
sharply to around 2% during the first 1.5 h with a rate of
0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Cu Loading / wt%
44 ppm minꢂ1
.
Fig. 10. Evolution of the catalytic activities at 60 °C of reduced Cu/TiO2 samples as a
function of the Cu loading at butadiene conversion around 20–30%.
4. Discussion
are metallic Cu. The HAADF results are therefore consistent with
the XRD results obtained with the as-prepared samples (Fig. 5a)
since both show that the 4.24 wt% Cu/TiO2 sample contains large
The evolution of the Cu loading on titania as a function of pH
and of the deposition time reported in Fig. 3 shows that the process

104
Z. Wang et al. / Journal of Catalysis 340 (2016) 95–106
of Cu deposition on the TiO2 surface mainly occurs at pH lower
than the PZC of TiO2 (ꢀ6). It starts at the very beginning of the
DPu, since a small amount of copper is already deposited on TiO2
at pH < 3.8 (Fig. 3), i.e., at a pH at which the TiO2 surface is not sup-
posed to interact with cationic complexes such as Cu(H2O)26+ or Cu
(NH3)26+ (Fig. 1). Cationic CuII complexes interaction with titania at
low pH has been already reported by Bourikas et al. in a recent
review on titania [50]. They mentioned that although literature
related to the studies of the deposition of CuII complexes on TiO2
is not very abundant and not recent, there are several papers
reporting the adsorption of CuII aqua complexes on TiO2 surface
at pH of 2 or 4, i.e., below the PZC of titania (ꢀ6) [51,52].
itation in solution (pH 5.2 in the absence of TiO2), and it corre-
sponds to the inflection point of the pH curves in Fig. 4. This
result is consistent with the principle of the method of deposi
tion–precipitation [1–5], in particular, the precipitation of
Cu2(OH)3NO3 onto the titania support, also attested by the fact that
the particles of the copper hydroxynitrate precipitate are much
larger than those formed in the presence of titania. One can
therefore propose that the Cu2+ species already adsorbed on TiO2
at low pH act as nucleation sites for the growth of the precipitate
onto the support rather than into solution, lowering the pH
of precipitation, i.e., lowering the pH of the supersolubility of
Cu2(OH)3NO3aq, and preventing its precipitation into solution.
As the pH of the solution increases during DPu at 80 °C, instead
of reacting with the dissolved support as for silica and forming
copper silicate [56] since TiO2 is not known to dissolve in the 3–
8 pH range, Cu2(OH)3NO3 gradually transforms into CuO; this is
accompanied by a change of the suspension color from green to
gray. The evolution of the nature of the Cu compound on TiO2 is
also attested by the different colors taken by the as-prepared sam-
ples as a function of the Cu loading (Fig. 2) and the DPu duration
(Table 2). The green color of the samples with less than 1.6 wt%
Cu is due to the presence of stable Cu2(OH)3NO3 and the more
grayish color observed for the higher Cu loadings is related to the
additional presence of CuO. The formation of CuO is attested by
XRD (Fig. 5) and by the presence of the broadband at 450–
800 nm in the UV–visible spectra (Fig. 6a). However, Cu2(OH)3NO3
is more stable on titania than in the absence of support since the
Cu/TiO2 suspension (<3 wt% Cu) remains green after 5 h of DPu
whereas it turns blackish without support (Table 1). The grayish
color taken by the sample during DPu when more Cu is deposited
(4.24 wt% Cu) (Table 1) or when DPu duration increases (Fig. 2 and
Table 2), which is due to the presence of large particles of CuO, can
be explained as follows. It probably initially results from a phe-
nomenon of Ostwald ripening of the Cu2(OH)3NO3 particles during
DPu as the particle size of metallic Cu increases with DPu time
(Fig. 5b). In addition, very small Cu0 particles were also observed
by TEM (Fig. 8). These results may therefore indicate that Ostwald
ripening has occurred during sample preparation by DPu, produc-
ing a bimodal size distribution. Moreover, as the Cu2(OH)3NO3 par-
ticles increase in size, they become instable and transform into
CuO that continue to grow. It is interesting to note that at variance
with the silica and alumina supports (see Section 1), no mixed
compound, i.e., no copper titanate formation was detected. Sankar
et al. [57] showed that in the case of a parent system, Ni/TiO2,
NiTiO3 was detected by EXAFS after calcination but at temperature
higher than 600 °C and only when the samples were prepared by
impregnation (not by cation adsorption).
The titania P25 used in this study contains two phases, anatase,
the main one, and rutile. In these two different crystal structures,
each Ti4+ ion is surrounded by six oxygen forming octahedral enti-
ties (TiO6), and each oxygen is bonded to three Ti4+ ions (Ti3–O) by
two short Ti–O bonds (1.946 Å in rutile, and 1.937 Å in anatase)
and one longer (1.984 Å in rutile and 1.964 Å in anatase) [53]. Sur-
face Ti4+ ions have lower coordination than in the bulk, and the
surface of TiO2 possesses in principle three types of surface oxy-
gens, namely singly coordinated surface oxygen (TiO), doubly coor-
dinated one (Ti2O) and triply coordinated one (Ti3O). On the basis
of different surface ionization models for these surface sites, Bour-
ikas et al. [54] proposed a derived model of protonation reactions
for TiO and Ti2O that they consider more realistic than those in
the former models:
K1
TiOꢂ0:35 þ Hþ () TiOHþ0:65 logK1 ¼ 7:8
ð4Þ
ð5Þ
K2
Ti2Oꢂ0:57 þ Hþ () Ti2OHþ0:43 logK2 ¼ 4:6
Note that the oxygen of Ti3O is not protonated (logK3 = ꢂ3.96)
[55] and that the TiO sites are negatively charged (TiOꢂ0.35
,
Eq. (4)) even at pH lower than 3, based on experimental results
of Bourikas et al. [54].
We can therefore propose that at the very beginning of the DPu
preparation, i.e., at pH < 4 before precipitation starts (Fig. 3), part of
the Cu2+ complexes in solution can interact electrostatically with
the TiOꢂ0.35 sites of TiO2 surface and then could form Ti–O–Cu spe-
cies according to [52]. The formation of inner sphere complexes is
attested by the presence of the band at 420 nm in the UV–visible
spectra of the wet Cu/TiO2 samples (Fig. 6b). At the stage of the
present study, it is not possible to discard another hypothesis as
the one proposed by Petala et al. [41], which is a direct surface
reaction between Cu2+ complexes and surface OHs of titania that
would also lead to the formation of inner sphere complexes. It is
interesting to mention that titania behaves differently from silica
support. In a former study of one of us [28], dealing with adsorp-
tion of [Cu(NH3)4(H2O)2]2+ complexes on silica, the formation of
Si–O–Cu bonding was observed only after the sample has been
dried at RT and not when it was still wet. In any case, one can pro-
pose that once CuII complexes have interacted with the negatively
charged sites of TiO2 or some surface OHs or when pH has reached
4, the remaining of the Cu complexes in solution start precipitating
on titania. The nature of the Cu precipitate is probably the same as
the one obtained in the absence of support (Experiment B2 in Sec-
tion 3.1.2), i.e., copper hydroxynitrate (Cu2(OH)3NO3) and not cop-
per hydroxide. In spite of the fact that the diffractogram of as-
prepared Cu/TiO2 (Fig. 5a) does not show the corresponding
diffraction peaks, this hypothesis is consistent with the fact that
the samples have the same green color in the presence or in the
absence of TiO2, that their UV–visible and IR spectra are similar
(Figs. 6 and S1) and that Cu2(OH)3NO3 has been identified previ-
ously in Cu/SiO2 samples prepared by DPu [56]. Moreover, the
pH ꢀ 4 of Cu precipitation in the presence of TiO2, associated with
the color change of the suspension is lower than the pH of precip-
From this discussion, one can now propose an interpretation for
the TPR profiles (Fig. 7). The first TPR peak can be attributed to the
reduction of small NPs of Cu2(OH)3NO3 while the second one at
higher temperature can be attributed to the reduction of large
NPs of CuO; this peak becomes predominant as the Cu loading
increases.
The number of Cu nucleation sites is limited; thus, when the Cu
loading increases, the particles of Cu2(OH)3NO3 grow on these
sites, and increase in size. When the Cu concentration is high
enough (>1.6 wt%), Cu2(OH)3NO3 becomes instable and transforms
into CuO that also increases in size. These results are consistent
with the different metal particle sizes obtained after reduction at
350 °C, i.e., a progressive increase of the particle size and broaden-
ing size distribution with the Cu loading (Fig. 8). One can conclude
that the various particle size ranges observed by TEM are caused by
the different Cu loadings and Cu species present on TiO2 in the as-
prepared samples and also to Ostwald ripening.
The catalytic results can now be discussed in the light of these
characterization results. Fig. 10 shows that the sample containing

Z. Wang et al. / Journal of Catalysis 340 (2016) 95–106
105
0.88 wt% Cu, which has the smallest metallic particle size, exhibits
the highest catalytic activity. It is interestinꢂg1to note that the activ-
20 h of DPu. The results clearly show that the green sample con-
taining smaller Cu particles after reduction is more active than
the gray one.
ity is almost 4 times higher (25
l
mol sꢂ1 gCu) than that of 1.1 wt%
ꢂ1
Au/TiO2 catalyst tested in the same conditions [58] (4.5
l
mol sꢂ1
g
)
Au
(or twice as much active if activity is expressed per molmetal),
which is also a highly selective catalyst at full conversion of buta-
diene. Note that compared to Pd, the reference catalyst in industry,
copper is less active, but its cost is also much lower and its selec-
tivity to butenes is high at 100% conversion of butadiene. The activ-
ity of the Cu/TiO2 catalysts decreases as the Cu loading and the
particle size increase. One can also note that the activity clearly
decreases between the 1.55 wt% Cu/TiO2 sample and the 1.62 wt
% one although the loadings are very close (Fig. 10). As reported
before, the 1.55 wt% Cu/TiO2 sample is green before reduction
whereas the 1.62 wt% one is gray (Fig. 2). Although metal particle
sizes have not been determined for these samples, one can explain
the different activities by the fact that the green 1.55 wt% sample
contains no or less CuO than the 1.62 wt%, and therefore should
contain smaller metal particles after reduction, resulting in a
noticeable increase in activity. The stability experiment reported
in Fig. 11 shows that the rate of deactivation increases as the Cu
loading increases. According to former works on Cu catalysts
[25,59] oligomerization with the formation of ‘‘green oil” is respon-
sible for deactivation. XRD (Fig. S3) and thermogravimetry com-
paring the weight loss of the Cu/TiO2 sample after stability test
at 105 °C for 20 h with a fresh one (Fig. S4) confirm these results,
i.e., no apparent change in metal particle size and an important
weight loss during calcination of the spent catalyst. One can tenta-
tively propose that the higher catalytic stability of the Cu/TiO2 cat-
alyst containing the smaller particle size may result from the
presence of smaller Cu ensembles on which oligomerization reac-
tion takes place. In a way, this interpretation is consistent with a
recent study that showed that atomically dispersed Pd on gra-
phene was very stable in selective hydrogenation of butadiene
against deactivation compared to Pd nanoparticles [60].
5. Conclusions
In this contribution, the preparation of Cu/TiO2 samples (up to
5 wt% Cu) by the method of deposition–precipitation with urea
was explored as a function of the Cu loading, DPu time and pH.
We observed that the Cu2+ complexes in solution start to adsorb
on TiO2 at pH lower than PZC of TiO2 (<6) probably because of
the presence of some negatively charged Ti–O sites, forming
Ti–O–Cu bridges, then most of Cu2+ ions are deposited as
Cu2(OH)3NO3 on TiO2 surface in the first 2 h of the DPu process.
The characterization study led us to propose that the adsorbed
Cu2+ species act as nucleation sites for the growth of particles of
Cu2(OH)3NO3 (green samples). As DPu duration proceeds or Cu
loading increases, these particles increase in size through a mech-
anism of Ostwald ripening, then the large Cu2(OH)3NO3 transforms
into CuO (gray samples) depending on the DPu duration and the
final Cu loading, e.g., 2 h for 4.2 wt% Cu/TiO2 and more than 5 h
for 1.6 wt% Cu/TiO2. The presence of CuO in the as-prepared
sample has a great effect on the particle size of metallic Cu after
reduction at 350 °C: small particles (<3–5 nm) were obtained in
the green samples, which mainly contained Cu2(OH)3NO3, whereas
large ones (>10 nm) were also present in the gray samples initially
containing CuO. Moreover, the study of the Cu/TiO2 catalytic
performance in selective hydrogenation of butadiene in excess of
propene revealed that the supported Cu catalysts are able to fully
convert butadiene with almost 100% selectivity to butenes at
rather mild temperature, and that the catalytic activity and
stability are clearly enhanced by decreasing the Cu particle size.
It can be concluded that DPu is an efficient preparation method
to produce active Cu catalysts in hydrogenation reaction provided
that the formation of CuO is avoided.
Since the large Cu0 particles result from the presence of CuO in
the as-prepared catalysts and since their presence induces lower
activity in butadiene hydrogenation, and stronger deactivation,
the catalytic results underline the fact that the formation of CuO
must be avoided during the deposition–precipitation process in
order to achieve high catalytic performance. In order to verify this
conclusion, an extra catalytic experiment was performed on the
green sample with 2.1 wt% Cu obtained after 3 h of DPu, (Table 2).
In Fig. 12, the catalytic performance is compared to that of a gray
sample with the same Cu loading (2.05 wt% Cu) but obtained after
Acknowledgments
Zhao Wang thanks the China Scholarship Council (CSC) for his
Ph.D. scholarship. The authors thank Dr. Guillaume Wang of the
Laboratoire Matériaux et Phénomènes Quantiques at University
Paris Diderot-Paris7 for the electron microscopy performed on
the JEOL ARM 200F microscope.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jcat.2016.05.011.
100
80
References
[1] J.W. Geus, Dutch Pat. Appl. 6705,259, 1967; 6813,236, 1968.
[2] J.A. Van Dillen, J.W. Geus, L.A. Hermans, J. van der Meijden, in: G.C. Bond, P.B.
Wells, F.C. Tompkins (Eds.), Proceedings of the 6th International Congress on
Catalysis, London, 1976, The Chemical Society, London, 1977, p. 677.
[3] L.A.M. Hermans, J.W. Geus, in: B. Delmon, P. Grange, P.A. Jacobs, G. Poncelet
(Eds.), Preparation of Catalysts II, Elsevier, Amsterdam, 1979, p. 113.
[4] P. Burattin, M. Che, C.J. Louis, Phys. Chem. B 101 (1997) 7060.
[5] P. Burattin, M. Che, C.J. Louis, Phys. Chem. B 102 (1998) 2722.
[6] C.J.G. Van Der Grift, P.A. Elberse, P.A. Mulder, J.W. Geus, Appl. Catal. 59 (1990)
275.
[7] V. Kaushik, C. Sivaraj, P.K. Rao, Appl. Surf. Sci. 51 (1991) 27.
[8] A. Sandoval, C. Louis, R. Zanella, Appl. Catal. B – Environ. 140 (2013) 363.
[9] L. Delannoy, G. Thrimurthulu, P.S. Reddy, C. Méthivier, J. Nelayah, B.M. Reddy,
C. Ricolleau, C. Louis, Phys. Chem. Chem. Phys. 16 (2014) 26514.
[10] R. Si, J. Raitano, N. Yi, L. Zhang, S.W. Chan, Flytzani-Stephanopoulos, Catal.
Today 180 (2012) 68.
60
40
2.10 wt% (DPu 3h green color)
20
2.05 wt% (DPu 20h gray color)
0
30
45
60
75
90
105
120
135
150
Temperature / oC
Fig. 12. Comparison of the evolution of the butadiene conversion between two Cu/
TiO2 samples containing the same Cu loading (2 wt% Cu) but prepared for different
durations, 3 h leading to a green sample after DPu and 20 h leading to a gray
sample.
[11] S.F. Peng, J.J. Ho, Phys. Chem. Chem. Phys. 13 (2011) 20393.
[12] V.G. Deshmane, R.Y. Abrokwah, D.J. Kuila, Mol. Catal. A – Chem. 408 (2015)
202.

106
Z. Wang et al. / Journal of Catalysis 340 (2016) 95–106
[13] H.C. Yang, H.Y. Lin, Y.S. Chien, J.C.S. Wu, H.H. Wu, Catal. Lett. 131 (2009) 381.
[14] M. Behrens, F. Studt, L. Kasatkin, S. Kuhl, M. Havecher, F. Ablid-Pedersen, S.
Zander, F. Girgsdies, P. Kurr, B.L. Kniep, M. Tovar, R.W. Fischer, J.K. Norskov, R.
Schlogl, Science 336 (2012) 893.
[15] J. Llorca, M. Dominguez, C. Ledesma, R.J. Chimentao, F. Medina, J. Sueiras, I.
Angurell, M. Seco, O. Rossell, J. Catal. 258 (2008) 187.
[34] N.L. Dias Filho, Microchim. Acta 130 (1999) 233.
[35] T. Toupance, M. Kermarec, C. Louis, J. Phys. Chem. B 104 (2000) 965.
[36] M. Liu, X. Qiu, M. Miyauchi, K. Hashimoto, Chem. Mater. 23 (2011) 5282.
[37] H. Irie, S. Miura, K. Kamiya, K. Hashimoto, Chem. Phys. Lett. 457 (2008) 202.
[38] H. Irie, K. Kamiya, T. Shibanuma, S. Miura, D.A. Tryk, T. Yokoyama, K.
Hashimoto, J. Phys. Chem. C 113 (2009) 10761.
[16] O.P. Vaughan, G. Kyriakou, N. Macleod, M. Tikhov, R.M. Lambert, J. Catal. 236
(2005) 401.
[39] I. Georgiadou, C. Papadopoulou, H.K. Matralis, G.A. Voyiatzis, A. Lycourghiotis,
C. Kordulis, J. Phys. Chem. B 102 (1998) 8459.
[17] R. Chimentao, F. Medina, J.L.G. Fierro, J. Llorca, J.E. Sueuras, Y. Cesteros, P.J.
Salagre, Mol. Catal. A – Chem. 274 (2007) 159.
[18] S. Pande, A. Saha, S. Jana, S. Sarkar, M. Basu, M. Pradhan, A.K. Sinha, S. Saha, A.
Pal, T. Pal, Org. Lett. 10 (2008) 5179.
[19] Q. Jia, D. Zhao, B. Tang, N. Zhao, H. Li, Y. Sang, N. Bao, X. Zhang, H.J. Liu, Mater.
Chem. A 2 (2014) 16292.
[20] F. Liao, Y. Huang, J. Ge, W. Zheng, T. Karaked, C. Paul, X. Hong, S.C. Tsang,
Angew. Chem. Int. Ed. 50 (2011) 2162.
[40] B. Choudhury, M. Dey, A. Choudhury, Int. Nano Lett. 3 (2013) 1.
[41] A. Petala, E. Ioannidou, A. Georgaka, K. Bourikas, D.I. Kondarides, Appl. Catal. B
178 (2015) 201.
[42] P. Platanitis, G.D. Panagiotou, K. Bourikas, C. Kordulis, J.L.G. Fierro, A.J.
Lycourghiotis, Mol. Catal. A 412 (2016) 1.
[43] J.Y. Kim, J.A. Rodriguez, J.C. Hanson, A.I. Frenkel, P.L. Lee, J. Am. Chem. Soc. 125
(2003) 10684.
[44] W.P. Dow, Y.P. Wang, T.J. Huang, J. Catal. 160 (1996) 155.
[45] G. Fierro, M.L. Jacono, M. Invesi, P. Porta, F. Cioci, R. Lavecchia, Appl. Catal. A –
Gen. 137 (1996) 327.
[21] B.M. Nagaraja, A.H. Padmasri, B. David Raju, K.S.J. Rama Rao, Mol. Catal. A –
Chem. 265 (2007) 90.
[22] J.J. Phillipson, P.B. Wells, G.R. Wilson, J. Chem. Soc. A: Inorg. Phys. Theor. (1969)
1351.
[46] J.N. Nian, S.A. Chen, C.C. Tsai, H.J. Teng, Phys. Chem. B 110 (2006) 25817.
[47] F.S. Delk II, A. Vavere, J. Catal. 85 (1984) 380.
¯
[23] E. Nishimura, Y. Inoue, I.B. Yasumori, Chem. Soc. Jpn. 48 (1975) 803.
[24] R.A. Koeppel, J.T. Wehrli, M.S. Wainwright, D.L. Trimma, N.W. Cant, Appl. Catal.
A – Gen. 120 (1994) 163.
[48] X.F. Yu, N.Z. Wu, Y.C. Xie, Y.Q. Tang, J. Mater. Chem. 10 (2000) 1629.
[49] G. Fierro, M. Lojacono, P. Inversi, P. Porta, R. Lavecchia, F. Cioci, J. Catal. 148
(1994) 709.
[25] N.J. Ossipoff, N.W. Cant, J. Catal. 148 (1994) 125.
[50] K. Bourikas, C. Kordulis, A. Lycourghiotis, Chem. Rev. 114 (2014) 9754.
[51] W.F. Bleam, M.B. McBride, J. Colloid Interface Sci. 110 (1986) 335.
[52] M.F. Ottaviani, E.M. Ceresa, M. Visca, J. Colloid Interface Sci. 108 (1985) 114.
[53] D.T. Cromer, K. Herrington, J. Am. Chem. Soc. 77 (1955) 4708.
[54] G.D. Panagiotou, T. Petsi, K. Bourikas, C.S. Garoufalis, A. Tsevis, N. Spanos, C.
Kordulis, A. Lycourghiotis, Adv. Colloid Interface Sci. 142 (2008) 20.
[55] L. Pauling, J. Am. Chem. Soc. 1010 (1929) 51.
[56] C.J.G. Van Der Grift, P.A. Elberse, A. Mulder, J.W. Geus, Appl. Catal. 59 (1990)
275.
[57] G. Sankar, C.N. Ramachandra Rao, T. Rayment, J. Mater. Chem. 1 (1991) 299.
[58] A. Hugon, L. Delannoy, C. Louis, Gold Bull. 41 (2008) 127.
[59] M.R. Stammbach, D.J. Thomas, D.L. Trimm, M.S. Wainwright, Appl. Catal. 58
(1990) 209.
[26] D.M. Fernandes, R. Silva, A.A. Winkler Hechenleitner, E. Radovanovic, M.A.
Custodio Melo, E.A. Gomez Pineda, Mater. Chem. Phys. 115 (2009) 110.
[27] A. Hugon, L. Delannoy, J.M. Krafft, C. Louis, J. Phys. Chem. C 114 (2010) 10823.
[28] L. Trouillet, T. Toupance, F. Villain, C. Louis, Phys. Chem. Chem. Phys. 2000
(2005) 2.
[29] M.K. van der Lee, A. Jos van Dillen, J.H. Bitter, K.P. de Jong, Am. Chem. Soc. 127
(2005) 13573.
[30] C. Louis, in: J.R. Regalbuto (Ed.), Catalyst Preparation: Science and Engineering,
Taylor& Francis/CRC Press, New York, 2007, p. 319.
[31] J. Park, J.R. Regalbuto, J. Colloid Interface Sci. 175 (1995) 239.
[32] D.C. Pereira, D.L.A. de Faria, V.R.L. Constantino, J. Braz. Chem. Soc. 17 (2006)
1651.
[33] S.A. Kondrat, P.J. Smith, P.P. Wells, P.A. Chater, J.H. Carter, D.J. Morgan, E.M.
Filordaliso, J.B. Wagner, T.E. Davies, L. Lu, J.K. Bartley, S.H. Taylor, M.S. Spencer,
C.J. Kiely, G.J. Kelly, C.W. Park, M.J. Rossernsky, G.J. Hutchings, Nature 531
(2016) 83.
[60] H. Yan, C. Cheng, H. Yi, Y. Lin, T. Yao, C. Wang, J. Li, S. Wei, J. Lu, J. Am. Chem.
Soc. 137 (2015) 10484.