www.elsevier.com/locate/apcatb

Applied Catalysis B: Environmental 77 (2007) 125–134

The nature of surface acidity and reactivity of MoO3/SiO2 and

MoO3/TiO2–SiO2 for transesterification of dimethyl oxalate

with phenol: A comparative investigation

Yue Liu a, Xinbin Ma a,*, Shengping Wang a, Jinlong Gong b,**a Key Laboratory for Green Chemical Technology, School of Chemical Engineering and Technology, Tianjin University, Tianjin 300072, China

b Department of Chemical Engineering, University of Texas at Austin, Austin, TX 78712-0231, USA

Received 11 November 2006; received in revised form 25 May 2007; accepted 13 July 2007

Available online 19 July 2007

Abstract

This paper presents results of a comparative investigation of the transesterification of dimethyl oxalate (DMO) with phenol to produce methyl

phenyl oxalate (MPO) and diphenyl oxalate (DPO) over MoO3/SiO2 and MoO3/TiO2–SiO2 catalysts. The evaluation results show that MoO3/TiO2–

SiO2 is much more active and selective than MoO3/SiO2. The surface structure and acidity of MoO3/SiO2 and MoO3/TiO2–SiO2 were investigated

by a series of characterization approaches. XRD and FT-IR demonstrated that the incorporation of amorphous TiO2 could not only enhance the

interaction between MoO3 and SiO2, but also improve the dispersion state of MoO3 on the surface of SiO2. NH3-TPD and FT-IR of adsorbed

pyridine measurements indicated that amorphous TiO2 incorporation into and further interaction with MoO3 and SiO2 formed more weak acid sites

on the surface of the catalysts. However, Bronsted acid sites were also detected on the surface of the MoO3/TiO2–SiO2 catalysts, which further

motivated us to study the nature of Bronsted acid sites in detail. Several conventional Bronsted acids were also tested in the transesterification. The

results unexpectedly showed that conventional Bronsted acids have better reactivities than those of conventional Lewis acids from selective point

of view. In addition, the combination of reactivity tests, NH3-TPD spectra, and FT-IR measurements of adsorbed pyridine strongly suggest that the

strong Lewis acid is more responsible for the production of the by-product anisole than the strong Bronsted acid. The results indicated that the

improvement of catalytic efficiency of MoO3/TiO2–SiO2, compared to MoO3/SiO2, could be ascribed to the increased dispersion capacity of MoO3

and the dominant weak acid sites (including the synergistic effect of the weak Lewis acid sites with the weak Bronsted acid sites). A tentative

mechanism for the transesterification reaction over Bronsted acid was proposed.

# 2007 Elsevier B.V. All rights reserved.

Keywords: Phosgene-free; Diphenyl carbonate; Transesterification; Dimethyl oxalate; Diphenyl oxalate; Methyl phenyl oxalate; MoO3/TiO2–SiO2; MoO3/SiO2;

Bronsted acid; Lewis acid

1. Introduction

Polycarbonates (PCs) are important engineering thermo-

plastics with excellent mechanical and optical properties as

well as electrical and heat resistance properties. They have been

used in many fields as substitutes for glasses and metals. PCs

have been commercially prepared by the interfacial poly-

condensation of bisphenol-Awith phosgene. The disadvantages

* Corresponding author. Tel.: +86 22 2740 6498; fax: +86 22 2789 0905.

** Corresponding author. Tel.: +1 512 471 7988; fax: +1 512 475 7824.

E-mail addresses: xbma@tju.edu.cn (X. Ma), jlgong@mail.utexas.edu

(J. Gong).

0926-3373/$ – see front matter # 2007 Elsevier B.V. All rights reserved.

doi:10.1016/j.apcatb.2007.07.011

of this conventional phosgene process include the use of

large amounts of methylene chloride as the solvent, about ten

times the products’ weight, and highly toxic phosgene as a

reagent [1].

Over the past years, there has been increasing demand for

safe and environmentally friendly processes for PC synthesis.

One such process is the synthesis of diphenyl carbonate (DPC)

followed by transesterification between DPC and bisphenol-A

[2,3]. In this scheme, no toxic solvents are used and the by-

product, phenol, can be recycled.

However, DPC is prepared commercially by the reaction of

phenol and phosgene in the presence of bases such as sodium

hydroxide [4]. Obviously, this traditional process for DPC

synthesis has the same environmental concerns as mentioned

Y. Liu et al. / Applied Catalysis B: Environmental 77 (2007) 125–134126

above due to the use of phosgene. Thus, utilizing green DPC

synthesis has great potential to help avoid the social and

environmental effects of pollution.

So far, several non-phosgene alternatives for DPC synthesis

have been developed or proposed [5–14], e.g., the oxidative

carbonylation of phenol and transesterification reactions. Most

literature covering the catalytic oxidative carbonylation of

phenol process makes clear that it suffers from low phenol

conversions, poor diphenyl carbonate selectivity, and difficult

regeneration of the palladium catalyst [7–12]. Alternatively,

dimethyl carbonate (DMC) can be used as a substitute for

phosgene in the synthesis of DPC through transesterification

with phenol. Unfortunately, the involved reaction rate is low

and the formation of azeotropes between DMC and methanol

causes separation problems [4–6]. The transesterification of

dimethyl oxalate (DMO) with phenol to prepare diphenyl

oxalate (DPO), followed by the decarbonylation of DPO to

produce DPC, as shown in reactions (1) and (2), is an another

available route [13,14].

ðCOOCH3Þ2þ 2C6H5OH ! ðCOOC6H5Þ2þ 2CH3OH (1)

ðCOOC6H5Þ2 ! COðOC6H5Þ2þCO (2)

Technically, this method is more effective because no

azeotropes are formed in the reaction system and the co-

products, methanol and CO, can be separated easily. These, in

turn, can be reused in DMO production via oxidative

carbonylation of methanol as shown in reaction (3) [15].

2CO þ 2CH3OH þ 1=2O2 ! ðCOOCH3Þ2þH2O (3)

As a matter of fact, a pilot plant test in DMO production has

been completed by Ube Industries and the technology for large-

scale commercial production has been established. One of the

proposed applications of this process is to supply DMO for the

preparation of DPC [16].

With respect to DPC synthesis from the transesterification of

DMO with phenol, the decarbonylation of DPO to produce

DPC could be carried out easily over PPh4Cl catalyst, and the

yield of DPO could be up to 99.5% [17,18]. The synthesis of

DPO from the transesterification of phenol with DMO follows a

two-step reaction consisting of the transesterification of DMO

with phenol into methyl phenyl oxalate (MPO) (Reaction 4),

followed by the production of DPO via the disproportionation

of MPO (Reaction 5), as shown in the following reactions.

ðCOOCH3Þ2þC6H5OH ! C6H5OOCCOOCH3þCH3OH

(4)

2C6H5OOCCOOCH3 ! ðCOOC6H5Þ2þðCOOCH3Þ2 (5)

The thermodynamic equilibrium constants of reactions (4)

and (5) at 453 K are only 0.23 and 2.09 � 10�7, respectively,

based on thermodynamic calculations including group con-

tributions on liquid components. This result implies that

transesterification between DMO and phenol, especially the

disproportionation of MPO, is not thermodynamically favored.

The equilibrium conversion of DMO is only 32.4% based on

Eq. (4) [19].

The transesterification of DMO with phenol has been

generally performed in the liquid phase with homogeneous acid

catalysts like Lewis acids or soluble organic Pb, Sn, or Ti

compounds [13,14]. To our best knowledge, few studies have

been reported regarding the development of Bronsted acids

catalysts for the reaction. Therefore, some Bronsted acids were

tested in the present study to extend the scope of homogeneous

acid catalysts in the transesterification reaction. The results

unexpectedly showed that the conventional Bronsted acids had

better catalytic reactivities than the homogeneous acid catalysts

mentioned above regarding product selectivities. However, from

an industrial point of view, the difficulty of separating catalysts

from products in a homogeneous system and the potential

environmental impact must be considered. Based on these

concerns, the development of heterogeneous catalysts with

excellent catalytic performance is highly desirable, and some

efforts have been made in this direction [19–24]. Gong et al.

presented results of an investigation into the reactivities of

supported MoO3 catalysts in the transesterification of DMO with

phenol [20]. Due to low DPO selectivity and yield obtained, the

development of novel improved transesterification catalysts with

desirable general activity and excellent DPO selectivity is

necessary. This motivated some of the studies we present here.

Recently, TiO2 supported MoO3 catalysts have attracted

increased attention in many industrially relevant reactions such

as hydrodesulfurization, because of their high catalytic activity

and the strong interaction of active species with the support

[25,26]. However, TiO2 supports have relatively small specific

surface areas compared to those of silica, alumina, etc., and are

difficult to make into pellets. Moreover, the active anatase

structure possesses only low thermal stability, which makes TiO2

support alone unsuitable for industrial applications [27]. In order

to make good use of these advantages while overcoming the

barriers, we prepared composite TiO2–SiO2 supports, where

MoO3 was deposited by slurry impregnation, which is an

environmentally friendly process and a simple, clean, effective

alternative to the conventional preparation using a solution of

(NH4)6Mo7O24 [28]. The synthesis of DPC from the transester-

ification of DMO with phenol over MoO3/SiO2 and the novel

MoO3/TiO2–SiO2 catalysts was well investigated and compared.

Specifically, to gain a better understanding of the relationships

between the dispersion of surface species, distribution of the

surface acid sites (Bronsted and Lewis acidities), and the

catalytic activity, MoO3/SiO2 and MoO3/TiO2–SiO2 catalysts

were characterized by XRD, FT-IR, NH3-TPD, and FT-IR of

adsorbed pyridine measurements. We also present results of

reactivity measurements of conventional Bronsted acids to

examine their effects on our MoO3/TiO2–SiO2 catalysts. A

tentative mechanism for the transesterification reaction over a

Bronsted acid has been proposed.

2. Experimental

2.1. Catalyst preparation

Commercial silica (average particle size 4 mm, Jiangyan

City Chemical Auxiliary Factory of China) was used as a

Y. Liu et al. / Applied Catalysis B: Environmental 77 (2007) 125–134 127

support material. Prior to impregnation it was calcined at 393 K

for 2 h. This treatment resulted in silica with a surface area of

146 m2/g and an average pore size of 100 A, as measured by

nitrogen adsorption.

The TiO2–SiO2 composites were prepared by a conventional

impregnation method using an ethanol solution of Ti(OBu)4 in

excess of solution. Fifteen grams of the silica support were

impregnated with 100 ml of the salt solution until dry. After

impregnation, the samples were dried and calcined in a muffle

furnace at 393 K for 4 h and then 823 K for 5 h. The TiO2–SiO2

composites are given in a weight percentage based on

TiO2 contents (TiO2/(TiO2 + SiO2)) and labeled as 2% Ti–Si,

4% Ti–Si, 8% Ti–Si, 10% Ti–Si and 12% Ti–Si, respectively.

MoO3 (green powder, Tianjin No. 4 Chemical Reagents

Factory of China) was ground in an agate mortar before use.

Mixtures of the appropriate amounts of MoO3, pure SiO2 or

TiO2–SiO2 composite, and about 150 ml distilled water were

prepared in a set of rotary flasks. Each mixture was heated

under a rotary evaporator to 363 K for several hours depending

on the MoO3 loading amount. After impregnation, the water

was vacuum-removed at 353 K until dry followed by calcining

samples in a muffle furnace at 433 K for 6 h. MoO3 loadings

ranged from 1 to 32 wt.% and the composition of MoO3 is

expressed based on the weight percent of MoO3 contents

(MoO3/(MoO3 + SiO2)). The resultant catalysts were named as

MoO3/SiO2 and MoO3/Ti–Si, respectively.

2.2. Transesterification of DMO with phenol

The reaction was conducted in a 250-ml glass flask

equipped with a thermometer, a distillation apparatus, and a

rotor under refluxing conditions at atmospheric pressure. The

top of distillation column was kept at 353 K by flowing

recycled hot water through it to remove methanol from the

reaction system. Thus, the reaction equilibrium limitation in

reaction (1) was broken and the reaction was accelerated in the

desired direction. The reaction mixture contained 0.1 mol

DMO, 0.3 mol phenol and 1.8 g catalysts (0.01 mol for

homogeneous catalysts). After the raw materials and catalysts

were placed into the batch reactor, nitrogen gas was flowed at

30 SCCM to purge the air from the reaction system. After

10 min, nitrogen flow was stopped and the flask was heated at a

rate of 8 K min�1. The reaction was conducted at 453 K at

atmospheric pressure. Quantitative analyses of reaction

products and distillates were carried out on a SP3420 gas

chromatogram equipped with a flame ionization detector

(FID). An HP-5 capillary column (Hewlett-Packard Company,

15 m � 0.53 mm � 1.5 mm) was used to separate products for

GC analysis. The products were mainly DPO, MPO and AN.

An internal standard qualitative analysis method was used

where ethyl benzoate was chosen as an internal standard

reagent. The conversions were reported on the basis of the

limiting reagent, DMO, and were defined as the ratio of the

moles of converted DMO to the moles of DMO fed initially to

the reactor. The selectivities to MPO and DPO were expressed

as the moles of MPO and DPO produced per 100 mol of

consumed DMO. The yields of MPO and DPO were obtained

from multiplication of DMO conversion by the selectivities to

MPO and DPO, respectively.

2.3. Characterization of catalysts

Powder X-ray diffraction crystalline phases were deter-

mined at room temperature. A PANalytical X’Pert Highscore

(Holland) diffractometer, equipped with Co Ka radiation anode

(k = 1.78901 A, 40 kV and 40 mA), was used for these

measurements. Intensity data were obtained by step scanning

with a scanning rate of 128 min�1 from 2u = 58 to 2u = 808. The

Fourier transform infrared spectroscopy (FT-IR) spectra were

recorded on a Nicolet AVATAR360 single beam spectrometer at

ambient conditions using KBr disks, with a nominal resolution

of 4 cm�1. The mixed samples were pressed into a 10 mg/cm2

self-supporting wafers before measurements were conducted at

room temperature in the range of 4000–400 cm�1 32 times. In-

situ Fourier transform infrared spectroscopy (in-situ FT-IR) of

adsorbed pyridine (Py) was carried out on a Bruker

VECTOR22 FT-IR instrument with a 4 cm�1 resolution and

a 4000–500 cm�1 scanning range. The nature of the acid sites

was investigated using pyridine as a probe molecule. Prior to

each experiment, the sample was pressed into a 10 mg/cm2 self-

supporting wafer followed by evacuation (1 Pa) at 693 K for

1.5 h. The sample was cooled down to 303 K for 2 h, and then

exposed to pyridine with a base pressure of 30 Torr for 30 min.

The pre-treated sample was degassed and the spectra were

recorded at room temperature. The temperature-programmed

desorption of ammonia (NH3-TPD) was measured employing a

chemical adsorption spectrometer (Model 2910, Micromeritics

Co.). The catalyst was heated to 873 K in an Ar environment for

1 h, and then cooled to room temperature. Adsorption of

ammonia was carried out at 323 K upon saturation followed by

heating the sample to 873 K at 10 K min�1 to desorb NH3.

3. Results and discussion

3.1. Activities of acid catalysts and conventional ester

exchange catalysts

Table 1 shows the catalytic activities of different homo-

geneous catalysts for the transesterification of DMO with

phenol. A screened experiment showed that in the absence of

any catalysts, no yield of DPO was gained and only a trace

amount of DMO was converted (entry 1), suggesting that direct

transesterification was difficult to shift the equilibrium from the

raw materials to the products as reported previously [20].

Several typical Bronsted acids were tested in the transester-

ification reaction (entries 2–6). Generally, some protic acids

were very effective and selective for the present reaction

(entries 2–4). It is notable that the strong Bronsted acid, H2SO4,

showed a high conversion rate of DMO (61.6%) and a desirable

selectivity to DPO (27.1%), despite the formation of a small

amount of anisole (AN) (9.6% selectivity). In addition, as a

weak acid, H3PO4 exhibited a significantly high DMO

conversion of 60.9%, as well as a trace yield of AN. Another

interesting feature shown in our study is that Bronsted acids

Table 1

Activities of various catalysts for the transesterification of dimethyl oxalate with phenola

Entry Catalysts DMO

conversion (%)

Selectivity (%) Yield (%)

AN MPO DPO MPO DPO

1 None 1.3 0 82.1 17.9 1.1 0.2

2 H2SO4 61.6 9.6 62.8 27.1 38.7 16.7

3 HCl 49.2 0.1 94.9 5.0 46.7 2.5

4 H3PO4 60.9 0.2 91.7 8.1 55.8 4.9

5 HAc 41.6 0 100 0 41.6 0

6 H3BO3 33.7 0 99.5 0.5 33.5 0.2

7 ZnCl2 37.8 11.3 46.9 10.0 17.7 0.4

8 AlCl3 79.8 60.4 30.8 2.1 24.6 1.7

9 MgCl2 30.5 39.6 19.7 0 10.9 0

10 Ti(OC4H9)4b 92.5 0 20.6 2.8 19.0 2.6

11 SnOBu2b 25.7 2.2 80.8 12.5 20.8 3.2

12 C32H64O4Snb 54.1 1.8 21.0 13.7 11.4 7.4

MPO: methyl phenyl oxalate; DPO: diphenyl oxalate; AN: anisole.a Reaction conditions: 0.1 mol DMO, 0.3 mol phenol, 0.01 mol catalyst, conducted at 453 K for 2 h.b Conventional homogeneous catalysts for the transesterification of DMO with phenol.

Y. Liu et al. / Applied Catalysis B: Environmental 77 (2007) 125–134128

such as HAc and H3BO3 demonstrate total selectivities up to

100% (entries 5 and 6), though they were less active than strong

Bronsted acids from the viewpoint of DMO conversion.

Another aspect of this study that needs to be addressed is the

reactivity differences between conventional Bronsted acid

catalysts and Lewis acid catalysts. The experiments that were

performed to acquire the data shown in Entries 7–9 were an

attempt to address such an issue by using chloride salts of

Zn(II), Al(III), Mg(II) [25]. All the chloride salts showed high

DMO conversion but with high selectivity to AN rather than

MPO and DPO. As reported elsewhere, the strong acid sites

favor formation of AN [22]. However, the catalytic perfor-

mances of the strong Bronsted acids appear to be less

responsible for the formation of AN than strong Lewis acids.

Conventional homogeneous catalysts such as Ti(OC4H9)4,

SnOBu2, and C32H64O4Sn were also tested (entries 10–12)

[21]. Generally, they were somewhat effective for the formation

of MPO and DPO, but simultaneously gave anisole, benzyl

alcohol and methyl phenol as the main products. Consequently,

the selectivities to MPO and DPO were relatively low.

Table 2

The catalytic performances of MoO3/SiO2 and MoO3/Ti–Si catalystsa

Entry Catalysts Conversionb (%)

1 1% MoO3/SiO2 50.8

2 4% MoO3/SiO2 60.1

3 8% MoO3/SiO2 63.3

4 16% MoO3/SiO2 63.9

5 24% MoO3/SiO2 58.1

6 32% MoO3/SiO2 56.4

7 16% MoO3/2% Ti–Si 66.1

8 16% MoO3/4% Ti–Si 67.7

9 16% MoO3/8% Ti–Si 69.4

10 16% MoO3/10% Ti–Si 65.2

11 16% MoO3/12% Ti–Si 64.3

MPO: methyl phenyl oxalate, DPO: diphenyl carbonate, AN: anisole.a Reaction conditions: 0.1 mol DMO, 0.3 mol phenol, 1.8 g catalyst, conductedb Based on DMO converted.

The results above suggest that Bronsted acids are active and

selective in the transesterification of DMO with phenol, and

also form less of the by-product AN. Despite the fact that

conventional Bronsted acids are not expected to be applied in

most industrial reactions due to their separation and environ-

mental pollution problems, our experimental results provide an

important hint that Bronsted acid sites on solid active catalysts

would also exhibit desirable activity and selectivity towards

transesterification reactions.

3.2. Performances of MoO3/SiO2 and MoO3/Ti–Si catalysts

Table 2 presents the comparative results of catalytic

activities of a series of MoO3/SiO2 catalysts with different

MoO3 loadings and 16% MoO3/Ti–Si catalysts with different

TiO2 contents in the transesterification of DMO with phenol.

MoO3/Ti–Si catalysts performed better than MoO3/SiO2

catalysts regarding DMO conversions and DPO yields. It is

notable that in the case of MoO3/SiO2 catalysts, the DMO

conversions increased steadily with increasing MoO3 loadings

Selectivity (%) Yield (%)

AN MPO DPO MPO DPO

0.2 80.9 18.6 36.9 8.5

0.3 80.3 19.0 38.7 9.2

0.4 82.0 17.3 43.6 10.9

0.4 84.3 15.3 53.9 9.8

0.5 88.7 10.8 51.5 6.3

0.8 90.1 9.1 50.8 5.1

0.4 72.9 26.7 48.2 17.6

0.5 67.0 32.5 45.4 22.0

0.7 60.2 39.1 41.8 26.4

0.8 65.7 33.5 42.8 21.8

0.9 74.8 24.3 48.1 15.6

at 453 K for 2 h.

Y. Liu et al. / Applied Catalysis B: Environmental 77 (2007) 125–134 129

up to 16 wt.%, followed by a decrease in DMO conversions at

higher MoO3 loadings. As can be seen, the selectivity for DPO

only increased when MoO3 loadings were below 8 wt.% and

decreased afterwards. However, different performances have

been shown over MoO3/Ti–Si catalysts, where DMO conver-

sion, the selectivity of DPO, and the yield of DPO increased

remarkably with the addition of TiO2 below 16 wt.%. The

maximum values of 64.3, 39.1, 26.4% were reached at 8 wt.%

TiO2 loading, respectively, suggesting that not only was more

DMO converted into MPO but also more MPO was further

reacted to produce DPO over a MoO3/8% Ti–Si catalyst. As

reported previously, the TiO2/SiO2 catalytic system shows low

selectivities to DPO similar to MoO3/SiO2 [19,22]. Therefore,

we speculate that the incorporation of TiO2 can effectively

improve the catalytic activity and selectivity of MoO3/SiO2 in

the transesterification of DMO with phenol. This result can be

attributed to the synergetic effect of MoO3 with TiO2, which

will be explained in more detail later.

3.3. X-ray powder diffractograms

As mentioned earlier, a small amount of MoO3 or TiO2

played a key role improving catalytic performances, whereas a

large amount of MoO3 or TiO2 showed a negative effect. Hence,

XRD measurements were taken here to localize the dispersion

states of MoO3 on MoO3/SiO2 and MoO3/Ti–Si samples. As the

XRD spectra of all the samples shown in Fig. 1, no apparent

diffraction peaks associated with the bulk MoO3 phase were

detected at low MoO3 loadings of 1 and 4 wt.% over MoO3/

SiO2 samples. However, increasing sharp diffraction peaks of

orthorhombic MoO3 were observed when MoO3 loadings were

higher than 8 wt.%, which implies a weak interaction of MoO3

with SiO2. As for 16% MoO3/Ti–Si samples, the MoO3 peak

Fig. 1. XRD spectra of MoO3/SiO2 and MoO3/Ti–Si catalysts: (A) 1% MoO3/

SiO2; (B) 4% MoO3/SiO2; (C) 8% MoO3/SiO2; (D) 16% MoO3/SiO2; (E) 16%

MoO3/4% Ti–Si; (F) 16% MoO3/8% Ti–Si; (G) 16% MoO3/12% Ti–Si.

intensity was even weaker than those of 16% MoO3/SiO2 and

8% MoO3/SiO2 samples. Furthermore, the MoO3 phase peak

intensity decreased sharply with TiO2 content up to 8 wt.% and

then remained nearly constant independent of TiO2 loadings,

suggesting that the addition of TiO2 makes a better dispersion

for MoO3 species. On the other hand, since weak XRD peak of

anatase was only detected at 12 wt.% TiO2 content, it can be

deduced that TiO2 was highly dispersed on the silica surface

below 12 wt.% TiO2 content.

From the viewpoint of the monolayer dispersion theory [29],

amorphously dispersed TiO2 has a much stronger interaction

with the SiO2 support and other metal oxides than crystalline

TiO2. This prediction agrees with the experimental observa-

tions shown here that the amorphous TiO2 on the SiO2 could

effectively enhance the interaction between MoO3 and SiO2

and then improve the dispersion capacity of MoO3 on the

surfaces of carrier, whereas the crystalline TiO2 does not

contribute significantly to the strong interaction between MoO3

and SiO2. Thus, a conclusion can be drawn that the amorphous

TiO2 preferentially promotes the dispersion of MoO3 on the

surfaces of Ti–Si composite support.

It is notable that, based on the results of reactivity test and

XRD measurements, crystalline MoO3, which is more like bulk

MoO3, is active with respect to DMO conversion, but less

selective for the further disproportionation of MPO into DPO

compared to amorphous MoO3. Therefore, in the case of MoO3/

SiO2 catalysts, a DPO selectivity maximum was obtained at

4 wt.% MoO3 loading, strongly dependent on the amount of

surface amorphous MoO3. On the other hand, incorporation of

TiO2 favors surface dispersion of amorphous MoO3 by

providing more catalytic activity centers on the surfaces of

the carriers. Consequently, more DMO were converted and

more DPO were obtained over MoO3/Ti–Si.

3.4. Fourier transform infrared (FT-IR) spectroscopy

With the purpose of getting more detailed information

concerning the surface structure of MoO3 on the pure SiO2

surface and Ti–Si surface, IR spectroscopic analysis on SiO2,

MoO3/SiO2 and MoO3/Ti–Si samples was performed. As

shown in Fig. 2, IR spectra of all the samples produce low

frequency region bands at 1220–1020, 800 and 467 cm�1 due to

asymmetric stretching, symmetric stretching and bending

modes of bulk Si–O–Si, respectively [30,31]. A band at

1633 cm�1 that appears on all spectra can be assigned to water

[32]. These indicate that both MoO3/SiO2 and MoO3/Ti–Si

samples had similar chemical structures with SiO2.

Upon loading increasing amounts of MoO3, the spectra of

MoO3/SiO2 samples exhibited some new features that

depended on the MoO3 loadings. Absorption bands at 956

and 908 cm�1 are such examples, which can be ascribed to the

terminal Mo = O groups in the MoO3 phase and Mo–O–Si

vibration, respectively [31,33]. As shown in Fig. 2, both bands

at 956 and 908 cm�1 appear and increase in intensity with

increase of MoO3 loadings. However, the band due to the

vibration of Mo = O groups in the MoO3 phase at about

908 cm�1 did not appear in all the samples, probably because of

Fig. 3. NH3-TPD profiles of the MoO3/SiO2 and MoO3/Ti–Si catalysts: (A)

SiO2; (B) 4% MoO3/SiO2; (C) 8% MoO3/SiO2; (D) 16% MoO3/SiO2; (E) 16%

MoO3/4% Ti–Si; (F) 16% MoO3/8% Ti–Si; (G) 16% MoO3/12% Ti–Si.

Fig. 2. IR spectra of SiO2, MoO3/SiO2 and MoO3/Ti–Si catalysts: (A) SiO2; (B)

4% MoO3/SiO2; (C) 8% MoO3/SiO2; (D) 16% MoO3/SiO2; (E) 16% MoO3/4%

Ti–Si; (F) 16% MoO3/8% Ti–Si; (G) 16% MoO3/12% Ti–Si.

Y. Liu et al. / Applied Catalysis B: Environmental 77 (2007) 125–134130

the low content of molybdenum species [31,33]. Another

interesting feature is that the intensity of the bands at 908 cm�1

increased with increased MoO3 loadings, indicating an

enhancement of association between MoO3 and SiO2 within

the range of MoO3 loadings studied. Moreover, the increase of

another weak band at 956 cm�1 with the increase in Mo loading

suggests that the attributable species depend on the MoO3

loading and/or the formation of a Mo–O–Si bond, which was

shown by El Shafei et al. to be a prerequisite for the formation

of surface MoO3 phases [31].

As for the spectra of MoO3/Ti–Si samples, no new bands

appeared while the bands at 956, 908 cm�1 were much intenser

than those of MoO3/SiO2. Strong features for dispersed MoO3

were revealed, partially explaining the remarkable increase in

intensity of Mo–O–Si (908 cm�1) when TiO2 content was below

8 wt.%, Although IR spectroscopy is not very sensitive to

crystalline TiO2 [34], considering the XRD results, it can be

shown that the incorporation of amorphous TiO2 is responsible

for the enhancement of the weak association between MoO3 and

SiO2, and then the improved dispersion capacity of MoO3 on the

surface of SiO2. However, the crystalline TiO2 may not

effectively contribute further to the interaction of MoO3

with SiO2. Unfortunately, the small bands attributable to the

TiO2 vibrations (Ti–O–Si around 930 cm�1, Ti–O–Ti around

400–600 cm�1) [35] and Mo–O–Ti bond vibrations (around

785 cm�1) [36] were not observed in the spectra of MoO3/Ti–Si

samples. This result is probably due to saturation by more

intense SiO2 bands, which reduced the signal-to-noise ratio of

MoO3/Ti–Si bands. However, the results of FT-IR measurements

further manifested that amorphous TiO2 played a crucial role in

the enhancement of the interaction between MoO3 and SiO2.

Established knowledge of the microstructure of MoO3/Ti–Si

from literature provides us direct evidence to further address the

observation that incorporation of TiO2 improved the dispersion

of MoO3 on SiO2. As is well known, the surface of SiO2

support, unlike Al2O3, TiO2 and so on, contains abundant

weakly basic/neutral hydroxyl groups, which leads to the weak

interaction of the deposited MoO3 with support surfaces and

causes three-dimensional oxide agglomeration. Amorphous

TiO2 may work as a mediator between MoO3 and SiO2, i.e., be

present at the interface, and provide the hydroxyl groups on the

surface of the composite support [37]. On the other hand, as

proposed by Wachs et al., the dispersion of MoO3 on TiO2 can

be regarded as the dehydration of surface OH groups by MoO3

and TiO2 [38]. Considering that MoO3 is an acidic oxide and

that OH groups on the MoO3 surface would show acidic

properties, we propose here that the OH groups on MoO3

surface would preferentially interact with the basic OH groups

on TiO2 to form isolated tetrahedral species during the drying

procedure [39].

3.5. Temperature-programmed desorption of NH3 (NH3-

TPD)

NH3-TPD spectra were taken to survey the acid strength of

MoO3/SiO2 and MoO3/Ti–Si. In the NH3-TPD curves, peaks

are generally distributed into two regions: low-temperature (LT,

T < 673 K) and high-temperature (HT, T > 673 K) regions

[40,41]. The peaks in the HT region are ascribed to the

desorption of NH3 from strong Bronsted and Lewis type acid

sites, while the peaks in the LT region are assigned as the

desorption of NH3 from weak acid sites. As shown in Fig. 3, the

peaks appear only in the low temperature region, which

confirms only weak acid sites are on the surface. In addition, the

maximum temperature offset of these peaks is 15 K, implying

that the incorporation of TiO2 had a weak impact on the strength

of the surface acid of MoO3/Ti–Si.

Fig. 4. Amount of NH3 desorbed at low temperature from the MoO3/SiO2 and

MoO3/Ti–Si catalysts: (A) 1% MoO3/SiO2; (B) 4% MoO3/SiO2; (C) 8% MoO3/

SiO2; (D) 16% MoO3/SiO2; (E) 16% MoO3/4% Ti–Si; (F) 16% MoO3/8% Ti–

Si; (G) 16% MoO3/12% Ti–Si.

Fig. 5. FT-IR spectra of adsorbed pyridine on the MoO3/SiO2 and MoO3/Ti–Si

catalysts: (A) 1% MoO3/SiO2; (B) 4% MoO3/SiO2; (C) 8% MoO3/SiO2; (D)

16% MoO3/SiO2; (E) 16% MoO3/4% Ti–Si; (F) 16% MoO3/8% Ti–Si; (G) 16%

MoO3/12% Ti–Si.

Y. Liu et al. / Applied Catalysis B: Environmental 77 (2007) 125–134 131

Fig. 4 shows the amount of NH3 desorbed at low temperature

from MoO3/SiO2 and MoO3/Ti–Si catalysts. For the MoO3/

SiO2 catalysts, the formation of total surface acid sites

following MoO3 deposition was measured, suggesting MoO3

loadings had an impact on the amount of the total acid sites.

Specifically, the 16% MoO3/Ti–Si catalyst had more total acid

sites than the 16% MoO3/SiO2 counterpart. Also the amount of

desorbed NH3 increased with TiO2 content up to 8 wt.%, and

decreased slightly with further addition of TiO2. Therefore,

these data provide reliable evidence that the catalytic activity

has close relation with the total weak acid sites amount, namely,

the more the total weak acid sites amount is, the more active the

catalyst is. Furthermore, based on the results from XRD and

NH3-TPD measurements, for MoO3/Ti–Si catalysts with TiO2

content ranging from 2–8 wt.%, the amount of the total weak

acid sites increases with the amorphous TiO2 amount. This

indicates that the addition of amorphous TiO2 contributes a

number of new weak acid sites. Although crystalline TiO2 can

provide a few acid sites by itself, the number of acid sites it

provides is less than it occupies on SiO2, which resulted in the

decrease of total acid sites on MoO3/Ti–Si catalysts at the

highest TiO2 content with the presence of crystal TiO2. On the

other hand, since the surface acidity is a function of surface

structure, i.e. the coordinative environment of surface hydroxyl

groups rather than the number of sites present, it is not

surprising to find that the changes in the total amount of acid

sites did not result in the corresponding changes in surface acid

strength [42–47].

As reported elsewhere, weak acid sites are responsible for

the formation of MPO and DPO, while strong acid sites favor

the formation of the by-product AN [22]. Therefore, the results

of NH3-TPD validate an explanation for high selectivities to the

target products over MoO3/Ti–Si catalysts. Additionally,

combining the catalytic activity with the total acid amount

over the MoO3/Ti–Si catalysts, one may conclude that the total

acidity shows a direct influence on the activity of these catalysts

such that more weak acid sites provide more active centers.

3.6. In situ Fourier transform infrared (FT-IR)

spectroscopy of adsorbed pyridine (Py)

FT-IR measurement of adsorbed Py allows a clear

distinction between Bronsted and Lewis acid sites as shown

in Fig. 5. The absorption bands at 1540 cm�1 can be ascribed to

pyridine interaction with the Bronsted acid sites. The bands at

1450 cm�1 can be assigned to the pyridine coordinated with

Lewis acid sites. Another feature appearing at 1490 cm�1 can

be ascribed to the combined contribution of pyridine adsorbed

in Bronsted acid and Lewis acid sites [48–51]. Fig. 5 shows the

FT-IR spectra of pyridine adsorption at 1300–1700 cm�1 for

the MoO3/SiO2 and MoO3/Ti–Si catalysts. The IR pyridine

absorption spectra of MoO3/SiO2 produced bands at 1450 cm�1

and 1490 cm�1 but not at 1540 cm�1, suggesting that only

Lewis acid sites were on MoO3/SiO2 catalysts and MoO3

loading had little influence on acidity variety. In particular, as

shown in Fig. 6, the Lewis acid sites concentration reached a

maximum at low MoO3 loadings and decreased slightly with

further deposition of MoO3. The result is in good agreement

with a study by Rajagopal et al. [37], which showed that

although the acidity of SiO2 is too weak to chemisorb pyridine

at 473 K, acid sites from the deposited MoO3 can adsorb

pyridine. In our case, it can be concluded that amorphous MoO3

provides more available Lewis acid sites on the catalyst surface

than crystal MoO3, and thus the appearance of crystalline MoO3

at high loading decreases the Lewis acid sites concentration.

Since Lewis acid sites are the only acid species on the surface of

Scheme 1. Cluster models of the octahedral MoO3 species and tetrahedral

MoO3 species proposed to exist on the surface of the Ti–Si complex support.

These models were derived from the formal reaction of MoO3 and the most

abundant types of hydroxyls identified and classified by Knozinger and

Ratnasamy [39].

Fig. 6. Lewis and Bronsted acid amounts on the MoO3/SiO2 and MoO3/Ti–Si

catalysts: (A) 1% MoO3/SiO2; (B) 4% MoO3/SiO2; (C) 8% MoO3/SiO2; (D)

16% MoO3/SiO2; (E) 16% MoO3/4% Ti–Si; (F) 16% MoO3/8% Ti–Si; (G) 16%

MoO3/12% Ti–Si. aCalculated from the ratio of IR absorption peak area to

catalyst mass.

Y. Liu et al. / Applied Catalysis B: Environmental 77 (2007) 125–134132

MoO3/SiO2 catalysts, it is very likely that they are the active

centers for the transesterification of DMO with phenol to

produce MPO and DPO. This conclusion is consistent with

results reported elsewhere [23].

Interestingly, for the adsorbed pyridine (FT-IR) spectra of

the MoO3/Ti–Si catalysts, a new feature at 1540 cm�1 is shown

in Fig. 5 besides the pyridine adsorption bands at 1450 and

1490 cm�1, suggesting that both Lewis and Bronsted acid sites

existed on the surface of MoO3/Ti–Si catalysts. Furthermore, as

shown in Fig. 6, an increase in the amount of TiO2 incorporated

in the catalyst produced a decrease in Lewis acid sites

concentration while an increase in Bronsted acid sites

concentration was observed. To better understand these

observations and the relationship between the surface acidity

and the catalytic activities of these samples, here we will first

refer to established knowledge of the microstructure of

supported MoO3.

Considering the MoO3/SiO2 system first, Zhao et al.

proposed that there are two kinds of surface MoO3 species

on the support. One is a tetrahedrally coordinated surface

species, and the other is an octahedrally coordinated surface

species [51]. The crystalline MoO3 is an tetrahedrally

coordinated species and the surface amorphous MoO3 on

SiO2 is mostly octahedrally coordinated. In addition, Zhao et al.

confirmed that amorphous MoO3 cannot disperse on SiO2 in the

form of a tetrahedrally coordinated monolayer surface species

[52]. The Lewis acid sites formed on the surface of MoO3/SiO2

samples are mainly contributed by the octahedrally coordinated

MoO3 species (Scheme 1, Species I), rather than the

tetrahedrally coordinated MoO3 species.

Since MoO3 is preferably dispersed on the surface of the

monolayer dispersed TiO2 surface, it can be deduced that, in the

case of MoO3/Ti–Si, besides the octahedrally coordinated

species (Scheme 1, Species I), some surface MoO3 on Ti–Si is

also present as the tetrahedrally coordinated species (Scheme 1,

Species II), similar to the case of MoO3/TiO2 [51]. As reported

elsewhere, TiO2 deposited on SiO2 surface can generate only

Lewis acid sites rather than Bronsted acid sites [19]. In

addition, Mathieu et al. have shown that the OH groups of

crystalline TiO2 did not show protonic character based on the

adsorbed pyridine measurements [53]. Thus, we can speculate

that Bronsted acid sites were formed on the 16% MoO3/Ti–Si as

the new acidic surface species in the following way. As the

MoO3 loadings were high compared to the TiO2 contents, the

density and basicity of the unoccupied hydroxyl groups of TiO2

cannot fully satisfy the requirements of MoO3. Consequently,

during impregnation, the MoO3 species reacted with TiO2

hydroxyls, mainly forming a monodentate species (Scheme 1,

Species III) that was acidic. Such an argument is supported by

experimental evidence showing similar monodentate tetrahe-

dral species over MoO3/Al2O3 and WO3/Al2O3, respectively

[50,54]. Further, with the formation of species III, the

monodentate groups can also appear on top of the multilayers

of amorphous Mo oxide, leading to the acidic species IV

(Scheme 1). Therefore, based on the results presented above, it

is very likely that the formation of Bronsted acid sites is related

to the interaction between MoO3 and TiO2. Also, the reduced

Y. Liu et al. / Applied Catalysis B: Environmental 77 (2007) 125–134 133

amount of Lewis acid sites is primarily due to the onset of new

Bronsted sites.

The tetrahedrally coordinated surface MoO3 species has

lower coordination numbers than the octahedrally coordinated

bulk MoO3, which is ‘‘coordinatively unsaturated’’ [55]. The

coordinatively unsaturated MoO3 species played a significant

role in catalytic activities as explained by Knozinger [39]. The

surface atom coordination sphere may be completed by

adsorbed molecules, which may also be activated for catalytic

transformations [55]. From this point of view, one can explain

why the amorphous MoO3 on Ti–Si is more active and

selective than the crystalline MoO3. Also, it can be further

deduced that the tetrahedrally coordinated MoO3 species is

more active than the octahedrally coordinated MoO3 species.

It is notable that there were both tetrahedrally and octahedrally

coordinated MoO3 species on the surface of MoO3/Ti–Si

samples while only octahedrally coordinated MoO3 species

were on the surface of MoO3/SiO2 samples. This result is

probably another reason that MoO3/Ti–Si shows higher

activities than MoO3/SiO2.

To summarize, the incorporation of dispersed TiO2 on

MoO3/SiO2 favors not only forming Bronsted acid sites but also

suppressing Lewis acid sites. Both Bronsted and Lewis acid

sites are catalytically active centers for the transesterification of

DMO with phenol. Further, the tetrahedrally coordinated MoO3

species was more active than the octahedrally coordinated

MoO3 on the surface of Ti–Si.

3.7. Proposed reaction mechanism over Bronsted acid

Based on these experimental results, it is apparent that both

Lewis and Bronsted acid sites are the active centers for the

transesterification of DMO and phenol over MoO3/Ti–Si. Since

much attention has been paid to the Lewis-acid-catalyzed

transesterification reaction (the mechanism of the transester-

ification over Lewis acid sites has already been proposed

elsewhere [19,20]), here we only focus on the mechanism of the

transesterification over Bronsted acid.

As proposed in Scheme 2, which is somewhat and similar to

the mechanism over Lewis-acid, in the presence of proton, the

Scheme 2. A proposed reaction mechanism.

reaction course is determined by cleavage of acyloxy bond. The

carbonyl in DMO molecule is not easily attacked by a weak

nucleophilic reagent, phenol, leading to its low activity.

Furthermore, the charges between the carbon atom in carbonyl

and the oxygen atom are not distributed evenly. Consequently,

the proton combines with a negatively charged oxygen atom,

which converts the subsequent carbonyl into a carbenium ion.

The carbon atom in carbonyl possesses the higher positive

charge, causing preferential attack of phenyl oxalate.

4. Conclusions

In this work, we reported the successful synthesis of MPO

and DPO over MoO3/Ti–Si catalysts and compared their

catalytic performances to the MoO3/SiO2 catalysts in the

transesterification of DMO with phenol. The evaluation results

showed that, at some point, conventional Bronsted acids were

more efficient than the conventional Lewis acids from the

viewpoint of total selectivity to the target products. Further-

more, the strong Bronsted acid was less favorable for the

formation of the by-product AN than the strong Lewis acid. Ti–

Si supported MoO3 catalysts were more favorable for the

transesterification than the pure silica supported MoO3

catalysts based on the experimental results regarding DMO

conversion, and, particularly, DPO selectivity. XRD and FT-IR

measurements verified that incorporation of amorphous TiO2

could enhance the interaction of MoO3 with SiO2, which further

improved the dispersion state of MoO3 on the Ti–Si composite.

Additionally, the amorphous MoO3 were more responsible for

the increase of DMO conversion, and especially the DPO

selectivity, than the crystalline MoO3. NH3-TPD and FT-IR of

adsorbed Py characterizations indicated that the addition of

amorphous TiO2 created more weak acid sites on the surface of

the catalysts. Interestingly, the total surface acid sites showed a

direct influence on the activities of either MoO3/SiO2 or MoO3/

Ti–Si catalysts. Both Lewis acid sites and Bronsted acid sites

were present on the surface of MoO3/Ti–Si. The generation of

Bronsted acid sites was closely related to the association

between MoO3 and TiO2. In addition, there was more

tetrahedrally coordinated MoO3 on the surface of MoO3/Ti-

Si, which was more active than the octahedrally coordinated

MoO3 on the surface of MoO3/SiO2 catalysts. Evidence has

been shown to explain the significant catalytic improvement of

MoO3/Ti–Si catalyst by the incorporation of TiO2. Specifically,

the desirable catalytic efficiency of MoO3/Ti–Si, especially the

selectivity to DPO, can be ascribed to the highly dispersed

MoO3 and increased weak acid sites (including the synergistic

effect of the weak Lewis acid sites with the weak Bronsted acid

sites).

Acknowledgements

Financial support from the National Natural Science

Foundation of China (NSFC) (Grant no. 20276050), the

Program of Introducing Talents of Discipline to Universities

(Grant no. B06006), and the Program for New Century

Excellent Talents in University (NCET-04-0242) are gratefully

Y. Liu et al. / Applied Catalysis B: Environmental 77 (2007) 125–134134

acknowledged. The authors thank M.C. Akin and R.A. Ojifinni

for helpful discussion and for proofreading the manuscript.

References

[1] D. Freitag, U. Grico, P.R. Muller, W.H.F. Nouvertne, in: J.I. Kroschwitz

(Ed.), Encyclopedia of Polymer Science and Engineering, vol.11, Wiley-

Interscience, New York, 1987, p. 648.

[2] Y. Ono, Pure Appl. Chem. 68 (1996) 367.

[3] J.L. Gong, X.B. Ma, S.P. Wang, Appl. Catal. A 316 (2007) 1.

[4] M. Janatpour, S.J. Shafer, EP 228672 (1987).

[5] Z.H. Fu, Y. Ono, J. Mol. Catal. A 118 (1997) 293.

[6] W.B. Kim, J.S. Lee, Catal. Lett. 59 (1999) 83.

[7] I. Gabriello, R. Ugo, T. Renato, Ger. Offen., 2528412 (1976).

[8] W. Akinobu, O. Yoshiyuki, T. Hideaki, JP 11279126 (1999).

[9] M.H. Oyevaar, B.W. To, M.F. Doherty, US Patent 6093842 (2000).

[10] H. Hideaki, O. Yoshiyuki, M. Atusi, EP 684221 (1995).

[11] A.J. Chalk, Ger. Offen. 2738520 (1978).

[12] K. Kongen, C. Arihito, S. Kashun, JP 08198815 (1996).

[13] K. Nishihira, S. Tanaka, K. Nishimura, R. Sugise, US Patent 5834651

(1998).

[14] K. Nishihira, S. Tanaka, K. Harada, R. Sugise, A. Shiotani, K. Washio, US

Patent 5811573 (1998).

[15] T. Matsuzaki, A. Nakamura, Catal. Surv. Jpn. 77 (1997) 1.

[16] S. Uchiumi, K. Ataka, T.J. Matsuzaki, J. Organomet. Chem. 576 (1999) 279.

[17] K. Harada, R. Sugise, K. Kashiwagi, T. Dio, S. Niida, T. Kurafuji, US

Patent 5792883 (1998).

[18] S.P. Wang, X.B. Ma, Z.H. Li, G.H. Xu, Nat. Gas Chem. Eng. (China) 27

(2002) 1.

[19] S.P. Wang, X.B. Ma, H.L. Guo, J.L. Gong, J. Mol. Catal. A 214 (2004)

273.

[20] J.L. Gong, X.B. Ma, S.P. Wang, J. Mol. Catal. A 207 (2004) 215.

[21] X.B. Ma, H.L. Guo, S.P. Wang, Fuel Process. Technol. 83 (2003) 275.

[22] S.P. Wang, X.B. Ma, J.L. Gong, J. Mol. Catal. A 222 (2004) 183.

[23] X.B. Ma, J.L. Gong, S.P. Wang, N. Gao, Catal. Commun. 5 (2004) 101.

[24] S.P. Wang, X.B. Ma, J.L. Gong, Ind. Eng. Chem. Res. 43 (2004) 4027.

[25] S. Yoshinaka, K. Segawa, Catal. Today 45 (1998) 293.

[26] K.Y.S. Ng, E. Gulari, J. Catal. 95 (1985) 33.

[27] K. Segawa, M. Soeya, D.S. Kim, Res. Chem. Intermed. 15 (1991) 129.

[28] M. Zdrazil, Catal. Today 65 (2001) 301.

[29] Y.C. Xie, Y.Q. Tang, Adv. Catal. 37 (1990) 1.

[30] M.M. Mohamed, T.M. Salama, T. Yamaguchi, Coll. Surf. A 207 (2002) 25.

[31] G.M.S. El Shafei, M.M. Mokhtar, Coll. Surf. A 94 (1995) 267.

[32] S.S. Hong, M.S. Lee, S.S. Park, G.D. Lee, Catal. Today 87 (2003) 99.

[33] S.R. Seyedmonir, S. Abdo, R.F. Howe, J. Phys. Chem. 86 (1982) 1233.

[34] C. Rocchiccioli-Deltcheff, M. Amirouche, M. Che, J. Catal. 125 (1990) 292.

[35] S.K. Samantaray, K. Parida, Appl. Catal. A 220 (2001) 9.

[36] L.X. Sang, S.H. Zhong, X.F. Xiao, Chem. J. Chin. Univ. 25 (2004) 1115.

[37] S. Rajagopal, J.A. Marzari, R. Miranda, J. Catal. 151 (1995) 192.

[38] I.E. Wachs, R.Y. Saleh, S.S. Chan, C.C. Chersich, Appl. Catal. 15 (1985)

339.

[39] H. Knozinger, Adv. Catal. 25 (1976) 184.

[40] F. Lonyl, J. Valyon, Micropor. Mesopor. Mater. 47 (2001) 293.

[41] M. Sawa, M. Niwa, Y. Murakami, Zeolites 10 (1990) 532.

[42] H. Knozinger, P. Ratnasamy, Catal. Rev. 17 (1978) 31.

[43] V. Barron, J. Torrent, J. Colloid Interf. Sci. 177 (1996) 407.

[44] C. Contescu, J. Jagiello, J.A. Schwarz, J. Phys. Chem. 97 (1993) 10152.

[45] W. Stumm, R. Wollast, Rev. Geophys. 28 (1990) 53.

[46] W. Stumm, Johowiley & Sons Inc. (1992).

[47] P.J. Pretorius, P.W. Linder, Chem. Spec. Bioavailab. 15 (2003) 127.

[48] T. Barzetti, E. Selli, D. Moscotti, L. Forni, J. Chem. Soc., Faraday Trans.

92 (1996) 1401.

[49] T.R. Hughes, H.M. White, J. Phys. Chem. 71 (1967) 2192.

[50] C.A. Emeis, J. Catal. 141 (1993) 347.

[51] B.Y. Zhao, Q. Xu, L.L. Gui, Y.Q. Tang, Acta Chim. Silica 48 (1990) 227.

[52] F.G. Zhao, Y.H. Duan, J.Q. Li, Y.L. Fu, Chin. J. Chem. Phys. 4 (1991) 276.

[53] M.V. Mathieu, M. Primet, P. Pichat, J. Phys. Chem. 75 (1971) 1221.

[54] J. Bernhole, H.A. Horsley, L.L. Murrell, J. Phys. Chem. 91 (1987) 1526.

[55] H. Knozinger, Science 287 (2000) 1407.