Fuel 108 (2013) 740–748

Contents lists available at SciVerse ScienceDirect

Fuel

journal homepage: www.elsevier .com/locate / fuel

Oxidative dehydrogenation of ethylbenzene with CO2 for styrene productionover porous iron-based catalysts

Antonio J.R. Castro a,1, João M. Soares b,2, Josue M. Filho a,1, Alcineia C. Oliveira a,⇑, Adriana Campos c,3,Édwin R.C. Milet c,3

a Universidade Federal do Ceará, Campus do Pici-Bloco 922, Departamento de Física/Departamento de Química Analítica e Físico-Química, Fortaleza, Ceará, Brazilb Universidade do Estado do Rio Grande do Norte, BR 110, km 48, R. Prof. Antônio Campos, Costa e Silva, Mossoró/RN, CEP 59625-620, Brazilc CETENE, Av. Prof. Luiz Freire, Cidade Universitária, Recife, Pernambuco, Brazil

h i g h l i g h t s

" Styrene production over porous iron-based catalysts." XRD, 57Fe-Mössbauer and Raman spectroscopy, TPR and N2 adsorption–desorption measurements characterizations." FeTi showed any tendency to sintering or phase transformation whereas the other solids suffered from hard carbon deposition.

a r t i c l e i n f o

Article history:Received 9 December 2012Received in revised form 6 February 2013Accepted 7 February 2013Available online 27 February 2013

Keywords:Porous iron oxidesCharacterizationsDehydrogenationEthylbenzeneCarbon dioxide

0016-2361/$ - see front matter � 2013 Elsevier Ltd. Ahttp://dx.doi.org/10.1016/j.fuel.2013.02.019

⇑ Corresponding author. Tel./fax: +55 85 3366 90 0E-mail address: alcineia@ufc.br (A.C. Oliveira).

1 Tel./fax: +55 85 3366 90 08.2 Tel./fax: +55 84 3315 2196.3 Tel./fax: +55 81 33347224.

a b s t r a c t

Porous iron-based catalysts with different promoters (Zr, Ti or Al) have been tested in oxidative dehydro-genation of ethylbenzene with CO2 for styrene production. The catalysts were characterized by X-ray dif-fraction (XRD), 57Fe-Mössbauer and Raman spectroscopy, temperature-programmed reduction (TPR) andN2 adsorption–desorption measurements, before and after the catalytic evaluation. The reactivity of iron-based catalysts toward styrene production was dependent on the structural and textural features of thesolid as well as the nature of the promoter. a-Fe2O3 and rutile TiO2 present on FeTi were converted in situinto FeTiO3, Fe2TiO5 and FeTi2O5, and these phases revealed a high styrene yield (up to 50%) in the firststage of the reaction, but lower selectivity than that exhibited by their FeZr and FeAl counterparts. How-ever, FeTi performed much better in terms of stability showing no tendency to sintering or phase trans-formation whereas the other solids suffered from hard carbon deposition.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Oxidative dehydrogenation of ethylbenzene with CO2 (ODH) is apromising alternative to the industrially applied catalytic non-oxida-tive dehydrogenation due to the strong market incentive for styreneproduction [1]. As industrial demand for styrene grows, its produc-tion via ODH of ethylbenzene is assuming great importance. Indeed,styrene is of interest to the petroleum industry for the production ofpolystyrene, styrene–butadiene rubber, fibers and resins [2]. How-ever, the major challenges associated with the dehydrogenationare requirement of developing selective and stable catalysts, sincethe deactivation of the solids due to coke formation is inevitable.

ll rights reserved.

8.

In another context, iron-based compounds have been recog-nized as important materials due to their wide range of applica-tions, such as ion exchangers, material science, pharmaceuticals,biotechnology, adsorbents, catalysts and recently in nanomedicine[2–7]. In the field of catalysis, many studies have been conductedconcerning the use of iron-based compounds as catalysts or cata-lytic supports [8–10] due to their desirable properties, includingredox abilities, acidic features and low-cost.

Among the various types of iron-based compounds, porous ironoxides are suitable as catalytic active sites for a variety of reactions[10–13]. Moreover, they have a much higher surface area than con-ventional microporous iron oxides and possibilities a larger poresize distribution and a more availability for their surface function-alities [11]. Indeed, the pores are often used as catalytic reactors[10]. Therefore, research efforts have been mainly focused ondeveloping porous iron monoxides or iron-based compounds toimprove the resistance against the deactivation resulting from car-bon deposition and increase the adsorption properties [8–12].

A.J.R. Castro et al. / Fuel 108 (2013) 740–748 741

However, porous binary iron oxides are not easy to obtain owing tothe facile deposition of the second labile species to be added duringthe synthesis on the iron specie, which could act as a support. Thisis a drawback concerning the production of mixed iron-based oxi-des and requires special synthesis conditions or reactants for con-trolling pore size and shape and avoiding pores’ blocking.

In case of ODH to obtain styrene by using iron-based catalysts,nearly all of the transition metals and lanthanides have been stud-ied as electronic, structural and textural promoters, giving rise togood catalytic activity [9–14]. Titanium, zirconium and aluminumare leading the subjects in this area of research. Although Fe ismore susceptible to reduction, and therefore to deactivation thanmetals oxides such as Al, Zr and Ti, a combination of the latter spe-cies in a porous iron oxide would be useful to improve the catalyticperformance. Most of the studies are devoted to the synthesis con-ditions and characterizations, and only a few of them deeply inves-tigated the resultant binary iron oxides porosity effect in thedehydrogenation of ethylbenzene with CO2.

The influence of the active promoters on the physicochemicalproperties of iron-based catalysts has been investigated in thisstudy. The effect of iron oxide porosity and phases formed on thestyrene production explain the different catalytic behavior of thesolids.

2. Experimental

2.1. Preparation of the solids

Aluminum tri-sec-butoxide (Al(OC4H9sec)3) and ferric nitrateFe(NO3)39H2O, were used as precursors to prepare the FeAl solid,according to a previously published work [15]. The hydrolysis reac-tion took place by introducing aluminum tri-sec-butoxide into ex-cess of ethanol at 60 �C under vigorous stirring. Briefly, thesynthesis was carried out by adding in a drop wise manner a mix-ture of 2.9 mol of water, ferric nitrate and 6.5 mol of absolute eth-anol to the stirred mixture of aluminium, to obtain the clear sol,which turned into a gelatinous precipitate within few minutes.The reactants were maintained under constant stirring and reflux-ing for 24 h. The gel was afterwards, washed with ethanol, dried atroom temperature and calcined at 600 �C under air flow at a heat-ing rate of 5 �C min�1 during 2 h. The abovementioned methodol-ogy was used to obtain the FeZr and FeTi, in which the zirconiumoxychloride, ZrOCl2�8H2O, and titanium (IV) isopropoxide, Ti(OiPr)4

were the active component precursors. The metal contents mea-sured by chemical analyses were 80:20 wt%, respectively for ironand the second metal added to the solid.

2.2. Characterization of the solids

X-ray powder diffraction (XRD) patterns were recorded in aPANalytical X’PERT HighScore’s diffractometer. The Cu Ka radia-tion was used and diffractrograms were collected with a 2h stepof 0.02 and a counting time of 10 s per step. Diffraction peaks re-corded between 3� and 80� have been used to identify the structureobtained at 40 kV and 30 mA. Particles size were calculated byScherrer formula (D = Kk/B cosh), where K = 0.9, k = 0.15418 nm, his the Bragg angle, and B is the full width at half maximum of dif-fraction peaks. The diffractograms were compared to that of ICDDdatabase (International Centre for Diffraction Data).

The Brunauer–Emmett–Teller (BET) method was employed tomeasure the specific surface of oxides through the nitrogenadsorption–desorption isotherms. The measurements were madeat �196 �C using a Micromeritics instrument. The samples wereoutgassed for 6 h at 200 �C under vacuum, prior to the sorptionanalyses.

The temperature-programmed reduction experiments (H2-TPR)were carried out in home-made equipment. About 80 mg of cata-lyst was embedded in a fixed-bed quartz tube and heated undernitrogen at 100 �C for 2 h. Subsequently, the reactor was cooleddown to room temperature and was then heated from room tem-perature to 1000 �C using a heating rate of 10 �C min�1 in the pres-ence of a 8% H2/N2 mixture.

Raman spectroscopy was used to obtain information about thestructural features of the solids under ambient conditions on a al-pha 300 microscope from Witec spectrometer. The confocalmicroscopy was used with a 532 nm laser line for the spectral exci-tation and a power of 10 mW.

Mössbauer spectra were measured on powdered spent solids atroom temperature with the spectroscopy system from Wissel. Themeasurements were carried out by standard transmission geome-try, using a constant acceleration spectrometer with a radioactivesource of 57Co in Rh matrix and activity of 50 mCi. The spectrawere fitted using the Fit routine, which makes use of a set ofLorentzian profile peaks. The analyses allowed the calculation ofamplitude and width (G) of each peak, isomer shift (d), electricquadrupole splitting (D) and hyperfine magnetic fields (BFH). Inaddition, all isomer shifts (d) refer to metallic iron (a-Fe) at roomtemperature.

2.3. Catalytic evaluation

Catalytic reaction of dehydrogenation of ethylbenzene usingCO2 was carried out in a fixed bed quartz reactor by using100 mg of catalyst. Ethylbenzene (EB) was fed to the reactor bypassing the gas feed e.g., nitrogen (11 mmol h�1), and carbon diox-ide (58 mmol h�1) over an ethylbenzene saturator vessel with a EBfeed rate of 1.9 mmol h�1. The reaction tests were performed at550 �C under atmospheric pressure and a CO2/EB = 30:1. The prod-ucts formed were analyzed with an FID gas chromatograph (SimpleChrom) using a capillary column. A detailed description of the cat-alytic tests evaluation is given in the references [10,11,16]. Theperformance of catalysts was evaluated by means of the EB conver-sion, the styrene selectivity and styrene yield and the formulae areshown in the papers previously published [8,16].

3. Results and discussion

3.1. Structural features of the fresh catalysts

3.1.1. XRD and Raman measurementsXRD patterns of the catalysts are shown in Fig. 1. The diffraction

lines for fresh catalysts corresponding to the (110), (012), (113)and (300) reflections of rhombohedral symmetry of hematite (a-Fe2O3, space group R3c, D6

3d, ICDD no. 33-664) and those havingcharacteristic reflections along (220), (311), (400) and (440) fromcubic maghemite (c-Fe2O3, space group of P4332, ICDD no. 39-1346) [17] phase are observed for all solids. Indeed, XRD patternsare broad and weak due to the nanocrystalline features of the sol-ids, as summarized in Table 1.

For FeZr, tetragonal ZrO2 (space group P42/nmc, ICDD no. 37-1484) is the sole zirconium specie observed in the XRD pattern, be-sides the aforesaid iron-based monoxides. In case of FeAl, eventhough FeAl2O4 formation could be likely under the synthesis con-ditions [15,18], the XRD pattern of the fresh solid reveals only c-Fe2O3, c-Al2O3 (space group and ICDD no. 29-0063) and a-Fe2O3

phases and does not show any Fe–Al–related species. FeAl difracto-gram is shown in our previous paper [15].

The diffractogram of FeTi shows not well resolved diffractionpeaks at interplanar spacings of 3.24, 1.89 and 1.66 Å correspond-ing to the (110), (111) and (211) characteristic reflections of the

10 20 30 40 50 60 70 80

100

150

200

10 20 30 40 50 60 70 8020

40

60

80

100

120

FeZr

Inte

nsity

(a.u

.)In

tens

ity (a

.u.)

2θ (degree)

FeTi

2θ (degree)

200 400 600 800 1000 1200 1400 1600 1800180

200

220

240

260

280

300

320

200 400 600 800 1000 1200 1400 1600 1800

200

250

300

350

400

411296225 607

1053

146

1318

847660

503

FeZr

Wavenumber (cm-1)

Wavenumber (cm-1)

Ram

am in

tens

ity (a

.u.)

FeTi

Ram

am in

tens

ity (a

.u,)

(a)

(b)

Fig. 1. (a) X-ray diffraction patterns and (b) Raman spectra for iron-based catalysts.

742 A.J.R. Castro et al. / Fuel 108 (2013) 740–748

rutile phase of TiO2 in accordance with the ICDD no. 21-1276,space group P42/mnm. Low intensities peaks mainly at 2h equalto 48.1�, which is indexed to the (200) plane arises from thetetragonal TiO2 in anatase form (ICDD no. 21-1272). Also, somepeaks of a-Fe2O3 and c-Fe2O3 are found over FeTi. It is known thatFeTiO3 ilmenite structure is derived from that of a-Fe2O3 and theilmenite is possibly formed in an either amorphous or crystallinestate by replacing every other layer of Fe3+ (ionic radius 0.64 Å)atoms in the (001) planes by a layer of Ti4+ (ionic radius 0.68 Å)atoms during the co-precipitation of Fe/Ti mixed oxides [19]. How-ever, no XRD peaks of ilmenite are observed for FeTi due to itsbroad diffraction lines.

Raman spectrum of FeZr (Fig. 1b) shows the typical vibrationmodes of hematite. As stated above, a-Fe2O3 belongs to the D6

3d

space group and the phonon lines at about 212, 247, 293, 299,412, 498 and 613 cm�1 and additional broad modes at 1050 and1319 cm�1 should appear in the Raman spectrum of hematiteand maghemite [15,20]. Indeed, earlier reports have shown that

the latter modes are due to two magnons scattering created onclose antiparalell spin sites [19,20]. Thus, the modes appear around146, 225, 296, 411, 503, 607, 660, 869, 1050 and 1319 cm�1

confirming the presence of hematite and maghemite. In contrast,c-Fe2O3 is an inverse spinel and its structure can be read as an irondeficient form of magnetite (Fe3O4), possessing three broad vibra-tional modes at about 350, 500 and 700 cm�1 [15]. Thus, c-Fe2O3

modes could be assigned in the observed spectrum, in accordancewith XRD results. In addition, the tetragonal form of ZrO2 (D15

4h

space group) has Raman modes at about 270, 315, 455, 602 and645 cm�1 whereas the monoclinic ones possesses modes at 192,335, 347, 382, 476, 617 and 638 cm�1 [21]. Thus, the presence ofZrO2 in both tetragonal and monoclinic forms is suggested for FeZrby Raman measurements. These results are in line with the ZrO2 intetragonal phase found by XRD.

For FeAl and FeTi, two broad and not well defined bands are ob-served in the 200–600 cm�1 and 700–900 cm�1 ranges. Thesemodes overlap each other, and thus their assignations correspond

Table 1Physicochemical and catalytic properties of the solids in dehydrogenation ofethylbenzene with CO2.

Catalyst Phasesa BETSurfacearea(m2 g�1)

Porevolumeb

(cm3 g�1)

Poresizeb

(Å)

% EBconversionc

FeAl a-Fe2O3 384 0.55 30 27.4c-Fe2O3

c-Al2O3

FeZr a-Fe2O3 418 0.60 29 19.8c-Fe2O3

t-ZrO2

FeTi c-Fe2O3 rutileTiO2a-Fe2O3

anatase TiO2

421 0.67 33 50.5

a From XRD analysis.b BJH method.c Reaction conditions: CO2/EB = 30; Temperature = 550oC.

0.0 0.2 0.4 0.6 0.8 1.00

20

40

60

80

100

120

140

160

180

200

20 40 60 80 100

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

Pore size ( )

dV/lo

gD (c

m3 .g

.-1A-1

)

Relative pressure (P/po)

Adso

rbed

vol

ume

(cm

3 .g

-1)

FeAl

1000

1500

2000

2500

0 100 200 300 400 500 6000

1

2

3

4

5

dsor

bed

volu

me

(cm

3 .g-1

)

FeZr

Pore size ( )

dV/lo

gD (c

m3 .g

.-1A-1

)

A.J.R. Castro et al. / Fuel 108 (2013) 740–748 743

to the superimposed modes values for c-Fe2O3 and a-Fe2O3. More-over, Raman modes of either Al2O3 or TiO2 are not observed at all.Considering that there are few Raman studies on the influence ofthe Ti and Al promoters in porous iron oxide catalysts availablein literature, it is difficult to explain the exact meaning of thesuperimposition of the modes in this study. Nevertheless, it is pos-sible to explain that the Fe–O vibrations were perturbed by thevariations the chemical bonds energy, besides other effects causedby the presence of the promoters.

0.0 0.2 0.4 0.6 0.8 1.0

Relative pressure (P/po)

0.0 0.2 0.4 0.6 0.8 1.0

Relative pressure (P/po)

500dV

/logD

(cm

3 .g

.-1A-1

)

AAd

sorb

ed v

olum

e (c

m3 .g

-1)

100

200

300

400

500

50 100 150 200

0.4

0.8

1.2

1.6

2.0

FeTi

Pore size ( )

Fig. 2. N2 adsorption–desorption isotherms of the solids. The included figures arethe pore size distribution curves of the catalysts.

3.1.2. N2 adsorption–desorption analysesTextural properties of the solids are evaluated by N2 adsorp-

tion–desorption experiments. The isotherms as well as the corre-sponding pore size distribution curves are plotted in Fig. 2.

As it can be seen, the isotherm of FeAl is of type IV, with a hys-teresis in the 0.48 P/Po region, which is typical of mesoporousmaterials. Moreover, the well-defined hysteresis loop associatedwith irreversible capillary condensation on mesopores from 0.4to 1.0 indicates the existence of a mesoporosity arising fromnon-crystalline voids and spaces formed by interparticle contactsin the catalysts. FeZr exhibits a type II isotherm, with a more pro-nounced capillary condensation step that shifts to a higher P/Po andposses a hysteresis loop between H3 and H4 [22]. BET surface areaof FeTi and FeAl are calculated to be 421 and 418 m2 g�1 (Table 1),respectively. The BJH pore size distribution confirms that FeZr pos-sesses a well-developed mesoporosity besides meso and macrop-ores (Fig. 2b included), FeZr displays the representative type IIand IV curves, with a capillary condensation step at P/Po 0.4–0.6 re-gion, similar to that obtained for sol–gel based-zirconia solids [23].Additionally, the pore size distribution of FeAl is narrower thanthat of FeTi, implying that the latter has a more open pore structurewith mesopores size and volumes of 33 Å and 0.67 cm3 g�1, respec-tively. Different from FeTi and FeAl, the prominent hysteresis loopof FeZr is mainly characterised by a micro-meso-macroporousstructure. Indeed, the pore sizes of FeZr are centred at 12, 33 and62 Å, with respect to its broad pore size distribution for micro,meso and macroporoes, respectively. Moreover, the micropore areais not reported because either the micropore volume is negative orthe calculated external surface area is larger than the total surfacearea. The relatively low textural parameters of FeAl were expectedto be related to its crystalline feature. Due to this fact, the non-crystallized or tiny particles of the FeAl2O4 phase would blockthe inter-particle spaces among crystalline a-Fe2O3 and c-Fe2O3,which eventually resulted in the decrease of exposed internal sur-face area of entire solid.

3.1.3. H2-TPR profilesTPR analyses are used to predict the reducibility of the solids.

TPR profiles (Fig. 3) reveal, not unexpectedly, two reduction stages,which are typical of iron-based solids. The low hydrogen consump-tion peaks observed at temperatures of 200–400 �C characterizesthe Fe3+ to Fe2+ reduction while the high hydrogen consumptionpeaks above 600 �C are attributed to the reduction of Fe2+ to Feo

[8,24].Based on XRD and Raman characterizations, the former peak

can be assigned to the reduction of the free c-Fe2O3 and a-Fe2O3

Fig. 3. TPR profiles of the solids.

744 A.J.R. Castro et al. / Fuel 108 (2013) 740–748

species, while the latter peaks are attributed to the reduction of thehard reducibility ZrO2 and perovskite FeTiO3 phases, respectivelyfor FeZr and FeTi. In case of FeTi, the high temperature TPR peakmay also be associated to the reduction of TiO2. However, TiO2 iswell known to be more difficult to reduce and reduction of bulkoxygen of TiO2 has been reported to occur above 600 �C [25,26].Besides, the Fe–Ti–O system displays a very rich phase diagram,with solid solutions including pseudobrookite (Fe1+xTi2�xO5),ulvöspinel-magnetite (Fe3�xTixO4) and ilmenite-hematite (Fe2�xTix

O3). Such FeTiO3, Fe2TiO5 and Fe2TiO5 have been suggested to re-duce only at high temperatures (>750 �C) [27]. Therefore, the highhydrogen consumption peak above 600 �C could be attributed toboth TiO2 and perovskite reduction. In addition, ZrO2 weakly inter-acts with iron species and provides a lesser resistance to reductionin FeZr compared to that of FeTi, thus causing the shifting of thereduction of Fe3+ to lower temperatures. TPR profile of FeAl hasbeen described previously [15] and it can be seen that the ironstates of reduction are identical in terms of profiles, however, theFeAl2O4 phase formation is facilitated under hydrogen environ-ment at about 600 �C.

3.2. Catalytic results

The results on the ethylbenzene dehydrogenation with CO2 areshown in Table 1 and Fig. 4.

The catalytic conversion of FeAl at the beginning of the reactionis slightly higher than that of FeZr and lesser than half of that ofFeTi (Table 1). Moreover, the styrene yield is high in the first min-utes of the reaction for all solids (Fig. 4). It can be seen that the nat-ure of promoter present on the different iron phase is related to theporous structure and the catalytic performance, being titaniumwell suited to be added to the iron oxides. The fact that c-Fe2O3

and a-Fe2O3 active phases are detected over all catalysts by XRDand Raman results suggest that the elevated initial conversion ofthe catalysts could be due to the presence of these phases. How-ever, our earlier report has demonstrated that unprompted bulkiron oxides in either hematite or maghemite forms have conver-sion values lower than 2% in the steady state condition. This wasattributed to the inactivity and/or ease reducibility of these oxidesin the EB and CO2 environments [10]. Burri et al. also reported thatEB conversion over either ZrO2 or TiO2 is inferior to that of pro-moted binaries catalysts counterparts [28]; this lies to the fact thatpure TiO2 gives very low styrene yield in the ODH reaction [29].Also, alumina itself performed badly in the dehydrogenation ofethylbenzene with CO2, as shown by the findings [30–32]. SinceFe2O3, ZrO2, TiO2 alone or Al2O3 itself are not highly actives inthe reaction, the interaction between iron oxide and the aforesaidpromoters might be necessary to obtain high activity.

The fast decline of the selectivity of FeTi along the time onstream with respect to that of FeAl indicated an aluminum actionas structural promoter of the iron oxide and this enhanced the sty-rene yield greatly [32]. This is supported by the fact that there areno diffraction lines of alumina in the patterns indicating an incor-poration of Al into the iron oxide to form a nanocrystalline and sta-ble FeAl2O4. Furthermore, FeAl improves the selectivity to styreneabout 29% in the steady state, whereas the addition of Zr to theiron-based catalyst contributes to increase the selectivity onlyaround 18% at iso-conversion. The detection of t-ZrO2 phase overFeZr reveals that zirconia particles do not stabilize the iron-basedphases over the course of the reaction, which in turn resulted ina smaller activity and stability. This is confirmed by TPR resultsthat show the easy reducibility of FeZr catalyst. The same effectis observed over Ti–Zr and Mn–Zr based catalysts for CO2 dehydro-genation of ethylbenzene [28,33]. The styrene yield is improved byadding aluminum to the iron oxide, but selectivity to styrene isconsidered low, in comparison to other aluminum-based iron oxi-des catalysts [32]. The selectivity to benzene and toluene increaseshowing that cracking reactions are favored at high time on streamfor FeAl and FeZr.

The best performance of FeTi for styrene production may berationalized from the information obtained from the characteriza-tion results. Previous studies on oxidative dehydrogenation of pro-pane reaction reveal that the TiO2 phase itself is active onethylbenzene conversion while this effect is significant over binaryTiO2–ZrO2 catalyst [34]. The iron titanates phases are ilmenite(FeTiO3), a spinel phase (Fe2TiO4) and pseudo-brookite (Fe2TiO5)[35]. Among them, FeTiO3 (further shown by spent catalysts char-acterizations) is found to remain stable in the steady state, evenwhen rutile TiO2, c-Fe2O3 and a-Fe2O3 were consumed, initially.Any structural change is observed (latter confirmed) and thisbehavior is attributed to stability of the ilmenite phase and themesopores, which remains accessible to ethylbenzene and CO2

reactants, after the reaction.

3.3. Structural features of the spent solids

3.3.1. 57Fe Mössbauer spectroscopyAll the catalysts exhibited structural changes after the reaction.

Mössbauer spectra of the solids are shown in Fig. 5.57Fe Mössbauer spectroscopy is a technique largely used to

study the valence states of iron and its occupation at the unit cellssites in iron-containing crystal structures.

FeAl Mössbauer spectrum is fitted with two doublets. The firstdoublet possessing D = 1.01 mm s�1 e d = 0.35 mm s�1 is attributedto the presence of hematite (a-Fe2O3), which shows superpara-magnetic behavior, as a result of its nanosized particles. The second

20

40

60

80

100

(%) S

t Sel

ectiv

ity

50 100 150 200 250 300 350 400 450

Time (min.)

50 100 150 200 250 300 350 400 450

Time (min.)

50 100 150 200 250 300 350 400 450

Time (min.)

(%) S

t Yie

ld

FeAl

20

40

60

80

100

20

40

60

80

100

(%) S

t Yie

ld

(%) S

t Sel

ectiv

iyFeZr

20

40

60

80

100

20

40

60

80

100

(%) S

t Yie

ld

20

40

60

80

100

(%) S

tyre

ne s

elct

ivity

FeTi

Fig. 4. Styrene yield and selectivity of the catalysts studied.

95.0

96.0

97.0

98.0

99.0

100.0

FeAl

-12 -9 -6 -3 0 3 6 9 12

Velocity (mm/s)

Tran

smis

sion

(%)

-12 -9 -6 -3 0 3 6 9 12

Velocity (mm/s)

FeZr

Tran

smis

sion

(%)

98

99

100

96

97

98

99

100

Data Fit Doublet Fe3+

Doublet Fe2+

Fe3O4 (A)

Fe3O4 [B]

FeTi

Fig. 5. 57Fe Mössbauer spectra of the spent solids: FeTi, FeZr and FeAl.

A.J.R. Castro et al. / Fuel 108 (2013) 740–748 745

doublet with hyperfine parameters of ca. D = 1.71 mm s�1 andd = 0.92 mm s�1 corresponds well to a paramagnetic spinel typecrystal structure, where only the Fe2+ ions occupy tetrahedral sites,as for FeAl2O4 structure [36]. Therefore, FeAl has Fe3+ with 55.5% ofthe relative abundance arising from hematite while 44.5% is Fe2+

originated from hercynite (FeAl2O4), which crystallizes after thecatalytic test.

For FeZr, Mössbauer spectrum is adjusted with one doublet andtwo sextets. The hyperfine parameters of the doublet (d =0.38 mm s�1 and D = 1.01 mm s�1) are approximately the sameto those found for FeAl. Taking in account this result, it can be sug-gested that superparamagnetic nanoparticles, like ZrFe2O4 spinelcould be present on FeZr. However, the hyperfine parameters of

sextets are typical magnetite, as futher observed for FeTi. Thus,FeZr has half of the iron ions e.g., 50.6% in Fe2+ state whereas theother half, 49.4% is in Fe3+enviroment.

Mössbauer spectrum of spent FeTi is adjusted with two dou-blets and two sextets. The two well-resolved doublets indicatethe presence of paramagnetic high-spin Fe2+ and Fe3+ species.Mössbauer parameters obtained from the spectra are shown inTable 2.

The first doublet of FeTi exhibits relative large isometric shift(d = 1.07 mm s�1) and quadrupole splitting (D = 0.70 mm s�1),which corresponded to a Fe2+ environment. The second doubletdisplays relative small isometric shift (d = 0.35 mm s�1) and quad-rupole splitting (D = 0.58 mm s�1), which could be assigned to aFe3+ environment. Both doublets and their hyperfine parametersare characteristic of ilmenite (FeTiO3), in accordance with the find-ings [37,38]. In addition, the two sextets are characterized byhyperfine parameters of ca. HHF = 48.93T, D = �0.06 mm s�1 andd = 0.41 mm s�1 for the first sextet and these parameters were typ-ical Fe2+ in the tetrahedral site of a spinel structure. The other sex-tet with the HHF = 45.96T, D = 0.02 mm s�1 and d = 0.58 mm s�1 isdue to Fe3+ in octahedral environment. Since the inverse spinel

Table 2Parameters obtained from refinements at room temperature of Mössbauer spectra ofFe-containing samples. Electric quadrupole splitting (D), isomer shift (d), hyperfinemagnetic field (BHF) and Relative spectral area (R.A.) parameters of the Fe-containingsamples.

Samples Electric quadrupole andisomer shift

Average of BHF over the distribution

D (mm/s)

d (mm/s)

R.A.(%)

HHF

(T)D (mm/s)

d (mm/s)

R.A.(%)

FeAl 1.01 0.35 55.51.71 0.92 44.5

FeTi 0.58 0.32 29.6 48.93 �0.06 0.41 19.20.70 1.07 39.7 45.96 0.02 0.58 11.5

FeZr 1.01 0.38 14.6 49.71 0.00 0.28 35.046.64 0.04 0.68 50.4

746 A.J.R. Castro et al. / Fuel 108 (2013) 740–748

magnetite structure, namely, [Fe3+]tetra[Fe3+, Fe2+]octaO4, the iron issituated in the two crystallographically inequivalent tetrahedral Aand octahedral B sites [15], it can be concluded that magnetitecoexists with ilmenite structure for FeTi. Also, Fe2+ represents58.9% of the whole area while Fe3+ composes 41.1% one.

3.3.2. Raman and XRDRaman measurements (Fig. 6a) provide additional evidences for

the catalytic behavior of the spent catalysts.Noteworthy, the modes of spent FeAl increase their intensity

and shifts to lower wavenumbers compared to the fresh analogous.On the contrary of the 57Fe Mössbauer analyses, Raman modes cor-responding to Fe3O4 appear at 200, 305, 530 cm�1 and a sharpmode at 668 cm�1 [15,20]. It is assumed that due to the modesof c-Fe2O3 appears at about 350, 500 and 700 cm�1 [20], such char-acteristics modes are also observed in the Raman spectrum of FeAl.

200 400 600 800 1000 1200 1400 1600 1800180

200

220

240

260

280

300

320(a)

200

250

300

350

400

Ram

an In

tens

ity (a

.u.)

Wavenumber (cm-1)

200 400 600 800 1000 1200 1400 1600 1800

Wavenumber (cm-1)

FeZr

Ram

an In

tens

ity (a

.u.)

Fig. 6. (a) Raman and (b) XRD an

Broad bands in the 200–600 cm�1 and 700–900 cm�1 ranges arethe main features in the FeTi and FeZr spectra, indicating thatthe hematite phase changed after the reaction.

Raman spectra at high wavenumbers display much more in-tense and blunter modes than in low wavenumber for all solids.The modes at 1300–1550 cm�1 are associated to the D and Gbands, as for carbon nanotubes [16]. This result suggests that thefast deactivation of FeZr and FeAl is due to the heavy coke deposi-tion whereas the slight decrease of the styrene yield on FeTi iscaused by the formation of labile carbon deposits on the latter.

XRD patterns of the solids are shown in Fig. 6b. After the cata-lytic test, FeAl displays low-intensity reflections corresponding tothe FeAl2O4 phase (ICDD no. 34-0192), where the (311), (440),(511), (400) planes corresponds to the 2h equal to 35.8, 44.2,55.0 and 64.3�, respectively. Peaks assigned to magnetite, maghe-mite and the two main peaks of carbon graphite at 2h = 31� and64� (ICDD no. 41-1487) can be identified too, which accord withthe results obtained from Mössbauer. Indeed, peaks related to c-Al2O3 are not observed. Judging from the broadness of the diffrac-tion lines, nanocrystalline dimensions of ca. 12–17 nm from the(311) plane of FeAl2O4 can be found for spent FeAl, in good agree-ment with the textural properties results (Table 1). FeAl2O4 is anormal spinel, where one eighth of the tetrahedral sites are occu-pied by Fe2+ cations whereas one half of the octahedral sites areoccupied by Al3+ cations [39]. These obtained results are compati-ble with the fact that the solid state reaction involving Al2O3 andFe2O3 in presence of CO to produce FeAl2O4 and carbon species[40] occurred during the ODH reaction.

Thus, the FeAl nanoparticles are confirmed to be crystalline, andsintering does not strongly affect the solid during the reaction.However, the yield of styrene decays to approximately 30% in thesame trend as selectivity due to the coking during the reaction.Some other studies found that the selectivity to styrene dropped

0

100

200

300

400

200 400 600 800 1000 1200 1400 1600 1800

Wavenumber (cm-1)

FeAl

Ram

an In

tens

ity (a

.u.)

alyses of the spent catalysts.

(b)

Fig. 6. (continued)

A.J.R. Castro et al. / Fuel 108 (2013) 740–748 747

along the reaction in reason of styrene oligomerization reactionsthat results in coke formation [41,42]. This confirms the Ramanspectroscopy measurements that detected the carbon deposits for-mation over the solid. Both sintering and coke formation are thecritical factors to deactivate FeAl catalyst along the reaction.

The XRD pattern of spent FeZr confirms the existence of thesame phases observed for the fresh analogous solid, including mag-netite, which was similar to that shown by Mössbauer. Indeed, thespent solid is more crystalline than the fresh analogue. This resultindicates that the change in valence state of Fe3+ (hematite) tomagnetite (mostly Fe2+) damaged the solid and sharply decreasedthe catalytic performance of FeZr. These results also imply thatthe poorer selectivity of FeZr towards ODH of ethylbenzene iscaused by coking and the reduction of the catalyst.

The spent FeTi exhibits the well crystalline TiO2 rutile and Fe3O4

monoxides besides the mixed FeTiO3, Fe2TiO5 and FeTi2O5 phases.It suggests that Fe2O3 reacts with TiO2 rutile during the reaction toform the mixed Fe–Ti phases [27] due to the reductive conditionsoffered by the ODH. Thus, styrene yield drops off at the beginningof the reaction due to sintering of the active species and the mes-opores may become occluded close to saturation by the coke fromethylbenzene cracking or styrene oligomerization. However, thepromoting effect of CO2 in reoxidizing the FeTiO3, Fe2TiO5 andFeTi2O5 phases and leaching the coke have probably avoided thedeactivation of the solid, as for Fe-based catalysts [43].

Studies on the activity of the solids by varying the CO2 feed con-tent as well as the promotion of the binary oxide by a third activecomponent are in course.

4. Conclusions

Porous iron-based oxides were produced by sol–gel method.Modification of the iron solids by adding Ti, Al or Zr resulted in dif-ferent mixed phases such as non-crystalline spinel and ilmienite,besides the segregated a-Fe2O3, c-Fe2O3, rutile TiO2 and t-ZrO2

monoxides. Based on XRD, Raman, Mössbauer and textural proper-ties analyses, it is clear that the effect of the porosity on the phasetransformation and physicochemical properties of mesoporous sol-ids was correlated to the catalytic activity of the solids. Catalyticevaluation for styrene production showed that the styrene yieldis high when Ti and Al are added to the solids whereas styreneselectivity decreases due to cracking of ethylbenzene reactions.The higher catalytic performance of porous FeTi is attributed tothe in situ formation of FeTiO3, Fe2TiO5 and FeTi2O5 active species.These phases were resistant against phase transformation and sin-tering but promoted coke formation due to both ethylbenzenecracking or styrene oligomerization reactions.

Acknowledgments

The authors gratefully acknowledged to CNPq (Process no.482364/2010-6) for supporting this research project and CETENEfor N2 adsorption–desorption measurements.

References

[1] Cavani F, Trifiro F. Alternative processes for the production of styrene. ApplCatal A 1995;133:219–39.

[2] Chen S, Qin Z, Wang G, Dong M, Wang J. Promoting effect of carbon dioxide onthe dehydrogenation of ethylbenzene over silica-supported vanadiumcatalysts. Fuel; in press. doi: 0.1016/j.fuel.2012.06.004.

[3] Xu P, Zeng GM, Lian Huang D, Feng CL, Hu S, Zhao MH, et al. Use of iron oxidenanomaterials in wastewater treatment: a review. Sci Total Environ2012;424:1–10.

[4] Mahmoudi M, Sant S, Wang B, Laurent S, Sen T. Superparamagnetic iron oxidenanoparticles (SPIONs): development, surface modification and applications inchemotherapy. Adv Drug Deliver Rev 2011;63:24–46.

[5] Yang C, Rait A, Pirollo KF, Dagata JA, Farkas N, Chang EH.Nanoimmunoliposome delivery of superparamagnetic iron oxide markedlyenhances targeting and uptake in human cancer cells in vitro and in vivo.Nanomed: Nanotechnol, Biol, Med 2008;4:318–29.

[6] Figuerola A, Corato R, Manna L, Pellegrino T. From iron oxide nanoparticlestowards advanced iron-based inorganic materials designed for biomedicalapplications. Pharmacol Res 2010;62:126–43.

[7] Park S. Preparation of iron oxides using ammonium iron citrate precursor: Thinfilms and nanoparticles. J Solid State Chem 2009;182:2456–60.

[8] Batista HEM, Honorato SB, Ayala AP, Filho JM, de Sousa FW, Pinheiro AN, et al.Ethylbenzene to chemicals: catalytic conversion of ethylbenzene into styreneover metal-containing MCM-41. J Mol Catal A: Chem 2010;315:86–98.

[9] Nigrovski B, Zavyalova U, Scholz P, Pollok K, Muller M, Ondruschk B.Microwave-assisted catalytic oxidative dehydrogenation of ethylbenzene oniron oxide loaded carbon nanotubes. Carbon 2008;46:1678–86.

[10] Araujo JCS, Freire FNA, Souza CBA, Oliveira AC, Ayala AP, Oliveira AC.Dehydrogenation of ethylbenzene with CO2 to produce styrene over Fe-containing ceramic composites. Appl Catal A: Gen 2010;377:55–63.

[11] Batista HEM, Ramos FSO, Pinheiro TB, Lima CL, de Sousa FF, Barros EBD, et al.Mesoporous MAl2O4 (M = Cu, Ni, Fe or Mg) spinels: characterisation and

748 A.J.R. Castro et al. / Fuel 108 (2013) 740–748

application in the catalytic dehydrogenation of ethylbenzene in the presenceof CO2. Appl Catal A: Gen 2010;382:148–57.

[12] Tang S, Wang L, Zhang Y, Li S, Tian S, Wang B. Study on preparation of Ca/Al/Fe3O4 magnetic composite solid catalyst and its application in biodieseltransesterification. Fuel Process Technol 2012;95:84–9.

[13] Ohishi Y, Kawabata T, Shishido T, Takaki K, Zhang Q, Wang Y, et al. Mg–Fe–Almixed oxides with mesoporous properties prepared from hydrotalcite asprecursors: Catalytic behaviour in ethylbenzene dehydrogenation. Appl CatalA: Gen 2005;288:220–31.

[14] Hossain MM, Atanda L, Al-Yassir N, Al-Khattaf S. Kinetics modeling ofethylbenzene dehydrogenation to styrene over a mesoporous aluminasupported iron catalyst. Chem Eng J 2012;207–208:308–21.

[15] Castro AJR, Marques SPD, Filho JM, Saraiva GD, Soares JM, Oliveira AC.Nanosized aluminium derived oxides catalysts prepared with differentmethods for styrene production. Chem Eng J 2012;209:345–55.

[16] Nogueira IM, Sabadia GQ, Moreira AA, Filho JM, Oliveira AC. Investigation ofthe deactivation of iron nanocomposites by coking in the dehydrogenation ofethylbenzene. J Mol Catal A: Chem 2011;351:81–92.

[17] Wang B, Zhao H, Zheng Y, Liu Z, Yan R, Zheng C. Chemical looping combustionof a Chinese anthracite with Fe2O3-based and CuO-based oxygen carriers. FuelProcess Technol 2012;96:104–15.

[18] Wang B, Yan R, Zheng Y, Zhao H, Zheng C. Mechanistic investigation ofchemical looping combustion of coal with Fe2O3 oxygen carrier. Fuel2011;90:2359–66.

[19] Chen YH. Synthesis, characterization and dye adsorption of ilmenitenanoparticles. J Non-Crystalline Solids 2011;357:136–9.

[20] Varadwaj KSK, Panigrahi MK, Ghose J. Effect of capping and particle size onRaman laser-induced degradation of c-Fe2O3 nanoparticles. J Solid State Chem2004;177:4286–92.

[21] Sobol AA, Voronko YK. Stress-induced cubic-tetragonal transformation inpartially stabilized ZrO2: Raman spectroscopy study. J Phys Chem Solid2004;65:1103–12.

[22] Liu BS, Wan ZY, Zhan YP, Au CT. Desulfurization of hot coal gas over high-surface-area LaMeOx/MCM-41 sorbents. Fuel 2012;98:95–102.

[23] Minesso A, Genna F, Genna F, Finotto F, Baldan A, Benedetti A. Synthesis andcharacterisation of sulfated zirconia sol–gel systems. J Sol–Gel, Sci Technol2002;24:197–206.

[24] Bomfim HEL, Oliveira AC, Rangel MC. Effect of zinc on the catalytic activity ofhematite in ethylbenzene dehydrogenation. React Kinet Catal Lett2003;80:359–64.

[25] Zhang C, He H, Tanaka K-I. Catalytic performance and mechanism of a Pt/TiO2

catalyst for the oxidation of formaldehyde at room temperature. Appl Catal B:Environ 2006;65:37–43.

[26] Watanabe S, Ma X, Song C. Characterization of structural and surfaceproperties of nanocrystalline TiO2�CeO2 mixed oxides by XRD, XPS, TPR, andTPD. J Phys Chem C 2009;113:14249–57.

[27] Corbella BM, Palacios JM. Titania-supported iron oxide as oxygen carrier forchemical-looping combustion of methane. Fuel 2007;86:113–22.

[28] Burri DR, Choi K-M, Han S-C, Burri A, Park S-E. Selective conversion ofethylbenzene into styrene over K2O/TiO2–ZrO2 catalysts: Unified effects of K2Oand CO2. J Mol Catal A: Chem 2007;269:58–63.

[29] Sivaranjani K, Verma A, Gopinath CS. Molecular oxygen-assisted oxidativedehydrogenation of ethylbenzene to styrene with nanocrystalline Ti1�xVxO2.Green Chem 2012;14:461–71.

[30] Mimura N, Saito M. Dehydrogenation of ethylbenzene to styrene over Fe2O3/Al2O3 catalysts in the presence of carbon dioxide. Catal Today 2000;55:173–8.

[31] Nederlof C, Talay G, Kapteijn F, Makkee M. The role of RWGS in thedehydrogenation of ethylbenzene to styrene in CO2. Appl Catal A: Gen2012;423–424:59–68.

[32] Oliveira AC, Fierro JLG, Valentini A, Nobre PSS, Rangel MC. Non-toxic Fe basedcatalysts for styrene synthesis: the effect of salt precursors and aluminumpromoter on the catalytic properties. Catal Today 2003;85:49–57.

[33] Burri DR, Choi KM, Han D-S, Koo JB, Park S-E. CO2 utilization as an oxidant inthe dehydrogenation of ethylbenzene to styrene over MnO2–ZrO2. Catal Today2006;115:242–7.

[34] Irún O, Sadosche S-A, Lasobras J, Soler J, Francés E, Herguido J, Menéndez M.Catalysts for the production of styrene from ethylbenzene: redox anddeactivation study. Catal Today, in press. doi.org/10.1016/j.cattod.2012.07.019.

[35] Tan S, Örs T, Aydınol MK, Öztürk T, Karakaya I. Synthesis of FeTi from mixedoxide precursors. J Alloys Compod 2009;475:368–72.

[36] Ji M, Chen G, Wang J, Wang X, Zhang T. Dehydrogenation of ethylbenzene tostyrene with CO2 over iron oxide-based catalysts. Catal Today 2010;158:464–9.

[37] Cristóbal AA, Ramos CP, Bercoff PG, Botta PM, Porto López JM. Effects of the Ti/Fe ratio on the phase composition and magnetic properties of mechanochemicallyactivated Ti–Fe2O3 mixtures. Mat Chem Phys 2012;133:971–6.

[38] Wu X, Qin S, Dubrovinsky L. Structural characterization of the FeTiO3–MnTiO3

solid solution. J Solid State Chem 2010;183:2483–9.[39] Dutta DP, Sharma G. Synthesis and magnetic behavior of spinel FeAl2O4

nanoparticles. Mat Sci Eng B 2011;176:177–80.[40] Duraes L, Costa BFO, Santos R, Correia A, Campos J, Portugal A. A Fe2O3/

aluminum thermite reaction intermediate and final products characterization.Mat Sci Eng A 2007;465:199–210.

[41] Rao R, Zhang Q, Liu H, Yang H, Ling Q, Yang M, et al. Enhanced catalyticperformance of CeO2 confined inside carbon nanotubes for dehydrogenation ofethylbenzene in the presence of CO2. J Mol Catal A: Chem 2012;363–364:283–90.

[42] Nederlof C, Kapteijn F, Makkee M. Catalysed ethylbenzene dehydrogenation inCO2 or N2—carbon deposits as the active phase. Appl Catal A: Gen2012;417:163–73.

[43] Badstube T, Papp H, Kustrowski P, Dziembaj R. Oxidative dehydrogenation ofethylbenzene with carbon dioxide on alkali-promoted Fe/active carboncatalysts. Catal Lett 1998;55:169–72.