Bimetallic Fe-Ni/SiO2 catalysts for furfural hydrogenation: Identification of the interplay between Fe and Ni during deposition-precipitation and thermal treatments

Article abstract of DOI:10.1016/j.cattod.2018.11.041

Supported Fe-Ni catalysts have been reported for their activity and selectivity in the hydrogenation of unsaturated organic molecules. However, the control of the size and composition of the bimetallic nanoparticles remains a bottleneck when oxide-supported catalysts are prepared by impregnation, and alternative procedures should be investigated. Starting with Ni(II) and Fe(II) sulfates as precursor salts, deposition-precipitation with urea (DPU) on SiO₂ in an inert atmosphere initially leads to the formation of an ill-crystallized Fe-containing Ni(II) 1:1 phyllosilicate, which reduces under hydrogen at 700 °C into bimetallic fcc Fe-Ni nanoparticles of 5.4 nm in average. Compared with the composition of the DPU solution (50 Fe at %, 50 Ni at %), an excess of Ni is detected on the catalyst (38 Fe at %, 62 Ni at %), due to the preferential reaction of Ni²⁺ ions with silica. In situ X-ray absorption spectroscopy and ⁵⁷Fe M?ssbauer spectroscopy show that the reduction of Fe ions to the metallic state is triggered by the formation of reduced Ni centers above 350 °C, and, from then, proceeds progressively, resulting in an excess of Fe in the outer shells of the bimetallic particles. The composition of individual Fe-Ni particles evidences a standard deviation of 8%. The bimetallic Fe-Ni/SiO₂ catalyst gives high yields in furfuryl alcohol in the hydrogenation of furfural, in contrast with an analog Ni/SiO₂ catalyst that favours side-reactions of etherification, hydrogenolysis and hydrogenation of the furan ring.

Full text of DOI:10.1016/j.cattod.2018.11.041

Accepted Manuscript
Title: Bimetallic Fe-Ni/SiO2 catalysts for furfural
hydrogenation: identification of the interplay between Fe and
Ni during deposition-precipitation and thermal treatments
Authors: Dichao Shi, Qifeng Yang, Christi Peterson,
´
´
Anne-Felicie Lamic-Humblot, Jean-Sebastien Girardon, Anne
Griboval-Constant, Lorenzo Stievano, Moulay T. Sougrati,
´
´
Valerie Briois, Paul A.J. Bagot, Robert Wojcieszak, Sebastien
Paul, Eric Marceau
PII:
S0920-5861(18)31011-3
DOI:
Reference:
https://doi.org/10.1016/j.cattod.2018.11.041
CATTOD 11775
To appear in:
Catalysis Today
Received date:
Revised date:
Accepted date:
8 July 2018
14 September 2018
20 November 2018
´
Please cite this article as: Shi D, Yang Q, Peterson C, Lamic-Humblot A-Felicie,
´
Girardon J-Sebastien, Griboval-Constant A, Stievano L, Sougrati MT, Briois V,
Bagot PAJ, Wojcieszak R, Paul S, Marceau E, Bimetallic Fe-Ni/SiO2 catalysts
for furfural hydrogenation: identification of the interplay between Fe and Ni
during deposition-precipitation and thermal treatments, Catalysis Today (2018),
https://doi.org/10.1016/j.cattod.2018.11.041
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

Bimetallic Fe-Ni/SiO2 catalysts for furfural hydrogenation:
identification of the interplay between Fe and Ni during deposition-
precipitation and thermal treatments
Dichao Shi a, Qifeng Yang b, Christi Peterson c, Anne-Félicie Lamic-Humblot c, Jean-
Sébastien Girardon a, Anne Griboval-Constant a, Lorenzo Stievano d, Moulay T. Sougrati d,
Valérie Briois e, Paul A. J. Bagot b, Robert Wojcieszak a, Sébastien Paul a, Eric Marceau a,c ,*
a Univ. Lille, CNRS, Centrale Lille, ENSCL, Univ. Artois, UMR 8181 - UCCS - Unité de Catalyse et
Chimie du Solide, F-59000 Lille, France
b Department of Materials, University of Oxford, Parks Road, Oxford OX1 3PH, United Kingdom
c Sorbonne Université, CNRS UMR 7197, Laboratoire de Réactivité de Surface, LRS, 4 place Jussieu,
F-75252 Paris cedex 05, France
d Institut Charles Gerhardt Montpellier, CNRS UMR 5253, Université de Montpellier, Pl. E. Bataillon,
F-34095 Montpellier cedex 5, France
e Synchrotron SOLEIL, CNRS-UR1, L’Orme des Merisiers, BP48, Saint-Aubin, F-91192 Gif-sur
Yvette, France
______________________________________________________________________
*Corresponding author: Dr. Eric Marceau, at the address above,
e-mail: eric.marceau@univ-lille.fr
1

Graphical abstract
Highlights
 Co-deposition of Ni(II) and Fe(II) by DPU requires the absence of oxidizing species.
 The result is an ill-crystallized Fe-containing Ni(II) 1:1 phyllosilicate phase.
 Fe reduction is triggered by Ni reduction but needs high temperatures to be complete.
 fcc Fe-Ni particles are formed with a standard deviation in composition of 8 at %.
 A surface enrichment in Fe is evidenced by Mössbauer spectroscopy.
Abstract
Supported Fe-Ni catalysts have been reported for their activity and selectivity in the
hydrogenation of unsaturated organic molecules. However, the control of the size and
composition of the bimetallic nanoparticles remains a bottleneck when oxide-supported
catalysts are prepared by impregnation, and alternative procedures should be investigated.
Starting with Ni(II) and Fe(II) sulfates as precursor salts, deposition-precipitation with urea
2

(DPU) on SiO2 in an inert atmosphere initially leads to the formation of an ill-crystallized Fe-
containing Ni(II) 1:1 phyllosilicate, which reduces under hydrogen at 700°C into bimetallic fcc
Fe-Ni nanoparticles of 5.4 nm in average. Compared with the composition of the DPU solution
(50 Fe at %, 50 Ni at %), an excess of Ni is detected on the catalyst (38 Fe at %, 62 Ni at %),
due to the preferential reaction of Ni2+ ions with silica. In situ X-ray absorption spectroscopy
57
and Fe Mössbauer spectroscopy show that the reduction of Fe ions to the metallic state is
triggered by the formation of reduced Ni centers above 350°C, and, from then, proceeds
progressively, resulting in an excess of Fe in the outer shells of the bimetallic particles. The
composition of individual Fe-Ni particles evidences a standard deviation of 8%. The bimetallic
Fe-Ni/SiO2 catalyst gives high yields in furfuryl alcohol in the hydrogenation of furfural, in
contrast with an analog Ni/SiO2 catalyst that favours side-reactions of etherification,
hydrogenolysis and hydrogenation of the furan ring.
Keywords
bimetallic catalysts; furfural; X-ray absorption spectroscopy; Mössbauer spectroscopy; in situ
measurements
3

1. Introduction
Noble metals have long been used in catalysis for the hydrogenation of organic
molecules, but their natural abundance is low and, consequently, their cost is high and subject
to large variations. Substitutes should be found among more abundant and cheaper transition
metals. Fe is abundant, but it is also poorly active and poorly stable. In contrast, Ni is known
for its activity, but it faces deactivation by carbon, and can favour hydrogenolysis or
overhydrogenation side-reactions. However, associating Fe and Ni in supported bimetallic
catalysts has been found to be beneficial in terms of activity and selectivity for reactions
between H2 and ketones [1,2], aldehydes [2-5], esters [6,7], phenols [8] or nitriles [1,9].
Fe25Ni75/SiO2 (formulations are herein given in molar proportions) was first mentioned by
Khumbar et al. to be more active than Ni/SiO2 in the hydrogenation of acetophenone and nitriles
[1]. More recently, Yu et al. have reported that 5-hydroxymethyl furfural was almost
quantitatively hydrogenated into the corresponding diol at 110°C with Fe-Ni catalysts, while
the selectivity was lower than 80% with a monometallic Ni catalyst [4].
The structure of reduced Fe-Ni nanoparticles is strongly dependent on the Fe/Ni ratio.
Fe-rich nanoparticles exhibit a bcc structure while Ni-rich alloys belong to the fcc type [10].
Furthermore, as recently reviewed by Tomishige et al. [11], the particles sizes are often not
uniform when the catalyst is prepared by co-impregnation, and can exceed 10 nm. The control
of the Fe/Ni ratio in individual nanoparticles is also difficult. Monometallic nanoparticles, and
bimetallic Fe-rich and Ni-rich nanoparticles, may co-exist in the catalyst [1,4,12-15]. If the
association between the two metals is not effectively controlled, it can be difficult to link the
best catalytic results to precise Fe and Ni proportions in the catalyst formulation.
The thermal treatment of mixed Fe-Ni phases such as hydrotalcites appears to lead to
well-defined systems [11,16,17]. In a few instances, deposition-precipitation has been
4

implemented to yield oxide-supported Fe-Ni nanoparticles of controlled size. The
decomposition of co-precipitated Fe and Ni cyanides allowed van de Loosdrecht et al. to attain
particles sizes smaller than 4 nm on TiO2 [18]. Deposition-precipitation involving the
decomposition of urea (DPU) was recently tested on alumina by Mutz et al. [17], resulting in a
particle size distribution ranging from 2 to 7 nm for the Fe25Ni75 formulation, but the speciation
of the metals in the first stages of the DPU process was not described.
DPU has been thoroughly investigated as a method to prepare well-dispersed
monometallic Ni nanoparticles on silica, even at metal loadings as high as 40 wt.% [19-22]. Its
extension to Fe-based bimetallic nanoparticles raises the question of the association between
the two metals upon deposition and thermal treatments. We discuss here the choice of the
synthesis parameters required to avoid early metal segregation during DPU, and we evidence
the interplay existing between Fe and Ni-containing species along the preparation of a Fe-
Ni/SiO2 catalyst. A combination of in situ characterization techniques (X-ray diffraction –
XRD; X-ray absorption spectroscopy – XAS; Mössbauer spectroscopy - MössS) is associated
to overcome the difficulty to distinguish the two poorly Z-contrasting metals, as conventional
transmission electron microscopy or EXAFS alone could not differentiate between two
elements with close numbers of electrons. Quick-X-ray absorption spectroscopy recorded
simultaneously at the Ni and Fe K edges in in situ conditions will provide information on the
evolution of the oxidation state and chemical environment of the two metals after DPU and
during reduction, while Mössbauer spectroscopy will give valuable data on the speciation of
ionic and metallic Fe at all stages of the preparation.
5

2. Experimental
2.1 Catalysts preparation
A Fe50Ni50/SiO2 catalyst (in which 50/50 represents the relative atomic proportions of
the two metals in the initial solution) was prepared using a modified DPU procedure originally
developed for Ni/SiO2 catalysts [19-23]. Sipernat-50 silica (Degussa; surface area, 400 m2 g-1,
pore volume, 1.4 cm3 g-1) was used as a support. Metal salts used were iron (II) sulfate
heptahydrate (Alfa Aesar, 98%), nickel (II) nitrate hexahydrate (Merck, 99%) and nickel (II)
sulfate hexahydrate (Aldrich, 99%). Other reactants were nitric acid (Prolabo, 70%), sulphuric
acid (Alfa Aesar, 93-98%) and urea (Alfa Aesar, 99-100%). All reactants were used without
further purification.
150 mL of aqueous solution of 0.84 mol L-1 urea, 0.02 mol L-1 H+ introduced via a strong
acid, 0.14 mol L-1 Ni(II) salt, 0.14 mol L-1 Fe(II) salt, were mixed with 1.14 g of silica in a
three-necked flask allowing Ar bubbling. The suspension was continuously stirred and slowly
heated to 80°C, resulting in the decomposition of urea, increase of the pH to 5.3, and deposition-
precipitation of metallic compounds onto the support. The suspension was kept at 80°C for 22
h. After cooling to room temperature (RT), the solid was recovered by centrifugation, washed
three times with distilled water and dried at RT. The choice of the reactants, their order of
introduction, and the conditions of washing, drying and storage are discussed in section 3.1.
The catalyst was reduced by heating in a flow of 10% H2/He (100 mL min-1) at 5°C min-1 up to
700°C (with a plateau of 2 h), without intermediate calcination.
A Ni100/SiO2 catalyst was prepared as a reference following the same procedure (Ni(II)
concentration in solution: 0.28 mol L-1). It was verified by transmission electron microscopy
that, in line with formerly reported results on this system [21,22], the sizes and distribution on
the support of metal particles were similar to those of Fe50Ni50/SiO2. Two reference systems
providing other Fe/Ni proportions (0.21 mol L-1 Fe(II) sulfate, 0.07 mol L-1 Ni(II) sulfate; 0.07
6

mol L-1 Fe(II) sulfate, 0.21 mol L-1 Ni(II) sulfate), and a reference Co-Ni/SiO2 system prepared
from 0.14 mol L-1 Ni(II) nitrate hexahydrate and 0.14 mol L-1 Co(II) nitrate hexahydrate (Sigma
Aldrich, 98%), were prepared for the sake of comparison, for the characterization of the dried
systems (section 3.1) and to support a discussion on metal contents (section 3.2), respectively.
2.2 Characterization
Fe, Ni and Si elemental analysis was performed by inductively coupled plasma-optic
emission spectroscopy using a 720-ES ICP-OES spectrometer (Agilent) with axial viewing and
simultaneous CCD detection. N2 physisorption isotherms at liquid nitrogen temperature were
carried out on a TriStar II Plus gas adsorption analyzer (Micromeritics) after activation at 150
°C in vacuum overnight. The specific surface area was evaluated with the Brunauer-Emmett-
Teller (BET) model over the range P/P0 = 0.05–0.30. The total pore volume was measured from
the volume of N2 adsorbed at P/P0 = 0.95. Temperature-programmed reduction (TPR) was
followed using an Autochem apparatus (Micromeritics) under H2 (5%)/Ar (flow rate: 50 mL
min-1; ramp rate: 7.5 °C min-1). Attenuated Total Reflection Infra-Red (ATR-IR) spectra were
recorded using a Nicolet Magna iS50 spectrometer equipped with a diamond crystal and a
DTGS detector (10 scans; resolution: 4 cm-1).
X-ray diffraction patterns (XRD) were recorded in ambient conditions using the Cu Kα
radiation ( = 1.5418 Å; 40 kV, 30 mA) on a Siemens D5000 diffractometer, with a 0.05° scan
step and 2 s time step. In situ measurements were conducted under H2 (3%)/N2 on a D8 Bruker
diffractometer equipped with an Anton Paar XRK900 reactive chamber. Data were acquired in
the Bragg-Brentano geometry, using the Cu Kα radiation (40kV, 40mA). Diffractograms were
recorded every 25 ˚C from 50 to 700 ˚C. A peak from the holder at 2 = 27.2° was used for
calibration between different experiments. The average size of metal nanoparticles was
evaluated by the Laue-Scherrer equation applied to the (111) reflection.
7

57Fe Mössbauer spectra (MössS) were measured with a source of 57Co embedded in a
rhodium metal matrix. Measurements were performed in the transmission mode, with the
source at room temperature. The measured sample, in the form of a self-supported pellet, was
mounted in an in situ oven working at controlled temperatures up to 700°C under diluted H2
flow. The gas flow was maintained during cooling to 25°C for recording of the spectrum at RT.
Low temperature spectra (-268 °C) were measured after transferring the sample pellet to a
helium gas flow cryostat in an Ar-filled glove box (O2, H2O < 1 ppm), strictly avoiding any
contact with ambient air. The spectrometer was operated with a triangular velocity waveform,
and a gas-filled proportional counter was used for the detection of the gamma rays. Velocity
calibration was performed with an -Fe foil. The isomer shift δ, the electric quadrupole splitting
, the magnetic hyperfine field BHF, the linewidth at half maximum  and the relative
absorption area of the different components were determined by fitting the spectra with the
appropriate combinations of Lorentzian profiles.
Time-resolved XAS data were measured in the transmission mode at the ROCK
beamline [24] using the SOLEIL home-made Quick-EXAFS monochromator equipped with a
Si (111) channel-cut crystal and described in ref. [25]. The channel-cut oscillates with a ± 1.95°
amplitude around 14.6 ° and with a frequency of 2 Hz, allowing simultaneous recording of the
Fe (7112 eV) and Ni (8333 eV) K-edge spectra in 250 ms. Successive spectra collected with
upward Bragg angles during the in situ monitoring of catalysts were merged to improve the
signal-to-noise ratio, leading to a time resolution of 5 s per spectrum effectively analyzed.
Ionization chambers provided by Ohyo Koken Kogyo Co. Ltd. filled with a 2:1 mixture of N2
and He were used for measurements. The beam size was 2.5 x 0.5 mm. Normalization of spectra
was performed by using the Python normal_gui graphical interface developed at SOLEIL for
the fast handling of Quick-XAS data. The proportions of the different Fe and Ni species were
determined by a multivariate data analysis, the MCR-ALS method [26-28]. MCR-ALS GUI 2.0
8

developed by the Tauler group [29], and freely available as a Matlab toolbox, was used for
MCR-ALS minimization. The EXAFS signal extraction and Fourier transformation of the
EXAFS spectra were done using the Athena graphical interface software [30]. EXAFS fitting
of distances, coordination numbers and Debye–Waller factors was performed with the Artemis
interface to IFeFFIT using least-squares refinements [31], firstly checked on NiO, Ni metallic
foil and Fe2O3 references. Fourier-transformed EXAFS signals are presented without phase
correction. The temperature-programmed reduction of the Fe50Ni50/SiO2 sample was performed
using a dedicated cell [32] connected to the gas-feeding system installed on the ROCK
beamline. The cavity of the sample holder (2 mm thick) was filled with the powder catalyst (20
mg) pre-mixed with boron nitride (BN, 10 mg). The in situ catalyst reduction was performed
by heating the cell from RT to 500 °C under a H2 (5%)/He flow (50 mL min-1) with a heating
rate of 5°C min-1, followed by 2 h of isothermal treatment.
TEM observations were made at University of Oxford, using a JEOL JEM-2100
microscope fitted with a LaB6 source (JEM-2100) for bright field observation, and on a JEM-
3000F FEGTEM microscope (JEM-3000) fitted with a spherical aberration STEM probe
corrector and a Centurion EDX detector equipped with a super atmospheric thin window
(SATW) detector and an INCA processor, for elemental analysis. E-beam size was selected to
1.2 nm for EDX point scan measurements on individual particles.
2.3 Catalytic tests
Fe50Ni50/SiO2 and Ni100/SiO2 catalysts were tested in the hydrogenation of furfural using
a Screening Pressure Reactor (SPR) system from Freeslate equipped with 24 parallel stainless
steel batch reactors of 6 mL each. 10 mg of catalysts, reduced at 700°C (section 2.1), were
placed after transfer in ambient air into each of the 24 reactors and exposed in situ to pure H2
at 400°C (1 bar, 2 h) prior to reaction. 2 mL solutions of furfural (0.20, 0.56 or 1.00 mmol) in
9

isopropanol were then injected under the controlled atmosphere of a glove box. Tests were
performed at 150°C for 2 h under 20 bar H2 and a continuous stirring of 700 rpm. Reaction
products were analyzed using Shimadzu GC-2010 Plus and GC-FID-MS-QP2010 Ultra EI gas
chromatographs, equipped with a FID detector, a ZB-5MS column (30 m × 0.25 mm × 0.25
μm), and for the latter, with a QP2010 Ultra mass spectrometer. It was verified by quantifying
acetone that in these experimental conditions, hydrogen transfer from isopropanol to furfural
did not occur to a significant extent (less than 3% of converted furfural).
3. Results and discussion
3.1 Choice of the experimental parameters for the synthetic procedure and
characterization of dried catalysts
Given the solubility products of iron and nickel hydroxides (pKs Fe(OH)3 = 38.8;
Fe(OH)2 = 15.1; Ni(OH)2 = 15.2 at 25°C), Fe2+ ions are expected to deposit in the same pH
range as Ni2+ ions, and Fe3+ ions at much lower pH. Fe(II) sulfate was thus selected as the iron
precursor salt for DPU, in preference to Fe(III) nitrate. In spite of this, initial attempts of DPU
with nickel nitrate as precursor salt and nitric acid as acidifier led to the oxidation of Fe(II) to
Fe(III) upon heating, as evidenced by the change of colour of the suspension from green to
orange, and to the early, separate precipitation of well-crystallized goethite FeIIIO(OH) at pH
4.8, from which large particles of metallic bcc Fe would later grow upon reduction (diffraction
peaks at 2 = 44.8 and 65.0°, (110) and (200) reflections; size: 40 nm). In order to eliminate
oxidizing species, nickel (II) sulfate and sulphuric acid were instead used as reactants, but the
preparation still suffered from a lack of reproducibility. Diffraction peaks of Fe(III)-containing
layered hydroxysulfate phases [33-35] were observed after drying, with ulterior formation of
large Fe-Ni particles upon reduction (Fig. S1, Supplementary Material). It was next chosen
to degas with Ar the suspension containing silica, sulphuric acid, nickel (II) sulfate and urea for
10

2 h, and to introduce Fe (II) sulfate only after degassing. Ar bubbling was maintained during
the whole heating and cooling steps. In these conditions, deposition took place at pH 5.3 and
the suspension remained green. Centrifugation and washing were also carried out with degassed
water.
Drying at RT in ambient air caused a rapid change of colour of the solid from green to
ochre, evidencing the oxidation of Fe(II) to Fe(III). Such oxidation is prone to happen to Fe(II)-
containing layered hydroxides [36-38]; it has been shown that the presence of Ni(II) ions
inhibits their further transformation to Fe3O4 [39,40]. Drying the solid under Ar flow in the
three-necked flask, and storage under Ar, preserved the green colour longer, but oxidation of
iron to Fe(III) was nevertheless seen to develop over a period of six weeks. Using XRD and
XAS, it was checked by reducing dried solids either kept under Ar or aged in air that, unlike Fe
oxidation during DPU, Fe oxidation upon drying and storage had no detrimental effect on the
formation, nature and size of the reduced metal nanoparticles.
The XRD pattern of all dried solids exhibits a broad diffraction peak at about 22˚,
characteristic of amorphous siliceous solids, a broad asymmetrical line at 2 = 34˚ and a narrow
symmetrical one at 2 = 60˚ (Fig. S1). As previously shown for Ni/SiO2 catalysts prepared by
DPU, the latter two peaks can be assigned to structures based on brucitic layers, either MII(OH)2
turbostratic hydroxides or ill-crystallized phyllosilicates [19].
MössS confirms the exclusive presence of octahedral Fe(III) ions in dried Fe50Ni50/SiO2
after exposure to air (Fig. 1a, Table 1). The signal recorded at RT is asymmetric and must be
fitted with at least two different quadrupole split contributions, as is often done for ions present
in a continuum of environments. Given the relatively narrow range of isomer shifts and
quadrupole splittings for octahedral Fe(III), it is virtually impossible to distinguish between
Fe(III) oxyhydroxides and silicates [41]. However, the absence of a sextet allows discarding
the presence of well-crystallized goethite FeOOH, which is expected to exhibit magnetic order
11

at room temperature [42,43]. Moreover, the relatively low isomer shift of the narrow doublet,
close to that measured for octahedrally coordinated iron in mixed metal Ni-Fe ferrites, suggests
the formation of mixed-metal species [44]. To test this hypothesis, Mössbauer spectra were
recorded on dried solids prepared following the same protocol but with two other Fe and Ni
nominal contents (Fig. S2, Table S1). The significant changes in the quadrupole splittings, as
well as the slight decrease of the average isomer shift when changing the metal contents,
suggests that Fe(III) ions are co-deposited with Ni(II) ions. Finally, electric quadrupole
splittings are noticeably larger here than in well-defined oxyhydroxides and silicates, especially
that of the second doublet. This is usually interpreted as the sign of distorsions in the chemical
environment: distorted octahedra, surface sites, proximity to tetrahedral sites or to divalent
cations, such as Ni2+, in hydroxides or silicates [39,42,45-47].
According to the analysis of EXAFS oscillations, the local environment of both Fe3+
and Ni2+ ions is consistent with the octahedral sites of brucitic layers, common to the structures
of divalent hydroxides and phyllosilicates [19] (Fig. 2, Fig. S3, Table 2). The first shell is
satisfactorily fitted using 6 oxygen atoms (R = 1.99 Å for Fe3+, R = 2.05 Å for Ni2+), probably
in a more distorted symmetry in the case of Fe(III) as shown by the higher value of the Debye-
Waller factor. The best fits for the second shell of neighbours involve 6 Ni atoms at R = 3.11
Å, characteristic of edge-sharing octahedra (it must be noted though that, due to the closeness
of scattering phase and amplitude functions for Fe and Ni, the nature of the second neighbours
as backscatterers cannot be unambiguously discriminated by EXAFS). If existing, iron
environments involving two very different Fe-Fe distances above 3 Å, as encountered in
FeOOH phases [46,48,49], are not predominant.
3.2 Evolution and composition of the phases deposited during DPU
12

The nature of the deposited phases and the metal contents were investigated as a
function of DPU duration, from 1 to 22 h. While the X-ray diffractograms did not exhibit major
changes besides the gradual broadening of the large halo associated to silica (Fig. S4), the
evolution of ATR-IR bands (Fig. 3a) revealed that, following the pattern reported for
monometallic Ni/SiO2 catalysts [19], the brucitic MII(OH)2 phase formed during the first 4 h of
DPU evolved into an ill-crystallized 1:1 phyllosilicate. This is proved by (1) the progressive
growth of the asymmetrical doublet of bands at 1007 and 1083 cm-1 (υ(Si-O) stretching
vibration in a 1:1 phyllosilicate); (2) the decrease of the weak asymmetrical υ(Si-OH) band
originating from silica at 974 cm-1; (3) the shift and decrease of the intensity of the υ(Si-O)
vibration band assigned to silica at ca. 800 cm-1; (4) the decrease of the (OH) band at 640 cm-
1, characteristic of turbostratic MII(OH)2, between 2 and 4 h, and, (5) from 4 h, the increase of
a broad band at 664 cm-1 assigned to a tetrahedral Si-O mode in the 1:1 phyllosilicate structure.
The poor crystallinity of the phyllosilicate phase may explain the absence of detection of Si
second-shell neighbours in the EXAFS fits.
ICP measurements indicate that, despite the equimolarity between Fe and Ni in solution,
nickel is the main metal initially depositing during the first hours (Fig. 3b). After a moderate
drop between 1 and 4 h, the nickel content increases by a two-fold between 4 and 8 h, when the
phyllosilicate phase forms. In contrast, the Fe content steadily increases throughout the process.
As a result, a significant excess of nickel with respect to iron is present after 22 h DPU in the
dried solid, compared with the atomic proportions in solution: 38 Fe at % and 62 Ni at % are
measured by ICP on the solid, corresponding to 42% of the amount of Fe and to 67% of the
amount of Ni initially introduced in solution.
This excess of nickel does not originate from its early introduction in the DPU solution,
as it was also observed when nickel and iron sulfates were added together in solution after Ar
degassing. For the sake of comparison, a reference system was prepared from nickel and cobalt
13

(II) nitrates (nominal proportions: 50 Co at % and 50 Ni at %), without degassing and separate
introduction of the reactants. The atomic proportions given by EDX measurements (34 Co at
%, 66 Ni at %) prove that nickel tends to react with silica preferentially, compared with the
other metal.
Finally, no IR band assigned to isocyanates or carbonates was observed in the spectra.
A shoulder at 607 cm-1 could be assigned to sulfate ions (4 vibration mode [50,51]), but it was
verified by ICP and XPS that sulfate ions were not detected on the dried solid.
In summary, the early precipitation of Ni(OH)2 onto silica during the first hours of DPU
is followed by the formation of a poorly organized 1:1 phyllosilicate, in which, after drying,
Fe3+ and Ni2+ ions occupy the octahedral layer. The (Fe+Ni)/Si ratio of 1.3 after 22 h of DPU
is consistent with a 1:1 phyllosilicate stoichiometry (M/Si ratio varying between 1 and 1.5 for
exclusively trivalent and divalent cations, respectively). In the absence of carbonates or sulfates,
charge compensation can be ensured by deprotonation of the layers, and formation of O2- or
OH- anions formed upon oxidation of iron during drying and storage. FeOOH as a separate
phase is not predominant, but the existence of minor domains cannot be excluded as MössS
evidences a distribution of environments around Fe3+ ions.
3.3 Reduction process
The temperature-reduction profile of Fe50Ni50/SiO2 (after storage in ambient air and
drying at 60°C to eliminate adsorbed water) evidenced three main reduction stages (Fig. S5): a
minor one between 270 and 350°C, and two major ones between 350 and 500°C, and between
500 and 700°C, respectively. In situ X-ray absorption spectroscopy and Mössbauer
spectroscopy were used to link these events to the reduction of Fe and Ni-containing species.
14

On the basis of the TPR profile, 700°C was selected as the reduction temperature at which both
metals were expected to be quantitatively reduced.
The reduction of air-stored Fe50Ni50/SiO2 was monitored simultaneously at the Ni and
Fe K edges by in situ XAS up to 500°C, the upper temperature limit allowed by the set-up (Fig.
4a and b). In line with the presence of a set of isosbestic points on the series of spectra recorded
at the Ni K edge, the quantitative analysis by MCR-ALS procedure (whose principle is recalled
in the Supplementary Material) requires two spectral components only (Fig. 4c, Fig. S6), that
recall the experimental spectrum of Ni2+ ions in the initial phyllosilicate, and that of metallic
fcc Ni as confirmed by the EXAFS analysis of the final state (Fig. 2b and c, Fig. S3b and c,
Table 2). The reduction of Ni2+ ions starts above 350°C, with an inflection point at 400°C.
Nickel reduction is complete after a plateau of 1 h at 500°C. The absence of an intermediate
dehydrated phase in the analysis is justified by the higher temperature range at which nickel
phyllosilicate would decompose to NiO (500-600°C) [22].
The progress of iron reduction, deduced from the spectra recorded at the Fe K edge, is
more complex to analyse. The lack of isosbestic points indicates that more than 2 components
are involved in the reduction, and the EXAFS signal of the final state could not be satisfactorily
fitted with a limited set of parameters. It was thus chosen in a first approach to follow the
decrease in energy of the Fe edge position, correlated to a decrease of the Fe oxidation state
(Fig. 4d). The position in energy of the edge was interpreted with respect to reference
compounds: hematite Fe2O3, Fe(II) sulfate, Fe foil. Fe3+ ions originally present inside the
layered structure underwent an accelerated reduction to Fe2+ in the 275-350 °C temperature
range. A more gradual decrease of the edge value above 350°C, starting when Ni reduces to the
metallic state, can be interpreted as a progressive reduction of Fe2+ iron to Fe metal triggered
by Ni reduction. A comparison between the final XANES spectrum and the spectrum of
15

reference metallic Fe shows that the reduction of Fe2+ is not complete at the end of the plateau
at 500°C.
The MCR-ALS analysis of Fe reduction provides five independent spectral components
(Supp. Mater., Fig. S7 and S8). The white lines of the first two components are identical, and
the EXAFS oscillations are found at the same energies, though damped for the second
component. The third component appears between 275 and 358°C. As seen above, the lower
edge energy is associated to the Fe(II) oxidation state. The EXAFS oscillations are similar to
those of the first component, though shifted to lower energies. The Fourier transforms of the
first three components resemble each other, apart from the decrease of intensity of the peaks
due to a temperature-induced increase of the Debye-Waller parameter. The Fourier transform
of the third component also differs from those recorded on oxides such as Fe3O4. This suggests
that the set of the first three components is related to the same phase, i.e., the Fe-containing
phyllosilicate, the second component reflecting mostly an increase of disorder with temperature
(data related to the first two components are provided in Fig. S9 and Table S2), and the third
component a reduced Fe2+ state of iron ions in the layered phase. The first event observed on
the TPR profile is thus linked to the reduction of Fe3+ to Fe2+.
The third component starts disappearing when nickel reduction starts, while the fourth
component appears exactly in parallel with the formation of metallic nickel. The position in
energy of the edge of the fourth component lets one assume that it is associated to Fe2+ ions,
though in a strongly altered environment: the shape of the first peak in the Fourier transform,
which seems to contain both Fe-O and Fe-Fe (or Fe-Ni) contributions, suggests both the
existence of an oxidic phase, and a proximity to metal particles. The last component, whose
weight increases gradually during the plateau at 500°C, gives a well-defined Fourier transform
identical to that of metallic Ni, and is interpreted as Fe(0) atoms located in the fcc environment
of Ni-containing nanoparticles. In view of this last result, the fourth component may be
16

understood as a Fe(II)-containing phase external to the Fe-Ni particles, from which Fe reduces
and diffuses into the bimetallic nanoparticles upon increasing temperature and time. These
processes and the concomitant reduction of Ni2+ to Ni(0) explain the second reduction stage on
the TPR profile.
The evolution of the X-ray diffractograms of Fe50Ni50/SiO2 during reduction under H2
is presented as a function of temperature on Figure 5. Only the diffraction peaks of the
phyllosilicate phase are visible up to 350°C. In accordance with XAS, three diffraction peaks
characteristic of fcc metal nanoparticles ((111), (200) and (220) reflections) are detected above
350°C and their intensity increases when temperature increases up to 700°C. No peak related
to bcc phases is observed (in particular the (200) reflection expected around 2 = 65.0°). The
average size of the crystalline domains increases from 3 to 5 nm from 350 to 700°C. The
diffraction peaks attributed to the phyllosilicate phase disappear above 500°C.
The unit cell parameter a is known to decrease linearly with increasing Ni content in
bimetallic fcc Fe-Ni alloys [52], as the atomic radius of Fe (0.172 nm) is larger than that of Ni
(0.162 nm). A linear Vegard correlation between a, measured at RT, and the Ni atomic content
was drawn from the JCPDS files of reference Fe-Ni alloys (Fig. S10); an uncertainty of 1 x 10-
4 nm on the value of a leads to an uncertainty of 1 Ni at % in the composition of the diffraction
domains. After correction from the temperature-induced shifts measured for monometallic
Ni100/SiO2 in the same conditions (Fig. S11 and S12), the positions of the (111) and (200)
reflections for Fe50Ni50/SiO2 give access to the values of a at RT after reduction at 500 and
700°C (0.3547 and 0.3557 nm, respectively) and, through the correlation established above, to
an assessment of the composition of the crystalline core of the particles (Fe22Ni78 and Fe31Ni69,
respectively). The latter value of a is confirmed after measurement back at RT under H2 (0.3555
nm, Fe29Ni71).
17

The 57Fe Mössbauer spectrum of the sample reduced at 500°C (Fig. 1b, Table 1) can
be interpreted in terms of three components. The isomer shift of the first two large quadrupole
split components is typical of high-spin Fe(II) ions, in line with XAS results. The component
with the larger quadrupole splitting is consistent with Fe(II) ions in silicates, since the hyperfine
parameters correspond to their typical range [53]. The quadrupole splitting of the second
component, however, is too small to correspond to a bulk silicate, and can be attributed to Fe(II)
ions in extremely distorted octahedral sites, or to lower coordination numbers. The third
contribution, covering 57% of Fe, consists of a singlet typical of Fe(0), with an isomer shift at
 = 0.06 mm.s-1 slightly more positive than that of iron metal, as reported for Fe-Ni alloys ( =
0-0.06 mm.s-1 at 25°C, the value of  increasing with the Ni content [54-56]). Based on the
overall composition determined by ICP (38 Fe at %, 62 Ni at %), the composition of the Fe-Ni
particles at 500°C is evaluated as Fe22Ni78, in good agreement with that determined by in situ
XRD. The absence of a sextet shows that the Fe-Ni nanoparticles are superparamagnetic at RT,
thus exhibiting a size smaller than 10 nm [55,56].
The third reduction stage on the TPR profile can be explained by comparing the
Mössbauer spectra recorded at 500 and 700°C. Reduction at 700°C (Fig. 1c and d, Table 1)
57
produces a complete modification of the Fe Mössbauer spectrum, which can now be fitted
with only two spectral components, a singlet and a quadrupole doublet. Both components are
often observed for reduced, superparamagnetic iron nanoparticles, and are usually attributed to
bulk and surface iron atoms, respectively [57]. The singlet at  = 0.04 mm.s-1 corresponds to
Fe(0) at the core of Fe-Ni nanoparticles. The decrease of  suggests an environment richer in
Fe compared with the reduced state at 500°C [54]. Since the hyperfine parameters of the
quadrupole doublet can be confused with those of Fe(III), the same sample was measured at -
268°C after transfer in a helium-flow cryostat. The low-temperature spectrum is dominated by
a magnetically split contribution, that can be fitted with at least two magnetic sextets,
18

comprising 90% of total Fe. The isomer shift, once taken into account the second-order Doppler
shift, and the magnetic field of these sextets are in the typical range of Fe(0) in Fe-Ni alloys
[58]. The two hyperfine fields are close to those measured for fcc Fe-Ni alloys in which Ni or
Fe is the major metal (29 and 33 T, respectively [59]). In addition to these two components, a
minor quadrupole split component accounting for about 10% of the total resonance area is still
present at the centre of the spectrum. At this stage, it can be interpreted as Fe(0) species not
ordered magnetically, either located at the periphery of bimetallic particles, or in highly
dispersed monometallic Fe nanoparticles.
3.4 Characterization of the metal nanoparticles by TEM and EDX after reduction at
500 and 700°C
After reduction at 700°C, the Fe50Ni50/SiO2 catalyst exhibited a specific surface area of
240 m2 gcatalyst-1, an apparently low value compared with pristine silica, but explained mainly
by the high metal loading ((Fe+Ni) loading = 52 wt%) and by the deep transformation of silica
during DPU. The pore volume was 0.6 cm3.g-1.
The metal nanoparticles on Fe50Ni50/SiO2 were characterized by bright field TEM after
reduction at 500 or 700°C, and exposure of the samples to ambient air (Fig. 6a and b). After
reduction at 500°C, they appeared as evenly distributed over the siliceous matrix and exhibited
an average diameter of 5.1 nm (standard deviation: 0.9 nm), with a high proportion of smaller
particles (Fig. S13a). Agglomeration was not observed after reduction at 700°C, and the
bimetallic nanoparticles were still evenly distributed on the silica substrate despite the high
metal loading, with a separation distance of 5 nm on average. The size distribution histogram
shows that the average size, 5.4 nm, has only slightly increased between 500 and 700°C (Fig.
19

S13b). The standard deviation was 1.3 nm and the size distribution was more symmetrical, with
a lower proportion of smaller particles.
The elemental distribution of Fe and Ni was examined by EDX mapping after reduction
at 500°C (Fig. 6c). No metal was detected on bare zones of the support. The composition of 28
individual particles was measured, resulting in an average composition of Fe31Ni69, and a
standard deviation of 8 at % (Tab. S3). Both elements were present in every particle and Fe
was never the major metal. It was not possible to perform point scans along the diameter of
particles, owing to their small size, or to evidence local metal segregation. Moreover, the
comparison with in situ measurements is not straightforward, as the distribution of the metals
may be perturbed by metal oxidation upon exposure to air [17].
DPU applied to the Fe-Ni system thus yields bimetallic nanoparticles of controlled size,
in the same range as for monometallic Ni/SiO2 catalysts [21], and resistant to sintering at high
temperature. The reduction above 350°C of the nickel ions from the phyllosilicate phase drives
the progressive destruction of the layered structure and the reduction of surrounding iron ions
to the metallic state, to form Ni-rich Fe-containing fcc nanoparticles. At 500°C, about 40% Fe2+
ions are still present in an oxidic environment, in close association with the metal nanoparticles.
Their gradual reduction between 500 and 700°C results in the diffusion of more Fe atoms inside
the Fe-Ni particles. The composition of the particles, consistent with the values presented in
sections 3.1 and 3.2, pertains to a limited range, with a standard deviation of 8 at %. In the
absence of monometallic and Fe-rich particles, both the detection of non-magnetically ordered
Fe by MössS, and the different values of the magnetic hyperfine field, suggest that a gradient
of Fe concentration exists in the bimetallic particles, as formerly postulated by Nagorny and
Bubert [59]. Due to the late reduction of Fe, the outer shells surrounding the bimetallic Fe-Ni
crystalline core would be enriched in Fe, in line with former interpretations from the literature
based on spectroscopic data and on beta emission from radioactive Ni [14,59-61].
20

3.5 Catalytic tests
Depending on the catalyst nature and reaction conditions, the hydrogenation of furfural
yields a slate of products resulting from the hydrogenation of the aldehyde function (furfuryl
alcohol, FFA), hydrogenation of the furan ring (tetrahydrofurfuryl alcohol, THFFA) and/or
hydrogenolysis of the C-O bond (2-methylfuran, MF; 2-methyltetrahydrofuran, MTHF)
(Scheme 1). In the present work, ethers resulting from reactions with the isopropanol solvent
were also identified by GC/MS (2-(isopropoxymethyl)furan, iPrOMF, m/z 81, 97, 139; 2-
(isopropoxymethyl)tetrahydrofuran, iPrOMTHF, m/z 71, 85, 101, 144). Etherification can
proceed with FFA or THFFA as a reaction intermediate, or via the direct reductive reaction of
furfural, and can be catalyzed by unreduced metal ions acting as Lewis acidic centers [62,63].
Fe50Ni50/SiO2 and Ni100/SiO2 catalysts were pre-reduced at 700°C, and underwent a
second stage of reduction at 400°C prior to reaction to counter the effect of transfer in air. A
characteristic TEM bright field micrograph of Ni100/SiO2, and the size histogram associated to
this catalyst, are presented on Fig. S14. The metal dispersion D (%) can be calculated for the
two catalysts by applying 97.1/Ø (nm) [64] to each class of diameter Ø from the two histograms
(Fig. 6b and S14), assuming that metal particles are hemispherical and that this calculation,
usually applied to fcc Ni, can be extended to fcc Fe-Ni alloys. A value of D = 20% is obtained
both for Fe50Ni50/SiO2 and Ni100/SiO2. The metal surface area exposed per gram of catalyst is
evaluated as 70 m2 gcat-1, assuming that one Fe or Ni atom occupies 6.3 Å2 [64].
In the given experimental conditions, both catalysts showed high activity in the
hydrogenation of furfural at 150°C (conversion: 97-100%), whatever the starting amount of
furfural in solution (Table 3). In contrast, yields were highly impacted by several factors.
Monometallic Ni100/SiO2 was rather selective to THFFA and MTHF, the over-hydrogenated
21

products, in good correlation with the literature in similar conditions of temperature and
pressure [5,65,66]. The formation of MF and MTHF by hydrogenolysis was favoured when the
initial amount of furfural was high. Etherification of the alcohols appeared prominent. As a
result, the yield of FFA from the hydrogenation of the carbonyl group remained very low, which
can be explained by its conversion to MF by hydrogenolysis.
Unlike monometallic Ni100/SiO2, the Fe50Ni50/SiO2 catalyst exhibited a high yield in
FFA whatever the initial amount of furfural, and low yields of THFFA, MF and MTHF. The
carbon balance was high, demonstrating that lower amounts of heavy products were formed
than with the monometallic Ni system. The fraction of ether formed from FFA slightly
increased with the amount of furfural, but remained minor except when the catalyst did not
undergo a pre-reduction at 700°C and thus contained iron mainly in an ionic form. As has been
reported in the literature, the hydrogenation of the carbonyl group reaction prevails on the
hydrogenation of ring and hydrogenolysis reactions on Fe-Ni catalysts [4,5]. The role of Fe
surface atoms in the suppression of Ni overhydrogenating and hydrogenolyzing properties is
thus confirmed here.
4. Conclusions
The preparation of bimetallic Fe-Ni/SiO2 catalysts by deposition-precipitation with urea
requires the use of a Fe(II) salt and the absence of oxidizing species such as nitrates or dissolved
oxygen. A layered structure best described as an ill-crystallized Fe-containing Ni(II) 1:1
phyllosilicate is formed after 22 h of DPU, with a lower proportion of Fe compared with the
composition of the DPU solution (38 Fe at %, 62 Ni at %, against 50 Fe at %, 50 Ni at %), due
to a preferential reaction of Ni2+ ions with silica. The oxidation of Fe2+ to Fe3+ upon drying and
storage does not challenge the subsequent formation of bimetallic particles upon reduction.
22

Reduction under H2 proceeds in several stages: (1) reduction of Fe3+ ions to Fe2+ above 275°C;
(2) reduction of Ni2+ to Ni(0) above 350°C, triggering the progressive reduction of Fe2+ ions to
the metallic state, and the insertion of a fraction of Fe atoms into the fcc structure of Ni-rich
particles; (3) deeper reduction of Fe2+ ions between 500 and 700°C, resulting both in the
diffusion of Fe atoms in the Fe-Ni core of the fcc particles, and in the enrichment in Fe of their
outer shells. After reduction at 700°C, the bimetallic particles are homogeneous in size (average
size: 5.4 nm) despite the elevated reduction temperature, and exhibit a standard deviation of
composition of 8 Fe at %. The enrichment in Fe of the surface of the Fe-Ni nanoparticles
suppresses hydrogenolysis and overhydrogenation side-reactions observed for Ni/SiO2 in the
hydrogenation of furfural, and provides a catalyst selective in furfuryl alcohol. Side-reactions
of etherification with the isopropanol solvent may be linked to the presence of unreduced metal
ions acting as Lewis acidic sites.
Acknowledgements
This work has benefited from the support of the CSC-Ecole Centrale de Lille PhD
scholarship program (D. S.), of the REU exchange program from the National Science
Foundation (C. P.), and of the French National Research Agency (ANR), through the
NobleFreeCat project (ANR-17-CE07-0022). Chevreul Institute (FR 2638), Ministère de
l’Enseignement Supérieur et de la Recherche, Région Hauts-de-France and FEDER are
acknowledged for supporting and funding partially this work.
Joelle Thuriot-Roukos (ICP), Jean-Charles Morin (ATR-IR), Olivier Gardoll (TPR),
Svetlana Heyte (SPR/GC-MS), Pascal Roussel and Laurence Burylo (in situ XRD) at UCCS,
as well as Alban Charlier (preparation of the reference Co-Ni catalyst) at LRS, are thanked for
their help.
23

XAS experiments were performed on the ROCK beamline, using the Chemistry support
lab, at SOLEIL Synchrotron, France (proposal number 20151218). The ROCK beamline is
benefiting from a public grant overseen by the French National Research Agency (ANR) as a
part of the ‘Investissements d’Avenir’ program (PIA) with the contractual reference ‘ANR-10-
EQPX-45’. Camille La Fontaine is thanked for his help on the beamline. The authors
acknowledge the European microscopy network ESTEEM 2 for funding access to the
microscope platform in Oxford (project 20150908), and Dr. Neil P. Young (University of
Oxford) for his help.
The REALCAT platform is benefitting from a state subsidy administrated by the French
National Research Agency (ANR) within the frame of the ‘Future Investments’ program (PIA),
with the contractual reference ‘ANR-11-EQPX-0037’. The European Union, through the ERDF
funding administered by the Hauts-de-France Region, has co-financed the platform. Centrale
Lille, the CNRS, and University of Lille as well as the Centrale Initiatives Foundation, are
thanked for their financial contributions to the acquisition and implementation of the
equipment.
The authors declare no competing interests.
References
[1] P. Khumbar, M.R. Kharkar, G.D. Yadav, R.A. Rajadhyaksha, J. Chem. Soc., Chem.
Commun. (1992) 584-586.
[2] G. Chieffi, C. Giordano, M. Antonietti, D. Esposito, J. Mater. Chem. A 2 (2014) 11591-
11596.
24

[3] S. Sitthisa, W. An, D.E. Resasco, J. Catal. 284 (2011) 90-101.
[4] L. Yu, L. He, J. Chen, J. Zheng, L. Ye, H. Lin, Y. Yuan, ChemCatChem 7 (2015) 1701-
1707.
[5] W.S. Putro, T. Kojima, T. Hara, N. Ichikuni, S. Shimazu, Catal. Sci. Technol. 7 (2017)
3637-3646.
[6] S. Leng, X. Wang, X. He, L. Liu, Y. Liu, X. Zhong, G. Zhuang, J. Wang, Catal. Commun.
41 (2013) 34-37.
[7] X. Yu, J. Chen, T. Ren, RSC Adv. 4 (2014) 46427-46436.
[8] H. Fang, J. Zheng, X. Luo, J. Du, A. Roldan, S. Leoni, Y. Yuan, Appl. Catal. A 529
(2017) 20-31.
[9] G.D. Yadav, M.R. Kharkara, Appl. Catal. A 126 (1995) 115-123.
[10] L J. Swartzendruber, V.P. Itkin, C.B. Alcock, J. Phase Equilibr. 12 (1991) 288-312.
[11] K. Tomishige, D. Li, M. Tamura, Y. Nakagawa, Catal. Sci. Technol. 7 (2017) 3952-
3979.
[12] E.E. Unmuth, L.H. Schwartz, J.B. Butt, J. Catal. 61 (1980) 242-255.
[13] C.N.R. Rao, G.U. Kulkarni, K.R. Kannan, S. Chaturvedi, J. Phys. Chem. 96 (1992)
7379-7385.
[14] L. Wang, D. Li, M. Koike, S. Koso, Y. Nakagawa, Y. Xu, K. Tomishige, Appl. Catal. A
392 (2011) 248-255.
[15] J. Ashok, S. Kawi, Appl. Catal. A 490 (2015) 24-35.
[16] S.M. Kim, P. Macarena Abdala, T. Margossian, D. Hosseini, L. Foppa, A. Armutlulu, W.
van Beek, A. Comas-Vives, C. Copéret, C. Müller, J. Am. Chem. Soc. 139 (2017) 1937-1949.
[17] B. Mutz, M. Belimov, W. Wang, P. Sprenger, M.A. Serrer, D. Wang, P. Pfeifer, W.
Kleist, J.D. Grunwaldt, ACS Catal. 7 (2017) 6802-6814.
[18] J. van de Loosdrecht, A.M. van der Kraan, A.J. van Dillen, J.W. Geus, Catal. Lett. 41
(1996) 27-34.
[19] P. Burattin, M. Che, C. Louis, J. Phys. Chem. B 101 (1997) 7060-7074.
[20] P. Burattin, M. Che, C. Louis, J. Phys. Chem. B 102 (1998) 2722-2732.
[21] P. Burattin, M. Che, C. Louis, J. Phys. Chem. B 103 (1999) 6171-6178.
[22] P. Burattin, M. Che, C. Louis, J. Phys. Chem. B 104 (2000) 10482-10489.
[23] L.A.M. Hermans, J.W. Geus, in Preparation of Catalysts II (B. Delmon, P. Grange, P.A.
Jacobs, G. Poncelet, Eds.), Elsevier, Amsterdam, 1979, p. 113.
[24] V. Briois, C. La Fontaine, S. Belin, L. Barthe, T. Moreno, V. Pinty, A. Carcy, R. Girardot,
E. Fonda, J. Phys.: Conf. Ser. 712 (2016) 012149.
25

[25] E. Fonda, A. Rochet, M. Ribbens, L. Barthe, S. Belin, V. Briois, J. Synchrotron Radiat. 19
(2012) 417–424.
[26] W.H. Cassinelli, L. Martins, A.R. Passos, S.H. Pulcinelli, C.V. Santilli, A. Rochet, V.
Briois, Catal. Today 229 (2014) 114–122.
[27] J. Hong, E. Marceau, A.Y. Khodakov, L. Gaberová, A. Griboval-Constant, J.S. Girardon,
C. La Fontaine, V. Briois, ACS Catal. 5 (2015) 1273-1282.
[28] A. Rochet, B. Baubet, V. Moizan, C. Pichon, V. Briois, C. R. Chim. 19 (2016) 1337–1351.
[29] J. Jaumot, A. de Juan, R. Tauler, Chemom. Intel. Lab. Syst. 140 (2015) 1-12.
[30] B. Ravel, M. Newville, J. Synchrotron Radiat. 12 (2005) 537–541.
[31] M. Newville, J. Synchrotron Radiat. 8 (2001) 322–324.
[32] C. La Fontaine, L. Barthe, A. Rochet, V. Briois, Catal. Today 205 (2013) 148–158.
[33] J.D. Bernal, D.R. Dasgupta, A.L. Mackay, Clay Miner. Bull. 4 (1959) 15-30.
[34] E.H. Nickel, J.E. Wildman, Mineral. Mag. 44 (1981) 333-337.
[35] L.H. Garófalo Chaves, J.E. Curry, D.A. Stone, M.D. Carducci, J. Chorover, Rev. Bras.
Ci. Solo 33 (2009) 1115-1123.
[36] J.M.R. Génin, R. Aïssa, A. Géhin, M. Abdelmoula, O. Benali, V. Ernstsen, G. Ona-
Nguema, C. Upadhyay, C. Ruby, Solid State Sci. 7 (2005) 545-572.
[37] J.M.R. Génin, A. Renard, C. Ruby, M. Abdelmoula, J. Phys.: Conf. Ser. 217 (2010)
012095.
[38] X. Wang, F. Liu, W. Tan, X. Feng, L. K. Koopal, Chem. Geol. 351 (2013) 57-65.
[39] P. Refait, H. Drissi, Y. Marie, J.M.R. Génin, Hyperfine Interact. 90 (1994) 389-394.
[40] K.W. Chung, Y.T. Kho, K.B. Kim, Corros. Sci. 44 (2002) 2757-2775.
[41] F. Menil, J. Phys. Chem. Solids 46 (1985) 763-789.
[42] M.L.M. de Carvalho-e-Silva, C.S.M. Partiti, J. Enzweiler, S. Petit, S.M. Netto, S.M.B. de
Oliveira, Hyperfine Interact. 142 (2002) 559-576.
[43] M. Usman, M. Abdelmoula, K. Hanna, B. Grégoire, P. Faure, C. Ruby, J. Solid State
Chem. 194 (2012) 328-335.
[44] N.A. Halasa, G. DePasquali, H.G. Drickamer, Phys. Rev. B 10 (1974), 154-164.
[45] E. Murad, U. Schwertmann, Am. Mineral. 65 (1980) 1044-1049.
[46] X. Wang, M. Zhu, L. K. Koopal, W. Li, W. Xu, F. Liu, J. Zhang, Q. Liu, X. Feng, D.L.
Sparks, Environ. Sci. Nano 3 (2016) 190-202.
[47] F. Baron, S. Petit, M. Pentrák, A. Decarreau, J.W. Stucki, Am. Mineral. 102 (2017)
1501-1515.
26

[48] S. Suzuki, T. Suzuki, M. Kimura, Y. Takagi, K. Shinoda, K. Tohji, Y. Waseda, Appl.
Surf. Sci. 169-170 (2001) 109-112.
[49] C.M. van Genuchten, J. Peña, S.E. Amrose, A.J. Gadgil, Geochim. Cosmochim. Acta
127 (2014) 285-304.
[50] S. Peulon, L. Legrand, H. Antony, A. Chaussé, Electrochem. Commun. 5 (2003) 208-213.
[51] A. Zegeye, G. Ona-Nguema, C. Carteret, L. Huguet, M. Abdelmoula, F. Jorand,
Geomicrobiol. J. 22 (2005) 389-399.
[52] D. Li, M. Koike, L. Wang, Y. Nakagawa, Y. Xu, K. Tomishige, ChemSusChem 7 (2014)
510-522.
[53] G. Bancroft, A. Maddock, R. Burns, Geochim. Cosmochim. Acta 31 (1967) 2219–2246.
[54] G.B. Raupp, W.N. Delgass, J. Catal. 58 (1979) 337-347.
[55] G.B. Raupp, W.N. Delgass, J. Catal. 58 (1979) 348-360.
[56] E.E. Unmuth, L.H. Schwartz, J.B. Butt, J. Catal. 61 (1980) 242-255.
[57] F. Bødker, I. Chorkendorff, S. Mørup, Z. Phys. D 40 (1997) 152–154.
[58] C. E. Johnson, M.S. Ridout, T.E. Cranshaw, Proc. Phys. Soc. 81 (1963) 1079–1090.
[59] K. Nagorny, S. Bubert, J. Catal. 108 (1987) 112-134.
[60] M. Matsuyama, K. Ashida, O. Takayasu, T. Takeuchi, J. Catal. 102 (1986) 309-315.
[61] T. Mizushima, K. Tohji, Y. Udagawa, M. Harada, M. Ishikawa, A. Ueno, J. Catal. 112
(1988) 282-289.
[62] A.B. Merlo, V. Vetere, J.F. Ruggera, M.L. Casella, Catal. Commun. 10 (2009) 1665-
1669.
[63] M. Koehle, R.F. Lobo, Catal. Sci. Technol. 6 (2016) 3018-3026.
[64] E. Marceau, A. Löfberg, J.M. Giraudon, F. Négrier, M. Che, L. Leclercq, Appl. Catal. A
362 (2009) 34-39.
[65] Z. Fu, Z. Wang, W. Lin, W. Song, S. Li, Appl. Catal. A 547 (2017) 248-255.
[66] Z. Li, Y.H. Huang, M. Zhu, X. Chen, H. Mei, J. Fuel Chem. Technol. 46 (2018) 54-58.
All figures must be printed in colour in the pdf version of the article
Captions
Fig. 1. 57Fe Mössbauer spectra recorded on Fe50Ni50/SiO2: a) after drying (RT); b) after
reduction at 500°C under H2 (RT); c) after reduction at 700°C under H2 (RT); d) after reduction
27

at 700°C under H2 (-268°C). Blue, red: Fe3+ (a) or Fe2+ (b) contributions. Orange, brown: Fe(0)
in Fe-Ni particles. Green: disordered Fe(0), in outer shells of Fe-Ni particles.
Fig. 2. EXAFS k3-weighted oscillations and results of k3 χ(k) EXAFS fitting in R-space: of
dried Fe50Ni50/SiO2, a) at the Fe K edge, b) at the Ni K edge; and of Fe50Ni50/SiO2 after
reduction at 500°C: c) at the Ni K edge (spectrum recorded at RT under H2). Black:
experimental; red or green: fit. Interatomic distances R are listed in Table 2.
Fig. 3. Evolution as a function of DPU duration: a) of the infrared bands of Fe50Ni50/SiO2
samples in the 500-1300 cm-1 region (ATR mode); b) of the Fe/Si and Ni/Si atomic ratios
measured by ICP.
Fig. 4. X-ray absorption spectra recorded during the reduction of Fe50Ni50/SiO2,
simultaneously: a) at the Ni K edge; b) at the Fe K edge. c) MCR-ALS analysis of Ni reduction.
d) Simultaneous evolution of the energy at the inflection point of the XANES spectra recorded
at the Fe K edge.
Fig. 5. X-ray diffractograms recorded as a function of temperature during the reduction of
Fe50Ni50/SiO2.
Fig. 6. Left: Bright field TEM micrographs of Fe50Ni50/SiO2: a) after reduction at 500°C and
exposure to air; b) after reduction at 700°C and exposure to air. Right: c) EDX mapping of
Fe50Ni50/SiO2 after reduction at 500°C and exposure to air (Fe: red; Ni: green).
Scheme 1. Reaction pathways in the hydrogenation of furfural: blue routes, hydrogenation;
red routes: hydrogenolysis of the C-O bond; green routes: etherification.
28

29