Synthesis, characterization, and catalytic performance of highly dispersed vanadium grafted SBA-15 catalyst

Journal of Catalysis 253 (2008) 74–90

www.elsevier.com/locate/jcat

Synthesis, characterization, and catalytic performance

of highly dispersed vanadium grafted SBA-15 catalyst

∗

Guoan Du, Sangyun Lim, Mathieu Pinault, Chuan Wang, Fang Fang, Lisa Pfefferle, Gary L. Haller

Department of Chemical Engineering, Yale University, P.O. Box 208286, New Haven, CT 06520-8286, USA

Received 9 July 2007; revised 23 October 2007; accepted 25 October 2007

Abstract

Vanadium oxide grafted on mesoporous silica SBA-15 has been synthesized using a controlled grafting process. Its structure has been thoroughly

investigated using different characterization techniques, including N -physisorption, X-ray diffraction, transmission electron microscopy (TEM),

2

Raman spectroscopy, H temperature-programmed reduction, X-ray absorption near-edge structure (XANES), and extended X-ray absorption ﬁne

2

structure (EXAFS). The spectroscopic results revealed that under dehydrated conditions, the grafted vanadium domains are highly dispersed on

the SBA-15 surface, composed predominately of isolated VO units with distorted tetrahedral coordination. The suggested (≡SiO) V=O sites on

4

3

the silica surface include one short bond (∼1.54 Å) and three long bonds (1.74 Å). Methanol oxidation was used as a chemical probe reaction to

examine the catalytic properties of these catalysts. At low vanadium loading, the vanadium species grafted on the surface show structural properties

similar to those of vanadium-incorporated MCM-41 catalyst. However, the present mesoporous V-SBA-15 catalysts in the oxidation of methanol

to formaldehyde show remarkable catalytic performance compared with that of VOx/SBA-15 catalysts synthesized through a conventional wet

impregnation method, which has been attributed to the homogeneous dispersion and uniformity of the catalytic vanadium species achieved on

the SBA-15 support with large pore diameter and surface area. The acidic properties of V-SBA-15 was investigated by pyridine temperature-

programmed desorption, which indicated the existence of both Lewis and Brönsted acid sites of the surface.

©

2007 Published by Elsevier Inc.

Keywords: Vanadium grafted SBA-15; Methanol oxidation; XRD; Raman; H -TPR; TEM; X-ray absorption (XANES and EXAFS)

2

1

. Introduction

high-surface area supports, such as silica, alumina, titania, and

zirconia, as well as the increased stability of active species on a

support. Although extensive efforts have been made to control

the dispersion on supported catalysts, the development of a sim-

ple methodology to prepare highly dispersed vanadium metal

oxide with high vanadium loading remains a challenge [13].

Many reports have addressed the synthesis and characteri-

zation of supported vanadium oxide species, and the chemical

and structural nature of the vanadia on the supports [14–17].

Improved catalytic performance is often attributed to isolated

tetrahedral monovanadate (VO4) species [16,18]. Using the in-

cipient wet impregnation method, the isolated, three-legged

Supported vanadium catalysts consisting of a surface vana-

dium oxide phase on a high-surface area oxide support have

been extensively reported to be active and selective for the

partial oxidation of hydrocarbons, oxidative dehydrogenation

of alkanes to alkenes, selective catalytic reduction of NOx,

and oxidation of SO2 [1–6]. The catalytic activity largely de-

pends on the dispersion of vanadium [7,8], the metal–support

interaction [5,9] and the nature of the vanadium active sites.

Isolated tetrahedral vanadium oxide species containing the ter-

minal V=O group have been proposed to be the active sites for

many reactions [10–12].

(

≡SiO)3V=O species was found to be the most favorable site at

low coverage on SiO2, Al2O3, TiO2, ZrO2, and HfO2 supports

7,10,19]. However, the precise control of vanadium-isolated

Recent approaches in developing catalysts have used the ho-

mogeneous dispersion of metal species on the surface of porous

[

active site dispersion has not been successfully achieved by

this method, resulting in numerous surface species, including

polymeric vanadia species and bulk crystalline V2O5, even at

*

Corresponding author.

E-mail address: gary.haller@yale.edu (G.L. Haller).

doi:10.1016/j.jcat.2007.10.019

G. Du et al. / Journal of Catalysis 253 (2008) 74–90

75

low V loadings [20,21]. Since the discovery of mesoporous

siliceous materials, such as MCM-41, MCM-48, and SBA-

been successfully used to prepare tungsten oxide catalyst sup-

ported on SBA-15 mesoporous silica [43]. Herein we report a

method of grafting vanadium oxide species onto the surface of

mesoporous silica SBA-15 using the method of atomic layer

deposition. This grafting approach was performed under strictly

anhydrous conditions and in an organic solvent system, to avoid

hydrolysis of the metal precursor and aggregation of the vana-

dia surface species to form oligomers after the grafting process.

To gain further insight into the dispersity and the nature of

vanadia species on the SBA-15 support, we performed detailed

characterization techniques, including N2-physisorption, X-ray

diffraction (XRD), transmission electron microscopy (TEM),

Raman spectroscopy, H2 temperature-programmed reduction,

X-ray absorption near-edge structure (XANES), and extended

X-ray absorption ﬁne structure (EXAFS). We investigated the

catalytic properties by the methanol oxidation reaction and

compared them with those of the vanadium catalysts synthe-

sized via conventional wet impregnation and with selected re-

ports in the literature.

1

5, extensive research efforts have focused on the applica-

tion of these materials as catalyst supports [22–24]. How-

ever, satisfactory single site dispersion still cannot be eas-

ily obtained. Through a modiﬁed Beck’s method, vanadium-

incorporated MCM-41 catalysts have been successfully syn-

thesized by atomic substitution of vanadium ions for Si in the

mesoporous framework, resulting in a catalyst with homoge-

neous, isolated vanadium active sites [25–27]. Through char-

acterization with ⁵¹V-NMR and X-ray absorption techniques,

the vanadium species have been proposed to be in a distorted

tetrahedral coordination local environment with a V=O dou-

ble bond exposed to the pore surface acting as the active sites

and three V–O–Si bonds to the silica framework [25,26]. This

type of vanadium-incorporated catalyst shows signiﬁcant ad-

vantages in various oxidation reactions, such as methanol ox-

idation [15] and methane partial oxidation [28]; however, the

low vanadium loading that incorporation can achieve limits fur-

ther increase in the product yields.

Several grafting methods have been widely used to pre-

pare catalysts on supports [29,30]. The chemical vapor depo-

sition (CVD) process using volatile molecular metal precur-

2. Experimental

2.1. Catalyst synthesis

i

sors [i.e., O=VCl3 [31,32], O=V(OC2H5)3 [9] or O=V(O Pr)3

[

32]] to modify the oxide surfaces and provide a way to con-

Mesoporous siliceous SBA-15 was prepared according to

a well-established procedure delineated by Stucky et al. [42]

using tetraethylorthosilicate (TEOS) and triblock co-polymer

poly(ethylene glycol)–block–poly(propylene glycol)–block–

poly(ethylene glycol) EO20PO70EO20 (Mav = 5800, from

Aldrich) as a template agent under acidic conditions. Typically,

a solution of EO20PO70EO20:HCl:TEOS:H2O = 2:60:4.25:12

(mass ratio) was prepared by dissolving 4 g of EO20PO70EO20

polymer in 80 g of 2 M HCl and 20 g of H2O under stirring

before the addition of 8.8 g of TEOS dropwise. The solution

mixture was stirred at 313 K for 20 h and then aged at 373 K

in an autoclave for 48 h. After cooling to room temperature, the

resulting solid was recovered by ﬁltration, washed in deionized

water, and dried under ambient conditions. The predried solid

was heated at a constant rate from room temperature to 813 K

over 17 h under He and held for 1 h under the same conditions,

followed by calcination at 813.2 K for 5 h with air to remove

the residual organic template materials, yielding the ﬁnal meso-

porous SBA-15 materials.

The vanadium-grafted SBA-15 catalysts were synthesized

following a controlled grafting process through atomic layer

deposition (ALD) [43]. The as-calcined SBA-15 sample was

suspended in anhydrous toluene (Aldrich, 99%) by reﬂuxing

under an inert N2 atmosphere for 5 h to remove any adsorbed

water. The synthesis of the vanadium precursor and the grafting

process were performed under an inert nitrogen atmosphere.

Commercial oxovanadium(V) chloride (Aldrich) was used to

prepare the vanadium precursor solution. Typically, 0.025 ml of

oxovadadium(V) chloride (for the 1.4V-SBA-15 sample) was

added dropwise via a syringe in 150 ml of toluene at room

temperature, followed by the addition of 20 ml of anhydrous

ethanol (Aldrich). The mixed solution was reﬂuxed for 8 h

with nitrogen bubbled from the bottom of the solution until

trol the dispersion of the active sites. The gas–solid reactions

of these metal complexes with the surface silica yields iso-

lated and uniform vanadium sites [31,32], whose structures

were characterized extensively by ⁵¹V magic-angle spinning

NMR spectroscopy [10,31,33–37] as well as the X-ray ab-

sorption spectroscopy (XANES and EXAFS) [38]. The grafted

vanadium is proposed to be in a distorted tetrahedral envi-

ronment with the molecular complex on the surface assigned

to ≡SiOVOX2 [38]. More recently, a novel so-called ther-

molytic molecular precursor (TMP) method has been developed

and used to control active site dispersion at a molecular level

[

30,39]. In this method, oxygen-rich metal siloxide complexes

are grafted onto an oxide support via a protonolysis reaction,

followed by low-temperature calcination to remove the remain-

ing hydrocarbons [40].

The new discovery of the family of siliceous mesostructured

material SBA-15 achieved by using a triblock copolymer as

structure-directing agent under acidic conditions [41,42] has

opened new opportunities in the ﬁeld of heterogeneous catal-

ysis. Mesoporous SBA-15 silica has become a promising oxide

support for many catalysts, owing to its high surface area (600–

2

1

000 m /g) and silanol concentration, allowing the immobi-

lization of active sites with high dispersion. It has a hexagonal

array of uniform tubular channels with pore diameters in the

range of 5–30 nm, which is much larger than that of MCM-41,

allowing much lower diffusion resistance. Moreover, the thick

pore wall (31–64 Å) relative to MCM-41 provides high ther-

mal stability and hydrothermal stability. Several studies have

used the mesoporous SBA-15 as a support for vanadium species

either by the conventional wet impregnation method or the ther-

malytic molecular precursor route. Recently, a novel controlled

grafting process through atomic layer deposition (ALD) has

76

G. Du et al. / Journal of Catalysis 253 (2008) 74–90

the oxovanadium chloride fully reacted with ethanol and the re-

leased HCl was dispelled from the system by the nitrogen ﬂow.

The solution was then cooled to room temperature. The excess

amount of ethanol was distilled from the precursor solution

before the pre-prepared SBA-15 toluene suspension (1.0 g of

SBA-15) was added, followed by reﬂuxing at 383 K overnight

were analyzed using the UWXAFS analysis package [47]. The

theoretical EXAFS function for vanadium oxide species gen-

erated by the FEFF6 program was used to ﬁt the experimental

data to calculate the V–O shell coordination numbers [48]. The

gas inlet, outlet, and heating and cooling unit allow the in situ

treatment of samples and reaction. The hydrated samples were

prepared by exposing the vanadium-grafted SBA-15 catalysts in

ambient conditions for 3 h. The dehydrated samples were pre-

pared by in situ heating of the sample wafer in ultra-high-purity

He at 623 K for 30 min. To test the reducibility and stability

of V-SBA-15, the sample was reduced in ultra-high-purity H₂

at 473, 623, and 773 K for 30 min, then reoxidized in air at

773 K for 30 min.

in a N atmosphere. The resulting mixture was cooled, ﬁltered,

2

and washed with anhydrous toluene several times, then dried at

room temperature overnight. The unreacted vanadium precur-

sor and byproducts from the grafting reaction were removed by

washing with toluene. The as-synthesized sample calcined in

the same process as above yielded the ﬁnal vanadium-grafted

SBA-15 catalysts.

The grafting process involved the reaction of O=V(OEt)₃

with surface silanol groups, causing cleavage of the V–O bond,

and readily formed in situ EtOH. The as-formed EtOH could be

easily removed from the toluene solvent system in the process

of reﬂuxing (383 K) due to its low boiling point, making the

grafting step highly favorable. Thus, it is reasonable to assume

that the three-legged (≡SiO)3V=O species is the most favor-

able species on the siliceous SBA-15 surface.

The impregnated sample was prepared by the traditional in-

cipient wet impregnation of siliceous SBA-15 or Cab-O-Sil

with vanadyl sulfate trihydrate (VOSO4·3H2O, Aldrich 99%)

aqueous solution. Then the sample was calcined at 813 K for 3 h

under air. The vanadium-incorporated MCM-41 catalyst was

synthesized as a comparison material using a modiﬁed Beck’s

method, as described in our previous work [4,26].

2.2.4. H TPR

2

The reducibility and the stability of V-SBA-15 and impreg-

nated V/SBA-15 samples were investigated by a TPR technique

using a gas chromatograph (Agilent 6890 Plus) with a ther-

mal conductivity detector (TCD). Approximately 50 mg of each

sample was loaded into a quartz cell. Before each TPR run, the

sample cell was purged by ultra-zero grade air at room temper-

ature; then the temperature was increased to 773 K at a rate of

5 K/min, soaked for 1 h at the same temperature, and cooled

to room temperature. This procedure produces a clean surface

before the TPR is run. The gas ﬂow was switched to 5 vol%

hydrogen balanced by argon. After baseline stabilization, the

sample cell was heated at a rate of 5 K/min to 1123 K and held

there for 1 h to ensure complete reduction. A frozen acetone

trap was installed between the sample cell and the TCD to con-

dense water produced by the sample reduction.

2

.2. Characterization

.2.1. Nitrogen physisorption

2.2.5. HR-TEM

Nitrogen adsorption–desorption isotherms at 77 K were

As a complementary technique, HR-TEM was used to visu-

alize the structure of V-SBA-15. It was performed on a Phillips

Tecnai F20 200-kV microscope.

measured with a Quantachrome Autosorb-3b static volumetric

instrument. Before measurement, the samples were outgassed

−

4

at 573 K to a residual pressure below 10 Torr. A Baratron

pressure transducer (0.001–10 Torr) was used for low-pressure

measurements. The speciﬁc surface areas were calculated fol-

lowing the BET method [44]. Pore size and pore size distribu-

tion were calculated by the BJH method [45] using the desorp-

tion isotherm branch.

2

.2.6. Raman spectroscopy

Raman spectra were recorded using an excitation wave-

length of 532 nm on a Jobin Yvon Horiba LabRam instrument

equipped with an Olympus confocal microscope. The samples

were pressed into wafers and dehydrated at 423 K for overnight

to remove surface moisture before each scan. The hydrated

samples were prepared through exposure to ambient conditions

for 2 h at room temperature. Spectra were collected at different

locations of the wafer and averaged to obtain the ﬁnal spectrum.

2

.2.2. XRD

The XRD measurements were conducted using a Bruker

AXS D8Focus diffractometer (CuKα, λ = 0.15, 4 nm, 40 KV,

3

0 mA).

2

.2.7. ICP analysis

The vanadium content of each sample was measured by ICP

2

.2.3. X-ray absorption

X-Ray absorption data were collected at the V K edge

(

5465 eV) according to the method described by Wong et

at Galbraith Laboratories, Inc.

al. [46], using Si (111) as the monochromator crystal at beam

line X23B (National Synchrotron Light Source, Brookhaven

National Laboratory). Typically, a ca. 100-mg sample was

pressed into self-supporting wafers and placed in a stainless

steel cell. Incidence and transmission absorbance of the X-ray

beam were measured by ion chambers ﬁlled with pure nitrogen.

EXAFS in the transmission mode was recorded from 200 eV

below to 900 eV above the V K edge. The spectra collected

2.2.8. TPD

The acid site strength and species of V-SBA-15 samples with

different vanadium content were estimated by TPD of pyri-

dine. The adsorption temperature of pyridine and the temper-

ature downstream of the gas chromatograph were set at 423 K,

to avoid the condensation of pyridine. After 30 min of pyri-

dine adsorption, the physisorbed pyridine was ﬂushed out by

G. Du et al. / Journal of Catalysis 253 (2008) 74–90

77

Fig. 1. Nitrogen physisorption of vanadium grafted SBA-15 samples with different vanadium loadings.

a high ﬂow rate of He for 1 h. The temperature was increased

to 1123 K at a rate of 5 K/min and held there for 1 h. The

signal was collected with a n Agilent 6890 Plus gas chromato-

graph equipped with a TCD. The carrier gas (He) ﬂow rate was

was achieved in a gas chromatograph column with temperature

programming. Blank runs were performed on the empty quartz

reactor, and no signiﬁcant conversion of methanol was found.

Before reaction, the catalysts were pretreated in situ in a

stream of 10 ml/min ultra-zero grade air at 773 K for 40 min

to obtain a clean surface with respect to hydrocarbon contami-

nation. After pretreatment, the catalysts were then ﬂushed with

ultra-high-purity He for 20 min. The methanol oxidation was

then carried out over a 100-K temperature range (573–673 K)

to obtain the conversion, selectivity, and space time yield data.

4

0 ml/min.

2

.3. Catalyst evaluation

In this work, we used the gas-phase partial oxidation of

methanol as a probe reaction to evaluate the catalytic perfor-

mance of the vanadium-grafted catalysts. Typically, a 50-mg

catalyst sample was loaded into the quartz reactor, with the ther-

mocouple inserted into the catalyst bed for better control of the

actual pretreatment and reaction temperature. The reactor was

installed vertically inside a 127-mm-i.d. tubular furnace con-

trolled by a programmable PID temperature controller (Omega

CN2011J). All of the gases were controlled by calibrated mass

ﬂow controllers (Brooks). The pressure was adjusted with a

needle valve and measured on a pressure gauge (Matheson,

3

. Results and discussion

3

.1. Characterization

3.1.1. Nitrogen physisorption

As a primary measurement for the physical properties, ni-

trogen physisorption was conducted for a series of V-SBA-

15 samples with different vanadium loadings (nominal 1.4%–

21.6 wt%). The nitrogen adsorption/desorption isotherms and

the corresponding pore size distributions are shown in Fig. 1a.

All of the vanadium-grafted samples exhibited type IV iso-

0

–100 psi). The helium, with constant ﬂow rate controlled by

a mass ﬂow controller, was bubbled through a methanol satu-

rator to produce constant concentration of methanol vapor in

the feed. Oxygen (2 vol%) balanced with ultra-high-purity he-

lium was used to create a reactant feed mixture of methanol

therms with a H -type broad hysteresis loop [49], correspond-

1

ing to a typical large-pore mesoporous materials with 1D cylin-

drical channels. Capillary condensation of nitrogen with uni-

form mesopores occurred, causing a sudden step increase in

nitrogen uptake in the characteristic relative pressure (P/P₀)

range of 0.6–0.8 for all the grafted samples studied, suggest-

ing typical mesoporous structure with uniform pore diameters.

Because SBA-15 exhibited a uniform hexagonal array of meso-

pores connected by smaller micropores, the broad hysteresis

loop in the isotherms for all of the vanadium-grafted samples

(

1.9 vol%), oxygen (2.7 vol%), and helium (95.4 vol%) bal-

ance. The downstream of the reactor to the gas chromatograph

was heated at 393 K, to avoid condensation of the products. The

feed and products were analyzed by an online programmable

gas chromatograph (HP 6890) equipped with a TCD, a ﬂame

ionization detector with a methanizer, and a Hayesep T column

(

100/120 mesh 5 ft). Separation of the products and reactants

78

G. Du et al. / Journal of Catalysis 253 (2008) 74–90

Table 1

Physical properties of synthesized V-SBA-15 catalysts compared with the blank SBA-15 and various silica supported vanadium catalysts

Sample ID

V-loading

wt%)

Surface density

(VOx/nm )

BET

(m /g)

Pore diameter

(nm)

Mesopore volume

(cc/g)

Total pore

volume (cc/g)

2

(

SBA-15

–

1024

956

899

847

783

716

569

679

550

1130

188

7.46

7.26

7.00

6.98

6.87

6.74

6.77

6.94

6.78

2.83

–

1.38

1.33

1.23

1.18

1.14

1.02

0.96

1.07

0.88

1.00

0.03

1.54

1.47

1.43

1.32

1.30

1.15

1.08

1.21

1.15

2.70

2.01

1

2

5

8

1

2

6

1

2

6

.4V-SBA-15

.7V-SBA-15

.4V-SBA-15

.1V-SBA-15

0.8V-SBA-15

1.6V-SBA-15

.0V/SBA-15

0.0V/SBA-15

.5V-MCM-41

1.4

2.8

4.8

6.1

7.0

9.2

6.0

0.17

0.36

0.66

0.92

1.16

1.91

1.04

2.15

0.16

3.77

10.0

1.5

.0V/SiO

6.0

2

is an indication of long mesopores, limiting the emptying and

ﬁlling of the accessible volume [50]. The amount of nitrogen

adsorbed decreased with increasing V content in V-SBA-15.

As shown by the pore size distribution in Fig. 1b, the pore

size of the catalyst decreased with increasing vanadium loading,

with the decrease slowing as the vanadium loading approached

monolayer coverage (∼11 wt%).

creasing vanadium loading demonstrated no diffraction peaks,

excluding the presence of any agglomerate vanadium species

on the surface of silica SBA-15 of particle size >ca. 3 nm. This

XRD result indicates that not only did the atomic layer depo-

sition by the reaction of the vanadium precursor (EtO) V=O

3

with the surface hydroxyl groups not result in aggregation of the

vanadium to form large particles, but also the vanadium-grafted

SBA-15 samples maintained their vanadium surface structure

even after the high-temperature (813 K) calcination process.

This ﬁnding further afﬁrms the high thermal stability of the

surface vanadium species. In comparison, XRD features of the

The physical properties of the series of vanadium-grafted

SBA-15 and vanadium-impregnated SBA-15 catalysts, as well

as the vanadium-incorporated MCM-41 catalysts, are summa-

rized in Table 1. The BET surface area of the vanadium-grafted

samples decreased compared with that of the bare SBA-15. This

trend continued with the further increases in V-loading on the

surface. Decreases in total volume (from 1.47 to 1.08 cc/g) and

mesopore volume (from 1.33 to 0.96 cc/g) and a decline in BET

impregnated samples (10V/SBA-15 and 6V/SiO ), as shown

2

in Fig. 2b, indicate that V2O5 crystallites formed during the

impregnation. At low vanadium loading for 6V/SBA-15, no sig-

niﬁcant diffraction peaks indicative of crystallites were seen.

2

surface area (from 956 to 569 m /g) occurred with increasing

vanadium loading. The systematic decrease in average pore size

from 7.5 to 6.8 nm reﬂects a homogeneous monolayer covering

of the inner pores of SBA-15, which is consistent with the de-

ceased mesopore volumes. The V/SBA-15 samples synthesized

by conventional wet impregnation showed smaller BET sur-

face area, lower mesopore volume and lower total pore volume

compared with the V-SBA-15 samples with similar V loadings

3

.1.3. TEM

Fig. 3 compares TEM images obtained from different ori-

entations of vanadium-grafted SBA-15 prepared by differ-

ent methods. For the low-magniﬁcation images shown in

Figs. 3a–3c, the TEM images of the bare SBA-15, the vanadium-

grafted SBA-15 and the conventional impregnated VO /SBA-

1

x

5 samples all clearly show well-dispersed silica particles with

(

6.0V/SBA-15 compared with 8.1V-SBA-15; 10.0V/SBA-15

uniform morphology, typically showing a 2-D hexagonal shape

with a diagonal of approximately 1 µm (as shown in Figs. 3a–3c

for the 8.1V-SBA-15 sample), due to the self-assembly of the

tetraethylorthosilicate (TEOS) as the silica source in the process

of synthesizing mesoporous SBA-15. The high-magniﬁcation

TEM images of all of the samples clearly reveal the highly

ordered hexagonal arrays of one-dimensional mesopores as a

characteristic of SBA-15 support materials, typically repre-

sented by 8.1V-SBA-15, as shown in Figs. 3d–3f. These im-

ages for the vanadium-supported silica materials prepared by

atomic layer deposition clearly show no evidence of bulk V2O5

crystallites condensed on the SBA-15 surface, which further

conﬁrms that the vanadium species are thoroughly dispersed

on the SBA-15 mesopore walls after the grafting process and

high-temperature thermal treatment. Figs. 3g–3i show the TEM

images for 6.0V/SBA-15 synthesized by wet impregnation. Un-

like the vanadium-grafted SBA-15, the samples prepared by

compared with 21.6V-SBA-15). Using Cab-O-Sil as a support

of vanadium species (6.0V/SiO ) produced a much higher av-

2

erage surface density of VOx (3.77 VOx/nm ) due to the much

2

lower surface area of SiO2 (∼200 m /g), which resulted in

facile surface vanadium aggregation.

3

.1.2. XRD

The powder XRD pattern of as-calcined mesoporous sil-

ica SBA-15, vanadium-grafted SBA-15, and vanadium-impreg-

nated silica samples are shown in Fig. 2. Typically, the low-

angle diffraction pattern shows evidence of three reﬂections

◦

at 2θ values of 0.5 –3 , including one strong peak (100) and

two weak peaks (110) and (200), corresponding to a highly or-

dered hexagonal mesoporous silica framework (Fig. 2a). The

vanadium-grafted samples showed the same pattern, indicating

that the long-range order of the SBA-15 framework was well re-

tained after the grafting and calcination process. The high-angle

diffraction pattern shown in Fig. 2b for the samples with in-

impregnation show bulk V O crystallites in the framework of

2

5

G. Du et al. / Journal of Catalysis 253 (2008) 74–90

79

Fig. 2. X-ray diffraction patterns of vanadium grafted SBA-15 at (a) low angle and (b) high angle range.

SBA-15, some of which are large enough (>10 nm) to partially

block the pores.

41, 1.5 wt%) are shown in the inset of Fig. 4b. The V-SBA-15

sample shows only one distinct peak in the range of 673–

1

123 K, indicating that the samples synthesized by surface

3

.1.4. TPR

grafting are likely to have a single species on the SBA-15

pore surface. The incorporation method for the synthesis of

V-MCM-41 allows substitution of the vanadium (+5) into the

silica framework of MCM-41 mostly occupying isolated dis-

torted tetrahedral coordination sites with one V=O exposed on

the surface of the pore walls and three V–O bonds rooted in

the pore walls [25,28]. This allows the presence of both the di-

rect exposed vanadium species and those partially occluded by

surface silica, resulting in a distinct peak and a broad shoulder

at higher reduction temperatures. In contrast, the atomic layer

deposition method used in this study, grafting the vanadium

species after the synthesis of the SBA-15 framework (postsyn-

thesis), excludes the partial occlusion effect of the silica; as a

result, only distinct peaks can be obtained at lower temperature

(∼40 K lower than V-MCM-41 at similar vanadium loadings).

H -TPR is a useful characterization technique for investigat-

2

ing the chemical properties of a metal catalyst. TPR can reveal

not only the reducibility and stability of the metal-supported

or unsupported catalysts, but also the more profound surface

chemical information, that is, metal species, metal distribution,

and, even quantitatively, the loading of different metal surface

forms. In this study, TPR was carried out between 298 and

1

123 K for all of the vanadium-grafted SBA-15 samples with

increasing vanadium loading, as shown in Fig. 4. For compari-

son, the impregnated V/SBA-15 and V/SiO samples, as well as

the vanadium-incorporated MCM-41 samples, also were mea-

sured. The main feature observed in the V-SBA-15 TPR proﬁles

2

(Fig. 4a) was a linear increase in the height and width of the re-

duction peaks for V grafted SBA-15 with increasing vanadium

content. A systematic change in the temperature of the maxi-

mum reduction rate and of the initial reduction with increasing

vanadium content grafted onto the surface of SBA-15 also could

be seen. As shown in Fig. 4a, there is a synchronous increase

The TPR proﬁles of the impregnated 6.0V/SiO (6.0 wt%)

2

and 10.0V/SBA-15 (10.0 wt%) sample were compared with

those of the V-SBA-15 samples. Clearly, the impregnated

6.0V/SiO sample had a much lower reduction temperature

2

(

∼60 K) in temperature from the initial reduction to the max-

of ca. 638 K, which is 151 K lower than that of the grafted

V-SBA-15 sample with similar vanadium loading. This indi-

cates that using the conventional wet impregnation method,

imum reduction rate for all the grafted samples; however, this

small shift does not likely mean a change in the surface vanadia

structure, but likely is related to the difference in H -to-surface

vanadia ratio during reduction with increasing vanadium load-

ing.

TPR proﬁles of the grafted sample (1.4V-SBA-15, 1.4 wt%)

and the vanadium-incorporated MCM-41 sample (2.5V-MCM-

2 5

aggregated or crystallized V O can be formed on the surface of

2

silica (Cab-O-Sil) with low surface area (∼200 m /g), resulting

in a low reduction temperature. On the other hand, the impreg-

nated V/SBA-15 sample showed similar reduction temperatures

(806 K for 6.0V/SBA-15 and 821 K for 10.0V/SBA-15) at a

80

G. Du et al. / Journal of Catalysis 253 (2008) 74–90

Fig. 3. The TEM images for low magniﬁcation SBA-15 structure (a–c), vanadium grafted 8.1V-SBA-15 (6.1 wt%) sample (d–f) and impregnated 6.0% V/SBA-15

sample (g–i).

maximum reduction rate to those of grafted V-SBA-15 catalyst

with similar vanadium loadings; however, the integral peak area

of the impregnated sample was much smaller than that of the

grafted sample, suggesting that polymeric vanadium species

deposited on the high-surface area SBA-15 support might be

more difﬁcult to reduce to the same valance with V-SBA-15 un-

der the TPR conditions. The reduction temperature (806 K) for

the impregnated catalyst (6.0V/SBA-15, 6 wt% V) was slightly

shown in Fig. 5, along with samples prepared by conventional

wet impregnation. The silica support exhibits Raman features

at 800 cm (siloxane bridges), 480 and 600 cm (three- and

−

1

−1

−

1

fourfold siloxane rings), and 980 cm (surface silanol groups)

−

1

[10]. A sharp Raman band at ∼1040 cm , which has been as-

signed to the symmetric V=O stretching vibration of isolated

VO species [10,19], can be seen for the V-SBA-15 samples

4

(spectra b–f), with increased intensity with increasing vana-

(

∼15 K) lower than the data reported by Fornés [23], likely due

dium loading. The intensity of the two broad bands at ∼1070

−

1

to the different temperature ramp rate and/or hydrogen concen-

tration in the reducing stream during TPR.

and ∼915 cm increased with the vanadium loading. These

−

two bands, which are characteristics of Si–O and Si(–O )

2

functionalities [7,52] corresponding to the perturbation in sil-

ica vibrations, may be an indication of the V–O–Si formation.

Its increasing intensity with the V loading suggests that more

V–O–Si is formed during the grafting process, consuming more

3

.1.5. Raman spectroscopy

Laser Raman spectroscopy was used to elucidate the molec-

ular nature of vanadium domains. Although some have claimed

that Raman spectroscopy may not be reliable for discriminating

monomeric and polymeric vanadyl species [51], it still can be

used to identify crystalline V O . The Raman spectra of dehy-

drated V-SBA-15 samples with different vanadium loadings are

−1

Si–OH hydroxyl groups. No signiﬁcant band at ∼141 cm

corresponding to crystalline V2O5 could be observed even with

the V-SBA-15 with the highest vanadium loading (21.6V-SBA-

2

5

−1

15, 9.2 wt%). The absence of a typical band at 995 cm of

G. Du et al. / Journal of Catalysis 253 (2008) 74–90

81

Fig. 4. Temperature programmed reduction (TPR) results of (a) grafted V-SBA-15 with different vanadium loading and (b) the comparison of V-SBA-15 with

V-MCM-41, V/SBA-15, and V/SiO samples.

2

V2O5 in spectra b–f implies that no polymeric crystalline V2O5

was formed in the grafted V-SBA-15 samples. These obser-

vations suggest that the vanadium domains on the surface all

existed in an isolated environment, without aggregation to form

crystalline V2O5. Similar Raman spectra pattern have been re-

ported elsewhere for the vanadia-silica with relatively low vana-

dium loadings through the direct grafting of two precursors,

OV[OSi(O Bu)3]3 and OV(O Bu)3 [17].

All of the V2O5 characteristic Raman bands can be clearly

observed for the V/SiO2 samples (6 and 10 wt%), showing

a series of peaks at 995, 698, 527, 482, 407, 302, 282, 194,

and 141 cm . The band located at 995 cm is due primarily

−

1

−1

to the symmetric stretch of V=O groups in the vanadium oxide

−

1

bulk [53]. A broad shoulder at ∼1020 cm was observed be-

−

1

side the sharp ∼993 cm band. For the vanadium-impregnated

SBA-15 catalysts, however, at relatively low vanadium loading

−

1

(6.0 wt%), only a band at ∼1040 cm could be seen as a char-

acteristic of the V=O isolated species on the surface, due to the

high surface area of mesoporous SBA-15 support. This is con-

sistent with the absence of the feature of crystallization in the

t

82

G. Du et al. / Journal of Catalysis 253 (2008) 74–90

Fig. 5. Raman spectra of (a) SiO (SBA-15), (b) 2.7V-SBA-15, (c) 5.4V-SBA-15, (d) 8.1V-SBA-15, (e) 10.8V-SBA-15, (f) 21.6V-SBA-15, (g) 6.0V/-SBA-15,

2

(

h) 10.0V/SBA-15, (i) 6.0%V/SiO , (j) 10.0%V/SiO .

2 2

XRD result for 6.0V/SBA-15. On the other hand, a weak Ra-

−

1

man band appears at ∼141 cm , which likely indicates a small

amount of crystalline V2O5 is formed at this loading [7]. When

the vanadium content is increased to 10.0 wt%, all the features

of the crystalline V2O5-like structure were obtained, resulting

in a similar Raman spectra with that of V/SiO2 samples.

For the partially hydrated V-SBA-15 sample with high vana-

dium loading (21.6V-SBA-15, 10 wt%) and the vanadium im-

pregnated SBA-15 sample (6.0%V/SBA-15) as shown in Fig. 6,

some characteristic broad bands were seen at ∼702, ∼509,

−

1

∼

271, and ∼154 cm , which have been assigned to hydrated

amorphous polymeric vanadium oxide species [10]. However,

each spectrum exhibited a sharp peak at ∼1040 cm instead

−

1

−

1

of a broad band at ∼1020 cm as another characteristic band

of V2O5-like species, which might indicate that the V=O iso-

lated structure can be substantially sustained by the introduc-

tion of H2O molecules. Bonding of the H2O molecule to the

center V atom only changed the domain size by changing the

coordination number, or aggregated the neighboring vanadium

center to form V–O–V bonds. The 6.0%V/SiO2 hydrated sam-

ple showed similar spectra with the dehydrated sample with

V2O5 surface structure (not shown here), indicating that hy-

dration in ambient condition did not change the local structure

of the impregnated sample at this loading (6.0 wt%) because of

the already-occupied polymeric vanadium crystalline domain

with square pyramidal environment. Nevertheless, these char-

acteristics were not signiﬁcantly present in the V-SBA-15 with

relatively low loading (8.1V-SBA-15, 6.1 wt%). This result in-

dicates that at similar vanadium loadings, the grafted sample

Fig. 6. Raman spectra of hydrated samples (a) 8.1V-SBA-15,

b) 21.6V-SBA-15, (c) 6.0V/SBA-15.

(

was more structurally stable against polymerization compared

with the impregnated sample on hydration.

G. Du et al. / Journal of Catalysis 253 (2008) 74–90

83

Fig. 7. Comparison of vanadium K-edge XANES regions for V grafted SBA-15 catalysts and the reference compounds.

3

.1.6. XANES

with increasing vanadium loading; however, it still was much

higher than that of the octahedral coordination species com-

pared with PbV2O6. The distance between the pre-edge peak

and the main edge is a measurement of the oxidation state of

vanadium [46]. The Eedge–Epre-edge value has been reported

to be between 12.4 and 12.8 for the tetrahedral vanadium(V)

species, compared with 9.5 eV for the square pyramidal vana-

dium (V) as in V2O5 [46]. For compounds with lower oxidation

states, this value decreased signiﬁcantly as for VOSO4·3H2O

(Eedge–Epre-edge = 8.4 eV). For the energy difference of the

pre-edge peak to the main edge position as shown in Table 2,

the Eedge–Epre-edge of the grafted samples varied from 11.6 to

13.0 eV, suggesting a tetrahedral coordination V(V) species.

This value is much larger than that of V2O5 with a square pyra-

midal coordination, which excludes the existence of measurable

aggregated vanadium oxides on the surface. This result sug-

gests that the structure of vanadium clusters in V-SBA-15 is

most likely distorted tetrahedron, which does not signiﬁcantly

change with vanadium loading up to the limit of monolayer

coverage. The XANES results suggests that the samples syn-

thesized through a grafting method preserve a local structural

uniformity, and the increase of V-loading does not result in se-

rious aggregation of the vanadium species on the SBA-15 wall

surface. This result is consistent with Raman results showing

the presence of only monomeric vanadium species on the SBA-

15 surface under dehydrated conditions.

The ﬁrst-row transition metal elements exhibited well-

deﬁned site symmetry spectra in the XANES analysis. Normal-

ized V K-edge XANES spectra of dehydrated vanadium-grafted

SBA-15 catalyst and the vanadium reference compounds as

well as vanadium metal foil are shown in Fig. 7. The standard

edge energy was calibrated at the ﬁrst inﬂection point (K-edge)

in the foil calibration spectrum (5465.0 eV), which measures

the threshold of photoejection of the 1 s electron in vanadium

metal. The spectra of Fig. 7 show that the V K-edge XANES in

the samples exhibit a pre-edge absorption feature correspond-

ing to the forbidden transition 1s → 3d [46], which is caused

mainly by mixing of 4p orbitals of vanadium atoms and 2p or-

bitals of the oxygen atoms with the 3d π orbitals of vanadium

atoms [46]. The XANES regions for all of the V-SBA-15 sam-

ples are indistinguishable, indicating that they may exhibit the

same electronic structures and site symmetries.

The XANES results on vanadium-grafted samples and ref-

erence compounds are summarized in Table 2. It is known that

the pre-edge height and intensity are closely related to the lo-

cal symmetry type, and they can be used to gain insight into the

coordination structure when comparing the model compounds

with a known coordination environment. The height of the pre-

edge peak decreases monotonically as the vanadium structure

changes from tetrahedral coordination to the square pyramidal

or octahedral coordination. As shown in Fig. 7, the local sym-

metry of the vanadium domain in the V-SBA-15 samples had a

smaller pre-edge height compared with the reference compound

Na3VO4 (tetrahedral) and similar pre-edge height to NH4VO3

Empirically, the strength of the pre-edge transition has been

found to be related to the size of the “molecular cage” [46].

Wong et al. [46] illustrated the correlation of the pre-edge in-

tensity (deﬁned by the height ∗ width at half-height) of dif-

ferent vanadium compounds and the domain size within the

same symmetry quantitatively, suggesting an inverse molecular

(

distorted tetrahedral), suggesting a distorted tetrahedral coor-

dination for the grafted samples with various vanadium load-

ings. It is observed that the pre-edge height decreased slightly

84

G. Du et al. / Journal of Catalysis 253 (2008) 74–90

Table 2

Energy positions of various spectral features in the V K-edge XANES spectra

a

a,b

Sample ID

Pre-edge peak position

eV)

Main-edge position

(eV)

E

–E

E

–E

pre-edge

Pre-edge peak height

(a.u.)

Pre-edge peak

intensity (a.u.)

pre-edge

0

edge

c

(

(eV)

Na VO

5469.4

5469.6

5469.9

5471.1

5469.3

5469.6

5469.5

5469.7

5469.6

5469.4

5469.7

5469.4

5482.1

5481.1

5481.4

5481.2

5477.7

5482.6

5481.9

5481.3

5481.4

5481.5

5481.4

5482.3

5481.1

4.4

4.6

4.9

6.1

4.3

4.6

4.5

4.7

4.6

4.4

4.7

4.4

12.7

11.5

10.1

8.4

13.0

12.4

11.6

11.8

11.9

12.0

12.6

11.7

0.87

0.73

0.57

0.64

0.43

0.74

0.72

0.67

0.68

0.67

0.65

0.72

0.68

1.60

2.06

1.71

2.42

1.17

2.28

1.99

1.93

2.02

1.70

1.86

2.15

2.02

3

4

NH VO

4

3

PbV O

2

6

V O

2

5

VOSO ·3H O

4

2

d

1

2

5

8

1

2

6

.4V-SBA15D

.7V-SBA15D

.4V-SBA15D

.1V-SBA15D

0.8V-SBA15D

1.6V-SBA15D

d

.5V-MCM41D

d

.0I-VSBA15D

a

The uncertainty in peak positions is estimated to be ± 0.2 eV.

b

c

d

Main edge position is chosen as the mid point of absorption jump.

Peak intensity is calculated by peak height ∗ width at half maximum.

Samples were dehydrated in He at 623 K for 30 min before measurements.

cage–pre-edge absorption intensity relationship. The pre-edge

peak intensity results for a series of V-SBA-15 samples sum-

marized in Table 2 indicate that the pre-edge intensities of the

samples, especially those with relatively low metal loading, re-

semble that of NH4VO3 with a distorted tetrahedral coordina-

tion, demonstrating that the vanadium domains of the V-SBA-

1

5 samples are of a similar cage size as those of NH4VO3.

It was previously mentioned that a color change (from white

or light yellow to dark yellow) occurred when the samples with

vanadium grafted on the surface were exposed to ambient con-

ditions. This change on hydration can be directly compared for

the V K-edge spectra in the XANES region. A typical XANES

spectrum of the V-SBA-15 sample (1.4 wt%) is illustrated in

Fig. 8, and the comparisons of the pre-edge feature between

the dehydrated and hydrated samples as a function of vanadium

content are shown in Fig. 9. The pre-edge heights of hydrated

vanadium grafted samples are smaller than those of the dehy-

drated samples, suggesting that the local structure of vanadium

surface species may change from mostly tetrahedral coordina-

tion to partially octahedral coordination due to the absorbed

H2O. As shown in Fig. 8, a dramatic difference in the pre-edge

feature can be observed, including a slight shift in the pre-edge

position and a relative increase in the 1s → 4p transition above

the edge at 5488.2 eV for the sample on hydration. As shown

in Fig. 9, both the pre-edge heights and intensities decrease

on hydration, suggesting an increase in the domain size or the

change of coordination from distorted tetrahedron for the hy-

drated samples to distorted octahedron or polymeric vanadium

structure with a V–O–V bridge, further conﬁrming the Raman

results on the hydration effect.

Fig. 8. Comparison of V K-edge XANES regions for dehydrated/hydrated sam-

ples and sample after methanol oxidation.

indicating that the local structure largely retained its tetrahe-

dral coordination. The vanadium-grafted sample shows resis-

tance to reduction up to 623 K under pure H atmosphere.

2

An increase in the reduction temperature to 773 K produced

a dramatic decrease in the pre-edge peak. The shift of the

main edge to lower energy indicates the reduction of vana-

+

dium from 5 to approximately 4 (E_e_d_g_e–E_p_r_e_-_e_d_g_e= 8.9 eV)

compared with the main edge position of VOSO ·3H O com-

4

2

pound (E –E

edge

pre-edge

= 8.4 eV). As in the TPR proﬁle of

10.8V-SBA-15, the vanadium reduction started at 740 K. It is

reasonable to deduce that vanadium grafted on the surface of

SBA-15 can be reduced only to 4 under the H -TPR condi-

+

The redox properties of the V-SBA-15 catalyst were studied

by in situ X-ray absorption. As shown in Fig. 10, reduction at

2

tions and remains in a square pyramidal coordination, without

further reduction to lower valance or vanadium metal. After re-

oxidation in air at 773 K for 30 min, interestingly, the pre-edge

peak increased to the level of the fresh sample, suggesting a

return of the local structure to the distorted tetrahedral coordi-

4

73 K in pure H2 for 30 min did not change the pre-edge fea-

tures of the dehydrated V-SBA-15 sample. With an increase in

reduction temperature to 623 K, the pre-edge peak height de-

creased only slightly, from 0.67 of the fresh sample to 0.61,

G. Du et al. / Journal of Catalysis 253 (2008) 74–90

85

1

Fig. 11. Fourier transformed k -weighted EXAFS of vanadium grafted

SBA-15: (a) 1.4V-SBA-15; (b) 2.7V-SBA-15; (c) 5.4V-SBA-15; (d) 8.1V-

SBA-15; (e) 10.8V-SBA-15; (f) 21.6V-SBA-15.

when the V species is reoxidized at the sample temperature,

the V–O–Si bond is reformed, resulting in reconversion to the

tetrahedral coordination. Therefore, the vanadium grafted on

the silica surface does not aggregate at the reduction condition,

and maintains its isolated site.

3.1.7. EXAFS

EXAFS can provide quantitative information about the

Fig. 9. Comparison of the pre-edge features including (a) pre-edge peak height

and (b) pre-edge intensity of hydrated and dehydrated samples.

atomic structure and ligand environment. Fig. 11 compares

1

the Fourier-transformed k -weighted EXAFS ﬁndings for the

V-SBA-15 samples. The spectra are quite similar in the range

of 1–3 Å, indicating similar ﬁrst coordination spheres for vana-

dium species. The size of the vanadium complex on the silica

surface was determined from the EXAFS spectra ﬁtted, con-

sidering only the ﬁrst-shell V–O coordination number with

average V–O bond length for all of the samples, as summa-

rized in Table 3. The average V–O coordination number varied

from 3.8 to 4.5, suggesting that the V ions were grafted mostly

on the silica surface tetrahedrally, conﬁrming the proposed VO₄

model, without formation of aggregated vanadium species (i.e.,

V–O–V). To reduce the number of independent variables in the

ﬁt, coordination numbers were subsequently ﬁxed at the ﬁxed

integer values of coordination, considering the well-accepted

isolated VO model with a terminal V=O and three bridging

4

V–O bonds. The ﬁtted parameters of the ﬁrst two V–O shells

are also listed in Table 3. The ﬁtted distances of V=O (1.54 Å)

and V–O (1.74 Å) were consistent with values reported previ-

ously [36,54]. Fig. 12 shows the agreement of the FT and ﬁt for

a typical vanadium-grafted sample (8.1V-SBA-15).

Fig. 10. In situ XANES study of the reducibility of V-SBA-15 sample

(10.8V-SBA-15).

nation and the reconstruction of the Si–O–V bonds on the sil-

ica surface after the high-temperature air treatment. This likely

suggests that after reduction at 773 K, the vanadium species

3.2. Catalyst performance in methanol oxidation

changed from the original three-legged (≡SiO) V=O to two-

The activities of the V-SBA-15 catalysts were explored un-

der steady-state conditions with constant ﬂow rate (78 ml/min)

and reaction pressure (2.5 psi) and a feed composition of

1.9 vol% methanol, 2.7 vol% of oxygen, and 95.4 vol%

helium. Fig. 13 shows the conversion of methanol, selec-

3

legged (≡SiO)2V=O species, accompanying the decrease in

+

valance of V from 5 to 4 . At 773 K, the two-legged V species

is more stable than the three-legged V species, and the vana-

+

dium is not likely to undergo further reduction to 3 . Thus,

86

G. Du et al. / Journal of Catalysis 253 (2008) 74–90

Table 3

EXAFS initial ﬁt results considering the average ﬁrst shell and the ﬁt of VO model of two oxygen shells (terminal and bridging)

4−

Sample ID

Average ﬁrst oxygen shell

Two oxygen shells

2

CN

R (Å)

σ

(Å )

R

V=O

V–O

CN

σ

(Å )

R

factor

CN

R (Å)

factor

1.4V-SBA15D

2.7V-SBA15D

5.4V-SBA15D

8.1V-SBA15D

10.8V-SBA15D

21.6V-SBA15D

2.5-VMCM41D

6.0IV/SBA15

4.54 ± 0.54

4.00 ± 0.40

3.78 ± 0.38

3.94 ± 0.43

3.95 ± 0.46

3.85 ± 0.39

4.04 ± 0.45

3.90 ± 0.44

1.729

1.740

1.744

1.740

1.744

1.738

1.743

0.00938

0.00778

0.00703

0.00782

0.00836

0.00752

0.00787

0.00881

0.006

0.007

0.008

0.006

0.007

0.006

1

1.544

1.542

1.543

1.542

1.544

1.542

1.543

3

1.739

1.737

1.738

1.737

1.738

1.737

0.00005

0.00013

0.00020

0.00043

0.00091

0.00053

0.00013

0.00157

0.005

0.006

0.005

0.006

this work is the exchange of the ethoxyl groups of the inter-

−

mediate (EtO) V=O with Si–O of the surface silanol groups,

3

ideally forming three Si–O–V bonds per vanadium domain on

the SBA-15 surface. This limits the maximum coverage of

vanadium species grafted on the SBA-15 surface to be well be-

low the monolayer (1.33 VOx/nm , considering the estimated

.0 Si–OH/nm silanol group density [57]), no matter how high

2

4

the concentration of the metal precursor used. We have demon-

strated that a maximum of 9.2 wt% vanadium on the SBA-15

(

21.6V-SBA-15) could be achieved with a nominal surface

2

vanadium density of 1.06 VOx/nm , due to the high surface

area of SBA-15, as is clearly shown in Fig. 14. Thus, for all of

the V-SBA-15 samples synthesized in this work, the monotonic

increase in activity with increasing vanadium content in the

temperature range studied indicates that the V-SBA-15 samples

were all within the dispersion-limit loading without aggregat-

ing to polymeric vanadium species, which is consistent with

the Raman and XANES results discussed above.

1

Fig. 12. k -weighted Fourier transform of the experimental χ(k) (solid) and the

calculated ﬁt (dashed) in R space for the ﬁrst two shells.

The formaldehyde selectivity proﬁles shown in Fig. 13b

demonstrate signiﬁcant differences with respect to vanadium

loading. For samples with low vanadium content (<3.0 wt%),

the selectivity increased with temperature within the 100-K

range. Increasing the vanadium loading in the samples pro-

duced maximum formaldehyde selectivity within the temper-

ature range studied and decreased the temperature of the great-

est HCHO selectivity. This phenomenon was previously ob-

served for V-MCM-41 catalysts on methane partial oxidation

to formaldehyde [28]. This has been suggested to be related to

the increase in highly dispersed isolated active sites with in-

creased vanadium loading, which has a pre-exponential effect

on the rate constant for the reaction, as we discussed in de-

tail in previous work [58]. Due to the consecutive oxidation

of the formaldehyde to carbon oxides for this series reaction,

decreased selectivity with increasing temperature can be ex-

pected. The 21.6V-SBA-15 (9.2 wt% V) exhibited the highest

formaldehyde selectivity of 86% at 598 K, due to its highest

surface vanadium active site density.

tivity of formaldehyde and dimethyl ether, and space-time

yield (STYHCHO) of the vanadium-grafted samples with dif-

ferent vanadium loadings in terms of reaction temperature for

the methanol oxidation reaction. Clearly, the conversion of

methanol depends strongly on the vanadium content and reac-

tion temperature. For 1.4V-SBA-15 with the lowest vanadium

loading, methanol conversion increased slowly with increasing

temperature up to 623 K, then rapidly thereafter. With increased

vanadium loading in the V-SBA-15 samples, the methanol con-

version proﬁles show a more signiﬁcant trend with respect to

reaction temperature, exhibiting a much lower “initiation” tem-

perature for the “jump” in conversion. In other words, the clear

increase in conversion shifted to lower temperatures with the

increase in active sites. For the V-SBA-15 samples with higher

vanadium loading, the methanol conversion approached 100%,

and the temperature for the maximum conversion decreased

with increasing vanadium loading. At high temperature, the

effect of the vanadium loading on the methanol conversion de-

creased due to the high activity of the isolated active sites and

the nearly complete consumption of the methanol in the feed,

as shown in the inset of Fig. 13.

As a major hydrocarbon product of methanol oxidation, the

selectivity of dimethyl ether (DME) is demonstrated in Fig. 13c.

This indicates the presence of acidic sites on V-SBA-15 cata-

lysts. Interestingly, no DME was observed for the blank SBA-

15 support. The DME selectivity proﬁles demonstrate that high

temperature and high vanadium loading decreased the DME

When the vanadium content exceeds the dispersion-limit

loading (e.g., the monolayer capacity), aggregated species pre-

dominate [55,56]. The grafting mechanism that we propose in

G. Du et al. / Journal of Catalysis 253 (2008) 74–90

87

Fig. 13. Conversion of methanol (a), selectivity of formaldehyde (b) and dimethyl ether (c), and space time yield of formaldehyde (d) as a function of reaction

temperature on V-SBA-15 catalysts with different vanadium content. (Reaction conditions: ﬂow rate 78 ml/min, 1.9 vol% CH OH:2.7 vol% O :95.4 vol% He,

3

2

pressure 2.5 psi.)

Fig. 13c also plots the space-time yield of formaldehyde

for the combination of methanol conversion and formaldehyde

selectivity. The ﬁgure shows that the yield of formaldehyde in-

creased monotonically with temperature for the samples with

vanadium loading up to 7.0 wt%. For the 21.6V-SBA-15 with

vanadium loading approaching monolayer coverage, the drop in

STY_H_C_H_Oat that high temperature from 648 to 673 K may be

related to the complete consumption of methanol (100% con-

version) and the secondary oxidation of the intermediate HCHO

to CO_x.

We also compared the catalytic properties of the grafted

V-SBA-15 catalyst and the V/SBA-15 catalyst with similar

vanadium content. As shown in Fig. 15, the grafted sample

exhibited much higher activity in the temperature range stud-

ied and demonstrated the maximum formaldehyde selectivity at

lower temperatures. Based on the Raman and XANES results,

the 6.0V/SBA-15 (6.0 wt% V) catalyst surprisingly showed

no features of polymeric vanadium, which means at this load-

ing, isolated vanadium species can predominate due to the

Fig. 14. The expected calculated V-loading compared to the actual vanadium

that grafted on the SBA-15 surface.

selectivity, indicating an inverse relationship between the vana-

dium active sites active in formaldehyde production and the

acidic sites active in DME production.

2

high surface area of SBA-15 (∼1000 m /g). However, the dra-

matic difference in catalytic performance strongly suggests a

88

G. Du et al. / Journal of Catalysis 253 (2008) 74–90

huge difference between the samples prepared through grafting

and those prepared through impregnation, demonstrating differ-

ences in local environment and dispersion. The grafted samples

had more isolated vanadium species than the impregnated sam-

ples, contributing to the catalytic reaction.

synthesized in this work and in similar work through a differ-

ent synthesis approach on the same reaction [59]. Compared

with that previous work, our HCHO selectivities are lower, but

our catalysts exhibit much higher activity. The lower selectivi-

ties likely can be attributed to high conversion; when compared

at the same conversion, the selectivities should be comparable

or higher. The difference in the two sets of results may be ex-

plained by the higher surface area of SBA-15 support, resulting

in slightly lower nominal vanadium surface density and thus

highly dispersed vanadia species.

Table 4 compares the catalytic performance (activity, selec-

tivity, TOFs) of highly dispersed vanadia supported on SBA-15

3

.3. Acidic properties of V-SBA-15

The acidic properties of V grafted SBA-15 samples were in-

vestigated by pyridine TPD and compared with those of pure

SBA-15. As shown in Fig. 16, the V-SBA-15 samples with dif-

ferent V-loading exhibited three pyridine desorption peaks at

around 523, 653, and 823 K, indicating different acid sites of

different acid strength on the surface of V-SBA-15. The broad

signal at around 1000 K is due to the impurities during the syn-

thesis of SBA-15. As reported by Lim and Haller in a study of

the acidic properties of V-MCM-41 monitored by FTIR [60],

the peak of pyridine TPD at relatively low temperature (523 K)

may correspond to the Lewis acid sites, whereas the peaks at

higher temperatures (653 and 823 K) may be attributed to Brön-

sted acid sites. The coexistence of both species of acid sites

results in the formation of dimethylether. The numbers of both

Lewis and Brönsted acid sites increase with increasing vana-

dium content grafted on the siliceous surface, accounting for the

increased DME selectivity in methanol oxidation at low temper-

ature.

(a)

It is generally agreed that the observation of DME as a

product of methanol oxidation implies acidity in our V-SBA-

1

5 catalysts, as has been conﬁrmed and partially characterized

by pyridine TPD. The fact that the acidity increases with V

loading suggests that this acidity may be an inherent prop-

erty of V embedded in silica for certain structures, but it is

very difﬁcult to rule out impurities. The fact that the acid site

density correlates with V for both V-MCM-41 [60] and V-

SBA-15 using different substrates, different V precursors, and

different processes for incorporation/grafting might argue for

an inherent acidity in some kinds of V sites. This is particu-

larly true for the SBA-15, which appears to have no inherent

acidity (see Fig. 16) but develops acidity progressively with in-

(

b)

Fig. 15. Comparison of grafted V-SBA-15 catalyst and V/SBA-15 synthesized

by conventional wet impregnation in terms of methanol conversion (a) and

formaldehyde selectivity (b).

Table 4

Methanol oxidation to formaldehyde over highly dispersed vanadia supported on SBA-15

a

Sample ID

V

VOx

(nm )

T

MeOH/O /He

2

(ml/min)

Conversion

(MeOH)%

Selectivity

(HCHO)%

TOF

Reference

2

3

−1

)

(mmol/g)

(K)

(×10 s

b

7

1

2

7

1

.2% V/SBA-15

0.8V-SBA-15

.3% V/SBA-15

.7V-SBA-15

.2% V/SBA-15

0.8V-SBA-15

1.4

2.3

1.16

0.6

0.36

2.3

1.16

623

673

3/7/90

2/3/95

3/7/90

2/3/95

3/7/90

2/3/95

39.0

81.0

64–70%

81.0

64–70%

100

93.3

75.4

88.1

64.0

88.7

58.4

–

2.1

11.2

4.4

5.5

2.6

[59]

1.37

0.45

0.55

1.4

This work

b

[59]

This work

b

[59]

1.37

This work

a

Apparent TOF of HCHO is calculated by the number of moles of methanol converted to formaldehyde per mole of surface vanadium atom per second.

A more completed summary of the catalytic performance of highly dispersed vanadia supported on various mesoporous silica materials can be found in the same

b

reference.

G. Du et al. / Journal of Catalysis 253 (2008) 74–90

89

a terminal V=O bonds and three V–O bonds. The better ac-

tivity of the V-SBA-15 catalysts compared with those synthe-

sized by impregnation is attributed to the homogeneous disper-

sion of vanadium active single sites. The Lewis and Brönsted

acid sites on the V-SBA-15 surface result in the production

of DME.

Acknowledgments

This work was supported by grants from NASA (NAG 04)

and the US DOE, Ofﬁce of Basic Energy Sciences (DE-FG02-

0

1ER15183). The authors thank the National Synchrotron Light

Source at Brookhaven National Laboratory for providing beam-

time for these experiments.

References

[

1] U.S. Ozkan, T.A. Harris, B.T. Schilf, Catal. Today 33 (1997) 57.

2] H. Bosch, F. Janssen, Catal. Today 2 (1988) 369.

3] G.C. Bond, S.F. Tahir, Appl. Catal. 71 (1991) 1.

4] G.A. Du, S. Lim, Y.H. Yang, C. Wang, L. Pfefferle, G.L. Haller, Appl.

Catal. A 302 (2006) 48.

[5] A. Khodakov, B. Olthof, A.T. Bell, E. Iglesia, J. Catal. 181 (1999) 205.

[6] I.E. Wachs, B.M. Weckhuysen, Appl. Catal. A 157 (1997) 67.

[7] X.T. Gao, S.R. Bare, B.M. Weckhuysen, I.E. Wachs, J. Phys. Chem. B 102

(1998) 10842.

[

8] G. Busca, G. Centi, L. Marchetti, F. Triﬁro, Langmuir 2 (1986) 568.

9] K. Inumaru, M. Misono, T. Okuhara, Appl. Catal. A 149 (1997) 133.

Fig. 16. Pyridine TPD patterns for V-MCM-41, SBA-15 and V-SBA-15 with

different V-loadings.

[

10] N. Das, H. Eckert, H.C. Hu, I.E. Wachs, J.F. Walzer, F.J. Feher, J. Phys.

Chem. 97 (1993) 8240.

11] S. Irusta, L.M. Cornaglia, E.E. Miro, E.A. Lombardo, J. Catal. 156 (1995)

167.

creasing V loading, because the grafting process uses a very

pure O=VCl3 vanadium source that is then converted to the

O=V(CH3CH2O)3 precursor, a virtual puriﬁcation process. On

the other hand, we have prepared V-MCM-41 using an impure

HiSil 233 silica course (and observed apparent acidity in the

methanol oxidation) and a pure Cab-O-Sil source where no

dimethyl ether was detected [15], suggesting that all acidity is

associated with impurities or that acidity associated with V oc-

curs only at certain (higher) loading in certain structures. The

latter hypothesis, although not impossible, seems somewhat im-

probable. Putting aside the structure/source of acidity, we note

that the 2.5V-MCM-41 (1.5 wt% V) has fewer acid sites than

the 1.4V-SBA-15 (1.4 wt% V), according to Fig. 16. This again

might argue for an impurity interpretation (which uncouples

acid site density from V site density), but may reﬂect differ-

ing distributions of V in sites that do and do not impart acidity.

Thus, we can neither conﬁrm the possibility of inherently acidic

V in silica sites nor identify the likely precursor if it is due to

an impurity, and therefore must leave the description of the acid

site as an open question.

12] M. Iwamoto, J. Hirata, K. Matsukami, S. Kagawa, J. Phys. Chem. 87

(

1983) 903.

13] B. Olthof, A. Khodakov, A.T. Bell, E. Iglesia, J. Phys. Chem. B 104 (2000)

516.

[14] J. Dobler, M. Pritzsche, J. Sauer, J. Am. Chem. Soc. 127 (2005) 10861.

1

[

15] Y.H. Yang, G.A. Du, S.Y. Lim, G.L. Haller, J. Catal. 234 (2005) 318.

16] J.M. Thomas, R. Raja, D.W. Lewis, Angew. Chem. Int. Ed. 44 (2005)

6456.

[

17] D.A. Ruddy, N.L. Ohler, A.T. Bell, T.D. Tilley, J. Catal. 238 (2006) 277.

18] M.M. Koranne, J.G. Goodwin, G. Marcelin, J. Catal. 148 (1994) 388.

[19] G.T. Went, S.T. Oyama, A.T. Bell, J. Phys. Chem. 94 (1990) 4240.

[20] C.B. Wang, R.G. Herman, C.L. Shi, Q. Sun, J.E. Roberts, Appl.

Catal. A 247 (2003) 321.

[

21] M.M. Koranne, J.G. Goodwin, G. Marcelin, J. Catal. 148 (1994) 369.

22] H. Berndt, A. Martin, A. Bruckner, E. Schreier, D. Muller, H. Kosslick,

G.U. Wolf, B. Lucke, J. Catal. 191 (2000) 384.

[23] V. Fornes, C. Lopez, H.H. Lopez, A. Martinez, Appl. Catal. A 249 (2003)

45.

3

[24] M. Baltes, K. Cassiers, P. Van Der Voort, B.M. Weckhuysen, R.A.

Schoonheydt, E.F. Vansant, J. Catal. 197 (2001) 160.

[

25] D. Wei, H. Wang, X.B. Feng, W.T. Chueh, P. Ravikovitch, M. Lyubovsky,

C. Li, T. Takeguchi, G.L. Haller, J. Phys. Chem. B 103 (1999) 2113.

26] S. Lim, G.L. Haller, J. Phys. Chem. B 106 (2002) 8437.

27] Y.H. Yang, S. Lim, C. Wang, D. Harding, G. Haller, Microporous Meso-

porous Mater. 67 (2004) 245.

[

4

. Conclusion

[28] G.A. Du, S.Y. Lim, Y.H. Yang, C. Wang, L. Pfefferle, G.L. Haller, Appl.

Catal. A 302 (2006) 48.

We have successfully synthesized vanadium-grafted SBA-15

[29] C. Coperet, M. Chabanas, R.P. Saint-Arroman, J.M. Basset, Angew.

Chem. Int. Ed. 42 (2003) 156.

catalysts with various vanadium content by an atomic layer de-

position method with high dispersions of isolated active sites.

Using different characterization techniques, the active vana-

[30] T.D. Tilley, J. Mol. Catal. A Chem. 182 (2002) 71.

[31] G.L. Rice, S.L. Scott, J. Mol. Catal. A Chem. 125 (1997) 73.

[32] G.L. Rice, S.L. Scott, Langmuir 13 (1997) 1545.

dium species have been assigned to isolated VO sites with

[33] W. Priebsch, D. Rehder, Inorg. Chem. 29 (1990) 3013.

4

90

G. Du et al. / Journal of Catalysis 253 (2008) 74–90

[

34] E.C.E. Rosenthal, F. Girgsdies, Z. Anorg. Allg. Chem. 628 (2002) 1917.

35] N.N. Ghosh, J.C. Clark, G.T. Eldridge, C.E. Barnes, Chem. Commun.

[49] R. Schmidt, E.W. Hansen, M. Stocker, D. Akporiaye, O.H. Ellestad,

J. Am. Chem. Soc. 117 (1995) 4049.

(

2004) 856.

[50] A. Vinu, D.P. Sawant, K. Ariga, K.Z. Hossain, S.B. Halligudi, M. Hart-

mann, M. Nomura, Chem. Mater. 17 (2005) 5339.

[

36] F.J. Feher, J.F. Walzer, Inorg. Chem. 30 (1991) 1689.

[

37] O.B. Lapina, M.A. Mats’ko, T.B. Mikenas, V.A. Zakharov, E.A. Paukshtis,

D.F. Khabibulin, A.P. Sobolev, Kinet. Catal. 42 (2001) 553.

[51] N. Magg, B. Immaraporn, J.B. Giorgi, T. Schroeder, M. Baumer,

J. Dobler, Z.L. Wu, E. Kondratenko, M. Cherian, M. Baerns, P.C. Stair,

J. Sauer, H.J. Freund, J. Catal. 226 (2004) 88.

[38] E.W. Deguns, Z. Taha, G.D. Meitzner, S.L. Scott, J. Phys. Chem. B 109

(2005) 5005.

[52] P. MacMillan, Am. Mineral. 69 (1986) 622.

[

39] K.L. Fujdala, T.D. Tilley, Chem. Mater. 14 (2002) 1376.

40] K.L. Fujdala, T.D. Tilley, J. Catal. 216 (2003) 265.

41] D.Y. Zhao, Q.S. Huo, J.L. Feng, B.F. Chmelka, G.D. Stucky, J. Am. Chem.

Soc. 120 (1998) 6024.

[53] L. Abello, E. Husson, Y. Repelin, G. Lucazeau, Spectrochim. Acta Part

A 39 (1983) 641.

[54] R. Rulkens, J.L. Male, K.W. Terry, B. Olthof, A. Khodakov, A.T. Bell, E.

Iglesia, T.D. Tilley, Chem. Mater. 11 (1999) 2966.

[42] D.Y. Zhao, J.L. Feng, Q.S. Huo, N. Melosh, G.H. Fredrickson, B.F.

Chmelka, G.D. Stucky, Science 279 (1998) 548.

[55] M. Faraldos, M.A. Banares, J.A. Anderson, H.C. Hu, I.E. Wachs, J.L.G.

Fierro, J. Catal. 160 (1996) 214.

[

43] J.E. Herrera, J.H. Kwak, J.Z. Hu, Y. Wang, C.H.F. Peden, J. Macht, E.

Iglesia, J. Catal. 239 (2006) 200.

[56] M.A. Banares, J.H. Cardoso, F. Agullo-Rueda, J.M. Correa-Bueno, J.L.G.

Fierro, Catal. Lett. 64 (2000) 191.

[44] S. Brunauer, P.H. Emmett, E. Teller, J. Am. Chem. Soc. 60 (1938) 309.

[45] E.P. Barrett, L.G. Joyner, P.P. Halenda, J. Am. Chem. Soc. 73 (1951) 373.

[46] J. Wong, F.W. Lytle, R.P. Messmer, D.H. Maylotte, Phys. Rev. B 30 (1984)

[57] I.G. Shenderovich, G. Buntkowsky, A. Schreiber, E. Gedat, S. Sharif, J.

Albrecht, N.S. Golubev, G.H. Findenegg, H.H. Limbach, J. Phys. Chem.

B 107 (2003) 11924.

5

596.

47] E.A. Stern, M. Newville, B. Ravel, Y. Yacoby, D. Haskel, Physica B 209

1995) 117.

48] A.L. Ankudinov, B. Ravel, J.J. Rehr, S.D. Conradson, Phys. Rev. B 58

1998) 7565.

[58] G.A. Du, Y. Yang, W. Qiu, S. Lim, L. Pfefferle, G.L. Haller, Appl.

Catal. A 313 (2006) 1.

[59] C. Hess, I.J. Drake, J.D. Hoefelmeyer, T.D. Tilley, A.T. Bell, Catal.

Lett. 105 (2005) 1.

[

(

[60] S.Y. Lim, G.L. Haller, Appl. Catal. A 188 (1999) 277.

Article Doi

Synthesis, characterization, and catalytic performance of highly dispersed vanadium grafted SBA-15 catalyst

DOI: 10.1016/j.jcat.2007.10.019

Source and publish data:

Authors:

Article abstract of DOI:10.1016/j.jcat.2007.10.019

Full text of DOI:10.1016/j.jcat.2007.10.019

Products guided by the article

R&D Labs maybe for 115-10-6

Relevant to this article

Hot Product