Low temperature catalytic performance of

nanosized Ti–Ni–O for oxidative

dehydrogenation of propane to propene

Ying Wu *, Yiming He, Tong Chen, Weizheng Weng, Huilin Wan *

State Key Laboratory for Physical Chemistry of Solid Surfaces, Department of Chemistry and

Institute of Physical Chemistry, Xiamen University, 361005, China

Received 23 May 2005; received in revised form 5 July 2005; accepted 2 August 2005

Available online 12 September 2005

Abstract

Nanosized Ti–Ni–O catalysts prepared by a modified sol–gel method have been investigated in the oxidative dehydrogena-

tion of propane (ODP) to propene. At 300 8C the yield to propene of 12.1% was obtained on 9.1 wt.% Ti–Ni–O catalyst with the

selectivity of 43%. The continued variety of lattice parameter and variation of chemical value of nickel and titanium ion on the

surface indicates that there are strong interactions of TiO2 and NiO. The decreased low temperature oxygen desorption and the

weaker reducibility seems to be responsible for the decreased activity and enhanced selectivity of propane oxidative

dehydrogenation over Ti–Ni–O catalysts.

# 2005 Elsevier B.V. All rights reserved.

Keywords: Nanosized; Sol–gel; Oxidative dehydrogenation of propane; Ti–Ni–O

www.elsevier.com/locate/apsusc

Applied Surface Science 252 (2006) 5220–5226

1. Introduction

The oxidative dehydrogenation of propane (ODP)

to propene is one of the potentially important catalytic

processes for the effective utilization of light alkanes

[1,2]. Compared to the commercial production for

propene via non-oxidative dehydrogenation, the ODP

* Corresponding authors. Tel.: +86 592 2182440;

fax: +86 592 2182047

E-mail addresses: wuying@yanan.xmu.edu.cn (Y. Wu),

hlwan@xmu.edu.cn (H.L. Wan).

0169-4332/$ – see front matter # 2005 Elsevier B.V. All rights reserved

doi:10.1016/j.apsusc.2005.08.002

process is potentially advantageous, because the

reaction is exothermic and employs a cheaper and

more abundant raw material [3]. One of challenges for

the process is how to achieve a high selectivity and

yield to propene in the presence of oxygen at lower

temperature, because propene is more reactive than

propane and is prone to side reactions, such as

cracking and combustion. At moderate temperatures,

these processes are mainly heterogeneous. The

development of such a catalyst with a sufficiently

high activity and selectivity at low temperature is thus

highly desirable.

.

Y. Wu et al. / Applied Surface Science 252 (2006) 5220–5226 5221

The most studied catalytic systems for the reaction

are based on vanadia or molybdena [4–12]. Most of the

results obtained are at temperature higher than 500 8C,

and only few are concerned with 500 8C or lower

temperature [11]. Among other oxide systems, those

containing nickel are also reported as active and

selective [13–15]. Therein Ni–Ce–O catalyst showed

relatively better low temperature catalytic perfor-

mance with 11% propene yield at 300 8C [14].

On the other hand, ultrafine particle of nanometer

dimension is paid more attention as a kind of new

material. Due to their large proportion between

surface and bulk atom numbers, nanosized materials

are used in many fields including catalysis [16–18].

However, the reports on nanostructured catalysts

being used in the ODP reaction are rare. The

nanostuctured two-component oxide catalysts are

prepared often by co-precipitation way. In order to

getting a strong possibility of interactions between

metal species, however, the higher calcination

temperature is needed and leads to the larger obtained

particles size [19]. Compared with it, modified sol–gel

way is an alternate method to synthesis small and well-

distribution nanoparticles [20].

In this work, a series of TiO2 doped NiO catalysts

were prepared by modified sol–gel method. And the

prepared catalysts were studied with regard to the

behavior for the propane oxidative dehydrogenation to

propene. X-ray diffraction (XRD), high-resolution

transmission electron microscopy (HRTEM) and X-

ray photoelectron spectroscopy (XPS) were used to

determine the nature of the Ti species at various

weight doping. Temperature programmed reduction

(H2-TPR) and temperature programmed desorption

(O2-TPD) were used to study changes in redox

capability and desorption behavior as a function of

TiO2 content.

2. Experiment

2.1. Catalysts preparation

Catalysts were prepared using a modified sol–gel

method. Ni(NO3)2�6H2O was dissolved in stoichio-

metric amount of de-ionized water, respectively. A

series of calculated tetra-n-butyl titanate

((C4H9O)4Ti) and nickel nitrate solution were

simultaneously added drop by drop into measurable

citric acid solution, used as ligand, to generate a

completely homogeneous and transparent solution.

Then the solution was subject to slow evaporation at

70 8C until a highly viscous residual was formed.

After drying at 110 8C, a gel precursor was developed,

then calcinated at 400 8C for 4 h in muffle furnace. For

comparison purposes, pure nickel oxide and titania

were also prepared by the same method. Large size Ti–

Ni–O catalyst was prepared by the same method but

calcinated at 600 8C. What’s more, 9.1 wt.% Ti–Ni–O

catalyst was also prepared by co-precipitation method

and calcinated at 400 8C.

2.2. Activity testing

Catalytic activities were measured at atmospheric

pressure in a quartz fixed-bed micro-reactor

(i.d. = 6 mm). For each testing, 0.1 g catalyst was to

be used. The total flow rate of gaseous reactants was

9000 mL/(h gcat) with C3H8/O2/N2 molar ration of

1.1/1/4. The outlet mixture was analyzed on-line by a

gas chromatograph with impregnated squalane and

carbon molecular sieves being the columns.

Before catalyst testing an empty tube loaded with

nanosized TiO2 catalyst prepared by modified sol–gel

method was used on ODH of propane reaction and

showed inactive below 350 8C.

2.3. Catalyst characterization

BET surface area measurements and nitrogen

adsorption–desorption isotherms were recorded using

a TriStar3000 instrument. Phase identification and

structural analysis were performed by X-ray powder

diffraction (Model D/Max-C, Rigaku), with Cu target

(l = 0.15406 nm) and scanning rate of 8 8C/min. The

morphologies of the nanosized NiO and Ti–Ni–O

catalysts were observed by HRTEM (TECNAI F-30

FEG). H2-TPR was performed using a laboratory-

made chromatogram. 5% H2/Ar mixed gas was used as

reduction gas and its flow rate was 17 mL/min. The

heating rate was 10 8C/min. O2-TPD was conducted

on a mass spectrometer (Balzer Omni Star 200). The

catalysts (0.1 g) were first pretreated in air for 0.5 h at

400 8C and then cooled to room temperature in air.

After the system was purged with a flow of He to

smooth baseline, the sample was heated to 600 8C in

Y. Wu et al. / Applied Surface Science 252 (2006) 5220–52265222

Table 1

Catalytic performance of 9.1 wt.% Ti–Ni–O catalyst for oxidative dehydrogenation of propanea

Catalyst Temperature (8C) Conversion (%) C3H8 Selectivity (%) Yield (%) C3H6 STYc3H6(mol h�1 gcat�1)

C3H6 Crackb COxc

Ti–Ni–O-400d 250 22.7 40.5 0.3 59.2 9.2 0.149

275 27.3 42.2 0.4 57.4 11.5 0.187

300 28.4 42.5 0.6 56.9 12.1 0.196

Ti–Ni–O-600d 300 0.3 61.1 – 38.9 0.2 0.003

STY: space time yield.a GHSV = 9000 mL h�1 g�1; C3H8:O2:N2 = 1.1: 1: 4.b Crack = C2H6 + C2H4 + CH4.c COx = CO + CO2d Ti-Ni-O-400 and Ti-Ni-O-600 mean the 9.1 wt.% Ti–Ni–O catalysts prepared by modified sol–gel method and calcinated at 400 and 600 8C,

respectively.

He at a constant rate of 15 8C/min. X-ray photoelec-

tron spectroscopy was performed using a VG ESCAL

AB/Auger spectrometer, which was ideally suited to

probe the presence of different oxidation states via the

chemical shift observed. Spectra were corrected using

the C1s signal, located at 284.5 eV.

3. Results and discussion

3.1. Catalytic performance

The catalytic results for the oxidative dehydro-

genation of propane on 9.1% Ti–Ni–O catalyst are

presented in Table 1. It can be seen from this table that

the catalytic activity of nanosized Ti–Ni–O catalyst

calcinated at 400 8C is clearly better than the large size

one with the same doping calcinated at 600 8C, whose

average particle size is larger than 100 nm from the

Table 2. The propane conversion on nanosized Ti–Ni–

O catalyst (28%) outclassed than on the large size one

(0.3%). On the other hand, the catalytic results

Table 2

Average sizes, surface areas of Ti–Ni–O catalysts

TiO2 content (wt.%) NiO 4.7 9.1c 9.1d

Surface area (m2/g) 90 166 210 156

Particle size (nm) 9.9a 6.7a 5.8a 7.1a

a (A) 4.1860 4.1849 4.1838 4.18

a Calculated by Scherrer Formula.b Observed by HRTEM.c Prepared by modified sol–gel method.d Prepared by co-precipitation method.e Calcinated at 600 8C.

presented here show that the catalytic performance of

the catalyst strongly depends on the reaction

temperature. With reaction temperature rising, the

propane conversion increased while the propene

selectivity was almost constant with 42%. Since the

oxygen in the feeds were almost completely consumed

in the reactions at 300 8C, the propane conversions

increased unremarkably from 300 8C. The yield to

propene of ca. 11.5% with propane conversion of

27.3% have been obtained at 275 8C, considerably

lower than the temperature of V, Mo-based catalysts.

The reaction temperature is also lower than that

reported by Boizumault-Moriceau et al. for Ce–Ni–O

(300 8C) when the equivalent yield to propene is

obtained.

Fig. 1 presents the values of conversion of propane,

selectivity and yield to propene at 300 8C as a function

of the doped titania content. Propane and COx were the

main reaction products during the oxidation of

propane on Ti–Ni–O, whereas little cracking product

(<0.7%) and no oxygenated products other than COx

were observed. The catalysts showed a general

9.1e 16.7 28.6 50 TiO2

11 281 249 202 73

>100b 3.7a 4.5a �6b �8b

54 – 4.1821 4.1805 4.1788 –

Y. Wu et al. / Applied Surface Science 252 (2006) 5220–5226 5223

Fig. 2. XRD patterns of nanosized Ti–Ni–O catalysts with different

TiO2 content. (TiO2 content: (a) NiO, (b) 4.8 wt.%, (c) 9.1 wt.%, (d)

16.7 wt.%, (e) 28.6 wt.%, (f) 50 wt.%, (g) TiO2).

Fig. 1. Catalytic data obtained at 300 8C for Ti–Ni–O catalysts as a

function of the TiO2 content.

increase in propene selectivity with increasing

amounts of TiO2 except 50 wt.% Ti–Ni–O. On the

other hand, the conversion of propane decreased with

the increasing of TiO2 doping. While increasing TiO2

content in the catalysts, the yield to propene decreased

monotonously. A maximum yield to propene (12.1%)

was obtained on 9.1% Ti–Ni–O catalyst. Since TiO2 is

inactive at 300 8C, the catalytic performance is

ascribed to the synergic effect of doped TiO2 to NiO.

3.2. Physical properties

The physical properties of the catalysts are given in

Table 2. High-resolution transmission electron micro-

scopy on both NiO and catalysts containing varying

amounts of TiO2 calcinated at 400 8C shows spherical

nanoparticles. For catalysts containing up to

16.7 wt.% TiO2, particle sizes decreased and surface

areas increased with increasing TiO2 content due to

the dispersion of doped TiO2. From 28.6% doping

amount of TiO2 the surface area deceased gradually.

Comparison of the ODP data and surface area of the

catalysts, it can be found that catalytic performance is

incompletely dependent of surface area.

3.3. XRD studies

Fig. 2 shows the effect of doped TiO2 on the XRD

patterns of NiO catalysts prepared by modified sol–gel

method. A weak anatase phase of TiO2 was observed

at 258 in nanosized TiO2. For the series of Ti–Ni–O

catalysts three distinct diffraction peaks at 2u = 37, 43,

638 were observed. These peaks positions correspond

well with three most intense peaks of pure cubic NiO

[21]. No XRD peaks corresponding to titanium oxide

have been observed. That means the doped TiO2 is

well dispersed in NiO. The half-width of nanosized

Ti–Ni–O peaks gets wider with the increasing of TiO2

content. The average crystalline sizes of the catalysts

determined by XRD and calculated by Scherrer

Formula (T = 0.89l/b cos u) was smaller than the

values observed by HRTEM, which is ascribed to that

observations by imaging techniques such as HRTEM

often gives the size of the particle while the X-ray

diffraction discloses the size of crystalline. Unfortu-

nately, due to the larger width of peak the crystalline

size of catalyst with more than 28.6% TiO2 couldn’t be

evaluated through Scherrer formula because it is

applicable to the particles whose sizes are between 3

and 200 nm [22]. The particles sizes of 50% Ti–Ni–O

and TiO2 catalysts were observed only by HRTEM

directly.

Besides, the lattice parameters of nanostructured

Ti–Ni–O were calculated by peaks positions through

Debye-Scherrer method. The lattice parameter a of

nanostructured Ti–Ni–O, listed in Table 1, mono-

tonously decreased with the increasing doping amount

of TiO2. Such decrease indicates that some of smaller

Ti4+ (ionic radius is 0.61 A) [23] could enter the lattice

of nickel oxide and substitutes smaller size Ni2+

(0.69 A) [23] site, which results in the lattice

distortion. The continued variety of lattice parameter

Y. Wu et al. / Applied Surface Science 252 (2006) 5220–52265224

clearly demonstrates the existence of the solid solution

of Ti4+ ions into NiO, though no peaks from the solid

solution phases are observed due to the less amount.

It’s suggested that there are strong interaction between

TiO2 and NiO. The surface area of Ti–Ni–O prepared

by co-precipitation were smaller than Ti–Ni–O

prepared by modified sol–gel method, however, and

the lattice parameter of the former were close to that of

pure NiO, which proved that the distribution of

nanoparticles prepared by sol–gel way was better than

co-precipitation.

3.4. XPS studies

To study the surface states of catalysts, XPS was

performed over samples with various TiO2 doping

amount. The Ni2p3/2 XPS spectra (Fig. 3(a)) showed

there were a peak located at 853.4 eV and a satellite at

855.3 eV for nanosized NiO, which is related with

Ni2O3 [24–26]. The Ni2p3/2 binding energy (BE) was

seen shifting to higher values with increasing TiO2

content, from 853.9 eV (4.7 wt.% Ti–Ni–O) to

856.5 eV (50 wt.%). The BE of the latter one is near

that of Ni3+. It can be speculated that nanosized Ti–

Ni–O contains more Ni2+d contents, so the BE was

more close to that of Ni2O3.

As Fig. 3(b) showed, in O1s spectra the binding

energies of Ti–Ni–O catalysts were between that of

NiO and TiO2. The BE was seen to shift from 529.5 eV

for 4.7 wt.% Ti–Ni–O to 530.4 eV for 50 wt.% Ti–Ni–

Fig. 3. XPS profile of nanosized Ti–Ni–O catalysts with different Ti wt

16.7 wt.%, (e) 28.6 wt.%, (f) 50 wt.%, (g) TiO2).

O approaching that of TiO2 (ca. 531.4 eV). This is

because higher valance Ti4+ is easier to induce O2�

than Ni2+, then the electron density around O2�

decreases and the BE of catalyst increases. The BE

shifted to higher value with the TiO2 content increases.

The Ti2p3/2 BE showed in Fig. 3(c) was seen

shifting to lower values with TiO2 doping increases.

The BE of 4.7 wt.% Ti–Ni–O (457.5 eV) was 0.6 eV

less than that of nanosized TiO2 (458.1 eV). The

electron transfer is easier to occur from lower valence

Ni2+ to higher valence Ti4+ through the approximate

bridge Ti–O–Ni interaction species. Then the Ti site

may be an electron-rich species and shifts to lower BE

values.

Though only the peaks of NiO were detected in

XRD patterns, the XPS results clearly indicate that the

chemical valence of nickel and titanium ion on the

surface has distinct variation compared with pure

nanosized NiO or TiO2. It further confirms that there

are strong interactions between doped TiO2 and NiO,

which may play an important role on the catalytic

performance.

3.5. O2-TPD-MS studies

The O2-TPD-MS profiles for Ti–Ni–O catalysts

with different TiO2 content are shown in Fig. 4. For

comparison, desorption profiles for the nanosized NiO

and TiO2 have also been shown. For nanosized TiO2

prepared by modified sol–gel method, no desorption

.% doping. (TiO2 content: (a) NiO, (b) 4.8 wt.%, (c) 9.1 wt.%, (d)

Y. Wu et al. / Applied Surface Science 252 (2006) 5220–5226 5225

Fig. 5. H2-TPR profiles of nanosized Ti–Ni–O catalysts with dif-

ferent TiO2 content. (TiO2 content: (a) NiO, (b) 4.8 wt.%, (c)

9.1 wt.%, (d) 16.7 wt.%, (e) 28.6 wt.%, (f) 50 wt.%, (g) TiO2).

Fig. 4. O2-TPD-MS spectra of nanosized Ti–Ni–O catalysts with

different TiO2 content. (TiO2 content: (a) NiO, (b) 4.8 wt.%, (c)

9.1 wt.%, (d) 16.7 wt.%, (e) 28.6 wt.%, (f) 50 wt.%, (g) TiO2).

peak were observed in this temperature range. For

nanosized NiO, there were two distinct peaks at about

300 and 480 8C, respectively, which means that at

least two different types of oxygen exist. With the

doping amount of TiO2 increases, the peak areas of Ti–

Ni–O catalysts around 300 8C were decreased, the

area of the second peak was also smaller than that of

NiO. And the desorption temperatures of the second

peaks were shifted to low temperature. At higher

dopings (50 wt.% TiO2) no obvious desorption peaks

were observed.

It seems that the desorption temperature of the first

peak is coincide with the reaction temperature, which

indicates that more oxygen species in Ti–Ni–O

catalysts are much easy accessible to reactants and

may be related to their catalytic performance.

Combined with the catalytic results in Fig. 2, it can

be found that the more oxygen desorption at reaction

temperature, the higher the catalytic activity over the

catalyst. It’s coincide with the reported viewpoint that

the lower temperature desorption peak is involved

with the oxygen species which is related with high

reaction activity [27].

3.6. H2-TPR studies

Fig. 5 illustrated the H2-TPR profiles of nanosized

NiO and Ti–Ni–O catalysts with different doping

amount of TiO2. There was no distinct reduction peak

for nanosized TiO2 below 600 8C. For nanosized NiO

there were two narrower reduction peaks, of which the

low temperature peak is relevant to the reduction of

higher valence state NiO1+d on the surface, while the

higher temperature peak belongs to the reduction of

bulk nickel oxide. Compared with the H2-TPR curve

of nanosized NiO, the broader reduction peaks of Ti–

Ni–O were shifted to high temperatures. It’s obvious

that the addition of titania leads to that catalysts is

more difficult to be reduced due to the interaction of

nickel and titanium oxide. The deep oxidation of

propane to COx is restrained at a certain extent and

thus C3H6 selectivity on Ti–Ni–O is higher than that

on NiO. The interactions have complex effects on the

overall reducibility of the catalysts as indicated by the

shifted of temperatures of the reduction peak maxima

(Tmax). With the increase of TiO2 content, the Tmax of

catalyst is shift to higher temperature and propene

selectivity has been enhanced. Over 50 wt.% Ti–Ni–O

catalyst, however, the Tmax is shifted to lower value

and propene selectivity is decreased again.

4. Conclusion

Ti–Ni–O catalysts prepared by modified sol–gel

method were nano-sized and well-distributed. The

addition of TiO2 to NiO leads to the decreased lattice

parameter of the catalysts and the variation of

chemical value of nickel and titanium ion on the

surface, and it’s suggested that there are interactions

between NiO and doped TiO2. Compared with pure

NiO, over Ti–Ni–O catalysts the propene selectivity is

Y. Wu et al. / Applied Surface Science 252 (2006) 5220–52265226

remarkably enhanced due to the decreased reduci-

bility, and the activity of catalysts decreases with TiO2

content because of the decreased oxygen adsorption.

Therein the 9.1 wt.% Ti–Ni–O catalyst exhibited best

low-temperature catalytic performance for oxidative

dehydrogenation of propane, with 11.5% propane at

275 8C. The reaction temperature is lower than that

over the large size one or Ti–Ni–O catalyst prepared

by co-precipitation method with the same TiO2

content.

Acknowledgements

This work was supported by the Ministry of

Science and Technology (No. G1999022408) and

Natural Science Foundation of Fujian, China (No.

E02100002).

Reference

B. Silberova, M. Fathi, A. Holmen, Appl. Catal. A: Gen. 276

(2004) 17.

[2] D. Creaser, B. Andersson, R.R. Hudgins, P.L. Silveston, J.

Catal. 182 (1999) 264.

[3] Buyevskaya, D. Wolf, M. Baerns, Catal. Today 62 (2000) 91.

[4] T. Davies, S.H. Taylor, J. Mol. Catal. A: Chem. 220 (2004) 77.

[5] M. De, D. Kunzru, Catal. Lett. 96 (2004) 33.

[6] Z.S. Chao, E. Ruckenstein, Catal. Lett. 94 (2004) 217.

[7] J.B. Stelzer, J. Caro, M. Fait, Catal. Commun. 6 (2005) 1.

[8] R.B. Watson, U.S. Ozkan, J. Catal. 191 (2000) 12.

[9] E.V. Kondratenko, O.V. Buyevskaya, M. Baerns, Topic Catal.

15 (2001) 2.

[10] A. Pantazidis, S.A. Bucholz, H.W. Zanthoff, Y. Schuurman, C.

Mirodatos, Catal. Today 40 (1998) 207.

[11] Z.M. Fang, Q. Hong, Z.H. Zhou, S.J. Dai, W.Z. Weng, H.L.

Wan, Catal. Lett. 61 (1999) 39.

[12] K. Samson, A. Klisinska, I. Gressel, B. Grzybowska, React.

Kinet. Catal. Lett. 77 (2) (2002) 309.

[13] B. Zhaorigetu, W. Li, H. Xu, R. Kieffer, Catal. Lett. 94 (2004)

125.

[14] P. Boizumault-Moriceau, A. Pennequin, B. Grzybowska, Y.

Barbaux, Appl. Catal. A: Gen. 245 (2003) 55.

[15] B. Zhaorigetu, W. Li, R. Kieffer, H. Xu, React. Kinet. Catal.

Lett. 75 (2) (2002) 275.

[16] J.H. Adair, T. Li, K. Havey, J. Moon, J. Mecholsky, A.

Morrone, D.R. Talham, M.H. Ludwig, L. Wang, Mater. Sci.

Eng. R 23 (1998) 139.

[17] M. Devinder, G. Philipp, S. Ulrich, Catal. Commun. 4 (3)

(2003) 101.

[18] L.M. Hair, L. Owens, T. Tillotson, J. Wong, G.J. Thomas, D.L.

Medlin, M. Froba, J. Non-Crystal. Sol. 186 (1995)

168.

[19] M. Veith, M. Hass, V. Huch, Chem. Mater. 17 (2005) 95.

[20] B.S. Liu, C.T. Au, Catal. Lett. 85 (2003) 165.

[21] L. Xiang, X.Y. Deng, Y. Jin, Sci. Mater. 47 (2002) 219.

[22] W.J. Zhang, Petrochem. Technol. 30 (8) (2001) 651.

[23] J.A. Dean (ed.), Lange’s Handbook of Chemistry, the 15th

version.

[24] M.S. Hedge, M. Ayyoob, Surf. Sci. 173 (1986) L635.

[25] A.F. Carley, S.D. Jackson, M.W. Roberts, J. O’Shea, Surf. Sci.

454 (2000) 141.

[26] C.N.R. Rao, V. Vijayakrishman, M.K. Rajumon, G.U. Kulk-

arni, Appl. Surf. Sci. 84 (1995) 285.

[27] T. Chen, W. Li, C. Yu, R. Jin, H. Xu, Stud. Surf. Sci. Catal.

130B (2000) 1847.