Journal of Catalysis 299 (2013) 119–128

Contents lists available at SciVerse ScienceDirect

Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat

Preparation and characterization of WOx/ZrO2 nanosized catalysts with high

WOx dispersion threshold and acidity
Kunshan Song a, Hongbin Zhang a, Yahong Zhang a, Yi Tang a,⇑, Kangjian Tang a,b,⇑
a
Department of Chemistry, Shanghai Key Laboratory of Molecular Catalysis and Innovative Materials and Laboratory of Advanced Materials, Fudan University,
Shanghai 200433, China
b
Shanghai Research Institute of Petrochemical Technology, SINOPEC, Shanghai 201208, China

a r t i c l e i n f o a b s t r a c t

Article history: WOx/ZrO2 nanosized catalysts with high dispersion threshold were prepared via a novel two-phase inter-
Received 8 August 2012 face hydrolysis (TPIH) method and characterized by N2 physisorption, X-ray diffraction, transmission elec-
Revised 6 November 2012 tron microscopy, UV–vis diffuse reflectance spectroscopy, Raman spectroscopy, and potentiometric
Accepted 7 November 2012
titration with n-butylamine. It was found that the WOx/ZrO2 nanosized catalyst prepared by TPIH pos-
Available online 12 January 2013
sessed a higher dispersion threshold (ca. 0.372 g WOx/g ZrO2) than that obtained by a conventional impreg-
nation method, due to its higher specific surface area of 114 m2/g and tungsten surface density (6.2 W
Keywords:
atoms/nm2). Significantly, the catalyst at dispersion threshold by TPIH has a remarkable number of strong
WOx/ZrO2
Nanosized catalyst
acid sites with E > 400 mV and nearly double catalytic activity for n-pentane isomerization compared to
Two-phase interface hydrolysis that prepared by a conventional impregnation method. When this is combined with the results of spectral
Dispersion threshold characterization, the high activity of the catalysts obtained by TPIH is ascribed to the abundant strong
Isomerization acidic polytungstate species on ZrO2 at their dispersion threshold. Challenges for future work would be
developing the TPIH method with supports and exploring reaction conditions to reduce coke formation.
Ó 2012 Elsevier Inc. All rights reserved.

1. Introduction that the dispersion threshold should be at 6–8 W atoms/nm2

[22,24,26,27]. When the W loading is far lower than this value,
Solid acid catalysts play an important role in a great number of W usually exists in the form of isolated monotungstate species.
petroleum and chemical processes [1,2], such as catalytic cracking In contrast, when the W loading is close to or over the dispersion
[3], isomerization [4,5], alkylation [6], and hydration and dehydra- threshold, WOx would form polytungstate species and finally the
tion [7], because of their high activity, convenient separation, and crystallized WO3 [14,22]. It is believed that the maximum catalytic
lesser corrosion of reactors and environmental problems. Tungsten activity of WO3/ZrO2 appears at a W loading around the dispersion
oxide-based materials such as WO3/Al2O3 [8,9], WO3/TiO2 [10,11], threshold [14,22,24,28,29]. However, most of the experimental
WO3/SiO2 [12,13], and WO3/ZrO2 [14–18] are considered promis- values of catalysts corresponding to the dispersion threshold are
ing solid acid catalysts, owing to their strong acidity and thereby reported at 4–5 W atoms/nm2 [14,16,22,27,29,30]. Therefore, the
notable activity for various catalytic reactions. For example, new strategies for improving the dispersion threshold of WOx are
WO3/ZrO2, first reported by Hino and Arata [19], has a very strong still desired. On the other hand, there still exist some arguments
acidity of H0 < 14.52 and low-temperature catalytic activity for on the origin of catalytic activity. For example, Iglesia and co-work-
isomerizations of n-butane and pentane. Furthermore, compared ers [15,18,24,26] reported that catalytic activity depended on the
with other strong solid acids, for example, sulfated zirconia, Brønsted acid sites formed by the reduction of surface polytung-
WO3/ZrO2 is more stable under reduction conditions and shows state species, while Knözinger and co-workers [25,28] proposed a
less loss of active species [20,21]. So far, many scientists have been model of Zr-heteropolytungstates containing charge-compensat-
focusing their attention on this catalyst. Early studies demonstrate ing protons, which should be responsible for the strong Brønsted
that there is a maximum loading amount of WOx on ZrO2 before acidity of WO3/ZrO2. And Ross-Medgaarden et al. [22], Zhou et al.
crystallized WO3 is formed, that is, a dispersion threshold [22– [31], and Soultanidis et al. [14] attributed catalytic activity to Zr–
25] where WOx species are considered to cover the surface of WOx clusters.
ZrO2 in a monolayer state. Calculations of different models show In addition to W loading, the preparation route also affects the
structures and properties of WOx/ZrO2 catalysts. Impregnation
⇑ Corresponding authors. Fax: +86 21 65641740. [14,24], coprecipitation [20,21], and sol–gel [32,33] methods are
E-mail addresses: yitang@fudan.edu.cn (Y. Tang), tangkj.sshy@sinopec.com general ways to synthesize this type of catalyst. Previous studies
(K. Tang). by Santiesteban et al. [20] showed that coprecipitation could

0021-9517/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcat.2012.11.011
120 K. Song et al. / Journal of Catalysis 299 (2013) 119–128

improve the WOx dispersity and acidity compared with impregna- In addition, the reference samples of WOx/ZrO2 catalysts were
tion. Cortés-Jácome et al. [34] reported that the different condi- prepared by a conventional impregnation method referring to the
tions during the precipitation process led to different structures literature [36]. Briefly, aqueous ammonia (6 mol/L NH3H2O) was
and acidity. Boyse and Ko [32] investigated the effects of prepara- added slowly to an aqueous solution of ZrO(NO3)27H2O with
tion parameters on the physicochemical properties of WO3/ZrO2 0.2 mol/L Zr4+. After stirring and aging, the Zr(OH)4xH2O precipi-
and reported that a higher activation temperature was preferred tate was separated through centrifugation and dried. The as-
for sol–gel synthesis, since tungstate in the bulk was expelled onto prepared Zr(OH)4xH2O was impregnated with an aqueous solution
the surface of ZrO2, although the active species remained the same. of (NH4)6W12O3912H2O, followed by calcination at 973 K for 6 h as
Martínez et al. [29] illustrated the influence of preparation meth- described above. The WOx/ZrO2 catalysts obtained via this impreg-
ods (impregnation versus coprecipitation) on the optical and elec- nation method are named IM-WZ or IM-WZ(x), where x holds the
tronic properties of WOx domains on ZrO2 surfaces in the Pt- same meaning as above.
promoted WO3/ZrO2 catalysts. Most recently, a novel one-step syn-
thesis method by a two-phase interface hydrolysis (TPIH) method 2.2. Characterization
was reported for the preparation of high-quality ultrafine metal
oxide nanocrystals [35], which has the advantages of simplicity The chemical composition of the prepared WOx/ZrO2 catalysts
and easy recycling for reaction medium. Considering that ZrO2 pre- was characterized by inductively coupled plasma-atomic emission
pared by this technique can be controlled well in particle size dis- spectroscopy (ICP-AES) on a Hitachi P-4010 instrument. Nitrogen
tribution even below 10 nm [35], TPIH would be an effective sorption isotherms were obtained on a Micromeritics ASAP 2010
method of preparing multicomponent catalysts due to the rela- adsorption analyzer at 77 K. BET specific surface areas were calcu-
tively high specific surface area of the ultrafine nanosized ZrO2 lated based on adsorption data for a relative pressure range from
and its interaction with dispersed components during the confined 0.06 to 0.20. The crystal phases were investigated by X-ray powder
cohydrolysis process. diffraction (XRD) on a Rigaku D/MAX 2000 diffractometer using
Here, the TPIH method is employed for the first time to pre- CuKa radiation (k = 1.54056 Å). The average crystallite sizes were
pare WOx/ZrO2 catalysts, and a series of catalysts are obtained estimated according to Scherrer’s formula [37]: D = 0.9k/(bcos h),
with small particle size, high specific surface area, and high dis- where D was the average crystallite size, k the X-ray wavelength,
persion threshold of WOx compared to those prepared by con- b the peak width at half height, and h Bragg’s angle. The morphol-
ventional methods. Moreover, the WOx/ZrO2 nanosized catalysts ogy and structure information were obtained by field-emission
prepared by TPIH also exhibit some remarkable features with re- transmission electron microscopy (FETEM, JEM-2100F). UV–vis dif-
spect to the state of WOx species and acidity. On this basis, sys- fuse reflectance spectroscopy (UV–vis DRS) was recorded on a Shi-
tematic studies are pursued to explore the correlation among the madzu UV-2045 PC at room temperature with BaSO4 as reflectance
WOx surface species, acidity, and catalytic activity of WOx/ZrO2 standard. Raman spectroscopy was performed using a Dilor Lab-
catalysts. RAM-1B spectrometer with 632.8-nm laser excitation under ambi-
ent conditions.
The acidity of these WOx/ZrO2 catalysts was investigated by
2. Experimental potentiometric titration with n-butylamine. Briefly, a catalyst
was first dried in flowing air at 723 K for 2 h to desorb the adsorbed
2.1. Preparation of WOx/ZrO2 catalysts moisture. Then, about 120 mg of dried catalyst was dispersed in
100 mL of acetonitrile. After being agitated for 2.5 h, the suspen-
The WOx/ZrO2 catalysts were synthesized according to the TPIH sion was titrated by 0.02634 mol/L n-butylamine solution in aceto-
method under hydrothermal conditions reported previously [35]. nitrile with a ZDJ-5 automatic titrator (Lei-Ci Device Works,
Typically, about 5 g of zirconium butoxide (80 wt% solution in 1- Shanghai). Pyridine-adsorbed Fourier transform infrared spectros-
butanol, Aldrich) was diluted with toluene to 50 mL in a volumetric copy (FTIR) was performed on a Nicolet 380 FTIR spectrometer.
flask. Then, 6 mL of this diluted zirconium butoxide was trans- Brønsted (B) and Lewis (L) acid site concentrations were quantified
ferred into the inner chamber of a two-chamber Teflon-lined steel according to the Beer–Lambert law based on the FTIR bands at
autoclave. A proper amount of tungsten isopropoxide (5% w/v in 1540 cm1 and 1450 cm1 with molar extinction coefficients
isopropanol, Alfa Aesar) was then added into the above inner of 1.67 and 2.22 cmlmol1, respectively [38]. The samples were
chamber, followed by a moderate shaking to mix the two compo- first pressed into disks with a diameter of 15 mm and then pre-
nents thoroughly. After 10 mL of distilled water was injected into treated in situ at 673 K under vacuum for 2 h. After pyridine was
the outer chamber of the Teflon-lined steel autoclave, the auto- introduced to adsorb onto B and L acid sites at 473 K, the sample
clave was sealed and placed in an oven maintained at a tempera- disks were pretreated at 473, 573, and 673 K to desorb pyridine,
ture of 423 K for 24 h. After the autoclave was taken out and and the FTIR spectra were, respectively, recorded at different steps.
then cooled down in static air, the hydrolysis product was sepa-
rated simply by decanting residual toluene in the inner chamber 2.3. Catalytic studies
of the autoclave and dried at 353 K for 36 h to get a powder sam-
ple. Afterward, the as-prepared samples were heated at a ramp The catalytic performance was evaluated by the isomerization
rate of 2 K/min under an air flow and held at 773 K for 3 h followed of n-pentane on a downflow fixed-bed reactor. Typically, the cata-
by calcination at 973 K for 6 h. The WOx/ZrO2 catalysts prepared by lyst was pressed, crushed, and sieved to 20–40 mesh particle size,
TPIH are donated as TPIH-WZ, sometimes TPIH-WZ(x, q) or TPIH- and then, 0.1 g of that was mixed with quartz sand with a volume
WZ(x), where x refers to the tungsten oxide mass (g, quantified in ratio of 1:3. This mixture was then placed in the central region of
the form of WO3) per g of ZrO2, and q stands for the tungsten sur- the reactor, with each end filled with quartz sand. Before the reac-
face density (W atoms/nm2) calculated by x and BET surface area tion, the catalyst was pretreated at 673 K under flowing dry air for
of corresponding WOx/ZrO2 catalyst. 2 h to remove the adsorbed moisture. Then, the reactor was cooled
Ultrafine nanosized ZrO2 without supported tungsten oxide was to the reaction temperature under flowing high-purity nitrogen.
prepared by a similar process, in which just zirconium butoxide The reactant n-pentane was pretreated by outgassing and filtration
and toluene were added in the inner chamber of the autoclave and then introduced via a SSI Series II digital pump under a
and other conditions are the same as above. backpressure of 5 atm to keep a stable flow. The flow rate was
 K. Song et al. / Journal of Catalysis 299 (2013) 119–128 121

0.006 mL/min, corresponding to a WHSV of 2.25 h1. Then, the li- be the sample with the dispersion threshold of WOx. According
quid n-pentane was gasified in an evaporation chamber at 473 K. to the research results of previous researchers, the surface density
After that, the gaseous n-pentane underwent an isomerization of W at the dispersion threshold normally lies at about 4–5 W
reaction in the fixed-bed reactor. Under these conditions, the mass atoms/nm2 [14,16,22,27,29,30,43,44]. But, our results show that
and heat transfer can be neglected, since the Mears and Weisz- the W surface density at the dispersion threshold of the catalysts
Prater criteria are smaller than 0.15 and 1, respectively [39], and prepared by the TPIH method could reach as high as 6.2 W
the calculated temperature rise on the catalyst surface is less than atoms/nm2 (Table 1). This higher W surface density, together
9 K according to an adiabatic model. Finally, the products were with the enhanced BET surface area, ensures the outstandingly
analyzed by online gas chromatography with a HP-alumina capil- high WOx dispersion threshold of TPIH-WZ, 0.372 g/g ZrO2
lary column and a flame ionization detector (FID). The n-pentane [14,22,23,29].
conversion was calculated by peak area normalization of a chro- The average crystallite size of t-ZrO2 is estimated by Scherrer’s
matogram to evaluate the catalytic activity of various WOx/ZrO2 formula [25] from the (1 0 1) reflection of t-ZrO2 in XRD pattern.
catalysts. Fig. 2 presents the average crystallite sizes of t-ZrO2 as well as
the BET surface areas of TPIH-WZ versus the loading of WOx.
3. Results and discussion Clearly, the crystallite sizes of t-ZrO2 range between 7.4 and
8.6 nm. It is very interesting that there is a perfect match between
3.1. Chemical composition and BET surface area analysis the trends of crystallite sizes of t-ZrO2 and BET surface areas of
WOx/ZrO2. With increasing WOx loading, the crystallite size of t-
Table 1 lists the chemical compositions and BET surface areas of ZrO2 decreases monotonically, while the BET surface area of
the prepared TPIH-WZ nanosized catalysts. Compared with the WOx/ZrO2 increases until it reaches the WOx dispersion threshold,
nanosized ZrO2 without WOx loading, it is obvious that the addi- TPIH-WZ(0.372, 6.2). When the loading of WOx is above the
tion of a very small amount of WOx can greatly suppress the sinter- dispersion threshold, the crystallite size of t-ZrO2 begins to mono-
ing of ZrO2 and leave WOx/ZrO2 catalysts with very high BET tonically increase, while the BET surface area of WOx/ZrO2 de-
surface areas. Moreover, this suppression effect varies with the creases. The minimum of crystallite size and maximum of BET
amount of loaded WOx considering the trend of BET surface area surface area are both located at the point of the dispersion thresh-
variation. For the samples with WOx loading below 0.372 g WOx/ old of WOx. The crystallite size in our result is smaller than that in
g ZrO2, the BET surface area increases with increasing WOx loading. the literature [14,25], which should be due to the confined hydro-
However, above this loading, the BET surface area begins to de- lysis and lesser aggregation of nanoparticles for the TPIH method
crease gradually. Notably, the maximum value of the BET surface [35]. In the synthesis process, the droplets of the organic solvent
area of these WOx/ZrO2 catalysts (114 m2/g) is higher than that toluene could act as nanoreactors in which the hydrolysis of the
of WOx/ZrO2 catalysts prepared by a conventional impregnation reactant and growth of ZrO2 have been confined. This leads to
process under similar conditions, which is often around 90 m2/g ZrO2 particles with small size. Meanwhile, in the synthesis and
[14,22,25,36,40,41]. product collected process, capillary condensation induced by water
has been effectively avoided without contacting bulk water, which
3.2. XRD patterns prevents the further aggregation of nanocrystals. Moreover, the
cohydrolysis of zirconium butoxide and tungsten isopropoxide on
Fig. 1 shows the XRD patterns of TPIH-WZ nanosized catalysts. the interface of toluene droplets brings about stronger interaction
Similar to the observation of BET surface areas, the incorporation of between WOx and ZrO2 in a confined space and finally better
WOx affects the crystallization of ZrO2 dramatically. The bare nano- dispersity of WOx on ZrO2.
sized ZrO2 without WOx mainly exists in the monoclinic phase (m-
ZrO2, a stable phase under relatively high temperature) [22,42]. 3.3. TEM imaging
However, once WOx is added, ZrO2 dominantly crystallizes in the
tetragonal phase (t-ZrO2) [14,22,42], a metastable structure at TEM and high-resolution TEM (HRTEM) images of two typical
low temperature. At low WOx loadings, there is no diffraction line TPIH-WZ samples are displayed in Fig. 3. The crystallites in these
of the crystalline WO3 phase in the XRD patterns, implying that images present a diameter range between 6 and 10 nm, which is
WOx in these samples exists in a highly dispersed state on ZrO2 consistent with the average crystallite size estimated from XRD
surface. When the loading of WOx exceeds 0.372 g/g ZrO2, the pat- patterns. The HRTEM images of the catalyst with WOx dispersion
tern attributed to monoclinic WO3 (m-WO3) [14] emerges and be- threshold TPIH-WZ(0.372, 6.2) show only the crystal structure
comes distinct with further increases in WOx loading. It may be belonging to t-ZrO2, indexed to a crystal plane spacing of
deduced from this phenomenon that TPIH-WZ (0.372, 6.2) should 0.29 nm (Fig. 3a–c). In contrast, the HRTEM images of TPIH-
WZ(0.396, 6.6), with WOx loading just a little above the dispersion
Table 1 threshold, mainly display the crystal t-ZrO2 structure with tiny
Chemical composition, BET surface area, and tungsten surface density of TPIH-WZ WO3 structure with d-spacing 0.36 nm (Fig. 3d–f). Moreover, all
nanosized catalysts.
the HRTEM images show that the crystal lattice fringes inside the
Catalyst xa (g WOx/ SBET (m2/g) q (W t-ZrO2 grains extend directly to the surface without defects, indi-
g ZrO2) atoms/nm2) cating no formation of W-enriched domains within the t-ZrO2
Nanosized ZrO2 0 38 0 grain. So, it is reasonable to speculate that some kinds of noncrys-
TPIH-WZ(0.153, 4.4) 0.153 78 4.4 talline WOx species should exist on the surfaces of t-ZrO2 grains.
TPIH-WZ(0.192, 4.2) 0.192 100 4.2
TPIH-WZ(0.235, 4.7) 0.235 106 4.7
TPIH-WZ(0.292, 5.4) 0.292 108 5.4
3.4. UV–vis DRS
TPIH-WZ(0.337, 6.0) 0.337 109 6.0
TPIH-WZ(0.372, 6.2) 0.372 114 6.2 The absorption spectra of UV–vis DRS are illustrated in Fig. 4 in
TPIH-WZ(0.396, 6.6) 0.396 111 6.6 the form of the Kubelka–Munk function, F(R1) [24], versus incident
TPIH-WZ(0.421, 7.4) 0.421 104 7.4
photon wavelength. The spectrum for bare nanosized ZrO2 exhibits
TPIH-WZ(0.475, 8.9) 0.475 94 8.9
only one absorption band with the edge around 245 nm and max-
a
Calculated by ICP-AES results. imum near 210 nm [45]. For TPIH-WZ samples, another three
122 K. Song et al. / Journal of Catalysis 299 (2013) 119–128

Fig. 1. XRD patterns of various TPIH-WZ nanosized catalysts and bare nanosized ZrO2. Marks represent the peak positions of corresponding crystalline phases.

WOx loading below the dispersion threshold give no sign of crystal-

line WO3, as seen in XRD patterns (Fig. 1), these two bands must
originate from surface WOx species such as isolated WO4 species
or WO6-containing clusters [46]. Another absorption band, located
between 390 and 400 nm, only emerges in the spectra of the sam-
ples with WOx loading above the dispersion threshold of 0.372 g
WOx/g ZrO2, indicating that it should be ascribed to the crystalline
WO3. The appearance of a crystalline WO3 absorption band for the
sample over the WOx dispersion threshold further confirms the re-
sults of XRD experiments.
The UV–vis absorption edge energy (Eg) can well reflect the elec-
tronic structure and domain size of surface species [24,46,47]. To
investigate the specific WOx species of TPIH-WZ, their Eg values
are calculated by a linear extrapolation of the straight line of the
slope near the absorption edge in the plot of [F(R1)hm]1/2 versus
hm based on indirect electron transition [24]. The relationship of
Eg values to WOx loading is shown in Fig. 5. For the catalysts with
Fig. 2. BET surface area and average crystalline size of TPIH-WZ nanosized catalysts
WOx loading below the dispersion threshold, 0.372 g WOx/g ZrO2,
versus WOx loading.
just one Eg value is obtained at 3.26 ± 0.05 eV. On the basis of the
direct relation between Eg and domain size of WOx species, these
absorption bands appear with the band maxima between 240 and Eg values of 3.26 ± 0.05 eV correspond to the WOx species with
255, 280 and 290, and 390 and 400 nm. Two of these bands, in the polytungstate structure [24]. This indicates that polytungstate spe-
ranges 240–255 and 280–290 nm, appear in the spectra of all the cies exist in all those samples, even for those with very low WOx
TPIH-WZ samples. Taking into consideration that the samples with loading, although the existence of monotungstate species cannot

Fig. 3. TEM and HRTEM images of (a–c) TPIH-WZ(0.372, 6.2) and (d–f) TPIH-WZ(0.396, 6.6).
 K. Song et al. / Journal of Catalysis 299 (2013) 119–128 123

for the samples below the dispersion threshold. When the WOx
loading exceeds 0.372 g/g ZrO2, those three characteristic Raman
bands of crystalline m-WO3 are present and grow much more in-
tense with the increasing WOx loading. This is consistent with
the XRD and UV–vis DRS results.
In the region 810–1000 cm1, a broadband consisting of several
overlapping peaks is present. This band can be discussed in detail
as follows:

(i) The broad band located at 977–989 cm1 is attributed to the

stretch mode of monooxo W@O species [16,25,47]. This
band shifts from 977 cm1 for TPIH-WZ(0.153, 4.4) to
989 cm1 for TPIH-WZ(0.292, 5.4) and then keeps almost
the same wavenumber even with increasing WOx loading
(Fig. 6b). This shift for the catalysts with low WOx loading
suggests a growing domain size of the surface WOx species
[16,22,30,47], for example, from isolated monotungstate
species to polytungstate species. After that, the wavenumber
Fig. 4. UV–vis DRS of various TPIH-WZ nanosized catalysts and bare nanosized of the Raman band from monooxo W@O species tends to
ZrO2.
keep a similar value, indicating that for the samples with
WOx loading higher than 0.292 g/g ZrO2 [TPIH-
WZ(0.292, 5.4)], the surface species have similar polytung-
state structures.
(ii) The band from those overlapping peaks centered around
820 cm1 is most probably associated with the stretch mode
of WAOAW bonds [22,25,29,43,47]. The presence of the
820 cm1 band even in the sample with the lowest WOx
loading reflects the existence of WAOAW in low WOx-load-
ing catalysts. This 820 cm1 band of polytungstate species
finally evolves into the WAOAW vibration in crystalline
m-WO3 at 807 cm1 [22] in the samples over the WOx dis-
persion threshold.
(iii) The band at ca. 907 cm1 should correspond to the vibration
of WAOAZr bonds [22,25,46], indicating the occurrence of
interaction between the dispersed WOx species and the
ZrO2 support in all TPIH-WZ samples.

For the samples above the dispersion threshold, although the

Raman signals of crystalline WO3 would hide that of surface WOx
Fig. 5. UV–vis absorption edge energy (Eg) of TPIH-WZ nanosized catalysts with species [24], there are still obvious bands belonging to the vibra-
different WOx loading. tion of W@O and WAOAZr bonds of polytungstate species at 989
and ca. 907 cm1, even for the prepared catalyst with highest
WOx loading [TPIH-WZ(0.475, 8.9)].
be excluded here for the submergence of its absorption edge by that
of polytungstate species. The Eg values of the samples with WOx 3.6. Acid properties
loading over the dispersion threshold show two distinct values
(Fig. 5). The Eg values of about 2.60 eV, derived from the absorption The acid properties of WOx/ZrO2 catalysts are investigated by
band of 390–400 nm in the absorption spectra, are the characteris- potentiometric titration with n-butylamine in nonaqueous media
tic absorption edge energy of crystalline WO3 [24,29,46], and the [49]. The resultant titration curves are presented in Fig. 7. As a gen-
higher Eg values, 3.0–3.2 eV, still originate from the absorption of erally accepted criterion for interpreting the titration results, the
polytungstate species just as in the low WOx-loading region. Com- initial electrode potential (Ei) can be used as a sign of the maxi-
pared with the Eg values of polytungstate species of the catalysts mum strength of acid sites, and the trend of E with the amine titra-
with low WOx loading (3.26 ± 0.05 eV), the slightly lower Eg values tion amount in meq amine/g solid represents the acid strength
in this region should be due to the overlap of this absorption band distribution [50,51]. For comparison, a reference sample with
with that of crystalline WO3. WOx content of 0.193 g/g ZrO2 was also prepared by conventional
impregnation method and denoted as IM-WZ(0.193). This sample
3.5. Raman spectroscopy was reported at the WOx dispersion threshold of impregnation
and exhibited the best catalytic activity for n-pentane isomeriza-
The Raman spectra of TPIH-WZ catalysts recorded at room tem- tion [14]. The XRD and Raman spectra of IM-WZ(0.193) are dis-
perature under ambient conditions are depicted in Fig. 6. All of played in Fig. 8, which confirms that there is no crystalline WO3
them show the typical Raman signals of t-ZrO2 at 277, 312, 456, in IM-WZ(0.193). The XRD spectrum of Fig. 8a shows that ZrO2 in
and 645 cm1 without obvious bands of m-ZrO2 [22,42,48], which IM-WZ(0.193) also exists mainly in the form of t-ZrO2. In Fig. 8b,
conforms to the ZrO2 phase detected by XRD characterization. The the presence of Raman bands at 987 cm1 (associated with mono-
Raman bands marked by solid lines at 274, 715, and 807 cm1 in oxo W@O species in polytungstate structure) and 829 cm1 (asso-
Fig. 6a are the characteristic bands of the crystalline m-WO3 phase ciated with WAOAW bonds) indicates that WOx in IM-WZ(0.193)
[22,23,25]. Clearly, no crystalline m-WO3 in Fig. 6 can be observed forms polytungstate species.
124 K. Song et al. / Journal of Catalysis 299 (2013) 119–128

Fig. 6. Raman spectra of various TPIH-WZ nanosized catalysts. (b) is a partial enlargement of (a). The solid dots in (b) mark the positions of Raman bands attributed to
monooxo W@O stretch.

Fig. 9 depicts the Ei values of the catalysts as a function of WOx monotungstate species (Fig. 6). With the increasing WOx loading,
loading. Obviously, there is a sharp increase in Ei in the range of the rise of Ei values (Fig. 9) is accompanied with the transformation
low WOx loading for TPIH-WZ catalysts. The phenomenon of low of surface monotungstate to polytungstate species (Fig. 6). Thus, it
Ei value at low WOx loading implies that the incorporation of a can be deduced that the acid sites detected by n-butylamine poten-
small amount of WOx with ZrO2 could only bring about relatively tiometric titration should be related to the surface tungstate spe-
weak acid sites. When the WOx loading is higher than 0.235 g/g cies, and the larger polytungstate species could present higher
ZrO2, the increasing trend of Ei turns gentle gradually. With the acid strength. Moreover, for the sample with WOx dispersion
continuous increase in WOx loading to around the dispersion threshold by impregnation method, IM-WZ(0.193), although its
threshold, the corresponding Ei value climbs to 540–550 mV, a po- W content is lower than the dispersion threshold of TPIH-WZ, it
tential equal to or even higher than that of the reported sulfated presents similar Ei values to the latter. This may imply that the sur-
zirconia system [51–53]. face WOx species at the dispersion threshold from either the TPIH
The trend of Ei with WOx loading can be elucidated by compar- or the impregnation method have some similar features, for exam-
ison with the results of Raman spectroscopy. It is very interesting ple, domain size.
that the increasing trend of Ei values with WOx loading is consis- The cumulative acid site numbers at different acid strengths are
tent with that of the Raman band shift of monooxo W@O species, obtained from those potentiometric titration curves (Fig. 7) and de-
which reflects the growth of surface tungstate domains. The cata- picted as a function of WOx loading (Fig. 10). For the TPIH-WZ cat-
lyst with low WOx loading possesses a low Ei value (Fig. 9), while alysts with WOx loading below ca. 0.235 g/g ZrO2, the numbers of
most of the surface tungstate species exist in the form of isolated acid sites with different strengths all exhibit a sharp growing trend
 K. Song et al. / Journal of Catalysis 299 (2013) 119–128 125

Fig. 7. Potentiometric titration curves with n-butylamine of various WOx/ZrO2 Fig. 10. Acid site distribution of WOx/ZrO2 catalysts as a function of WOx loading.
catalysts. The insert shows the enlargement of the potentiometric titration curves The solid and hollow symbols represent catalysts prepared by TPIH and impreg-
higher than 0 mV. nation, respectively.

as WOx loading increases. Among them, the samples with very low
WOx loading, such as 0.153 and 0.192 g/g ZrO2, only possess acid
sites with potential below 400 mV. With the increase in WOx load-
ing, the acid sites with higher potential (e.g., above 400 mV) ap-
pear, and the number of them grows continuously but gradually
presents a gentle trend. After the WOx loading exceeds the disper-
sion threshold, the increase in WOx does not change the acid site
distribution any more. The superfluous tungstate in these samples
exists as a separated WO3 phase that makes no contribution to the
number of acid sites. With the same WOx loading, IM-WZ(0.193)
shows fewer weak acid sites than TPIH-WZ(0.192) but more strong
ones (and further increase in the WOx loading by impregnation
cannot obviously change the acid site number [cf. IM-
WZ(0.268)]). This can be explained by the state of WOx on ZrO2. Ra-
man spectra show that WOx in IM-WZ(0.193) (i.e., at the dispersion
threshold of impregnation) mostly exists in the form of polytung-
state species with strong acidity (Fig. 8), while W species on the
sample prepared by the TPIH method [TPIH-WZ(0.192)] still exist
mainly as monotungstate due to its high dispersibility on nano-
sized ZrO2 support (Fig. 6). However, when the WOx loading in
TPIH-WZ nanosized catalysts reaches its dispersion threshold at a
WOx loading of 0.372 g/g ZrO2, polytungstate species become dom-
Fig. 8. (a) XRD and (b) Raman spectra of IM-WZ(0.193).
inant, and WOx loading is about twice that at the dispersion
threshold of the impregnation method. Therefore, the sample pre-
pared by the TPIH method has many more acid sites for all acid
strengths than IM-WZ at their dispersion thresholds.
To further explore the nature of acid sites, pyridine-adsorbed
FTIR was performed for the dispersion threshold samples of
TPIH-WZ and IM-WZ. The calculation results for different acid site
concentrations are listed in Table 2. Both of them display Brønsted
acid sites as well as Lewis acid sites at different desorption temper-
atures. The total acid site numbers (B + L acid sites) are roughly
consistent with the results of potentiometric titration; that is, the
dispersion threshold sample of TPIH has many more acid sites than

Table 2
Brønsted and Lewis acidity of WOx/ZrO2 catalysts determined by pyridine-adsorbed
FTIR.

Catalyst B/L acid site concentrations (mmol/g catalyst)

473 Ka 573 K 673 K
TPIH-WZ(0.372) 0.098/0.127 0.071/0.062 0.043/0.033
IM-WZ(0.193) 0.059/0.058 0.045/0.035 0.027/0.024
Fig. 9. Initial electrode potential (Ei) of WOx/ZrO2 catalysts as a function of WOx
a
loading. Desorption temperature of pyridine after adsorption at 473 K.
126 K. Song et al. / Journal of Catalysis 299 (2013) 119–128

that obtained by the impregnation method. That is a distinctive Table 3

feature brought by the specific preparation method of TPIH, which Conversion of n-pentane under various reaction conditions catalyzed by TPIH-
WZ(0.360).
will be more obvious in the following section.
Reaction conditions Conversion (%)
3.7. Catalytic activity of the WOx/ZrO2 catalysts Temperature (K) WHSV (h1)
423 2.25 0.49
The catalytic conversion of n-pentane was employed to investi- 473 2.25 17.05
gate the catalytic performance of prepared WOx/ZrO2. As reported 523 2.25 24.35
[14,17,28,40], an induction period was involved at the beginning 523 6.75 3.24
623 2.25 11.15
of the reaction, followed by a maximum conversion. Then, the cata-
lysts would lose their activity as a result of the formation of coke (as
shown in Fig. 11). So, the maximum conversion is employed to eval-
uate the intrinsic catalytic activity of WOx/ZrO2 catalysts [28].
Various reaction temperatures and space velocities are tested
for the conversion of n-pentane (Table 3) catalyzed by TPIH-
WZ(0.360). For different temperatures, very low conversion of n-
pentane is observed at a low reaction temperature, for example,
0.49% at 423 K and 17.05% at 473 K. When the reaction tempera-
ture is higher than 523 K (e.g. 623 K), the observed catalytic activ-
ity is also lower than that at 523 K. The reason for lower activity at
higher temperature should be that catalyst is deactivated so
quickly and seriously at high reaction temperature that it is diffi-
cult to obtain catalytic activity close to the intrinsic properties of
the catalyst. A higher WHSV would also lower the observed cata-
lytic activity. Therefore, the reaction conditions of 523 K and a
WHSV of 2.25 h1 are chosen to conduct a comparison among cat-
alysts with different WOx loadings and preparation methods.
The conversion of n-pentane on WOx/ZrO2 catalysts prepared by
TPIH and impregnation is depicted in Fig. 12. The turnover rates
Fig. 12. Conversion of n-pentane catalyzed by various WOx/ZrO2 catalysts at 523 K
(TORw) in molecules of n-pentane reacted per W atom per second
with WHSV of 2.25 h1 and TORw for different reactions.
are also calculated for different catalysts (Fig. 12), and the product
distributions are displayed in Table 4. The catalytic data can be
analyzed in the following points: Table 4
Product distributions of n-pentane isomerizationa.
(i) A simple comparison between the catalytic activity (Fig. 12) Catalyst Temperature Product distribution (wt%)
and acid site distribution (Fig. 10) of TPIH-WZ catalysts (K)
could lead to the conclusion that the acid sites with strength 6C2 C3 C4 i-C5 n-C5 PC6
lower than 400 mV should display very little catalytic activ- TPIH-WZ(0.155) 523 0.01 0.02 0.15 1.01 98.55 0.07
ity. In other words, the possible acid sites responsible for the TPIH-WZ(0.192) 523 0.00 0.02 0.49 1.63 97.5 0.11
catalytic reaction of n-pentane should be those above TPIH-WZ(0.235) 523 0.01 0.27 2.06 6.31 90.2 0.79
TPIH-WZ(0.360) 423 0.00 0.01 0.05 1.00 98.7 0.05
400 mV. The trend of acid site number above 400 mV of
TPIH-WZ(0.360) 473 0.00 0.57 6.23 10.6 82.2 0.14
the TPIH-WZ catalysts matches the catalytic activity well. TPIH-WZ(0.360) 523 0.04 1.83 9.00 12.7 75.0 1.20
A low loading of WOx results in fewer sites with effective TPIH-WZ(0.360) 623 0.33 2.24 3.70 5.21 88.14 0.02
catalytic activity (i.e., the acid sites with E > 400 mV) and TPIH-WZ(0.408) 523 0.02 0.74 5.22 16.5 75.1 2.18
thus in low catalytic activity. The increase in WOx loading IM-WZ(0.193) 523 0.01 0.35 2.44 9.85 86.0 1.14
IM-WZ(0.268) 523 0.00 0.43 2.70 7.08 88.7 0.87
is accompanied by stronger acidic catalytic sites as well as
a
WHSV = 2.25 h1.

high activity till around the dispersion threshold. After that,

the number of strong acid sites is almost constant, which
makes the activity of catalysts similar.
(ii) For the samples with the same WOx loading but prepared by
different preparation methods, TPIH-WZ(0.192) exhibits
much less activity than IM-WZ(0.193). Raman spectroscopy
(Figs. 6 and 8b) and potentiometric titration (Figs. 9 and 10)
have demonstrated that TPIH-WZ(0.192) is far from the dis-
persion threshold of TPIH, and WOx dominantly exists as
monotungstate species with relatively weak acid strength,
while IM-WZ(0.193) has reached the dispersion threshold
of impregnation and most of WOx forms polytungstate spe-
cies with strong acid strength. That finally results in better
catalytic activity for IM-WZ(0.193). Furthermore, IM-
WZ(0.193) shows catalytic activity comparable to that of
TPIH-WZ(0.235). This can be attributed to the similar num-
Fig. 11. Time-dependent conversion of n-pentane catalyzed by TPIH-WZ(0.360) at bers of strong acidic catalytic sites with polytungstate struc-
523 K with WHSV of 2.25 h1. ture on TPIH-WZ(0.235) and IM-WZ(0.193).
 K. Song et al. / Journal of Catalysis 299 (2013) 119–128 127

(iii) For both of the catalysts around the dispersion threshold of confined hydrolysis process, TPIH method brings out a smaller
IM-WZ and TPIH-WZ, the catalytic activity of the catalysts average crystallite size and thus a larger BET surface area. Mean-
prepared by impregnation is only about half of the activity while, the W surface density of the sample at WOx dispersion
of the catalysts prepared by the TPIH method (Fig. 12), threshold turns out to be as high as 6.2 W atoms/nm2. The en-
which naturally results from the lower number of strong hanced BET surface area and high W surface density of the WOx/
acid sites with polytungstate structure in the IM-WZ sample ZrO2 sample prepared by TPIH together ensure the outstandingly
caused by its low loading of WOx at the dispersion threshold. high dispersion threshold of WOx on ZrO2. Acidity characterization
Even for the impregnation catalyst with higher WOx loading, by potentiometric titration of n-butylamine demonstrates that the
for example, IM-WZ(0.268), the catalytic activity turns out catalysts obtained by TPIH are able to achieve larger number of
to be slightly decreasing for the formation of crystalline strong acid sites than the impregnation catalyst due to the higher
WO3. This further proves the advantage of the TPIH method dispersion threshold of the former. The conversion of n-pentane
for its ability to generate more strong acidic polytungstate indicates that the catalysts around the dispersion threshold by
species with higher catalytic activity. TPIH show catalytic activity nearly double that for the impregna-
(iv) TORw in Fig. 12 is calculated based on the total W atom num- tion one. When this is combined with the results of UV–vis diffuse
ber in the catalyst, which could reflect the average catalytic reflectance spectroscopy and Raman spectroscopy, surface poly-
activity of each W atom. As shown in Fig. 12, with increasing tungstate species are suggested to be the sites with strong acidity
WOx loading, the TORw of the TPIH-WZ sample also and thus catalytic activity for the conversion of n-pentane. The in-
increases gradually below the dispersion threshold, suggest- crease in WOx dispersion threshold resulting from TPIH method
ing that the higher the WOx loading of catalysts, the stronger could increase the number of polytungstate species efficiently
catalytic capability each W atom displays. This confirms that and thereby elevate the catalytic activity of WOx/ZrO2 catalysts.
the polytungstate species formed at high WOx loading are There are still challenges left to improve the performance of cata-
the catalytically active sites. The monotungstate species in lysts such as developing an improved in situ two-phase interface
the samples with low WOx loading show little catalytic hydrolysis of the active components on supports or adding the
activity, so their TORw is very low. But in the sample around third component during the hydrolysis process, as well as explor-
the WOx dispersion threshold, W mainly exists as polytung- ing optimal reaction conditions.
state species with notable catalytic activity, which make
each W atom display a very high TORw. For the sample with Acknowledgments
WOx loading above the dispersion threshold, its TORw begins
to decrease because many W atoms are buried in crystalline This work is supported by the 973 program (2013CB934101,
WO3 but still treated as catalytic sites when TORw is calcu- 2009CB623506, and 2009CB930400), STCSM (10QB1404700,
lated. For both of the catalysts around the dispersion thresh- 09JC1400400, and 09DZ2271500), NSFC (20890122, 20903024,
old, TPIH-WZ(0.360) and IM-WZ(0.193), similar TORw are 21073237, 21073041, and 21171041), and PSF (200902263). The
present. Although they are very different in WOx loading authors also thank Liyuan Lu of SINOPEC for performing the pyri-
and conversion of n-pentane, the similar TORw indicates that dine-adsorbed FTIR.
WOx species at the dispersion threshold from TPIH and
impregnation method share the same surface polytungstate References
structure as suggested by acidity characterization (Fig. 9).
(v) As to product distribution in Table 4, it can be found that [1] K. Tanabe, W.F. Holderich, Appl. Catal. A Gen. 181 (1999) 399.
[2] H. Hattori, Top. Catal. 53 (2010) 432.
both the isomerization and the cracking reaction of n-pen- [3] N. Rahimi, R. Karimzadeh, Appl. Catal. A Gen. 398 (2011) 1.
tane can occur on the WOx/ZrO2 catalysts with the former [4] V. Adeeva, H.Y. Liu, B.Q. Xu, W.M.H. Sachtler, Top. Catal. 6 (1998) 61.
as main reaction. With increasing of acidity and conversion, [5] Y.D. Xia, W.M. Hua, Y. Tang, Z. Gao, Chem. Commun. (1999) 1899.
[6] B.M. Reddy, M.K. Patil, Chem. Rev. 109 (2009) 2185.
the selectivity for isomerization tends to decrease. It can be [7] A. Micek-Ilnicka, J. Mol. Catal. A Chem. 308 (2009) 1.
improved by optimizing reaction conditions such as lower- [8] X.Y. Chen, G. Clet, K. Thomas, M. Houalla, J. Catal. 273 (2010) 236.
ing reaction temperature. In the time-dependent conversion [9] V.M. Benitez, N.S. Figoli, Catal. Commun. 3 (2002) 487.
[10] H. Zhang, J. Han, X.W. Niu, X. Han, G.D. Wei, W. Han, J. Mol. Catal. A Chem. 350
of n-pentane of Fig. 11, the deactivation of WOx/ZrO2 cata-
(2011) 35.
lysts during the n-pentane isomerization seems obvious. [11] J. Engweiler, J. Harf, A. Baiker, J. Catal. 159 (1996) 259.
This should be owing to both the very strong acidity of [12] A. Spamer, T.I. Dube, D.J. Moodley, C. van Schalkwyk, J.M. Botha, Appl. Catal. A
WOx/ZrO2 catalysts and the reaction conditions chosen for Gen. 255 (2003) 153.
[13] C. van Schalkwyk, A. Spamer, D.J. Moodley, T. Dube, J. Reynhardt, J.M. Botha,
declaration of their difference in the catalytic reaction. The Appl. Catal. A Gen. 255 (2003) 121.
best possible approaches to improving the performance of [14] N. Soultanidis, W. Zhou, A.C. Psarras, A.J. Gonzalez, E.F. Iliopoulou, C.J. Kiely, I.E.
catalysts might be in situ two-phase-interface cohydrolyzing Wachs, M.S. Wong, J. Am. Chem. Soc. 132 (2010) 13462.
[15] R.D. Wilson, D.G. Barton, C.D. Baertsch, E. Iglesia, J. Catal. 194 (2000) 175.
the alkoxides of W and Zr on supports or adding the third [16] I.E. Wachs, T. Kim, E.I. Ross, Catal. Today 116 (2006) 162.
component, as well as trying different reaction conditions. [17] S. Kuba, P. Lukinskas, R.K. Grasselli, B.C. Gates, H. Knozinger, J. Catal. 216
(2003) 353.
[18] D.G. Barton, S.L. Soled, E. Iglesia, Top. Catal. 6 (1998) 87.
Thus, it can be concluded that the polytungstate species with [19] M. Hino, K. Arata, J. Chem. Soc. Chem. Commun. (1988) 1259.
very strong acidity in WOx/ZrO2 catalysts should be the catalyti- [20] J.G. Santiesteban, J.C. Vartuli, S. Han, R.D. Bastian, C.D. Chang, J. Catal. 168
cally active sites, and the increase in the WOx dispersion threshold (1997) 431.
[21] J.G. Santiesteban, D.C. Calabro, W.S. Borghard, C.D. Chang, J.C. Vartuli, Y.P. Tsao,
by TPIH could efficiently increase the number of polytungstate spe- M.A. Natal-Santiago, R.D. Bastian, J. Catal. 183 (1999) 314.
cies and thereby notably elevate the optimal catalytic activity. [22] E.I. Ross-Medgaarden, W.V. Knowles, T. Kim, M.S. Wong, W. Zhou, C.J. Kiely, I.E.
Wachs, J. Catal. 256 (2008) 108.
[23] B.Y. Zhao, X.P. Xu, J.M. Gao, Q. Fu, Y.Q. Tang, J. Raman Spectrosc. 27 (1996) 549.
[24] D.G. Barton, M. Shtein, R.D. Wilson, S.L. Soled, E. Iglesia, J. Phys. Chem. B 103
4. Conclusions (1999) 630.
[25] M. Scheithauer, R.K. Grasselli, H. Knozinger, Langmuir 14 (1998) 3019.
A novel two-phase interface hydrolysis (TPIH) method is used [26] D.G. Barton, S.L. Soled, G.D. Meitzner, G.A. Fuentes, E. Iglesia, J. Catal. 181
(1999) 57.
to prepare WOx/ZrO2 catalysts with exciting properties compared [27] N. Vaidyanathan, M. Houalla, D.M. Hercules, Surf. Interface Anal. 26 (1998)
with conventional impregnation method. Benefiting from the 415.
128 K. Song et al. / Journal of Catalysis 299 (2013) 119–128

[28] M. Scheithauer, T.K. Cheung, R.E. Jentoft, R.K. Grasselli, B.C. Gates, H. [40] S. De Rossi, G. Ferraris, M. Valigi, D. Gazzoli, Appl. Catal. A Gen. 231 (2002) 173.
Knozinger, J. Catal. 180 (1998) 1. [41] J.R. Sohn, M.Y. Park, Langmuir 14 (1998) 6140.
[29] A. Martinez, G. Prieto, M.A. Arribas, P. Concepcion, J.F. Sanchez-Royo, J. Catal. [42] L.D. Huy, P. Laffez, P. Daniel, A. Jouanneaux, N.T. Khoi, D. Simeone, Mater. Sci.
248 (2007) 288. Eng. B Solid 104 (2003) 163.
[30] T. Kim, A. Burrows, C.J. Kiely, I.E. Wachs, J. Catal. 246 (2007) 370. [43] I.E. Wachs, Catal. Today 27 (1996) 437.
[31] W. Zhou, E.I. Ross-Medgaarden, W.V. Knowles, M.S. Wong, I.E. Wachs, C.J. [44] G. Ferraris, S. De Rossi, D. Gazzoli, I. Pettiti, M. Valigi, G. Magnacca, C. Morterra,
Kiely, Nat. Chem. 1 (2009) 722. Appl. Catal. A Gen. 240 (2003) 119.
[32] R.A. Boyse, E.I. Ko, J. Catal. 171 (1997) 191. [45] A. Gutierrez-Alejandre, J. Ramirez, G. Busca, Catal. Lett. 56 (1998) 29.
[33] M.G. Falco, S.A. Canavese, N.S. Figoli, Catal. Today 107–108 (2005) 778. [46] E.I. Ross-Medgaarden, I.E. Wachs, J. Phys. Chem. C 111 (2007) 15089.
[34] M.A. Cortes-Jacome, J.A. Toledo, C. Angeles-Chavez, M. Aguilar, J.A. Wang, J. [47] T. Onfroy, G. Clet, M. Houalla, J. Phys. Chem. B 109 (2005) 3345.
Phys. Chem. B 109 (2005) 22730. [48] R.D. Purohit, S. Saha, A.K. Tyagi, Mater. Sci. Eng. B Solid 130 (2006) 57.
[35] K.J. Tang, J.N. Zhang, W.F. Yan, Z.H. Li, Y.D. Wang, W.M. Yang, Z.K. Xie, T.L. Sun, [49] R. Cid, G. Pecchi, Appl. Catal. 14 (1985) 15.
H. Fuchs, J. Am. Chem. Soc. 130 (2008) 2676. [50] L.R. Pizzio, M.N. Blanco, Micropor. Mesopor. Mater. 103 (2007) 40.
[36] N. Naito, N. Katada, M. Niwa, J. Phys. Chem. B 103 (1999) 7206. [51] A.I. Ahmed, S.A. El-Hakam, S.E. Samra, A.A. El-Khouly, A.S. Khder, Colloid Surf.
[37] I. Hyppanen, J. Holsa, J. Kankare, M. Lastusaari, M. Malkamaki, L. Pihlgren, J. A 317 (2008) 62.
Lumin. 129 (2009) 1739. [52] D.W. Zhai, Y.H. Yue, W.M. Hua, Z. Gao, Acta Phys. – Chim. Sin. 26 (2010) 1867.
[38] C.A. Emeis, J. Catal. 141 (1993) 347. [53] E.A. El-Sharkawy, S.S. Al-Shihry, Monatsh. Chem. 141 (2010) 259.
[39] S.T. Oyama, X.M. Zhang, J.Q. Lu, Y.F. Gu, T. Fujitani, J. Catal. 257 (2008) 1.