Contents lists available at ScienceDirect [630606]

Contents lists available at ScienceDirect
Applied Catalysis B: Environmental
journal homepage: www.elsevier.com/locate/apcatb
Support-induced e ffect on the catalytic properties of Pd particles in water
denitri fication: Impact of surface and structural features of mesoporous
ceria-zirconia support
P. Grangera,⁎, S. Troncéaa,b, J.P. Dacquina,⁎, M. Trentesauxa, V.I. Parvulescub,⁎
aUniv. Lille, CNRS, Centrale Lille, ENSCL, Univ. Artois, UMR 8181 –UCCS –Unité de Catalyse et Chimie du Solide, F-59000 Lille, France
bUniversity of Bucharest, Department of Organic Chemistry, Biochemistry and Catalysis, B-dul Regina Elisabeta 4-12, Bucharest 030016, Romania
ARTICLE INFO
Keywords:
PalladiumMesoporous materialsCe
0.5Zr0.5O2mixed oxides
Catalytic nitrite reduction
Drinking waterABSTRACT
The support eff ect on the catalytic properties of palladium particles has been investigated in the reduction of
nitrites by hydrogen at 20 °C in batch conditions. The support material was composed of ceria-zirconia mixed
oxide stabilized in the tetragonal structure according to an Evaporation Induced Self-Assembly method. Incomparison, a co-precipitation method leads to inhomogeneity in composition related to the partial segregation
of the cubic fluorite structure of CeO
2. Further textural properties obtained for the two synthesis routes di ffer as
well as changes in the reducibility. Surface analysis demonstrated a more extensive surface reduction of Ce4+to
Ce3+on the series prepared by coprecipitation correlated to a greater stabilization of metallic Pd particles.
However, the best catalytic performances were obtained on low loaded Pd samples supported on the ceria-zirconia support prepared by the Evaporation Induced Self-Assembly method which emphasizes the fact that
catalytic properties cannot be simply explained by the stabilization of zero valent precious metal at the vicinityof anionic vacancies. Further calculations of TOF and interfacial rates were achieved and compared to the
selectivity behavior suggesting that the localization of Pd in contact with the tetragonal or cubic structure of the
solid solution and CeO
2could play a key role in determining the catalytic properties.
1. Introduction
The European Community has set the maximum permitted level of
nitrate in drinking water to 50 mg/L because of its adverse health ef-
fects [1]. The U.S. Environmental Protection Agency has lowered that
level to 10 mg/L [2]. The removal of nitrates from drinking water is not
trivial because of the high stability of ionic species and soluble nature.
To comply with the legislation, a speci fic treatment of drinking water is
necessary. Up to now, biological and physicochemical treatments alloweffective removal of nitrates. On the other hand, they have several
economic and ecological disadvantages as reported elsewhere [3].
Catalytic processes are among the most promising technologies and the
liquid phase hydrogenation of nitrates to nitrogen by gaseous hydrogen
over precious metal based catalysts could represent a valuable alter-
native [3–5]. In this speci fic case, some important technical issues were
addressed such as the catalyst stability and selectivity in order to lowerthe production of ammonia as side product. Successful attempts have
been already reported combining the advantage of catalysis and ion
exchange leading to a complete suppression of ammonium [6]. The
shaping of the catalyst, most suited for separating the catalyst of thereaction mixture, is also an important outcome. Recently, Durkin et al.
[7]demonstrated that supported Pd –In particles on milled linen fibers
are robust and stable exhibiting a stable catalytic activity for nitratereduction after evaluation during four months. Negligible metal
leaching was observed emphasizing an innovative and scalable ap-
proach for water purifi cation.
Nitrate reduction can be described by consecutive and parallel re-
actions where nitrates are first reduced to nitrites in the presence of
hydrogen, then ultimately converted to gaseous nitrogen as targetproduct. However, the formation of ammonia as undesired by-product
can occur and must be limited with maximum concentration allowable
fixed at 0.5 mg/L [8–11]. The reduction of nitrates to nitrites is usually
recognized as structure-insensitive reaction [12]. On the other hand,
the sequential reduction of nitrites to N
2is a structure-sensitive reaction
depending on size and morphology of noble metal particles [13]. The
complete reduction of nitrates to nitrogen is generally activated on
bimetallic particles with high e fficiency when palladium is combined
with copper [5,13 –15]. Up to now, the catalysts used for such an ap-
plication are mostly characterized by high palladium loading re-
cognized as the most active precious metal [16 –18]. One of the reasons,
https://doi.org/10.1016/j.apcatb.2017.11.007
Received 26 July 2017; Received in revised form 13 October 2017; Accepted 4 November 2017⁎Corresponding authors.
E-mail addresses: pascal.granger@univ-lille1.fr (P. Granger), jean-philippe.dacquin@univ-lille1.fr (J.P. Dacquin), vasile.parvulescu@chimie.unibuc.ro (V.I. Parvulescu).Applied Catalysis B: Environmental 224 (2018) 648–659
Available online 08 November 2017
0926-3373/ © 2017 Elsevier B.V. All rights reserved.
T

of such high metal loading could be likely due to a strong deactivation
because of particle sintering in aqueous phase and loss of metallic
character of palladium particles. Local variations of the concentration
of OH−, produced during the reaction, inside the porous structure of
the host matrix could also favor the adsorption of hydroxyl groups overprecious metals inducing a strong inhibition of the reaction rate. A side
effect is also related to an alteration of the selectivity becoming more
favorable toward ammonia formation. Hence, open mesoporous net-work could facilitate mass transfer phenomena then minimizing gra-
dient concentrations and related detrimental e ffects on the catalytic
activity and selectivity. Some attempts were also reported by sub-stituting conventional alumina or silica supports by carbon based sup-
ports but, unfortunately, those systems were found more selective to-
ward ammonia formation [16]. Further improvements were reported
on reducible TiO
2support [19]: The creation of strong-metal support
interactions stabilize more active electron-rich active metal state via the
formation of partially reduced TiO 2-xaggregates. A careful monitoring
of the electron density, as reported elsewhere on Pt/CeO 2, could be of
great importance to prevent over hydrogenation process leading in-
evitably to the production of ammonia [20]. A strengthening of the
metallic character could prevent corrosion by water and particle ag-gregation phenomena.
This study reports the behavior of supported palladium catalysts on
ceria-zirconia support. CeO
2–ZrO 2mixed oxides in a mesoporous
structure containing crystalline walls are expected to provide enhancedcatalytic performances due to their large surface area and a certain
degree of size and shape selectivity. Sanchez et al. reported the first
successful fabrication of cubic mesostructured CeO
2–ZrO 2thinfilms
[21]. It is possible to adjust the textural properties of CeO 2–ZrO 2mixed
oxides by utilizing di fferent methods of preparation. Accordingly, a soft
templating method such as evaporation induced self-assembly (EISA) as
novel direct method for preparing mesoporous Ce 0.5Zr0.5O2with con-
trolled porosity is compared to more conventional co-precipitation
methods. The catalytic reduction of nitrites to nitrogen have been in-
vestigated on monometallic Pd/Ce 0.5Zr0.5O2with regard to the impact
of the mesostructure as well as the extent of the Pd-support interaction,modulated by the pre-activation thermal treatment, on the catalytic
activity and selectivity. Interestingly, it was found that the prerequisites
for developing the catalytic performances could di ffer according to the
reducibility of the support materials compared to conventional sup-ported silica or alumina systems as well as on the preparation method.
2. Experimental2.1. Catalyst preparation2.1.1. Ce
0.5Zr0.5O2support materials
Ce0.5Zr0.5O2was prepared according to two di fferent experimental
protocols. The EISA method was based on a sol –gel process combined
with evaporation-induced self-assembly process in ethanol using block
copolymer Pluronic P123 as template and cerium nitrate and zirconium
oxychloride as precursor salts [22]. 0.5 g of Pluronic P123 (supplied by
Aldrich) was dissolved in 10 mL of ethanol. Then the solution obtainedafter dissolution in ethanol of Ce(NO
3)3·6H 2O and ZrOCl 2·8H 2O (sup-
plied by Sigma Aldrich) precursor salts was added. After stirring for 3 hat room temperature, the homogeneous solution was transferred to an
oven under desired temperature and humidity (temperature 40 °C, and
relative humidity: 50%) and underwent solvent evaporation. After 48 h
aging, the gel product was dried first at 60 °C for 24 h and then at
100 °C for another 24 h, in order to favor the interaction between the
precursors and the polymer which originated the mesoporous structure
[22].
For the co-precipitation method (COP), Ce(NO
3)3·6H 2O and
ZrOCl 2·8H 2O were dissolved in distilled water under magnetic stirring
and then heated at 60 °C. Cerium and zirconium hydroxides were pre-cipitated by adding dropwise ammonia solution up to pH = 10 and thesuspension was stirred for 4 h. The solid precursor was washed withdistilled water several times before aging at room temperature for 48 h,
and drying at 120 °C for 24 h [23].
The solid oxy-hydroxide precursors obtained through these two
methods were calcined in air at 400 °C for 4 h. The temperature gra-
dually increased with a gradient temperature of 1 °C/min.
2.1.2. Impregnated Pd/Ce
0.5Zr0.5O2catalysts
The monometallic catalysts were prepared by incipient wetness
impregnation from aqueous solutions of palladium nitrate (Pd
(NO 3)2·2H 2O, Sigma Aldrich) with concentration adjusted to get 2.3 wt.
% or 0.46 wt.% Pd [24]. After impregnation, the precursors were dried
in air at 105 °C overnight, calcined in air at 400 °C for 2 h and finally
reduced in pure hydrogen at 300 °C or 500 °C for 4 h with a heating rateof 5 °C/min and a H
2flow of 60 mL/min. The reduced catalysts were
finally labeled xPd/CZ(EISA) and xPd/CZ(COP) where x stands for the
weight palladium loading equal to 0.46 or 2.3 wt.%.
2.2. Physicochemical characterization
2.2.1. Bulk characterization
Thermogravimetric analyses (TGA) were carried out on a TA
Instrument DSC-TGA SDT 2960 thermal analyzer in the temperature
range of 25– 900 °C under air. The heating rate was kept constant to
10 °C/min.
X-ray di ffraction (XRD) patterns were recorded on a HUBER G-670
diffractometer fitted with an anti-cathode CuK α(λ= 1.54178 Å). XRD
patterns were recorded over 2θ values ranging from 10° to 80°, with a
scanning rate of 5 s/step and 0.05 step sizes. Low-angle XRD patternswere recorded on a D8 advances X-ray di ffractometer (Bruker AXS)
fitted with a CuK α(λ= 1.54178 Å) radiation in the 2θ range 0.3 –6°
with a 0.02° steps.
Raman spectra were recorded in a Labram In finity Dilor spectro-
meter equipped with a frequency-doubled Nd:YAG laser correspondingto an excitation radiation of 532 nm. The silicon line at 521 cm
−1was
used for calibration. The exciting wavelength from an Ar ion laser witha power of 4 mW on the samples was 514.5 nm. The scanning range was
set between 100 and 1800 cm
−1.
The bulk redox properties were investigated from H 2-Temperature-
Programmed Reduction experiments (H 2-TPR) on a Micromeritics
Autochem II 2920 under a flow of 5 vol.% H 2in Ar and a gradual
heating rate of 5 °C/min. Scanning Emission Microscopy images were
recorded on a Hitachi SU-70, SEM-FEG microscope.
The elemental analysis was performed by inductively coupled
plasma-optic emission spectroscopy 720-ES ICP-OES (Agilent) with
axially viewing and simultaneous CCD detection. The ICP Expert ™
software (version 2.0.4) provided the concentration of metal in sampleallowing estimating the weight percentage of Pd.
2.2.2. Surface characterization
The textural properties were analyzed from nitrogen adsorption-
desorption experiments at −196 °C using a Micrometrics Tristar 3020
apparatus. Prior to adsorption, all samples were systematically de-
gassed at 200 °C under vacuum for 4 h. The speci fic surface area was
calculated from the BET equation and the pore volume (V
p) was esti-
mated using the adsorption branch of the nitrogen isotherm curve at P/
P0= 0.98 single point.
XPS experiments were performed under ultra-high vacuum (UHV)
(∼10−10Torr) using a Vacuum Generators Escalab 220XL spectro-
meter equipped with a monochromatized aluminum source for excita-
tion. Binding energy (B.E.) values were referenced to the binding en-
ergy of the C 1s core level at 285.1 eV. The relative quanti fication was
achieved using a mixed Gaussian/Lorentzian peak fit keeping binding
energies and half-width constant for all spectral decompositions.
Palladium dispersion was calculated from H 2chemisorption mea-
surements performed on a Micromeritics Autochem II 2920 apparatus.P. Granger et al. Applied Catalysis B: Environmental 224 (2018) 648–659
649

Prior to chemisorption, all samples were pre-reduced at 300 °C or
500 °C under a flow of 20 mL/min of pure hydrogen and outgassed at
the selected temperature in argon. H 2pulses with 5 vol.% H 2in Ar were
injected until saturation at 100 °C to avoid overestimation of H 2uptake
due to palladium hydride formation [25]. The dispersion was calculated
according to the stoichiometric ratio H/Pd = 1.
2.3. Catalytic measurements
The nitrite reduction reaction was performed in a 250 mL batch
reactor equipped with a magnetic stirrer operating at atmospheric
pressure. 80 or 400 mg of catalyst in powder form, with average grain
size of 250 μm, were introduced in the reactor and then purged with
pure H 2for 1 h at RT. In the second stage, 40 mL of ultrapure water was
add to the system under a flow of H 2(200 mL/min) for 1 h. Afterwards,
10 mL of nitrites solution (500 mg/L or 100 mg/L) were introduced andthe suspension was stirred with a the rotation speed set at 700 rpm. The
experiment was carried out at 20 °C with a H
2flow rate of 200 mL/min.
Nitrite concentration was monitored by an ionic chromatograph(Metrohm 844 UV/VIS Compact IC –Column Metrosep A Supp 16-250/
4.0). The concentration of ammonium ions were measured by using anion chromatograph (Metrohm 861 Advanced Compact IC –column
Metrosep C 6 –250/4.0).
3. Results and discussion
3.1. In fluence of the synthesis route of Ce
0.5Zr0.5O2on the physicochemical
properties: structural vs. textural properties
3.1.1. Thermal decomposition of solid precursors of Ce –Zr mixed oxides
prepared by coprecipitation and EISA
TG-DSC curves vs. temperature performed on the solid precursors of
Ce0.5Zr0.5O2prepared via the EISA and the coprecipitation methods are
reported in Figs. 1(a) and 2respectively. To complement those ob-
servations, the TG and heat flow curves vs. temperature curves recorded
on the precursors of the single oxides (CeO 2and ZrO 2) are recorded in
Fig. S1 in Supplementary Materials. The weight loss pro file recorded on
the EISA precursor is similar to that previously reported on Ce xZr1-xO2
according to the same preparation method [26]. Below 150 °C, the
endothermic water evaporation takes place. The strong discontinuity,
appearing in the weight pro file at 180 °C, can be ascribed to the com-
bustion of the copolymer template (pluronic 123) and could be cata-lyzed by the presence of cerium as discussed in Supplementary Mate-
rials.
The identi fication of processes occurring at 290 °C and 350 °C is
somewhat easier by comparing with in situ Raman spectroscopic mea-
surements in Fig. 1(b) on the precursor exposed to O
2/He mixture. Asseen, three broad Raman lines appear distinctly from 320 °C centered at
300 cm−1, 490 cm−1and 620 cm−1. The Raman line at 490 cm−1is
characteristic of the F 2gmode of the cubic fluorite-like structure
[27 –30]. This value is higher than that of the peak in the spectrum of
pure CeO 2, as a result of reducing cell parameters due to the insertion of
the Zr4+ion with a smaller radius than Ce4+(0.84 Å vs. 0.97 Å) [31].
The band at 620 cm−1can be attributed to a non-degenerate long-
itudinal optical (LO) mode of ceria induced by the oxygen vacancies in
the ceria lattice [32 –34]. The bands near 300 cm−1could be tentatively
related to the displacement of oxygen atoms from their ideal fluorite
lattice positions by the zirconium insertion into the CeO 2lattice with
the formation of the pseudo-cubic ( t”) phase [10,35] . Hence, based on
those observations and related assignments, a predominant growth of a
cubic structure would start at rather low temperature emphasizing a
predominant formation of solid solution in rather good agreement with
the absence of signi ficant signal above 450 °C ascribed to the crystal-
lization of the metastable tetragonal structure of ZrO 2(see Fig. S1(d)).
Let us note that, in this domain of composition, the stabilization of a
tetragonal structure for the Ce 0.5Zr0.5O2solid solution cannot be com-
pletely ruled out. TG analysis on the precursor obtained by coprecipi-
tation ( Fig. 2 ) also evidences processes occurring more smoothly in the
absence of surfactant.
3.1.2. Structural properties of Ce 0.5Zr0.5O2mixed oxides calcined at 400 °C
3.1.2.1. Wide angles X-ray di ffraction analysis . X-ray di ffraction pattern
on CeO 2inFig. 3(a) is characterized by re flections (111), (200), (220)
and (311) at 2θ values of 28.5°, 33.0°, 47.4°, 56.2°, corresponding to the
cubic structure of CeO 2(JCPDS 34-0394). For ZrO 2, the main
reflections at 2θ values of 30.2°, 35.3°, 50.2°, 60.1° are characteristic
of a tetragonal structure (JCPDS 81-15). In most cases, changes
observed in the structural features of the fluorite lattice of CeO 2are
highlighted by discernible shifts on the 2 θvalues when cerium is
partially substituted by zirconium. This could be attributed to
deformations inducing some modi fications of the interplanar spacing
when Ce4+(0.97 Å) is substituted by Zr4+(0.87 Å) having a smaller
ionic radius [31,33] . Hence, the 2θ values for Ce 0.5Zr0.5O2at 29.3° and
33.8° are in good agreement with previous observations [36] and
consistent with the aforementioned statements. In practice, both the
position and peak broadening observed in Fig. 3(a) cannot lead to
unequivocal assignments [36,37] . Indeed, for intermediate
compositions, the phase diagram of Ce xZr1-xO2suggests the
coexistence of stable and metastable tetragonal phases and apredominant cubic phase. An interesting observation is related to the
lower 2θ values observed on the X-ray pattern of Ce
xZr1-xO2prepared
by co-precipitation (see Fig. 3(b)). According to the aforementioned
explanation, this could re flect a weaker insertion of zirconium inside
the ceria lattice.
Fig. 1. (a) TG-DSC analysis with weight loss profi le in red, the derivative weight curve in blue and the heat flow pro file in black and (b) in situ Raman spectroscopic measurements
performed on the precursor of Ce 0.5Zr0.5O2prepared according to the EISA procedure –in situ Raman spectroscopic measurements were performed under a flow of oxygen diluted in He
(O2/He = 1.7) with a heating rate dT/dt = 5 °C/min. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)P. Granger et al. Applied Catalysis B: Environmental 224 (2018) 648–659
650

The crystallite sizes have been estimated from the Williamson-Hall
method leading to comparable values listed in Table 1.
3.1.2.2. Raman spectroscopy . The Raman spectra of CZ(EISA) and CZ
(COP) calcined at 400 °C are shown in Fig. 4. Raman spectroscopic
measurements can complement XRD data since this technique is also
sensitive to amorphous structures. Typically, 6 Raman active modes
(A1 g + 2B1 g + 3E2 g) are expected in the case of tetragonal structure
characterized by a P4 2/nmc space group instead of one for the F2 g
mode for a cubic fluorite structure with a space group Fm3 m
corresponding to an intense Raman line at 490 cm−1inFig. 4.A s
reported elsewhere, a signi ficant shift to higher wavenumbers is
observed compared to current values earlier reported on pure CeO 2
(465 cm−1)[37]. Additional weaker Raman lines located at 155, 200,
300 and 620 cm−1are discernible on both samples. For CZ(EISA) a
broadening of the peak from 490 cm−1can be observed with shoulder
at about 550 nm−1which is attributed to the tetragonal ( t) phase [27].
The bands near 300 cm−1and 180 cm−1for CZ(COP) and CZ(EISA) are
related to the displacement of oxygen atoms from their ideal fluorite
lattice positions by the zirconium insertion into the CeO 2lattice with
the formation of the pseudo-cubic ( t”) phase [10,35] . The presence of
the band at about 155 cm−1is related to the t-ZrO 2phase [27].
Moreover, all the bands exhibited by COP are red-shifted comparedwith EISA indicating a lower insertion of ZrO
2in the CeO 2lattice. This
seems to be consistent with a sharp broadening of the 490 cm−1causedby crystallites size and increasing concentration of crystal defectsessentially oxygen vacancies [38]. This explanation is in relative good
agreement with a more distinct observation of the 620 cm
−1Raman
line on the EISA sample previously assigned to a non-degenerate Raman
inactive longitudinal of CeO 2reflecting disturbances of the local Me O
bond chemistry leading to the relaxation of its symmetry [32 –34]. This
band, previously observed on Ce xZr1-xO2characterized by a larger
amount of defective sites, could suggest some improvement on the bulkredox properties of this materials [39].
Fig. 2. TG analysis performed on the precursor of Ce 0.5Zr0.5O2prepared
according to the coprecipitation method.
Fig. 3. Wide angles X-ray di ffraction patterns recorded on (a) single oxides reference and mixed Ce 0.5Zr0.5O2after calcination in air at 400 °C –CeO 2(EISA) in blue, ZrO 2(EISA) in green
and CZ(EISA) in red. (b) X-ray di ffraction patterns of Ce 0.5Zr0.5O2prepared by coprecipitation (blue) and evaporation self-assembly method (red). (For interpretation of the references to
colour in this figure legend, the reader is referred to the web version of this article.)Table 1
Elemental and textural analysis of Ce 0.5Zr0.5O2prepared by coprecicipation and the EISA
method.
Sample Ce
(at.%)Zr (at.%) Speci fic
surf. area(m
2g−1)Total porevol.(cm
3g−1)Averagepore size(nm)Crystallitesize (nm)
CZ(EISA) 0.35 0.65 98 0.10 4.5 5.6CZ(COP) 0.43 0.57 100 0.10 6.5 5.2Ce(EISA) 14 0.04 11.9Ce(COP) 52 0.12 9.3Zr(EISA) 62 0.10 6.7
Zr(COP) 132 0.17 5.3P. Granger et al.
Applied Catalysis B: Environmental 224 (2018) 648–659
651

3.1.3. Textural properties of Ce 0.5Zr0.5O2mixed oxides calcined at 400 °C
Pore ordering quality of the materials was assessed by low angles
XRD (see Fig. 5). As illustrated, a well-de fined re flection centered at
0.55° is observed on the dried CZ sample prepared following EISA
protocol. This result evidences the genesis of an ordered mesophase
(P6mm hexagonal symmetry) with respect to the conventional pre-
cipitation route. However, the calcination step was partially detri-mental for this sample as observed by the broadening and shifting of the
X-ray line toward higher 2θ angle (1.15°). Partial shrinkage of the
mesostructure would indubitably occur during the decomposition of thesurfactant in the presence of inorganic species. Interestingly, the CZsample prepared by the coprecipitation route exhibits a broad X-ray line
following the calcination step that would probably originate from the
loose aggregation of particles at the nanoscale. Hence, mesopores for-
mation can be obtained following these two synthesis routes but arise
from di fferent processes.
Further information are given from nitrogen physisorptionmeasurements performed on CZ(EISA) and CZ(COP), as reported in
Fig. 6 . Both are showing adsorption-desorption pro files of type IV
characteristic of mesoporous systems [40]. As observed, the large
hysteresis obtained for CZ(COP) would more re flect the presence of
agglomerated particle solids as evidenced earlier by low angle XRD. Onthe other hand, the form of the hysteresis of type H2 for CZ(EISA)
would characterize the presence of partially uniform mesopores. The
specific surface area, pore volume and average pore size, calculated by
the BET and BJH methods (see Table 1) are comparable. On the other
hand, narrow pore size distribution obtained on CZ(EISA) in the range
2–6 nm drastically di ffer from that obtained on CZ(COP) exhibiting a
broader and bimodal distribution in the range 2 –12 nm. Similar dif-
ferences were also previously reported by comparing mesoporous Ce –Zr
solid solutions prepared with and without template [41] from respec-
tively the self-assembly of the pluronic template and disordered accu-mulation in the absence of template. Such di fferences correctly match
the observations from Scanning Electron Microscopy micrographs in200 400 600 800 1000 1200 EISA
Copp.
155
200300490
620Fig. 4. Raman spectra recorded on Ce 0.5Zr0.5O2prepared by EISA and
coprecipitation and calcined at 400 °C in air: CZ(COP) in blue and CZ
(EISA) in red. (For interpretation of the references to colour in this figure
legend, the reader is referred to the web version of this article.)
Fig. 5. Low angles XRD diagrams performed on dried samples (a) and
calcined samples (insight) (b) COP in blue and EISA in red. (For in-
terpretation of the references to colour in this figure legend, the reader
is referred to the web version of this article.)P. Granger et al. Applied Catalysis B: Environmental 224 (2018) 648–659
652

Fig. 7 with a sharp pore size distribution discernible for CZ(EIZA). As a
matter of fact, comparing the textural features obtained on single oxide,
especially the average pore size determined on CeO 2, one could suggest
that the presence of inhomogenities could be also the results of het-erogeneity in composition with a partial segregation of CeO
2in the
particular case of CZ(COP) in agreement with XRD and Raman spec-troscopic measurements.
3.2. Reducibility of bare Ce
0.5Zr0.5O2and doped with palladium
Preliminary observations from H 2-TPR experiments on CeO 2in Fig.
S2 (Supplementary information) reveal two distinct reduction processes
in the temperature range 420– 510 °C and above 800 °C previously as-
cribed to surface and bulk reduction processes of Ce4+to Ce3+[36,37] .
Fig. 8(a) shows the H 2uptake pro files vs. temperature recorded
during H 2-TPR experiments on CZ(EISA) and CZ(COP) calcined at
400 °C which slightly di ffers. As observed, an unique broad and intense
asymmetric signal is observed on CZ(EISA) with a maximum at∼575 °C (peak labeled γ) with a weakly discernible shoulder near
450 °C. Similar observations were previously reported showing, in mostcases, a broad signal developing below 700 °C and tentatively explained
by an improvement of oxygen ion mobility induced by the insertion of
zirconium inside the ceria lattice [37,42 –44]. Interestingly, the
shoulder coexisting with peak γon CZ(COP) intensi fies shifting to lowertemperature. In addition, an extra reduction process appears, with a
maximum around 800 °C, previously evidenced during the reduction of
pure ceria and suggesting a partial segregation of CeO
2on this sample
coexisting with the tetragonal Ce –Zr solid solution [36].
Subsequent incorporation of palladium drastically improves the
reduction of Ce 0.5Zr0.5O2with a sharp shift of peak βto lower tem-
perature. Let us note that the calculated H 2uptakes of 1.3 × 10−3and
1.8 × 10−3mol/g, respectively on 2.3Pd/CZ(EISA) and 2.3Pd/CZ
(COP), largely exceed the theoretical H 2consumption for the bulk re-
duction of PdO to Pd0corresponding to 0.2 × 10−3mol/g. Particular
attention was paid to the additional low temperature reduction process
taking place below 150 °C corresponding to the reduction of oxidic
palladium species. The reduction of PdO to metallic Pd particles was
previously observed in the temperature range 60 –90 °C on Pd/Ce –Zr–O
systems [33,45] . As a matter of fact, the reduction temperature of PdO
particles can be both closely related to the nature of support materials,
with which they interact, and the particle size. Indeed, numerous in-
vestigations found that well-dispersed PdO reduces above 100 °C
whereas the reduction of larger PdO particles would occur for
T≤50 °C [46 –48]. Hence, the absence of signal below 50 °C on Pd/CZ
(EISA) and Pd/CZ(COP) could suggest a strong interaction between PdOand Ce
0.5Zr0.5O2assuming a reduction of PdO partly hindered by the
reduction of Ce 0.5Zr0.5O2at least on Pd/CZ(EISA). As a consequence,
the appearance of the slight contribution (peak α) on Pd/CZ(COP)
could suggest a higher reducibility of PdO partly ascribed to the seg-regation of larger particles or to the existence of di fferent types of in-
teractions since Fig. 8 (a) clearly showed the segregation of Ce-Zr mixed
oxide with CeO
2.
3.3. Surface properties
Pd dispersion was estimated from H 2titration. Chemisorption
measurements were performed at 100 °C in order to avoid an over-estimation on H
2uptakes due to the current formation of bulk palla-
dium hydrides. As mentioned, a dissociative adsorption with H/Pd = 1has been taken into account leading to the results collected in Table 2.
The mean particle size has been roughly calculated assuming hemi-spherical metallic palladium particles. To check the validity of those
calculations the mean Pd particle size of Pd particle on 2.3Pd/CZ(EISA)
was measured from TEM analysis leading to a good consistency (7 nm
vs. 8.6 nm from H
2chemisorption). This comparison also underlines the
occurrence of weak spill-over e ffect previously underlined on Rh sup-
ported on ceria [44]. As explained, the presence of zirconia and the
relative low dispersion on Pd/CZ, could partly explain this observation.As seen, higher Pd dispersion are systematically obtained on the series
Pd/CZ(EISA) from direct comparison with their homologues preparedFig. 6. Nitrogen physisorption isotherms recorded at –196 °C on Ce 0.5Zr0.5O2calcined at
400 °C –CZ(COP) in blue and CZ(EISA) in red. (For interpretation of the references to
colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 7. Scanning electron microscopy images recorded on Ce 0.5Zr0.5O2calcined 400 °C –CZ(EISA) (a) and CZ(COP) (b).P. Granger et al. Applied Catalysis B: Environmental 224 (2018) 648–659
653

by co-precipitation.
To complement these measurements, ex situ XPS measurements
were performed on calcined samples and reduced in H 2at 400 °C. Let us
note that reduced samples were stored in ambient atmosphere prior to
XPS analysis. The characteristic Zr 3d, Ce 3d, O1s and Pd 3d core levels
were analyzed. The Binding Energy (B.E.) values obtained from the
photopeak Zr 3d 5/2in the B.E. range 182.4 –183.5 eV essentially evi-
dence the presence of Zr4+species. More complex spectral features
usually characterize the Ce 3d photopeak which can be decomposed
into di fferent components u, u ’,u”,u”’and v, v ’,v”,v”’as illustrated in
Fig. 9 [49] . In this particular case, the predominance of Ce3+or Ce4+
can be determined from the integration of the peak area relative to
these di fferent components leading to the estimation of the Ce3+/Ce4+
ratio in Table 2. Regarding the analysis of the Pd 3d photopeak, some
disturbances arose due to overlappings with the Zr 3p photopeak then
lowering the accuracy on the quantitative analysis, especially for lowloaded samples. Let us note that the B.E. values for the Pd 3d 5/2core
level essentially characterize the presence of oxidic palladium species
indicating a signi ficant reoxidation of ex situ reduced samples. How-
ever, an additional contribution can be isolated on Pd/CZ(COP) at335.5 –335.7 eV which could re flect the presence of Pd
0[50,51] and/or
PdxO clusters [52]. Let us note that in all cases the B.E. values diverge
than those currently reported for PdO in the range 336.1 –336.9 eV
[53]. The contribution appearing in the range 337.4 –338.4 eV could
correspond to the stabilization of Pd4+, possibly stabilized as PdO 2,a s
reported elsewhere [53 –55] and/or the formation of PdO ywith
1 < y < 2 caused by the di ffusion of oxygen in PdO [52]. As a matter
of fact, the observation of high B.E. can have di fferent origins re flecting
electron transfer processes governed by the extent of interaction with
the support and/or the in fluence of particle size then stabilizing elec-
tron defi cient Pdδ+species with δ>2 .
Semi-quantitative analysis provides important information
Fig. 8. H2-TPR experiments performed on bare Ce 0.5Zr0.5O2(a) and
Pd-doped Ce 0.5Zr0.5O2(b) In red CZ(EISA) and Pd/CZ(EISA) and in
black CZ(COP) and Pd/CZ(COP). (For interpretation of the references
to colour in this figure legend, the reader is referred to the web version
of this article.)P. Granger et al. Applied Catalysis B: Environmental 224 (2018) 648–659
654

revealing signi ficant divergences on the relative cerium composition at
the surface. As indicated in Table 2, bare and doped ceria-zirconia
supports prepared by the EISA method exhibit a zirconium surface
enrichment whereas cerium enrichment is systematically observed on
the series prepared by co-precipitation consistent with a partial segre-
gation of CeO 2. Regarding the palladium concentration, no signi ficant
deviation is observed after ex situ reduction which could suggest par-
ticle sintering. On the other hand, the calculation of the Ce3+/Ce4+
ratio led to signi ficant evolutions on the reduced Pd/CZ(COP) samples
with a sharp increase which could re flect a greater reducibility of sur-
face Ce4+stabilized as CeO 2instead of Ce xZr1-xO2solution.3.4. Catalytic properties of Pd/Ce 0.5Zr0.5O2in the reduction of nitrite by
hydrogen
3.4.1. Estimation of initial rates
Fig. 10 illustrates the conversion of nitrites, X(NO 2−) and the con-
centration of ammonium ions pro files vs. time in a stirred tank reactor.
Ammonia formed by extensive reduction of nitrites is stabilized as
ammonium ions in the reaction media. The in fluence of Pd loading and
pre-reduction temperature on the rate of nitrites conversion and am-monia formation was examined. Let us keep in mind that catalytic
measurements were performed in batch conditions with the sameTable 2
XPS analysis of Pd-doped Ce 0.5Zr0.5O2–influence of the reduction temperature on the composition and oxidation state of elements.
Catalyst Thermal pretreatment Pd Disp.aMean particle sizebB.E. (eV) Relative surface composition
(%) (nm) Zr 3d 5/2 Ce 3d 5/2 O1 s P d3 d 5/2 Zr/Ce Pd/Ce + Zr O/Ce + Zr Ce3+/Ce4+
CZ(EISA) none –– 183.5 883.9 530.9 – 1.21 – 2.25 0.57
CZ(COP) none –– 182.8 883.7 530.5 – 0.45 – 2.50 0.52
0.46Pd/CZ(EISA) none 182.5 882.9 530.1 338.1 1.80 0.70 × 10−22.20 0.42
0.46Pd/CZ(EISA) H 2/300 °Cc13 8.6 182.3 882.6 529.9 337.9 1.71 0.83 × 10−21.97 0.64
2.3Pd/CZ(EISA) none 183.0 883.7 530.4 338.4 1.37 5.66 × 10−22.34 0.29
2.3Pd/CZ(EISA) H 2/300 °Cc18.2 6.1 183.0 883.1 530.5 338.5 1.42 3.39 × 10−22.05 0.32
0.46Pd/CZ(COP) none 182.6 882.7 530.1 337.8 0.84 6.18 × 10−22.81 0.21
0.46Pd/CZ(COP) H 2/300 °Cc5.8 19.2 182.3 882.7 529.9 338.4/335.7 0.75 6.97 × 10−23.04 1.19
2.3Pd/CZ(COP) none 182.5 883.1 530.4 337.3 0.93 – 2.69 0.35
2.3Pd/CZ(COP) H 2/300 °Cc10.3 10.9 182.4 883.1 530.0 338.0/335.5 1.00 5.41 × 10−21.29 1.20
aFrom H 2titration at 100 °C on pre-reduced samples in pure H 2at 300 °C.
bCalculated from the palladium dispersion.
cEx situ reduced in pure H 2then stored in air at room temperature (XPS analysis were performed on sample stored in air).
Fig. 9. Typical example of the decomposition of the Ce 3d photopeak
recorded on 0.46Pd/CZ(EISA).P. Granger et al. Applied Catalysis B: Environmental 224 (2018) 648–659
655

amount of palladium by introducing inside the reactor 80 or 400 mg of
respectively 2.3 wt.% and 0.46 wt.% Pd/CZ corresponding to a total
number of 6.5 × 1019Pd atoms. The normalized initial speci fic rates
expressed per gram of palladium were calculated from the slope of thetangent in t = 0 leading to the rate values collected in Table 3.A s
shown, these values calculated on reduced sample at 300 °C are sensi-tive to the preparation method and the palladium loading. In these
operating conditions, the superiority of 0.46Pd/CZ(EISA) is clearly
demonstrated from the comparison of the normalized speci fic rates. Let
us note that this tendency is also conserved when the catalysts werepre-reduced at higher temperature, i.e., 500 °C, prior to reaction.
Further comparisons of normalized rates and TOF estimates from
previous investigations are not so evident because in most cases, kinetic
studies were performed at 25 °C while we selected a lower temperature,
i.e., 20 °C [56 –58]. Anyway, it is worthwhile to note that the TOF va-
lues on 0.46Pd/CZ(EISA) is one order of magnitude higher compared tothat measured on highly loaded 6.24 wt.% Pd supported on alumina
[56]. By comparing the normalized rate expressed per gram on 4.67 wt.
% Pd dispersed on pillared clays, 0.46Pd/CZ(EISA) still demonstrates itssuperiority (14.2 vs. 263 mmol min
−1gPd−1)[57]. Chinthaginjala and
Lefferts[58] reported relevant steady-state kinetic data measured in a
fixed bed flow reactor which account for the occurrence of deactivation
effects. In their operating conditions at 25 °C, the TOF is two order of
magnitude lower than that calculated on 0.46Pd/CZ(EISA). Despite
different reactor design (batch vs. continuous flow reactor), this latter
comparison suggests that 0.46Pd/CZ(EISA) could be potentially a goodcandidate for further practical developments.3.4.2. Selectivity towards the formation of ammonium ions
Significant information is also related to the selectivity behavior of
Pd/CZ samples with the formation of undesired products such as am-
monia. As exempli fied in Fig. S3, an important rise in ammonia con-
centration occurs at the initial stage of the reaction and then stabilizingon Pd/ZrO
2as reported elsewhere on di fferent catalytic systems
[59 –61]. As shown in Fig. 11, Pd/CZ behaves di fferently mimicking the
selectivity recorded on Pd/CeO 2with a volcano type curve suggesting
the occurrence of sequential processes involving the formation and theconsumption of ammonia (see Figs. S3(b) and 11). The literature re-
porting such a behavior is scarce [62,63] . Additional information re-
lative to the involvement of sequential processes can be obtained: (i)
Ammonium ions concentration gradually decreases while nitrites are
completely converted as exempli fied on 0.46Pd/CZ(EISA) which could
rule out an hypothetical direct redox reaction between NH
4+and NO 2−
contrarily to previous assumptions [63]. (ii) Qualitatively, the se-
quential consumption of ammonium ions seems to occur more readilyon sample reduced at 500 °C by comparing the ammonium concentra-
tion at the maximum and at the end after 6 h reaction. As exempli fied,
this sequential reaction process does not occur signi ficantly on 0.46Pd/
CZ(COP) reduced at 300 °C.
4. Discussion
This study clearly demonstrates the complexity for optimizing the
surface properties by combining reducible support materials and pal-
ladium particles. Indeed, previous investigations shown that catalytic
improvements in the reduction of nitrates can be obtained by
Fig. 10. Influence of the palladium loading and reduction temperature on the conversion profi les vs. time during the reduction of nitrites by hydrogen in batch conditions: (a) on Pd/CZ
(EISA) and (b) Pd/CZ(COP) –full symbol reduction at 300 °C, open symbol reduction at 500 °C –0.46 wt%Pd in blue and 2.3 wt.% Pd in green. (For interpretation of the references to
colour in this figure legend, the reader is referred to the web version of this article.)
Table 3
Initial rate measurements performed at 20 °C on Pd/Ce 0.5Zr0.5O2(0.46Pd/CZ and 2.3Pd/CZ) during the reduction of nitrite to nitrogen.
Catalyst T Red.(°C) d Pd(nm) Initial rate TOF (h−1) Interfacial rate NH4+aNH4+b
(mol h−1gPd−1) (10−12μmol s−1cm−1) at max at the end
2.3Pd/CZ(EISA) 300 6.1 5.0 × 10−229.3 8.4 15.8 3.0
2.3Pd/CZ(COP) 300 10.9 4.6 × 10−247.6 24.2 10.5 3.5
0.46Pd/CZ(EISA) 300 8.6 11.6 × 10−295.2 38.2 14.2 6.2
0.46Pd/CZ(COP) 300 10.2 1.9 × 10−233.8 30.3 6.1 6.3
2.3Pd/CZ(EISA) 500 19.9 3.4 × 10−264.4 75.1 16.0 1.7
2.3Pd/CZ(COP) 500 49.3 3.7 × 10−2175.9 499 16.0 1.7
0.46Pd/CZ(EISA) 500 31.7 15.8 × 10−2474.6 886 15.4 7.0
0.46Pd/CZ(COP) 500 30.1 2.6 × 10−274.0 130 24.0 3.3
aAmmonia concentration formed at maximum expressed in mg/L.
bResidual ammonia concentration formed at the end of the reaction expressed in mg/L.P. Granger et al. Applied Catalysis B: Environmental 224 (2018) 648–659
656

combining a reducible materials and the zero-valent states of precious
metals [14]. However, this statement is purely qualitative. At a first
glance, this concept could be widened in the reduction of nitrites since
the formation of anionic vacancies could facilitate the adsorption of
nitrites and metallic sites would be supposed to activate the dissociation
of hydrogen. Subsequent, di ffusion of dissociated hydrogen species at
the metal-support interface would finally reduce adsorbed nitrites.
However, particular attention should be paid to the nature of the crystal
phase which can in fluence oxygen mobility and the ease to create an-
ionic vacancies. Indeed, Ce xZr1-xO2mixed oxides with high zirconium
content, stabilizing a tetragonal structure, could have detrimental ef-
fects by suppressing the anionic mobility [64]. Moreover, in some ex-
tent the kinetic behavior of nano-sized metallic Pd particles can begoverned by structural requirements since the reduction on nitrites is
recognized structure sensitive [62] or particle size dependent [61,65] .
Generally speaking, the isolation of these two parameters is never aneasy task and could be somewhat con flicting.
As demonstrated in this study, the implementation of di fferent
methods for the preparation of Ce
0.5Zr0.5O2can lead to solids exhibiting
different bulk and surface properties which can potentially infl uence
the nature and the extent of interactions with palladium. In terms of
structural properties, the Evaporation Induced Self-Assembly method
leads mainly to the stabilization of the tetragonal structure for ceria-
zirconia mixed oxides with no detectable inhomogeneity in bulk com-
position. Importantly, bare and Pd-doped Ce 0.5Zr0.5O2samples pre-
pared according to the EISA procedure exhibit signi ficant surface zir-
conium enrichment. In contrast, the solids prepared by co-precipitation
are surface cerium enriched samples which can be simply explained by
a partial segregation of CeO 2proven by Raman spectroscopy with a
sharp intensi fication of the 490 cm−1Raman band related to the pre-
ferential growth of a cubic fluorite structure. These observations are
consistent with di fferent textural properties with the formation of a
broad bimodal pore size distribution emphasizing the coexistence of
different segregated phases on Pd/CZ(COP), i.e., the cubic structure of
CeO 2coexisting with the tetragonal structure of ceria-zirconia mixed
oxide. Those di fferent structural and textural features also coincide
with di fferent behavior in terms of reducibility with reduction processes
taking place more readily at the surface of the segregated CeO 2phase
characterizing Pd/CZ(COP) samples. As a consequence, the estimation
of higher Ce3+/Ce4+ratio from XPS analysis on ex situ reduced Pd/CZ
(COP) would be consistent with a higher density of anionic vacancieswith expected bene ficial effects in terms of catalytic properties com-
pared to the series prepared by the EISA method. XPS measurementsalso provide important information with respect to the oxidation stateof palladium related to the observation of metallic Pd species on theseries prepared by coprecipitation. In agreement with Kim et al. [19],
the promotion of stronger metal-support interactions on Pd/CZ(COP)
could stabilize electron-rich palladium particles. Hence, on the basis of
the statement postulated by Kim et al. [14] one can predict that the
series obtained by coprecipitation could exhibit higher catalytic per-formances in terms of activity if we assume the participation of metallic
sites at the vicinity of anionic vacancies.
As shown, the normalized rate expressed per gram of palladium
provide an overall information which undoubtedly show the superiority
of 0.46Pd/CZ(EISA) reduced at the lowest temperature, i.e., 300 °C,
which formally is not helpful to di fferentiate the particle size de-
pendency or the role played by the metal-support interface.Surprisingly the normalized rate expressed per gram of palladium
measured on reduced samples at 500 °C are comparable still verifying
the highest performances of 0.46Pd/CZ(EISA) while at higher reduction
temperature an increase of the density of anionic vacancies jointly to a
greater metallic character should enhance the reaction rate [14]. XPS
also reveal a greater metallic character on the reduced samples on Pd/CZ(COP) which could a priori bene fit the catalytic activity.
To go a little bit further the TOF were estimated on the basis of
chemisorption measurements. In addition we roughly estimated the
specific rate based on the length of periphery of Pd-Ce
xZr1-xO2interface
as earlier described [66,67] . In fact the length of the perimeter of
metallic Pd particles I 0can be calculated by assuming hemispherical
particles with a circular geometry at the interface and taking the Pddensity ρ
Pd= 12.02 g cm−3into account. The obtained values for I 0
oscillate in the range (0.3 –45.1) × 1011cm g−1leading to interfacial
rates expressed per cm in Table 3. Previous investigations [67] found
that the comparison between TOF and interfacial rates is useful toidentify di fferent kinetic regimes. Indeed, a rise in interfacial rate at
constant TOF values with increasing particle size was previously ob-served on Pt/LaFeO
3for the NO/H 2reaction and led the authors to the
conclusion that the metal-support interface would govern the kinetics.In the opposite way, the role of the Pt-LaFeO
3interface was found
negligible, recovering a structure sensitive reaction involving only
metallic Pt atoms as active sites on speci fic oriented surfaces. As shown
inTable 3, no decisive arguments are brought from the comparison of
the TOF and interfacial rates when Pd/CZ catalysts are reduced at
300 °C. Indeed the values likely vary within the margin of error and the
weak particle size dependency of the TOF correspond to a very narrow
range of particle size 6– 11 nm, (see Fig. 12 ) compared to previous in-
vestigations which demonstrated a structure-sensitivity on much largerparticles [65,68] . In fact, signi ficant di fferences arise when the catalysts
Fig. 11. Influence of the palladium loading and reduction temperature on the concentration profi les of ammonium ions vs. time formed during the reduction of nitrites by hydrogen in
batch conditions: (a) on Pd/CZ(EISA) and (b) Pd/CZ(COP) –full symbol reduction at 300 °C, open symbol reduction at 500 °C –0.46 wt% Pd in blue and 2.3 wt.% Pd in green. (For
interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)P. Granger et al. Applied Catalysis B: Environmental 224 (2018) 648–659
657

are reduced at 500 °C. As seen in Table 3, the range of particle sizes d Pd
widens up to 50 nm, in parallel, the surface is likely more reduced
corresponding to a greater density of anionic vacancies. As a result, a
sharp increase of the TOF and interfacial rates on Pd/CZ is observable
but in the same extent which is unable to discriminate between the
structure sensitivity and the prevalence of the metal-support interface
on the kinetics. On the other hand, this comparison accentuates the
peculiar properties of 0.46Pd/CZ(EISA).
As previously shown on precious metal catalysts supported on re-
ducible titania, strong-metal support interactions induce electron-rich
active metal state via the formation of partially reduced TiO 2-xag-
gregates which can in fluence both the rate and the selectivity.
Interestingly, a careful monitoring of the electron density on Pt/CeO 2
was found to avoid an over production of ammonia [20]. In practice,
the initial rate of production of ammonia shown in Fig. 11 is in rather
good agreement with the maximum concentration reported in Table 3
which emphasizes the fact that the production of ammonia seems to
occur more readily on samples reduced at 500 °C. Accordingly, thesubsequent comparison between 0.46Pd/CZ(EISA) and 0.46Pd/CZ
(COP) seems to be relevant with maxima in ammonium concentration
of respectively 15.4 and 24.0 mg/L. In agreement with previous state-
ments, such di fference could suggest di fferent extent in electron
transfer especially if Pd interacts di fferently with the tetragonal ceria-
zirconia solid solution and with CeO
2on Pd/CZ(COP). In this latter case
a more extensive electron donation could explain a sharp increase inammonia formation. This explanation seems consistent with a greater
stabilization of metallic Pd species on Pd/CZ(COP) reduced at 300 °C as
evidenced from XPS analysis. Alternately, a speci fic morphology of Pd
particles with preferential orientated surfaces due to di fferent interac-
tions with reduced CeO
2instead of Ce xZr1-xO2solid solution could be a
key parameter in monitoring the rate and the selectivity of the reactionin favor of nitrogen or ammonia. This statement has been earlier put in
evidence on Pt supported on LaFeO
3with epitaxial growth of Pt par-
ticles with preferential (111) surface [67]. Although, no decisive con-
clusion in this study regarding the nanostructure of such systems would
need more exhaustive advanced characterization, our results seems to
be more in favor of the explanation given by Epron et al. [59] showing
that the reduction of nitrites would predominantly take place on Ptparticles on CeO
2whereas the redox properties of the support would be
preferentially involved in the reduction of nitrates to nitrites. Based onthis, the changes observed on the rates and the selectivities recorded on
Pd/CZ(COP) and Pd/CZ(EISA) could be more rationalized in terms of
strength of metal-support interaction depending on the presence of
inhomogeinity modifying the electron density of palladium particles.5. Conclusion
This study reports the in fluence of the support materials on the
catalytic properties of Pd particles in the reduction of nitrites in aqu-eous phase performed at 20 °C. Ceria-zirconia mixed oxide
(Ce
0.5Zr0.5O2) as support was prepared according to co-precipitation
and evaporation-induced self-assembly methods and then calcined inair at 400 °C. Straightforward bulk and surface characterizations
showed a preferential tetragonal structure for the mixed-oxide prepared
from the EISA method exhibiting a surface zirconium enrichment
whereas the co-precipitation method leads to strong inhomogeneity in
composition with the segregation of the fluorite structure of CeO
2co-
existing with the mixed oxide. As a consequence some signi ficant de-
viations are observed on the surface composition with surface cerium
enrichment corresponding to a greater formation of anionic vacancies
reflected by higher Ce3+/Ce4+ratio on sample reduced at 300 °C. Such
structural changes have also some consequence on the textural prop-erties with monomodal pores size distribution on CZ(EIZA) whereas theCZ(COP) is characterized by a bimodal distribution with larger pores
likely associated to the formation of CeO
2. All those changes can alter
the impregnation of palladium which was found more poorly dispersedon CZ(COP) irrespective of the reduction temperature, i.e., 300 °C or
500 °C under pure hydrogen.
Catalytic measurements show a faster reduction of nitrites on
samples reduced at 500 °C with a prominent production of ammonia at
the early stage of the reaction at the expense of nitrogen the target
molecule. Unprecedented selectivity behavior is observed with a se-
quential ammonia consumption promoted on Pd/Ce
0.5Zr0.5O2irre-
spective of the preparation route. The changes observed in the reactionrates are strongly dependent on the pre-reduction leading to signi ficant
enhancements when the pre-reduction is performed at 500 °C. Thecomparison of TOF and interfacial rates expressed per cm, re flecting the
length of the perimeter of metallic Pd particles, versus the particle sizedoes not provide clear evidence to privilege the prevalence of structural
or electronic factors which could govern the catalytic properties but
highlight the superiority of 0.46Pd/CZ(EISA) compared to all other
samples. However, it seems that there is no strong evidence that the
support materials would directly participate to the reaction especially
on the EISA support stabilizing a tetragonal structure known for ex-
hibiting lower redox properties than the cubic form. This seems in re-
lative good agreement with previous investigation that showed nitrite
reduction would occur through classical mechanism involving only
metallic sites, the involvement of oxygen vacancies being more obvious
for the reduction of nitrates to nitrites.
Fig. 12. Comparison of TOF and interfacial rates vs. mean Pd particle size
on Pd/CZ.P. Granger et al. Applied Catalysis B: Environmental 224 (2018) 648–659
658

Acknowledgments
The authors greatly acknowledge the financial support from the
CCIFER – ANCS and the French Embassy (PhD fellowship awarded to S.
Troncéa who won the competition « Graine de Chercheur et Energie
Durable »). Chevreul Institute (FR 2638), Ministère de l ’Enseignement
Supérieur et de la Recherche, Région Nord –Pas de Calais and FEDER
are acknowledged for supporting and funding partially this work.Sandra Casale (UPMC, Paris VII) is acknowledged for conducting SEM
measurements.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in the
online version, at http://dx.doi.org/10.1016/j.apcatb.2017.11.007 .
References
[1]L.J. Puckert, Environ. Sci. Technol. 29 (1995) 408A –414A .
[2] U.S. Environmental Protection Agency (USEPA) (2008) Title 40, Part 141.54:425.
[3]K. Reddy, Water Res. 34 (2000) 995 –1001 .
[4]M. Al Bahri, L. Calvo, M.A. Gilarranz, J.J. Rodriguez, F. Epron, Appl. Catal. B
138 –139 (2013) 141 –148.
[5]F. Epron, F. Gauthard, C. Pineda, J. Barbier, J. Catal. 198 (2001) 309 –318.
[6]Y.N. Kim, M.Y. Kim, M. Choi, J. Chem. Eng. 289 (2016) 423 –432.
[7]D.P. Durkin, T. Ye, J. Choi, K.J.T. Livi, H.C. De Long, P.C. Turlove, D.H. Fairbrother,L.M. Haverhals, D. Shuai, Appl. Catal. B 221 (2018) 290 –301.
[8]S. Hörold, K.D. Vorlo, T. Tacke, M. Sell, Catal. Today 17 (1993) 21 –30.
[9]A. Suda, H. Sobukawa, T. Suzuki, T. Kandori, Y. Ukyo, M. Sugiura, J. Ceram. Soc.Jpn. 109 (2001) 177 –180.
[10] S. Letichevsky, C.A. Tellez, R.R. de Avillez, M.I.P. da Silva, M.A. Fraga, L.G. Appel,Appl. Catal. B 58 (2005) 203 –210.
[11] A.I. Kozlov, D.H. Kim, A. Yezerets, P. Andersen, H.H. Kung, M.C. Kung, J. Catal. 209
(2002) 417 –426.
[12] A. Pintar, J. Batista, J. Levec, T. Kajuchi, Appl. Catal. B 11 (1996) 81 –98.
[13] Y. Yoshinaga, T. Akita, I. Mikami, T. Okuhara, J. Catal. 207 (2002) 37 –45.
[14] M.-S. Kim, D.-W. Lee, S.H. Chung, J.T. Kim, I.-H. Cho, K.-Y. Lee, J. Mol. Catal. A 392(2014) 308 –314.
[15] J. Kugai, J.T. Miller, N. Guo, C. Song, Appl. Catal. B 105 (2011) 306 –316.
[16] X. Huo, D.J. Van Hoomissen, J. Liu, S. Vyas, Appl. Catal. 211 (2017) 188 –198.
[17] A.H. Pizarro, C.B. Molina, J.J. Rodriguez, F. Epron, J. Environ. Chem. Eng. 3 (2015)
2777 –2785 .
[18] J.K. Chinthaginjala, L. Le fferts, Appl. Catal. B 101 (2010) 144 –149.
[19] M.-S. Kim, S.-H. Chung, C.J. Yoo, M.S. Lee, I.-H. Cho, D.-W. Lee, K.-Y. Lee, Appl.
Catal. B 142 –143 (2013) 354 –361.
[20] N.
Barrabés, A. Da finov, F. Medina, J.E. Sueiras, Catal. Today 149 (2010) 341 –347.
[21] E.L. Crepaldi, G.J.D.A. Soler-Illia, A. Bouchara, D. Grosso, D. Durand, C. Sanchez,Angew. Chem. Int. Ed. 42 (2003) 347 –351.
[22] Q. Yuang, Q. Liu, W.G. Song, W. Feng, W.L. Pu, L.D. Sun, Y.W. Zhang, C.H. Yan, J.Am. Chem. Soc. 129 (2007) 6698 –6699 .
[23] M.A. Ebiad, Dalia R. Abd El-Ha fiz, R.A. Elsalamony, L.S. Mohamed, RSC Adv. 2
(2012) 8145 –8156 .
[24] S. Salasc, V. Perrichon, M. Primet, N. Mouaddib-Moral, J. Catal. 206 (2002) 82 –90.
[25] R. Di Monte, J. Kaspar, P. Fornasiero, A. Ferrero, G. Gubitosa, M. Graziani, Stud.Surf. Sci. Catal. 116 (1998) 559 –569.
[26] L. Xu, H. Song, L. Chou, Int. J. Hydrogen Energy 37 (2012) 18001 –18020.
[27] F. Zhang, C.H. Chen, J.C. Hanson, R.D. Robinson, I.P. Herman, S.W. Chan, J. Am.Ceram. Soc. 89 (2006) 1028 –1036 .
[28] M. Li, Z. Feng, G. Xiong, P. Ying, Q. Xin, C. Li, J. Phys. Chem. B 105 (2001)8107 –8111 .
[29] J. Serrano-Ruiz, J. Luettich, A. Sepúlveda-Escribano, F. Rodríguez-Reinoso, J. Catal.241 (2006) 45 –55.
[30] C. Tiseanu, V.I. Parvulescu, D. Avram, B. Cojocaru, M. Boutonnet, M. Sanchez-Dominiguez, Phys. Chem. Chem. Phys. 16 (2014) 703 –710.
[31] I. Atribak, B. Azambre, A. Bueno López, A. García-García, Appl. Catal. B 92 (2009)
126 –137.
[32] W. Mista, T. Rayment, J. Hanuza, L. Macalik, Mater. Sci. (Poland) 22 (2004)153 –170.
[33] B. Zhao, Q.Y. Wang, G.F. Li, R.X. Zhou, J. Alloys Comp. 508 (2010) 500 –506.
[34] B.M. Reddy, P. Bharali, P. Saikia, S.-E. Park, M.W. van den Berg, M. Muhler,W. Grünert, J. Phys. Chem. C 112 (2008) 11729 –11737.
[35] M. Yashima, H. Arashi, M. Kakihana, M. Yoshimura, J. Am. Ceram. Soc. 77 (1994)1869 –1874 .
[36] F.B. Passos, E.R. de Oliveira, L.V. Mattos, F.B. Noronha, Catal. Today 101 (2005)23–30.
[37] P.
Fornasiero, G. Balducci, R. Di Monte, J. Kaspãr, V. Sergo, G. Gubitosa, A. Ferrero,
M. Graziani, J. Catal. 164 (1996) 173 –183.
[38] M. Teng, L. Luo, X. Yan, Microporous Mesoporous Mater. 119 (2009) 158 –164.
[39] N.S. Priya, C. Somayaji, S. Kanagaraj, Proc. Eng. 64 (2013) 1235 –1241 .
[40] A. Trovarelli, J. Inorg. Chem. 20 (1999) 263 –284.
[41] L. Xu, H. Song, L. Chou, Int. J. Hydrogen Energy 37 (2012) 18001 –18020 .
[42] J. Kaspar, P. Fornasiero, M. Graziani, Catal. Today 50 (1999) 285 –298.
[43] C. Bozo, N. Guilhaume, E. Garbowski, M. Primet, Catal. Today 59 (2000) 33 –45.
[44] P. Fornasiero, J. Kaspar, M. Graziani, J. Catal. 167 (1997) 576 –580.
[45] B. Zhao, G.F. Li, C.H. Ge, Q.Y. Wang, R.X. Zhoua, Appl. Catal. B 96 (2010) 338 –349.
[46] H. Muraki, K. Yokota, Y. Fujitani, Appl. Catal. 48 (1989) 93 –105.
[47] G. Chen, W.-T. Chou, C.-T. Yeh, Appl. Catal. 8 (1983) 389 –397.
[48] I. Halasz, A. Brenner, M. Shelef, K.Y.S. Ng, Appl. Catal. A 82 (1992) 51 –63.
[49] F. Le Normand, J. El Fallah, L. Hilaire, P. Légaré, A. Kotani, J.C. Parlebas, SolidState Commun. 71 (11) (1989) 885 –889.
[50] M. Brun, A. Berthet, C.J. Bertolini, J. Electron Spectrosc. Relat. Phenom. 104 (1999)55–60.
[51] D. Briggs, M.P. Seah, 2nd ed., Practical Surface Analysis Vol. 1 John Wiley,
Chichester, 1999 (p. 599) .
[52] W. Huang, Z. Zuo, P. Han, Z. Li, T. Zhao, J. Electron Spectrosc. Relat. Phenom. 173(2009) 88 –95.
[53] E.M. Slavinskaya, O.A. Stonkus, R.V. Gulyaev, A.S. Ivanova, V.I. Zaikovskii,P.A. Kuznetsov, A.I. Boronin, Appl. Catal. A 401 (2011) 83 –97.
[54] K. Otto, L.P. Haack, J.E. de Vries, Appl. Catal. B 1 (1992) 1 –12.
[55] M. Moroseac, T. Skála, K. Veltruská, V. Matolín, I. Matolínová, Surf. Sci. 56 (2004)
1118 –1123 .
[56] X.
Huo, D.J. Van Hoomissen, J. Liu, S. Vyas, Appl. Catal. 211 (2017) 188 –198.
[57] A.H. Pizarro, C.B. Molina, J.J. Rodriguez, F. Epron, J. Environ. Chem. Eng. 3 (2015)
2777 –2785 .
[58] J.K. Chinthaginjala, L. Le fferts, Appl. Catal. B 101 (2010) 144 –149.
[59] F. Epron, F. Gauthard, J. Barbier, J. Catal. 206 (2002) 363 –367.
[60] J. Lee, Y.G. Hur, M.-S. Kim, K.-Y. Lee, J. Mol. Catal. A 399 (2015) 48 –52.
[61] F. Zhang, S. Miao, Y. Yang, X. Zhang, J. Chen, N. Guan, J. Phys. Chem. C 112 (2008)7665 –7671 .
[62] Y. Yoshinaga, T. Akita, I. Mikami, T. Okuhara, J. Catal. 207 (2002) 37 –45.
[63] B. Ileri, O. Ayyildiz, O. Apaydin, J. Hazard. Mater. 292 (2015) 1 –8.
[64] S.A. Ghom, C.Z. Zamani, S. Nazarpour, T. Andreu, J.R. Morante, Sens. Actuators B
140 (2009) 216 –221.
[65] J.K. Chinthaginjala, A. Villa, D.S. Su, B.L. Mojet, L. Le fferts, Catal. Today 183
(2012) 119 –123.
[66] C.M. Kalamaras, S. Americanou, A.M. Efstathiou, J. Catal. 279 (2011) 287 –300.
[67] Y. Wu, C. Dujardin, C. Lancelot, J.P. Dacquin, V.I. Parvulescu, M. Cabié, C.R. Henry,T. Neisius, P. Granger, J. Catal. 328 (2015) 236 –247.
[68] A. Miyazaki, T. Asakawa, Y. Nakano, I. Balint, Chem. Commun. (2005) 3730 –3732 .P. Granger et al.
Applied Catalysis B: Environmental 224 (2018) 648–659
659