Polyol synthesis of non-stoichiometric Mn-Zn ferrite nanocrystals: Structural /microstructural characterization and catalytic application

Article abstract of DOI:10.1039/c5ra07562a

The structure of polyol-made Mn-Zn ferrite powders was investigated by X-ray diffraction, X-ray photoelectron spectrometry and X-ray absorption spectroscopy. It was found to be a defect-spinel structure in relation with non-stoichiometry introduced by the mixed oxidation-state of Mn and Fe cations. The ionic defects were identified as cation vacancies mainly located in the octahedral sites. The microstructure of these ferrites was studied by X-ray-diffraction and transmission electron microscopy. They appear to be constituted by quasi-isotropic monodisperse nano-aggregates which consist of pseudo-single crystals. Considering their reduced size, high crystalline quality, and cation mixed valence state related to their non-stoichiometry, the produced particles appear to be particularly valuable for redox-based solid-gas catalytic reactions. They exhibit an interesting activity toward catalytic combustion of dimethyl ether.

Full text of DOI:10.1039/c5ra07562a

View Article Online
View Journal
RSC Advances
This article can be cited before page numbers have been issued, to do this please use: Z. Beji, M. SUN, L.
S. Smiri, F. herbst, C. Mangeney and S. Ammar, RSC Adv., 2015, DOI: 10.1039/C5RA07562A.
This is an Accepted Manuscript, which has been through the
Royal Society of Chemistry peer review process and has been
accepted for publication.
Accepted Manuscripts are published online shortly after
acceptance, before technical editing, formatting and proof reading.
Using this free service, authors can make their results available
to the community, in citable form, before we publish the edited
article. This Accepted Manuscript will be replaced by the edited,
formatted and paginated article as soon as this is available.
You can find more information about Accepted Manuscripts in the
Information for Authors.
Please note that technical editing may introduce minor changes
to the text and/or graphics, which may alter content. The journal’s
standard Terms & Conditions and the Ethical guidelines still
apply. In no event shall the Royal Society of Chemistry be held
responsible for any errors or omissions in this Accepted Manuscript
or any consequences arising from the use of any information it
contains.
www.rsc.org/advances

Page 1 of 40
RSC Advances
View Article Online
DOI: 10.1039/C5RA07562A
Polyol synthesis of non-stoichiometric Mn-Zn ferrite nanocrystals:
structural microstructural characterization and catalytic application.
Z. Beji,1,2 M. Sun,3 L. S. Smiri,2 F. Herbst,1 C. Mangeney,1 S. Ammar1,*
1
ITODYS, Université Paris Diderot, Sorbonne Paris Cité, CNRS UMRꢀ7086, Paris 75251,
France.
2 LSSMI, Faculté des Sciences de Bizerte, UR 99/12ꢀ30, Zarzouna 7021, Tunisia
3
School of Chemical Engineering and Light Industry, Guangdong University of Technology,
Guangzhou 510006, People’s Republic of China
(*) Email: ammarmer@univꢀparisꢀdiderot.fr
Address: 15 rue JeanꢀAntoine de Baïf 75205 Paris cedex 13 France
Phone: (33)1 44 27 86 09 / Fax: (33)1 44 27 61 37
ABSTRACT.
The structure of polyolꢀmade MnꢀZn ferrite powders was investigated by Xꢀray
diffraction, Xꢀray Photoelectron spectrometry and Xꢀray absorption spectroscopy. It was
found to be a defectꢀspinel structure in relation with a nonꢀstoichiometry introduced by
the mixed oxidationꢀstate of Mn and Fe cations. The ionic defects were identified as
cation vacancies mainly located in the octahedral sites. The microstructure of these
ferrites was studied by Xꢀrayꢀdiffraction and transmission electron microscopy. They
appear to be constituted by quasiꢀisotropic monodisperse nanoꢀaggregates which consist
of pseudoꢀsingle crystals. Considering their reduced size, high crystalline quality, and
cation mixed valence state related to their nonꢀstoichiometry, the produced particles
appear to be particularly valuable for redoxꢀbased solidꢀgas catalytic reactions. They
exhibit an interesting activity toward catalytic combustion of dimethyl ether.
1

RSC Advances
Page 2 of 40
View Article Online
DOI: 10.1039/C5RA07562A
KEYWORDS.
Ferrite spinels, Nanoparticles, Polyol process, HRTEM, XPS, EXAFS, Catalytic
combustion of dimethyl ether.
INTRODUCTION
Spinel ferrites with the formula of MFe2O4, where M represents a divalent metal
(such as Cu, Mn, Mg, Zn or a mixture of them) have attracted extensive attention
due to their versatile electromagnetic properties. They have been extensively used for
electronic and/or magnetic devices, such as microwave filters, power supplies, radioꢀ
frequency coils, transformer cores, permanent magnets among others [1]. When their
size is reduced to the nanometer scale, the electromagnetic properties of these ferrites
may be very useful for other applications. For instance, they have been considered for
biomedicine applications, such as contrast agents for magnetic resonance imaging, drug
carriers for magnetically guided delivery or heating sources for hyperthermia cancer
therapy [2]. They have been also considered for various organic catalysis reactions
since they exhibit improved reactivity under mild reaction conditions and can be easily
recovered from reaction systems and reused up to several runs almost without loss of
catalytic activity [3]. Spinel ferrite nanoparticles (NPs) are thus typically valuable for
catalytic decomposition of gaseous pollutants [4], water gas shift reaction [5] and
hydrogen production from oxygenated hydrocarbons [6]. To the best of our knowledge,
all these reactions are mainly driven by the chemical composition of the ferrite
nanocatalysts (the nature and the valence state of the M cation and its consequence on
the ferrite phase stoichiometry), their local structure (cation distribution toward the
octahedral and tetrahedral spinel subꢀlattices) and their microstructure (size, shape,
single or polycrystal character…). Indeed, if the effect of the two latter parameters were
intensively studied and more or less understood, a few observations were made on their
2

Page 3 of 40
RSC Advances
View Article Online
DOI: 10.1039/C5RA07562A
stoichiometry. Interesting results were obtained on ferrite NPs containing mixed
divalent or trivalent cations. For instance, the electroꢀcatalytic activity of nickel ferrite
NPs for hydrogen evolution reaction was found to increase in the order of Fe3O4 <
(Ni0.8Fe0.2)Fe2O4 < (Ni0.4Fe0.6)Fe2O4 [7]. It was also elucidated that the oxidationꢀ
reduction potential of methyl blue in aqueous solution, in the presence of cobaltꢀzinc
ferrite NPs at pH = 7 under natural sunlight irradiation, was negative and increased with
the increase of the Zn content, leading to a decrease of the methyl blue degradation rate
with the increase of the Zn content [8]. Also, the partial replacement of Fe3+ ions by the
Mn3+ ions in CoꢀZn ferrite NPs possessing cubic spinel structure (Co0.6Zn0.4)(Fe2ꢀ
xMnx)O4, 0<x≤1), resulted that the degradation of methyl orange dye was enhanced with
an increase in Mn3+ ions concentration [9]. These changes are due to the octahedral site
preference and higher redox potential of manganese ion as compared to that of iron.
Besides, the stoichiometry deviation in these mixed ferrite NPs deviation plays an
important role in all these catalytic reactions. Indeed, by comparing the catalytic activity
of stoichiometric (NixZn1−x)Fe2O4 NPs and their related oxygenꢀdeficient solid
solutions, (NixZn1−x)Fe2O4−δ, toward CO2 decomposition, the latter showed better
efficiencies [10]. This enhancement is mainly attributed to the mixed valence state of
iron cations (Fe2+/Fe3+) in the latest ferrites and their role in the electron exchange
between adsorbed CO2 and the nanocatalyst surface. High catalytic performance was
also observed on alkali (K+) doped CuFe2O4 nanoparticles toward NOx reduction into
N2, compared to nonꢀdoped ones [4]. Doping promoted mixed valence state of copper
and/or iron cations at the surface of the catalysts with their potassium enrichment,
improving the electron exchange process with the adsorbed NOx [4].
To take advantageous from these mixedꢀvalenceꢀstate cations spinel ferrite NPs for such
peculiar applications, it becomes crucial to perfectly tailor their chemical composition
and their local structure in term of cation distribution, while maintaining their size in the
nanometer scale and avoiding their aggregation and/or coalescence. To achieve this
3

RSC Advances
Page 4 of 40
View Article Online
DOI: 10.1039/C5RA07562A
control, a lot of works were devoted to the fabrication of nanoscaled ferrites, focusing of
course on cheap and economical processes. Two experimental approaches were
nowadays adopted to produce such nanopowders. The first one, and may be the most
explored one, consists of a postꢀsynthetic “soft” oxidation or reduction thermal
treatment to create defects in particles produced by conventional granular techniques
such as coꢀprecipitation, hydrothermal route or ballꢀmilling [10,11,12]. The second one,
and very poorly explored one, consists of a voluntary direct fabrication of nonꢀ
stoichiometric nanoscaled ferrites using wet chemical routes. This may be achieved by
introducing in the reaction medium a doping monovalent (Na+, K+…) or tetravalent
cations (Te4+…) with respect to its solubility limit into the spinel lattice [4] or by
balancing between a partial oxidation or reduction of metal cation dissolved in the
reactional medium during ferrite synthesis. Such a control may be easily achieved by the
polyol process. Indeed, polyols are efficient crystalꢀgrowth medium in which reductive
and oxidative reactions may take place at moderate temperature (up to the boiling point
of the solvent) [13]. Acting on the polyol nature, αꢀdiols or etherglycols, the former
being better reducing agents than the latter, the amount of added water [13] and the
reaction atmosphere, inert or oxidative [14], it is possible to prepare metal or oxide NPs
as it is possible to tune the oxidation state of the metal cations into the latter phases
[15].
In this context, we decided to produce nonꢀstoichiometric MnꢀZn ferrite NPs using the
polyol process, to check their chemical composition, cation distribution toward the
spinel lattice and their morphology as a preliminary study before the evaluation of their
redoxꢀbased reactivity toward catalytic combustion of dimethyl ether (DME). The
characterizations were carried out by combining several techniques, mainly Xꢀray
diffraction (XRD), inductive coupling plasmaꢀatomic emission spectroscopy (ICP)
analysis, Xꢀray photoelectron spectrometry (XPS), Xꢀray absorption spectroscopy
(XAS) and low and high resolution transmission electron microscopy (TEM).
4

Page 5 of 40
RSC Advances
View Article Online
DOI: 10.1039/C5RA07562A
Temperature programmed reduction of hydrogen (H2ꢀTPR) and temperature
programmed desorption of oxygen (O2ꢀTPD) were also performed on the produced
powders before to test them as catalysts for DME combustion.
EXPERIMENTAL SECTION
Chemicals. Mn(CH3COO)2.4H2O, Zn(CH3COO)2.2H2O, FeCl3 and NaCH3CO2.nH2O
(n = 0, 3) and ethyleneglycol (EG) were purchased from SIGMAꢀALDRICH and
ACROS ORGANICS, respectively, and used without further purification.
Synthesis. The MnꢀZn ferrite particles were prepared by forced hydrolysis in a polyol
solvent. In practice, hydrolysis and condensation reactions were carried out on dissolved
metallic cations in a polyol solvent. Hydrolysis reaction in favored in basic conditions.
So an excess of sodium acetate (Na+ cations are inert) was introduced to basify the
reaction medium. So, iron, zinc and manganese acetates were dissolved in 250 mL of
EG solvent within the desired nominal molar ratio. Typically, the nominal iron
concentration was fixed to 0.2 M and the nominal molar ratio of divalent cations to
ferric ions, namely [Zn+Mn]/[Fe], was fixed to 0.6. Required amounts of distilled water
amount and sodium acetate base were added (the nominal water and acetate to total
metal molar ratio h = H2O/(Zn+Mn+Fe) and A = Acetate/Zn+Mn+Fe were fixed to 9
and 3, respectively). The mixture is then heated (6 °C/min) under reflux for twelve
hours while mechanically stirred. The boiling temperature reached 160ꢀ170°C. After
cooling to room temperature, the solids were separated from the supernatant by
centrifugation, washed three time with a large volume of ethanol to remove organic
residue from the surface of the produced particles (mainly adsorbed polyol molecules).
This washing protocol is wellꢀestablished and routinely used. The recovered powders
were then dried in air at 50 °C overnight. Two samples were prepared, as representative
of the Zn1ꢀxMnxFe2.4O4 solid solution, namely a manganese poor and a rich one,
5

RSC Advances
Page 6 of 40
View Article Online
DOI: 10.1039/C5RA07562A
corresponding to x = 0.2 and 0.8, respectively. Hereafter, the samples are denoted as
P2M and P8M, respectively.
Structural characterization. XRD patterns of the crystalline phases were recorded on a
Brucker D8 Advance diffractometer, using Cu Kα radiation, in the 2θ range 10ꢀ70°,
with steps of 0.02° (2θ) and a step time of 10 s. the cell parameter and the size of
coherent diffraction domain (crystal size) were determined with the MAUD software,
which is based on the Rietveld method combined with Fourier analysis [16].
Polycrystalline strainꢀfree silicon was used as standard to quantify the instrumental
broadening contribution. ICP analysis of Mn, Zn and Fe was performed at the Service
Central d’Analyse of CNRS at the French facility (Vernaison). The produced powders
were prepared by acid digestion using concentrated ultraꢀhigh grade HNO3 solution
(69%). XPS measurements were performed using a Thermo VG ESCALAB 250
instrument equipped with a monochromatic Al Kα (hν = 1486.6 eV) 200 W Xꢀray
source. The Xꢀray spot size was 650 µm. The pass energy was set at 150 and 40 eV for
the survey and the narrow scans, respectively. Measurements were carried out on
powdered samples mounted on sample holders. Data acquisition and processing were
performed with the Avantage software, version 1.85. Extended Xꢀray Absorption Fine
Structure (EXAFS) data were collected for the Mn, Zn and Fe Kꢀedges of the particles at
77 K, using a transmission mode detection scheme with a Si (111) twoꢀcrystal
monochromator on the XAFS beamline of the ELLETRA storage ring at the Italian
facility (Trieste). Data were also similarly collected on commercial powdered samples
ZnFe2O4 (ALFA, 99,9%) and Fe3O4 (PROLABO, 99,9%) as standards for Zn and Fe,
and MnO (MERCK, >99%), Mn2O3 (PROLABO, >99%) and MnO2 (SIGMAꢀ
LADRICH, >99%) as standards for Mn. The samples were ground and homogeneously
dispersed in cellulose pellets. EXAFS data were analyzed using the IFEFFIT software
[17] with calculated amplitude and phase shift functions from FEFF7 program [18].
6

Page 7 of 40
RSC Advances
View Article Online
DOI: 10.1039/C5RA07562A
Microstructural analysis. Particle morphology was determined by TEM observations
using a JEOLꢀ100 CX II microscope operating at 100 kV. The particle size distribution
was obtained from the recorded micrographs by using SAISAM software (Microvision
Instruments) calculating the surfaceꢀaverage particle diameter of about 400 particles:
Redox and catalytic activity evaluation. H2ꢀTPR and O2ꢀTPD plots were measured in a
MICROMERITICS AutoChem2920 equipment. To clean their surface, the samples (20ꢀ
30 mg) were first pretreated in an argon stream (50 mL/min) up to 350 °C during 20
min, for TPR measurements, and in a helium stream (50 mL/min) at 200 °C during 25
min, for TPD ones. After cooling to room temperature, a mixture of 5 vol% H2 in Ar
was admitted onto the samples and temperature was raised up to 750 °C (10 °C /min) for
TPR experiments. For TPD experiment, the sample was exposed to O2 atmosphere for
20 min, and then a helium stream (50 ml/min) was introduced into the system and
o
o
temperature was raised at a constant rate of 10 C /min up to750 C. The amount of
consumed hydrogen and desorbed oxygen in the effluent gas were detected by a TCD.
The catalytic oxidation test was carried out in a fixedꢀbed continuous flow quartz
reactor (8 mm i.d.) at atmospheric pressure. The reactor was charged with 300 mg of
catalyst which had been pressed, crushed and sieved to 40~60 meshes. Two reactant
gases, DME and 20 vol.% O2/He, were coꢀfed into the reactor with a flow of 1 mL/min
and 50 mL/min, respectively. The gas hourly space velocity was 10.2 L.gꢀ1cat.hꢀ1. T10 and
T90 were used for description of the activity, which is the temperature acquired for 10%
and 90% DME conversion, respectively. The reactor effluents were analyzed by an
online gas chromatography (Agilent 6820) using a methyl silicone capillary column
(ATSE30) and a Porapak Q packed column connected to a flame ionization detector and
a thermal conductivity detector, respectively. All the connected lines between the reactor
and gas chromatograph were kept above 120 °C to prevent condensation.
7

RSC Advances
Page 8 of 40
View Article Online
DOI: 10.1039/C5RA07562A
RESULTS
Evidence of the non-stoichiometry of the produced powders. The chemical analysis
shows that oxygen, manganese, zinc and iron atoms are present in both samples. The
inferred compositions from ICP measurements are not found in accordance with the
expected manganese zinc ferrite stoichiometry. A large manganese and zinc atoms
deficiency are evidenced partly compensated by iron atoms excess (Table 1).
Zn (wt.ꢀ%)
Mn (wt.ꢀ%)
Fe (wt.ꢀ%)
Sample Expected Measured Expected Measured Expected Measured
P2M
P8M
21.9
5.6
12.5
3.5
4.6
1.9
9.0
46.7
41.0
49.7
45.5
18.9
Table 1: Determination of the chemical composition by ICP analysis of the manganeseꢀ
zinc ferrite samples. The expected Zn, Mn and Fe weight contents are calculated
assuming a stoichiometric Zn1ꢀxMnxFe2O4 with x = 0.2 and x = 0.8 phases.
The recorded XRD patterns confirm the formation of a pure spinel oxide (Figure 1)
without any foreign phase. They match quite well with those tabulated for the Zn1ꢀ
xMnxFe2O4 bulk solid solutions (ICDD n° 01ꢀ074ꢀ2398, 01ꢀ074ꢀ2399, 01ꢀ074ꢀ2400, 01ꢀ
074ꢀ2401, 01ꢀ074ꢀ2402 for x = 0.1, 0.2, 0.4, 0.6, 0.8, respectively). The considerable
broadening of all diffraction peaks of the samples suggests that they are constituted by
relatively small crystals. Indeed, Rietveld refinements permits us to estimate their
average size at <LXRD> = 17 nm and their average microꢀdeformation at <ε> = 0.2%.
They also permit us to determine their cell parameter at a = 8.425(5) Å. surprisingly,
there is no variation of the structural and microstructural parameters of the produced
powders while the zinc content changes from one sample to another. Moreover, a net
discrepancy is observed between the measured cell parameters and those expected for
the stoichiometric Zn1ꢀxMnxFe2O4 x = 0.2 and 0.8 solid solutions, namely a = 8,4616 Å
(ICDD n°01ꢀ074ꢀ2399) and a = 8,5136 Å (ICDD n°01ꢀ074ꢀ2402). The refined values
8

Page 9 of 40
RSC Advances
View Article Online
DOI: 10.1039/C5RA07562A
are clearly out the expected Vegrad Law cell range. (the cell parameters of bulk
ZnFe2O4 and MnFe2O4 are 8.44 and 8.54 Å, respectively [19]).
(311)
(440)
(220)
(511)
P2M
(400)
(111)
(222)
(422)
P8M
10
20
30
40
50
60
70
2θ (degree)
Figure 1: XRD patterns of P2M and P8M samples recorded in a θꢀθ BraggꢀBrentano
configuration.
The XRD and ICP results suggest a metal deficiency in the unit cell and in turn
deviation from stoichiometry. This spinel cation deficient structure might accommodate
with oxygen vacancies to equilibrate between the total positive and negative charges of
the spinelꢀlike cation and anion subꢀlattices, assuming manganese and iron cations as
exactly divalent and trivalent cations, respectively. Or it might accommodate with an
increase of the average oxidation state of its cations, mainly manganese ones, to
compensate the total negative charge of the fully occupied spinelꢀlike anion subꢀlattice.
In the two cases, the unit cell size decreases. In the first case, the solid oxide cohesion
requires a unit cell contraction. In the second case, the average cation radius contraction
decreases the cationꢀanion distances leading to the unit cell size reduction. The effect of
the oxidation state change of the constituting atoms of an oxide phase on its
9

RSC Advances
Page 10 of 40
View Article Online
DOI: 10.1039/C5RA07562A
stoichiometry and its unit cell size is something very often observed in solid state
chemistry [20].
To pursue our investigations, we decided to carry out XPS experiments to characterize
the electronic state of each constituting atom of these materials. Note XPS is a highly
surfaceꢀspecific technique. The information depth of XPS is considered to correspond to
the sample thickness from which 95% of the detected signal originates. This depth is
equal to 3λ cos θ, where λ is the inelastic mean free path (IMFP) of the photoelectrons
and θ the electron takeꢀoff angle with respect to the surface normal (equal to 0° in our
setup). The inelastic mean free path of Fe2p photoelectrons was calculated using the
Tougaard QuasesꢀIMFPꢀTPP2M program which is based on the algorithm proposed by
Tanuma, Powell, and Pen [21], and found equal to 1.6 nm, hence a depth analysis of ca.
4.8 nm. This value is lower than the crystal size inferred from Rietveld refinement (17
nm), but it is large enough to be considered quite deep to make the XPS analysis
valuable and representative of the bulk composition of the studied nanocrystals.
Spectral calibration was determined by setting the main C1s component at 285 eV [22].
Figure 2 displays the XPS survey spectra of P2M and P8M samples. The main peaks are
Mn2p, Fe2p, Zn2p, C1s and O1s centred at ca. 640, 720, 1030, 285 and 530 eV,
respectively. The carbon signal (about 5 wt.ꢀ% carbon content) is mainly assigned to
organic species, acetate and/or ethylene glycol, adsorbed on the surface of the particles.
10

Page 11 of 40
RSC Advances
View Article Online
DOI: 10.1039/C5RA07562A
Zn2p
Fe2p
O1s
Mn2p
C1s
P2M
P8M
1000
800
600
400
200
0
E (eV)
Figure 2: XPS survey spectra of P2M and P8M samples.
The other signals due to the mixed MnꢀZn ferrite particles appear to vary strongly in
their intensity from one sample to another, particularly the Mn and Zn ones. Indeed, the
comparison between the two spectra evidences clear modifications of the particle
chemical composition when the nominal atomic Mn:Zn content changes. As expected,
the Zn and Mn signal intensity increases and decreases, respectively, when the nominal
zinc content increases, in agreement with the results obtained previously by chemical
analysis.
Focusing now on the electronic state of each constituting cation, the high resolution
spectra of the two studied samples were recorded. Fe2p signal, shown in Figure 3, is
splitted into 2 main components, Fe2p3/2 (at ca. 710 eV) and Fe2p1/2 (at ca. 723 eV).
with an intensity ratio, I(Fe2p3/2)/I(Fe2p1/2) of 2:1, as determined by fitting the whole
Fe2p signal (see Figure 4 below). Each peak is mirrored by a shakeꢀup satellite on its
high binding energy side, caused by the configuration interaction due to relaxation of the
valence electrons. The high peak widths indicate that Fe is found in two different
oxidation states. Figure 3 allows comparing the Fe signal with reference samples, i.e.
Fe2O3 (which contains Fe3+ ions only) and Fe3O4 (which contains a mixture of Fe2+ and
11

RSC Advances
Page 12 of 40
View Article Online
DOI: 10.1039/C5RA07562A
Fe3+ ions). It is clear that the Fe electronic states in the MnꢀZn ferrite particles are closer
to those of Fe3O4 than Fe2O3: Fe thus appears to be in a mixed valence state, with the
presence of both Fe2+ and Fe3+ ions. Considering the two oxidation states for Fe (Fe2+
and Fe3+), and for each species, two main peaks (Fe2p3/2 and Fe2p1/2), each one
accompanied by one shakeꢀup, it is possible to fit the Fe signal with 8 components as
illustrated in figure 4 for P8M sample. However, one has to remember that a
characteristic feature of the oxide states is the multiplet splitting. It is caused by the
interaction between the 2p electrons and the unpaired 3d electrons, and it results in a
multitude of possible final states. The energy separation is usually so small that no
discrete peaks will appear with a typical spectrometer resolution. Consequently, the
multiplet splitting is seen in the spectrum as asymmetric broadening of the 2p peaks.
The fitting parameters for the Fe2p3/2 part of the Fe signal are summarized in Table 2.
From the ratio of the relative intensity of the Fe2+ and Fe3+ 2p3/2 components (I/ITotal
)
against the whole Fe2p3/2 signal (ΣI/ITotal), one can estimate the proportion of Fe2+ and
Fe3+, i.e., 35 % and 65 %, respectively. This result must be underlined, because a small
amount of Fe excess in the composition of ferrites by increasing the amount of Fe2+ ions
can decrease the electrical resistivity and enhance electron hopping between Fe2+ and
Fe3+ ions in the spinel unit cell. Reversely, a small amount of iron deficiency in the
composition by reducing the amount of Fe2+ ions can increase the electrical resistivity
and inhibit electron hopping between Fe2+ and Fe3+ ions.
The high resolution Mn2p spectra exhibits also two main peaks at ca. 640 (2p3/2) and
655 (2p1/2) eV. Satellite lines of manganese appear at much higher binding energies [23ꢀ
25]. The Mn 2p3/2 lines, shown in figure 4 for P8M sample, are fitted with three peaks
with binding energy at 640, 641.8 and 645.2 eV, assigned to Mn2+, Mn3+ and Mn4+,
respectively. These binding energy values are in agreement with many reported XPS
data of manganese containing compounds [23,26]. The Mn2+, Mn3+ and Mn4+ atomic
percent in the whole Mn2p3/2 signal are of ca. 50 %, 40 % and 10 %, respectively, for all
12

Page 13 of 40
RSC Advances
View Article Online
DOI: 10.1039/C5RA07562A
studied samples. Altogether, ICP and XPS results, suggest that the prepared MnꢀZn
mixed ferrites exhibit a defectꢀspinel structure exhibiting a large deviation from
stoichiometry with a none zero number of cation vacancies according the following
chemical formula for each sample (see supporting information):
2+
0.46
2+
0.75
2+
0.04
3+
1.39
3+
0.03
4+
0.01
(
Zn Fe Mn )(Fe Mn Mn
)l
0.32O4
2+
0.14
2+
0.77
2+
0.22
3+
1.44
3+
0.18
4+
0.04
and
(
Zn Fe Mn )(Fe Mn Mn
)l0.21O4
where
l
measures the deviation from stoichiometry.
711,2 eV
710,0 eV
Fe O
2 3
2p1/2
2p3/2
shake-up
shake-up
P8M
Fe O
3 4
740
730
720
710
E (eV)
Figure 3. Comparison of the Fe XPS signal of P8M sample with that of reference
compounds, i.e. Fe2O3 and Fe3O4.
13

RSC Advances
Page 14 of 40
2000
1500
1000
3500
3000
2500
2000
1500
2p
3/2
2p
1/2
2p3/2
Fe3+
Fe2+
shake-up
shake-up
Fe3+
Mn2+
Mn3+
2p1/2
Fe2+
Mn4+
656
652
648
644
640
636
735
730
725
720
715
710
705
E (eV)
E (eV)
Figure 4. Peakꢀfitted Fe2p and Mn2p3/2 coreꢀline spectrum, respectively, of P8M sample. The colored solid lines are the peaks fitted to the experimental
data.
14

Page 15 of 40
RSC Advances
View Article Online
DOI: 10.1039/C5RA07562A
BE (eV)
708.8
710.5
714.4
717.9
640.0
641.9
645.2
FWHM
3.0
L/G(%)a
I/Itotalb
Fe2+ (2p3/2)
Fe3+ (2p3/2)
30
30
0
0.18
0.33
0.06
0.08
0.27
0.23
0.06
3.9
Fe2+ shakeꢀup
Fe3+ shakeꢀup
Mn2+ (2p3/2)
Mn3+ (2p3/2)
Mn4+ (2p3/2)
3.7
3.6
0
2.9
30
30
30
3.0
2.9
a
The L/G parameter indicates the Lorentzian character in percentage in the Gaussianꢀ
Lorentzian function used to fit the spectra.
b
I is the intensity of the fitted component while ITotal is the overall intensity of the
analyzed signal: for Fe components, ITotal represents the sum of Fe2p3/2 and Fe2p1/2
signals while for Mn components, ITotal represents the sum of Mn2p3/2 and Mn2p1/2
signals.
Table 2: The fitting parameters for the Fe2p3/2 part of the XPS Fe signal of P8M sample
Such feature was already observed in manganese ferrite NPs prepared by other chemical
routes. A cell parameter largely smaller than that expected for the MnFe2O4
stoichiometry was measured in particles prepared by coꢀprecipitation [27,28] or by
thermal decomposition of oxalates [29]. Similar results were obtained for manganese
zinc ferrite NPs produced by reverse micelle route [30] or thermal decomposition of
acetylacetonates [31]. In all the cases, the authors claimed that the spinel structure
accommodates with cation vacancies in relation with a mixed valence state of Mn
and/or Fe atoms. Clearly, the polyol process conducts to the production of ferrite
particles with a defect spinel structure, cation vacancies being stabilized thanks to a
valence state change of their constituting metallic cations.
Local structure. Important structural informations can be obtained from the measured
cell parameters on these samples. Indeed, as previously mentioned, there is a basic
15

RSC Advances
Page 16 of 40
View Article Online
DOI: 10.1039/C5RA07562A
correlation between the cell parameter value and the anionꢀcation distance for a
particular site in the spinel structure. Tetrahedral and octahedral spinel sites are located
at (1/8,1/8,1/8) and (1/2,1/2,1/2), respectively. The oxygen atom lies on the cube body
diagonal at (
only two structural variables: (i) the unitꢀcell
. Assuming to be equal 1/4, the value can be completely determined using the
following equation [32]:
2.0995Rtetra + 5.8182Ro2cta −1.4107Rt2etra
u
,
u
,u
), where
u
is approximately 1/4. The cubic spinel structure thus has
a
and (ii) the oxygen positional parameter
u
u
a
a
=
(1)
where Rtetra and Rocta measure exactly the average value of the anionꢀcation distance on
the tetrahedral and octahedral sites. They depend on both the oxidation state of the
cations and their coordination, the covalent bonds in tetrahedral sites being shorter than
electrovalent bonds in octahedral sites. The Rtetra and Rocta distances in Eq. (1) are
calculated using the tabulated values of MnꢀO, FeꢀO and ZnꢀO [32ꢀ36] (see Table S1 in
the supporting information section) considering various cation distributions and taking
into account the previously obtained ICP and XPS results.
The calculated lattice parameter is then compared with the experimental one and the
best concordance between the theoretical and the experimental a values determine the
most realistic cations distribution in the solid solutions. We started our calculations
assuming a thermodynamically stable structure in which the majority of ionic defects are
located in the octahedral sites and the Mn2+, Mn3+, Mn4+, Fe2+, Fe3+ and Zn2+ cations are
located in the site for which their stabilization energy in the related electrostatic field is
the higher, agreeing thus with a strong Zn2+ tetrahedral coordination preference and a
strong Fe2+, Mn3+ and Mn4+ octahedral one, leading to a general formula:
+
u
+
+
+
+
+
+
[
Mn12− vZnu2 Fev3
]
(
Fe23−
z Fev2' Mnw3 Mnt4 lz
)O4
−
v'−w−t−
3+
or
[
Zn2+ Feu3+
]
(Fe
Fev2+Mny2+Mnw3+Mnt4+lz
)O4
1−u
2−v−y−w−t−z
16

Page 17 of 40
RSC Advances
View Article Online
DOI: 10.1039/C5RA07562A
where the brackets correspond to the tetrahedral sites and parentheses to the octahedral
ones, and the deviation from stoichiometry. For P2M sample, the calculation of the
l
cell parameter in both configurations gives exactly the same result, namely a = 8.45 ±
0.02 Å. This is not so surprising in regard to its very weak Mn content and obviously
Mn2+ one. A displacement of divalent manganese cations from octahedral to tetrahedral
sites does not affect significantly the a value. For P8M sample, there is a marked
difference between the two configurations, the former giving the closest a value to that
inferred from Riteveld refinements, meaning that the cation distribution in this sample
agrees with Zn2+ and Mn2+ tetrahedral coordination preference and a Fe2+ octahedral
one. Anyway, in all the cases, a partial transfer of Zn2+ from tetrahedral to octahedral
sites and a reverse transfer of Fe3+ lead to a progressive increase of the theoretical a
value. Similarly, a transfer of the cation vacancies from the octahedral to the tetrahedral
sites, accompanied with a decrease of the concentration of the Fe3+ ions in these sites
leads also to a significant a value increase.
To confirm this cation distribution toward the spinel lattice, EXAFS experiments were
carried out. The moduli of the Fourier transform (FT) of the total EXAFS data for the
different samples are given in figure 5 at the Fe and Zn Kꢀedges. The plot of the moduli
of the FT is not corrected from the electron phase shift. It is related to the radial
distribution function of backꢀscatterers around the absorbing atoms. By comparison with
that of standard ZnFe2O4 and Fe3O4, which adopt a normal and inversed spinel structure,
respectively, it is possible to identify the main FTꢀpeaks, at the Zn and Fe Kꢀedges. For
all the samples, the FT profile at Fe, and Zn Kꢀedges suggested that iron ions occupy
both tetrahedral (A) and octahedral (B) sites while zinc ones are located in tetrahedral
sites only. Indeed, at the Fe Kꢀedge, the relative intensity of the two peaks centered at
2.7 and 3.2 Å. These FT peak positions corrected from the electron phase shift which is
about 0.3ꢀ0.4 Å measure the distances between the Fe ions and their second nearest
metallic neighbors. They are close to those of standard Fe3O4, where the third of iron
17

RSC Advances
Page 18 of 40
View Article Online
DOI: 10.1039/C5RA07562A
ions is tetrahedrally coordinated and the other two thirds are octahedrally coordinated,
EXAFS analysis being not efficient to distinguish between the different oxidation states,
+2 and +3, of iron atoms. At the Zn Kꢀedge, a great proximity between the overall FT
profiles of the two samples and the standard ZnFe2O4, where zinc cations occupy
exclusively tetrahedral sites, is observed. The loss of intensity of the peak centered at
3.1 Å in the nanoparticles, corresponding to the distance between the Zn ions and their
second nearest metallic neighbors, might be assumed to be due to their nanocrystalline
character, the surface zinc ions of the particles having nearest metallic neighbors on one
side none on the other side [37]. Such feature is often observed in the case of metal
nanoparticles and used to solve their 3D structure [38].
Anyway, we performed least square modeling using single scattering Fe and Zn EXAFS
simulation on the first filtered Fe and Zn FT peak corresponding to the FeꢀO and ZnꢀO
shells, respectively, to determine the local symmetry of these cations. The fitting
procedure is based on the equation [17]:
Ni
Ri2
R
i
2
k
k
exp( k 2
exp(
f
k
σ i2
−
π
+ Φ
kR k ))
(
i i
(2)
χ
=
−
( )
2
)
)
i ( , ).sin(2
∑
λ
(k
i
where k is the wave vector, χ(k) the EXAFS function, Ni is the number of backꢀ
scatterers “i” at a distance Ri from the central atom and σi the Gaussian DebyeꢀWaller
factor associated with Ri. The energy threshold, E0, taken at the first inflexion of the
absorption edge is corrected in the fitting procedure by the ꢀE0 parameter.
Φi(k), are the amplitude and phase shift functions, respectively.
Ifi(π,k)I, and,
Two contributions have been necessary to reproduce the experimental EXAFS data, at
the Fe Kꢀedge, while only one has been used to simulate the experimental EXAFS
signal, at the Zn Kꢀedge (the quality of the fits is illustrated in figure S2 in the
supporting information section for P8M sample). The main structural fitted parameters
are listed in table 3. The refined interatomic cationꢀoxygen distances are consistent with
the tabulated ones, for tetrahedraly coordinated zinc cations and for tetrahedrally and
18

Page 19 of 40
RSC Advances
View Article Online
DOI: 10.1039/C5RA07562A
octahedrally coordinated iron ones (see Table S1). Using the fitted coordination
numbers N of Fe and Zn, the uncertainties being estimated to be about 15% [17], it is
possible to determine the occupancies of Fe and Zn cations in A sites, expressed in % of
the total Fe and Zn cations, respectively. They are found to be about 24 and 28% for Fe
and 100 and 100 % for Zn, in P2M and P8M samples, respectively.
At the Mn Kꢀedge, such quantitative analysis of the EXAFS data is not easy. In one
hand, the energy range of Mn Kꢀedge spectra are drastically shortened, due to the
recovering with the Fe Kꢀedge Xꢀray absorption signal. In other hand, EXAFS analysis
is not always efficient to distinguish between the different manganese oxidation states
+2, +3 and +4, particularly if their related distances to oxygen ions in tetrahedral and/or
octahedral polyhedra are close to each other. Nevertheless, the moduli of the FT at the
Mn Kꢀedge of the two samples, are almost identical and exhibit a first peak,
corresponding to MnꢀO interatomic distances, which is dedoubled. Such feature is
associated with two MnꢀO distances, a short one and a long one. Indeed, the peaks in the
FT moduli of the total EXAFS function weighted by k3, reported in Figure 5, correspond
to apparent interatomic distances. Two first peaks mean two distinct nn MnꢀO shells for
the nearest nn neigbor. Considering the listed distances in the supporting information
section (Table S1), one can attribute the shortest and the longest distances to the
octahedral Mn4+ꢀO2ꢀ and Mn2+ꢀO2ꢀ atomic pairs, respectively. A third distance, an
intermediate one comprised between the two previous ones, exists and corresponds
indifferently to the octahedral Mn3+ꢀO2ꢀ and tetrahedral Mn2+ꢀO2ꢀ atomic pairs.
Comparing the relative intensity of the two corresponding MnꢀO FTꢀpeaks, one may
assume that the MnꢀO distances are quite equivalently distributed between short,
medium and long ones in P2M, while they are in majority short or medium in P8M.
Using all these information, we can propose a probable cation distribution fixing first
the localization of Fe2+, Mn3+ and Mn4+ cations as well as the vacancies in the octahedral
sites, Zn2+ in the tetrahedral ones and Fe3+ in both octahedral and tetrahedral ones.
19

RSC Advances
Page 20 of 40
6
4
2
0
25
20
15
10
5
25
20
15
10
5
K(Mn)
P2M
P8M
K(Zn)
FeB-MB
K(Fe)
P2M
P8M
P2M
P8M
ZnFe2O
ZnA-M
B
ZnFe2O
4
4
Fe3O
4
FeA-MB
FeB-MA
FeA-O
FeB-O
ZnA-O
0
0
0
2
4
6
0
2
4
6
0
2
4
6
Radius (Å)
Radius (Å)
Radius (Å)
Figure 5. Mn, Fe and Zn Kꢀedges Moduli of the FT experimental EXAFS signals of P2M and P8M samples, compared to that of bulk ZnFe2O4 and Fe3O4.
The A and B indices correspond to the tetrahedral and octahedral spinel sites occupancy respectively. M corresponds to the metallic second nearest
neighbours.
20

Page 21 of 40
RSC Advances
K(Fe)
K(Zn)
Sample
P2M
First shell
FeAꢀO
FeBꢀO
[FeA/Fe] (%)
24
ZnAꢀO
ZnBꢀO
[ZnA/Zn] (%)
100
ρ
(%)
ρ (%)
N
0.98
1.93
0.058
13.1
4.52
2.02
0.088
9.8
3.94
1.97
0.059
4.7
0.00
ꢀꢀ
R (Å)
0.8
0.6
0.3
0.9
0.00
0.00
σ
(Å)
ꢀE0(eV)
P8M
N
1.10
1.93
0.047
16.1
4.36
2.02
0.086
9.1
4.02
1.97
0.063
3.5
0.00
ꢀꢀ
R (Å)
28
100
0.00
0.00
σ
ꢀ
(Å)
E0(eV)
Table 3. EXAFS results at the Fe and Zn Kꢀedges of the P2M and P8M samples, where
ρ the residue measures the quality. It must be the smallest possible.
The [FeA/Fe] and [ZnA/Zn] parameters correspond to the ratio of tetrahedrally coordinated Fe and Zn cations, respectively.
21

RSC Advances
Page 22 of 40
View Article Online
DOI: 10.1039/C5RA07562A
Starting from these assumptions, it becomes possible to share Mn2+ cations between the
two types of sites, the content of all tetrahedrally coordinated cations and that of all
octahedrally coordinated ones, including vacancies, leading to a value of 1.00 and 2.00
per formula unit, respectively:
+
+
+
+
+
+
+
[
Mn02.04Zn02.46Fe03.50
]
(
Fe03.89Fe02.75Mn03.03Mn04.01l0.32
O4 for P2M
)
+
+
+
+
+
+
+
and
[
Mn02.22Zn02.14 Fe03.64
]
(
Fe03.80Fe02.77 Mn03.18Mn04.04l0.21
O4 for P8M
)
Microstructural properties. The TEM images showed that the ferrite particles are
isotropic in shape and very uniform in size (Figure 6), with an average diameter varying
from 78±15 to 29±4 nm by changing the nominal atomic Mn:Zn ratio from 20:80 to
80:20 (see Figure S3 in the supporting information section), evidencing a size
dependency to the zinc content.
P2M
P8M
100 nm
100 nm
Figure 6. TEM images of an assembly of P8M and P2M particles.
A zoom on a representative particle from of each sample shows (Figure 7) that it
consists of a given number of smaller in size nanocrystals (less than 10 nm). Moreover,
their fast Fourier transformation (FFT) patterns (Figure 7) are of Laüe type. This means
that they are textured polycrystals or pseudoꢀsingle crystals, justifying thus the measured
22

Page 23 of 40
RSC Advances
View Article Online
DOI: 10.1039/C5RA07562A
discrepancy between the measured TEM and XRD sizes. Such a feature was already
observed for different oxide particles produced in polyol [39]. All these results suggest
that the particle formation in polyol is driven by an aggregation growth mechanism.
P2M
P2M
P8M
Figure 7. Zoom images on representative particle of each sample and its FFT patterns
indexed in the spinel structure. Note the double (400) reflections introduce the normally
forbidden (200) reflections observed for P8M sample.
The size variation between the zinc poorest and richest polycrystals is very probably
related to the variation of the number of nuclei. The largest polycrystals correspond to
the coalescence of a greater number of primary particles and then to the formation in the
reaction medium of a greater number of nuclei, and reversely, the smallest polycrystals
correspond to the coalescence of a smaller number of primary particles and then to the
formation in the reaction medium of a smaller number of nuclei. This is not so
surprising. Indeed, it is usually considered that the energy barrier for the formation of
ZnFe2O4 is much lower than that of MnFe2O4 because of the zinc ion’s strong
preference for tetrahedral sites. The main consequence is that the nucleation energy of
ZnFe2O4 or Mn1ꢀxZnxFe2O4 x ~ 1 crystals is smaller than that of MnFe2O4 or Mn1ꢀ
xZnxFe2O4 x ~0 [40,41], leading to a higher nuclei number
23

RSC Advances
Page 24 of 40
View Article Online
DOI: 10.1039/C5RA07562A
Redox and catalytic activity evaluation. Before testing the produced MnꢀZn ferrite NPs
for catalytic DME combustion, H2ꢀTPR and O2ꢀTPD characterizations were made to
study their redox activities, and the results are shown in Figure 8 and Table 4. Clearly,
the two samples exhibit three overlapped TPR peaks related to the reduction of Mn4+
into Mn3+ and Mn2+ and Fe3+ into Fe2+ or even Fe as already observed in manganese and
manganeseꢀzinc ferrite NPs produced by other routes [41], confirming the mixed
valence state of Mn and Fe atoms in the studied systems. P2M appears to be more easily
reduced than P8M. It consumes more hydrogen (12.33 mmol/g) than P8M (11.36
mmol/g), meaning that the average oxidation state of its constituting metal cations (Fe,
Mn and Zn) is higher than that in its P8M parent. Concerning the O2ꢀTPD profiles, once
again three oxygen desorption peaks are observed. They are located at around ~100°C,
200ꢀ450° and ~600°C and are assigned to the physisorption of oxygen (α peak), its
weak chemisorption (suprafacial oxygen, β peak) and its strong chemisorption (lattice
oxygen, γ peak), respectively [42]. P2M exhibits a more marked β peak than P8M,
suggesting its higher oxidation activity. Note the carried out pretreatment for TPR and
TPD experiments were realized under insert atmosphere at a temperature (350°C and
200°C, respectively) lower than the required one for the first reduction or oxidation
reaction in these particles, making the previously described TPR and TPD results
significantly representative of the intrinsic redox behavior of the asꢀproduced NPs.
Dimethyl ether (DME) is a kind of environmental energy with multiꢀsources and multiꢀ
purpose.
24

Page 25 of 40
RSC Advances
γ
3
(b)
α
(a)
P8M
2
β
3
1
γ
β
α
2
P8M
1
P2M
P2M
100
200
300
400
500
600
700
100
200
300
400
500
600
700
T (°C)
T (°C)
Figure 8. H2ꢀTPR (a) and O2ꢀTPD (b) plots of P8M and P2M samples.
25

RSC Advances
Page 26 of 40
View Article Online
DOI: 10.1039/C5RA07562A
H2 consumption
Peak 2
n (mmol/g) T (°C) n (mmol/g)
Peak 1
n (mmol/g) T (°C)
Peak 3
T (°C)
328
P8M
P2M
1.50
0.85
456
431
3.28
2.93
490
516
6.58
8.55
255
O2 release
Peak β
n (mmol/g) T (°C) n (mmol/g)
Peak α
n (mmol/g) T (°C)
Peak γ
T (°C)
76.6
P8M
P2M
0.121
0.106
ꢀ
ꢀ
655
662
0.193
0.122
76.5
342.5
0.235
Noted: Overlapped peaks were separated by fitting the curves to data using the software of AutoChem II .
Table 4. Results of H2ꢀTPR and O2ꢀTPD experiments for P8M and P2M samples.
To increase combustion efficiency and decrease the emission of hydrocarbons, catalytic
combustion of DME is often considered when it is used as fuel. Several catalyst systems
have been reported, eg, noble metal (Pd, Rh, Pt) [43], hexaaluminate [44], and
manganese oxides [45]. From the point view of practical application, the transition
metal oxides hold great potential for DME combustion because of their lowꢀcost and
relatively high activity. As for ferrite catalysts, they are reported as effective catalysts for
catalytic combustion of toluene [46] and of for DME [47]. Here, we explored the
catalytic combustion of DME over the polyol prepared ferrites. GC analysis confirms
that the combustion is total according the selected working conditions. No other
products than CO2 and H2O are formed. The combustion activity of P2M is a little
superior than that of P8M (Figure 9). For P2M sample, the T10 and T90 are 257 °C and
276 °C, respectively. However, the P8M sample has a T10 of 243 °C and a T90 of 291°C
(Figure 8). The catalytic performance (T10 = 257 °C and T90= 276 °C at SV = 10.2 L.gꢀ
1
cat.hꢀ1) over P2M sample is better than that (T50 = 356 °C and T90= 409 °C ) over cobalt
ferrite [47], that (T10 = 170 °C and T90= 370 °C) over BaNi0.2Mn0.8Al11O19ꢀ [44], that
δ
(T10 = 106 °C and T90> 320 °C) over Pd/Al2O3 [43a], but similar to or inferior to that
26

Page 27 of 40
RSC Advances
View Article Online
DOI: 10.1039/C5RA07562A
(T10 = 184 °C and T90= 220 °C) over
αꢀMnO2 [45b], that (T10 = 171 °C and T90=
180°C) over CuꢀOMSꢀ2 [45a], that (T10 = 149 °C and T90= 159 °C) over CeꢀOMSꢀ2
[45c].
100
T90~276
80
P2M
P8M
T90~291
60
40
20
0
T10~243
T10~257
100
150
200
250
300
350
T (°C)
Figure 9. DME conversion as a function of the temperature for P8M and P2M samples.
Note the homogeneous combustion of DME, under the same reaction conditions and
without catalysts, is efficiently achieved only at a temperature higher than 525 °C (see
Figure S4 in the supporting information section), largely higher than that required in
presence of P2M or P8M powders. These results must be underlined, they
unambiguously evidence the catalytic activity of the polyolꢀmade nonꢀstoichiometric
manganeseꢀzinc ferrite NPs toward DME combustion.
DISCUSSION
It appears clearly that the synthetic procedure developed in the present study offers
several important advantageous features for the synthesis of nanoscaled manganeseꢀzinc
ferrite powders. First, the synthetic process is economical, because it involves
inexpensive metallic salts (chloride and acetate) and need moderate heating (160ꢀ170
27

RSC Advances
Page 28 of 40
View Article Online
DOI: 10.1039/C5RA07562A
°C). Second, the particles are directly produced as stable vacancyꢀdeficient ferrites, as
established by XRD, XPS and ICP analyses, without going through a supplementary
appropriate heat treatment at oxidizing atmosphere, as it is often the case when this kind
of material is desired. Third, the ferrite nanoparticles are obtained with a local structure
as close as possible to their bulk counterparts, while, in the majority of the cases, a
redistribution of cations towards equilibrium conditions are observed only after a
prolonged annealing treatment. Finally, they exhibit a high crystalline quality and a fine
microstructure, as observed by TEM and HRTEM, which is of primary importance for
the desired application.
Catalytic tests consisting on DME combustion have shown that the produced ferrite
particles exhibit interesting catalytic feature, the manganese poor sample (P2M) being
more reactive than its manganese rich counterpart (P8M), while the former exhibits a
larger particle size than the latter (78 vs. 29 nm), suggesting that the particle size is not
the dominant factor affecting the DME combustion. This may be explained by the fact
that the crystal size remains almost equivalent in both samples (8ꢀ10 nm) according to
the HRTEM observations. In the same time, the structure doesn’t appear to be a limiting
parameter since both spinelꢀlike sample exhibit a cation distribution close to the
thermodynamically stable one, namely Fe2+, Mn3+ and Mn4+ octahedral and Zn2+
tetrahedral coordination preference, respectively. Moreover, focusing on octahedrally
coordinated iron cations, EXAFS analysis shows that their atomic ratios in both samples,
P2M and P8M, are very close to each other (76 vs. 72 at.ꢀ%), making the contribution of
the structural parameter toward the catalytic combustion activity of the two samples
quite equivalent. Note, the atomic ratio of octahedrally coordinated manganese cations
was also calculated for each sample, assuming, as a first approximation, that Mn2+
cations are exclusively localized in tetrahedral spinel sites. The obtained values are 67
vs. 50 at.ꢀ% for P2M and P8M, respectively. The difference between the two samples is
more marked here, but at this stage of our structural analysis, it is not easy to conclude,
28

Page 29 of 40
RSC Advances
View Article Online
DOI: 10.1039/C5RA07562A
since the manganese content of P2M is very week (about 2% of the total cation amount),
making this sample closest to a nonꢀstoichiometric zinc ferrite than a manganeseꢀzinc
ferrite. On these bases, one may suppose that the improved catalytic behavior of P2M
can be due to its own chemical composition, namely to its close to zero manganese
content. In the same time, looking at the relevant literature, manganese, cobalt and
manganeseꢀcobalt ferrite NPs are more often cited as good redox catalysts, including
DME combustion catalysts, than zinc ferrite ones [46,48], making this hypothesis
irrelevant.
Normally, local defects are very important for redox solid catalysts. The stabilization of
metal cations in a mixed valence state into the spinel lattice induces deviation from
stoichiometry and the creation of cation vacancies,
l
, with respect to the cation and
anion charge balance. This deviation from the perfect spinel structure corresponds to
about 0.3 and 0.2 vacancies per formula unit in P2M and P8M, respectively, slightly
more important in the former than in the latter. Their relative catalytic activity is giving
thus a slight advantage to the former.
Finally, reducibility is an important parameter to evaluate the redox activity which is
directly related with the catalytic activity. The TPR tests show that P2M powder reveals
a greater susceptibility to the reduction with a higher hydrogen consumption than that of
P8M. A higher hydrogen consumption means its average oxidation state is also higher,
which is thought to be helpful for catalytic activity [50]. The TPR results are consistent
with the catalytic performance. In parallel, the amount of desorbed oxygen and the
temperature at which oxygen is desorbed are thought to be directed related with the
capability to form surface vacancies, and are often considered to be a key factor
determining the catalytic activity [51]. Based on the TPD tests, the P2M sample
possesses more weakly chemisorbed oxygen than that of P8M sample, and this is in
accordance with their T90 order. Moreover, the suprafacial oxygen species over the
spinel ferrite dominates its catalytic activity.
29