Spectroscopic Characterization of Ni-Mo/γ-Al2O3-B2O3 Catalysts for Hydrodesulfurization of Dibenzothiophene

Article abstract of DOI:10.1006/jcat.1997.1730

The effects of boron on the dispersion of active metal species and hydrodesulfurization (HDS) activity of dibenzothiophene (DBT) over Ni-Mo/γ-Al₂O₃ containing boron were studied by means of BET surface area measurement, powder X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), Mo K-edge extended X-ray absorption fine structure (EXAFS), and B K-edge X-ray absorption near-edge structure. The HDS activity of DBT over Ni-Mo/ γ-Al₂O₃-B₂O₃ reaches a maximum at the B₂O₃ loading of about 1 mol%, in agreement with the basal dispersion and thus with the hydrogenation ability of the catalysts. It has been proven that the combination of XRD, XPS, and EXAFS techniques gives a complete description of the dispersion of active metal species in heterogeneous environmental catalysts.

Full text of DOI:10.1006/jcat.1997.1730

JOURNAL OF CATALYSIS 170, 357–365 (1997)
ARTICLE NO. CA971730
Spectroscopic Characterization of Ni–Mo/� -Al O –B O Catalysts
2
3
2
3
for Hydrodesulfurization of Dibenzothiophene
Dien Li, T. Sato, M. Imamura, H. Shimada, and A. Nishijima
National Institute of Materials and Chemical Research, Tsukuba, Ibaraki 305, Japan
Received November 11, 1996; revised March 10, 1997; accepted April 8, 1997
that the increase in the boria content resulted in an increas-
The effects of boron on the dispersion of active metal species and ing caprolactam selectivity. The control of the boria level
hydrodesulfurization (HDS) activity of dibenzothiophene (DBT) on the alumina surfaces would result in the optimum num-
over Ni–Mo/� -Al2O3 containing boron were studied by means of
BET surface area measurement, powder X-ray diffraction (XRD),
X-ray photoelectron spectroscopy (XPS), Mo K-edge extended
X-ray absorption fine structure (EXAFS), and B K-edge X-ray ab-
ber and strength of the acid sites and thus in an improved
caprolactam selectivity (25).
The effect of adding boron to these conventional
Co(Ni)–Mo(W)/� -Al2O3 catalysts on their catalytic activity
sorption near-edge structure. The HDS activity of DBT over Ni–Mo/
has been studied in the past. For example, the hydrogenol-
�
-Al2O3–B2O3 reaches a maximum at the B2O3 loading of about
ysis activity of Ni–Mo/� -Al2O3 catalysts was shown to in-
crease with the increasingboron content in the support (26).
Stranick etal. believed that the addition ofboron to � -Al2O3
1
mol%, in agreement with the basal dispersion and thus with the
hydrogenation ability of the catalysts. It has been proven that the
combination of XRD, XPS, and EXAFS techniques gives a com-
plete description of the dispersion of active metal species in hetero- can improve the dispersion of Co and change the chem-
geneous environmental catalysts.
�c 1997 Academic Press
ical states of Co in Co/� -Al2O3 catalysts (27). However,
Morishige and Akai indicated that boron addition de-
creases the dispersion of Mo in Mo/� -Al2O3 catalysts and
weakens the interaction between Mo species and the � -
Al2O3 surface (28). Chen et al. showed that the hydrodesul-
INTRODUCTION
Mo and W sulfide catalysts, promoted by Co or Ni and furization (HDS) activities of atmospheric gas oil (AGO)
supported on � -Al2O3, have been applied in industry for and residue oil over Co–Mo/alumina–aluminium borate
the hydrotreating processes of petroleum and coal liquefac- (prepared by coprecipitation) catalysts improve signifi-
tions and extensively investigated. The chemistry and struc- cantly compared to conventional Co–Mo/� -Al2O3 cata-
ture of these catalysts, particularly the Co–Mo/� -Al2O3 lysts, and the optimum HDS activity of AGO was found at
catalyst, have been reviewed (1–11). Several structural the Al/B ratio of about 3.5 (29, 30). Nevertheless, Ram ı´ rez
models for the active phases have been proposed; these in- et al. recently prepared Co–Mo/� -Al2O3–B(x) by wet im-
clude the monolayer model (12), intercalation model (13), pregnation and studied the effect of boron addition on the
contact synergy model (14, 15), and Co–Mo–S complex (16, activity and selectivity of the HDS reaction of thiophene
1
7). 57Co M o¨ ssbauer spectroscopy (7), more recently Co, over these catalysts. They found that the conversion rate
Ni, and Mo K-edge EXAFS results (e.g., 18–21), and ra- reached the maximum point at the loading of 0.8 wt% B
dioisotope 35S tracer studies (22) consistently indicated that (31).
the so-called “Co–Mo–S complex phases,” in which Co is
located at the edges of MoS2 are most likely to be the active been made by several different approaches (wet impregna-
sites for the Co–Mo/� -Al2O3 catalyst.
tion, coprecipitation, hydrolysis of alkoxides, and knead-
Co(Ni)–Mo(W)/� -Al2O3 catalysts containing B have
Aluminium borates are acid catalytic materials and were ing). Previous studies on the effect of boron on the
shown to have high catalytic activity for 2-propanol dehy- dispersion and catalytic activities of the catalysts are con-
dration. The presence and coordination of boron have a tradictory. The description of the dispersion of active metal
dramatic effect on the acidity of these materials, the high phases in heterogeneous MoS2 catalysts was incomplete,
acidity being closely related to the relative concentration because MoS2 has a layered and anisotropic structure. In
of BO4 (23). For alumina-supported boria, Sato et al. indi- this work, Ni–Mo/� -Al2O3–B2O3 catalysts were prepared
cated that these materials exhibited high catalytic efficiency by wet impregnation, and the HDS reaction of dibenzo-
for the vapor phase Beckmann rearrangement of cyclohex- thiophene (DBT) over these catalysts was carried out. DBT
anone oxime (24). Curtin et al. furthermore pointed out was chosen as model compound to test the catalytic activity
357
0021-9517/97 $25.00
Copyright �c 1997 by Academic Press
All rights of reproduction in any form reserved.

358
LI ET AL.
of the catalysts, because DBT is a main component con- Catalyst Characterization
taining sulfur in actual petroleum residues and coal lique-
BET surface areas of representative Ni–Mo/� -Al2O3–
factions, and the HDS of DBT includes hydrogenation and
hydrocracking reactions. The structure of the catalysts is
characterized by X-ray diffraction (XRD), X-ray photo-
electron spectroscopy (XPS), Mo K-edge extended X-ray
absorption fine structure (EXAFS), and B K-edge X-ray
absorption near-edge structure (XANES). The main pur-
pose of this work is to give a complete description of the
dispersion (both basal plane and edge dispersion) of ac-
tive metal species in the heterogeneous Ni–Mo sulfide cata-
lysts and to discuss the effects of adding boron to Ni–Mo/
B2O3 catalyst samples were measured using a Micromerit-
ics chemisorption controller (Shimadzu). XRD patterns of
�
-Al2O3–B2O3 supports and NMABx series catalysts were
taken using a Phillip MXP18 powder diffractometer (Ma-
terial Analysis and Characterization) at 40 kV and 50 mA
–
–
and with a scan rate of 3 /min and a sampling width of 0.02 .
The Cu Kfi (‚ D 1.5405 A˚ ) was used as the radiation source.
XPS spectra of the calcined and sulfided NMABx cata-
¡
9
lysts were collected at the chamber pressure of » 10 Torr
1 Torr D 1.33 £ 10¡ 4 MPa) using a Phi-5500 ESCA spectro-
meter (Perkin–Elmer). The monochromatized Al Kfi (h” D
486.7 eV) was used as the source of excitation. The sample
(
�
-Al2O3 catalysts on the dispersion of active metals and on
the HDS activity of DBT by using spectroscopic techniques.
1
preparations of freshly sulfided catalysts for XPS measure-
ments were performed in a N2 glove box; SO4 species in the
S 2p line of each sulfided sample were not observed and,
thus, the feasibility of this approach could be proven. The
EXPERIMENTAL
Catalyst Preparation
H3BO3 was dissolved in aqueous ammonia solution binding energy (BE) of Al 2p, B 1s, Mo 3d, Ni 2p, and S 2p
(
pH 9.7 § 0.2), and � -Al2O3 (VGS-1, BET surface area lines were calibrated using C 1s BE of 284.6 eV; these XPS
2
2
09 m /g, and pore volume 0.81 ml/g) was impregnated with lines were decomposed by fitting. Several constraints were
the H3BO3 solution to prepare the supports with different included in the decomposition of the Mo 3d envelope to
B2O3 loadings, depending on the initial concentration of give physical significance to the peaks obtained: (i) the spin-
B2O3 in the solution. The � -Al2O3–B2O3 slurries were orbit splitting of the Mo 3d peak is 3.15 eV; (ii) the Mo 3d5/2/
–
dried for 4 h in a rotary kiln and then calcined at 550 C Mo 3d3/2 area ratio was kept constant at the theoretical
for 3 h in air. The � -Al2O3–B2O3 materials were used as value of 1.5; (iii) the full width at half maximum of the
the supports and coimpregnated with a Ni(NO3)2 ¢ 6H2O Mo 3d5/2 and Mo 3d3/2 peaks of the Mo5 and Mo oxi-
and (NH4)6Mo7O24 ¢ 4H2O containing aqueous ammonia dation states was equal to the corresponding width of the
solution (pH 9.0 § 0.2). The loadings of NiO and MoO3 peaks of the Mo4 species (32). The atomic concentra-
were kept constant at 3 and 12 wt% , respectively, so that tion ratio on the surface of the catalysts was calculated
the molar NiO/MoO3 ratio was about 0.48 for all the using nA/nB D IASB/IBSA, where ni is the atomic number
catalysts. The Ni–Mo/� -Al2O3–B2O3 catalysts, designated of species i (i D A or B), Ii is the integrated intensity of
as NMABx (x D B2O3 mol% in supports), were dried at species i, and Si is the sensitivity factor given by XPS spec-
C
6C
C
–
–
2
00 C for 2 h in a rotary kiln and then calcined at 550 C tral measurement. The sensitivity factor depends mainly
–
for 3 h in air. Thereafter, the catalysts were dried at 400 C on the photoionization cross-section � i, but also on the ef-
for 2 h under N2 atmosphere and sulfided at the same fects of exciting X-ray energy, takeoff angle of measure-
temperature for another 2 h in a stream of H2–H2S mixture ment, detector efficiency, and kinetic energy; the formula
(
the volume percentage of H2S in the mixture is 5.04% ) used to calculate the atomic concentration is similar to the
before the catalytic activity tests.
Kerkhof–Moulijn model (33).
Mo K-edge EXAFS data were recorded at room temper-
ature using a channel-cut Si(311) double crystal monochro-
mator at the Photon Factory (BL10B) of the National Lab-
oratory for High Energy Physics. Each sample was ground
to a fine powder and pressed into a disk the thickness (x) of
which was such that the absorption „ x was 1, thus allowing
an EXAFS spectrum of optimal quality for collection by
transmission measurement. The disk was then sealed into
a plastic bag, so that the sample was kept in an inert at-
mosphere during the EXAFS measurement. Fourier trans-
Catalytic Activity
DBT was dissolved in decalin to give a solution with a
sulfur content of 1 wt% . The catalytic reactions of DBT
(
10 ml) over the NMABx catalysts (0.1 g) were performed
–
at 360 C and on H2 pressure of 6.92 MPa for 1 h using a
batch microreactor. The conversion rate was calculated on
the basis of the integrated areas for the reaction products,
diphenyl (DP) and cyclohexylbenzene (CHB), and on the
remaining DBT in the gas chromatography analysis:
3
formation of the k -weighted EXAFS data for 1 k D 9.0
(
3.9 • k • 13.55) was performed to obtain a radial distribu-
The total conversion rate D .AreaDP C AreaCHB/=
tion function around Mo. Reducing procedures of EXAFS
.
AreaDP C AreaCHB C AreaDBT/: data have been described elsewhere (34).

HDS ACTIVITY OF DBT OVER Ni–Mo/� -Al2O3–B2O3 CATALYSTS
359
B K-edge XANES spectra of catalysts NMAB1 and
NMAB2 were recorded by total electron yield (TEY) using
a Grasshopper beamline at the Aladdin storage ring, Uni-
versity of Wisconsin. This beamline uses a grazing incident
system with a grating of 1800 Grooves/mm and an energy
resolution of about 0.1 eV at 100 eV (35). The B K-edge
spectrum of each of these samples is the sum of three mea-
surements, the linear background being removed. The B K-
edge XANES spectra of samples NMAB5 and NMAB8
were collected by TEY using the BL13C beamline in the
Photon Factory (36). For all the B K-edge measurements,
the catalyst samples were carefully pressed, not ground,
into powder under static pressure, in order to minimize the
structural state change of tetrahedrally coordinated boron,
caused by grinding, on the surface of samples (see our un-
published data). The B K-edge XANES spectra were cali-
brated using the B K-edge peak of B2O3 at 194.0 eV (37).
RESULTS
Catalytic HDS Activity of DBT
Figure 1A shows the HDS conversion rate of DBT over
NMABx series catalysts (x from 0 to 17) against the B2O3/
Al2O3 ratio in the supports. The total HDS conversion
FIG. 2. The XRD patterns of some representative calcined Ni Mo/
–
-Al2O3–B2O3 catalysts.
�
(
filled squares and solid line) or DBT increases with the in-
creasing loadings of B2O3 and reaches the maximum point
at about 1 mol% of B2O3 in the support and then decreases
with a further increase in the content of B2O3 (up to 17
mol% ). The conversion of DBT to diphenyl (filled trian-
gles and dotted line) generally increases with increasing
loadings of B2O3. The general profile of the conversion
rate to cyclohexylbenzene (filled circles and dashed line)
with a loading of B2O3 is parallel to the total conversion of
DBT, indicating that the total HDS activity of DBT over
the NMABx catalysts depends mostly on the conversion
rate to cyclohexylbenzene.
The selectivity of diphenyl and cyclohexylbenzene was
calculated using the respective conversion rates over the to-
tal HDS activity and are plotted as a function of B2O3/Al2O3
in Fig. 1B. The selectivity of diphenyl (empty triangles
and solid line) increases and that of cyclohexylbenzene
(
empty circles and dashed line) decreases to about 10 mol%
with increasing loadings of B2O3; thereafter, the selectivity
for both diphenyl and cyclohexylbenzene remains nearly
constant.
XRD Patterns and Dispersion of Ni and Mo
Species in Bulk Catalysts
FIG. 1. The conversion rate (A) and selectivity (B) of DBT over
Ni–Mo/� -Al2O3–B2O3 catalysts as a function of the molar B2O3/Al2O3
ratio. (A) Filled squares and solid line, the total conversion rate; filled tri-
anglesand dotted line, conversion rate to diphenyl;filled circlesand dashed
line, conversion rate to cyclohexylbenzene. (B) Open triangles and solid
line, selectivity to diphenyl; open circles and dashed line, selectivity to
cyclohexylbenzene.
Figure 2 shows representative XRD patterns of the
NMABx catalysts. For the � -Al2O3–B2O3 supports, the
XRD patterns are similar to those of � -Al2O3 until the load-
ing of B2O3 reaches about 14 mol% at which point a

360
LI ET AL.
crystalline B2O3 phase first occurs; its diffraction peaks at
–
–
2
� D 13.58 and 27.98 become sharper and more promi-
nent with increasing loadings of B2O3. The XRD patterns
of NMAB0 and NMAB1 are very similar to that of the
original � -Al2O3, but when the loading of B2O3 reaches
–
about 5 mol% , a broad peak at about 2� D 27 first ap-
pears and becomes stronger with further loadings of B2O3.
This broad peak may be attributed to the MoO3 phase.
In addition, BET surface area measurements of represen-
tative NMABx catalysts indicated that the surface areas
of NMAB1, NMAB8, and NMAB17 are 183, 165, and
2
1
53 m /g, respectively, decreasing with the increasing load-
ings of B2O3.
XPS and Concentrations of Ni and Mo Species
on the Catalyst Surface
XPS spectra of the calcined and sulfided catalysts were
collected. The Ni 2p XPS spectra of a representative
NMAB5 catalyst are shown in Fig. 3. The spectra are de-
composed by fitting, in which the dotted line represents
the experimental data, the dashed lines are the Gaussian
components and the solid line the envelope of decomposed
FIG. 4. Mo 3d XPS spectra of calcined and sulfided Ni–Mo/� -Al2O3–
B2O3 catalyst (NMAB5) and their decomposition. Filled dotted lines,
experimental data; dashed lines, Gaussian components; solid lines, the
features. The two main peaks in the XPS spectrum of the corresponding envelopes.
calcined sample are assigned to the spin-splitting Ni 2p3/2
and Ni 2p1/2 and the two broad peaks to the envelopes of molybdenia, as we expected. The Ni 2p XPS spectrum of the
their corresponding satellite lines. The Ni 2p XPS spectrum sulfided catalyst is more complex, decomposing into two set
suggests that the nickel species in the calcined catalyst can of Gaussian components, one for the Ni oxide species, as
be assigned to Ni2 in interaction with alumina and/or with seen in the calcined catalyst, the other for the Ni sulfide
C
species. The Ni sulfide species are assumed to be Ni3S2 at
the sulfiding conditions and the so-called Ni–Mo–S phase
is assumed to be the active phase for the catalysis.
Mo 3d XPS spectra of the same catalyst are shown in
Fig. 4. For the calcined catalyst, it is well known that
two peaks are assigned to the spin-splitting Mo 3d5/2 and
Mo 3d3/2 lines of the Mo oxide species. The Mo 3d XPS spec-
trum of the sulfided catalyst decomposed into three sets of
doublets, corresponding to Mo6 , Mo , and Mo species
in the order of decreasing binding energy, and a broad peak
at about 226.3 eV assigned to S 2s line. The Mo6 species
are obviously Mo oxide species that were not completely
C
5C
4C
C
5C
sulfided. The Mo species may be a Mo oxy-sulfide species,
and the Mo4 species are MoS2 and the Ni–Mo–S phase.
C
As shown above, the Al 2p, Mo 3d, Ni 2p, and S 2p (for
sulfided samples only) of calcined and sulfided catalysts
were decomposed by fitting. The integrated areas for Al 2p,
Mo 3d5/2, and Ni 2p3/2 were used to calculate the atomic
concentration ratios of Mo and Ni over Al. The Mo 3d5/2/
Al 2p and Ni 2p3/2/Al 2p ratios at the surface of the calcined
and sulfided catalysts are also plotted against the molar
B2O3/Al2O3 ratio (Figs. 5 and 6). However, for the sulfided
catalysts, the integrated area ofsulfide speciesonlywasused
for the calculation. For both calcined and sulfided catalysts,
both Mo 3d5/2/Al 2p and Ni 2p3/2/Al 2p ratios reached the
maximum points at a B2O3 loading of about 1 mol% , in
FIG. 3. Ni 2p XPS spectra of calcined and sulfided Ni–Mo/� -Al2O3–
B2O3 catalyst (NMAB5) and their decompositions. Filled dotted lines,
experimental data; dashed lines, Gaussian components; solid lines, the
corresponding envelopes. In the spectrum of the sulfided catalyst, S stands
for sulfide and O for oxide species.

HDS ACTIVITY OF DBT OVER Ni–Mo/� -Al2O3–B2O3 CATALYSTS
361
FIG. 5. Ni/Al atomic concentration ratio on the surface of calcined
filled circles and solid line) and sulfided (open squares and dashed line)
(
catalysts as a function of the molar B2O3/Al2O3 ratio in the bulk catalysts.
The Ni/Al atomic ratio on the surface was calculated according to the
integrated area of the Ni 2p3/2 and Al 2p XPS lines.
good agreement with the total HDS activity of DBT and
the conversion of DBT to cyclohexylbenzene over these
catalysts (see Fig. 1). The S 2p3/2 BE is about 162.2 eV, close
to that of MoS2. The S/(0.67Ni C 2Mo) ratio varies from
FIG. 7. Fourier transforms (3.9 • k • 13.55) of the Mo K-edge EX-
AFS for sulfided Ni–Mo/� -Al2O3–B2O3 catalysts.
0
.98 (NMAB0), 1.04 (NMAB1), to 0.82 (NMAB5), indi-
cating that the sulfidation degree of the catalysts decreases
with the increasing loading of B2O3.
about 2.42 A˚ (after phase shift correction) and Mo–Mo scat-
tering at the bond distance of about 3.16 A˚ (after phase shift
correction). These results are very similar to those of X-ray
diffraction of MoS2 and those of Mo K-edge EXAFS anal-
–
yses of MoS2 crystal and Mo/Al2O3 (21, 39) and Ni Mo/C
catalysts (19). The Mo–Ni scattering may contribute to the
peak at about 2.0 A˚ . However, based on our previous EX-
AFS analyses (39), this contribution does not seem to be
very significant. The peak at about 2.8 A˚ tends to sharpen
Mo K-Edge EXAFS
Figure 7 shows the radial distribution functions de-
3
rived from the Fourier transforms of the k -weighted EX-
AFS oscillations at the Mo K-edge of sulfided Ni–Mo/
�
-Al2O3–B2O3 catalysts. The two dominant peaks, visible at
about 2.0 and 2.8 A˚ , are shown for all the samples, respec-
tively, corresponding to Mo–S shell at the bond distance of
with the addition of small amounts of boron (from NMAB0
to NMAB2) and broadens again with further loadings of
boron (from NMAB2 to NMAB17), indicating a slight
change in the disorder–order states. The Fourier transform
magnitude (FTM) of the two main peaks are correlated
with the coordination numbers and Debye–Waller factors
of the corresponding to Mo–S and Mo–Mo scattering and
may provide important information on the edge dispersion
of active Mo species in the catalysts (39).
The FTM (in arbitrary units) of the two notable peaks
corresponding to Mo–S (filled squares and solid line)
and Mo–Mo (filled circles and dashed line) shells in the
EXAFS analyses of the catalysts are plotted as a function
of B2O3/Al2O3 in Fig. 8A. The FTM of both Mo–S and
Mo–Mo shells dramatically increases on adding a small
amount of boron (0.5 mol% ) to � -Al2O3, and then grad-
ually decreases with increasing loadings of boron. The
FTM ratio for Mo–Mo and Mo–S shells (FTM(Mo–Mo)/
FTM(Mo–S)) is shown as a function of B2O3/Al2O3 in
FIG. 6. Mo/Al atomic concentration ratio on the surface of calcined
filled circles and solid line) and sulfided (open squares and dashed line)
catalysts as a function of the molar B2O3/Al2O3 ratio in the bulk catalysts.
The Mo/Al atomic ratio on the surface was calculated according to the
integrated area of the Mo 3d5/2 and Al 2p XPS lines.
(
Fig. 8B. The solid line best fits the experimental data (filled

362
LI ET AL.
rhombuses). The FTM(Mo–Mo)/FTM(Mo–S) ratio also in-
creases markedly when a small amount of boron was added
to � -Al2O3, and gradually decreases with further loadings
of boron, generally parallel to the XPS and HDS activity
results.
B 1s XPS and B K-Edge XANES
Figure 9 compares theoretical (solid line) and experi-
mental (filled circles and dash line) XPS peak intensity ra-
tios (IB 1s/IAl 2p) as functions of B2O3/Al2O3. The theoretical
values were calculated according to the Kerkhof–Moulijn
model, assuming that boron species form a monolayer on
alumina and that the difference in the kinetic energies of
B 1s and Al 2p electrons is not very large (33):
�
�
� B 1s fl 1 C e¡ fl
b � Al 2p 2 1 ¡ e¡ fl
IB 1s
p
s
D
;
IAl 2p
FIG. 9. Comparison of the theoretical (both thin and thick solid
line) and experimental (filled circles and dash line) XPS peak intensi-
ties (IB 1s/IAl 2p) as a function of the B2O3/Al2O3 ratio. The thin and thick
solid lines represent the theoretical equations when the surface area (S)
where IB 1s and IAl 2p are relative XPS intensities of the B 1s
and Al 2p electrons, (p/s)b is the bulk atomic ratio, � B 1s
0.486) and � Al 2p (0.356) are the photoionization cross sec-
(
2
2
of the support was used as 209 m /g (pure alumina) and 153 m /g (sample
tions at 1487 eV (40), and fl D 2/(‰ ¢ S ¢ ‚ ) in which ‰ is the NMAB17), respectively.
6
3
density of the alumina (3.7 £ 10 g/m ), S is the surface
area of the support, and ‚ is the escape depth of elec-
trons (2.68 nm) for the Al 2p electrons at a kinetic en-
Fig. 9); if the lower surface area of the borium-containing
samples is corrected, for example, the surface area of sam-
ple NMAB17 (153 m /g) is used in the calculation, the
2
ergy of 1390 eV (41). If the surface area of the alumina
2
(
209 m /g) is used in the calculation, the theoretical equa-
theoretical equation is IB 1s/IAl 2p D 1.50 ¢ (B2O3/Al2O3) (see
thick solid line if Fig. 9). The experimental values from XPS
measurements were calculated as shown above. The results
indicate that boron species are well dispersed and form
monolayers on alumina at low loadings of B2O3. At a B2O3
content of 17 mol% , however, the dispersion of the boron
species decreases and deviates from the monolayer model,
in good agreement with the previous XPS measurements
tion is IB 1s/IAl 2p D 1.42 ¢ (B2O3/Al2O3) (see thin solid line in
(
28) and our XRD results mentioned above.
In order to study the structural states of boron in the
catalysts, we measured B K-edge XANES spectra of the
NMABx catalysts before sulfiding (Fig. 10) and compared
them with those of model compounds B2O3 containing trig-
[3]
onally coordinated B ( B) only and BPO4 containing tetra-
[4]
hedrally coordinated B ( B) only. The sharp peak at about
94.0 eV is attributed to [ B in the boron species, and is fur-
3]
1
ther assigned to the transition of B 1s electrons to the unoc-
⁄
[3]
cupied B 2pz (… ) states for B with oxygen (37). However,
the B K-edge spectra show no evidence of the presence of
[4]
B with oxygen in the NMABx catalysts investigated. The
B species is most likely to be B2O3.
DISCUSSION
Speciation and Structural States of Additive Boron
FIG. 8. The relative heights (A) of Mo–S (filled squares and solid line)
and Mo–Mo shells (filled circles and dashed line) in the Fourier transforms
of the Mo K-edge EXAFS and the corresponding ratio (B) as a function
of B2O3/Al2O3.
The structure and chemical speciation of boron in alu-
minium borates which were prepared by the co-precipita-
tion technique, were studied by means of magic-angle

HDS ACTIVITY OF DBT OVER Ni–Mo/� -Al2O3–B2O3 CATALYSTS
363
the acidity of the catalysts, thus slightly enhancing the HC
activity. This is in good agreement with the presence of
only [ B in our samples, because the effect of boron on the
acidity of Al2O3–B2O3 is mainly related to the presence of
3]
[4]
B (23).
Dispersion of Active Metal Species
The chemical states and dispersion of active metal
species in the supported Co(Ni)–Mo(W)/� -Al2O3 catalysts
have been widely studied using chemisorption of probe
molecules (42), XRD (43), XPS (43, 44), EXAFS (39), in-
frared (45), M o¨ ssbauer (16, 17), transmission electron mi-
croscopy (46), and diffuse reflectance spectroscopy (31).
We did not attempt to investigate the structural states of Ni
and Mo in the calcined Ni–Mo/� -Al2O3–B2O3. However, it
is generally believed that both Ni2 and Mo may be tetra-
hedrally and octahedrally coordinated with oxygen and that
Ni and Mo species may enter the lattice of � -Al2O3 and/or
are present on the surface of � -Al2O3 particles (11).
As shown above, the majority of Mo atoms are present
as MoS2 in sulfided Ni–Mo/� -Al2O3–B2O3 catalysts, even
C
6C
though small amounts of oxide and oxy-sulfide species may
be present. It was assumed that the “Ni–Mo–S” phases,
nated B ( B) only and BPO4 containing tetrahedrally coordinated B ( B) in which Ni is located at the edges of the MoS2 crystal-
FIG. 10. B K-edge XANESspectra ofcalcined Ni–Mo/� -Al2O3–B2O3
catalysts, as compared with those of B2O3 containing trigonally coordi-
[3]
[4]
only.
lites, are responsible for the HDS catalytic activity. Thus,
it is important to investigate the dispersion of the active
metal phases. The term “dispersion” often simply means
spinning nuclear magnetic resonance (23). For aluminium the fraction of surface atoms relative to the total number
–
[3]
[4]
borates calcined at 500 C, both B and B species are of atoms. However, for anisotropic and layered MoS2, this
present, and the proportion of the tetrahedral boron term is not sufficient to describe the dispersion of metals.
species increases with the increasing B/Al ratio. The rel- The basal and edge dispersions must be determined, be-
ative strength of the acid sites is correlated with the cause for MoS2 particles with a fixed number of layers, the
higher proportion of the tetrahedral boron species. How- basal plane dispersion is similar, irrespective of the lateral
ever, the structure and chemical speciation of boron in � - dimension of the particles; in this case, however, the num-
Al2O3–B2O3 materials, prepared by wet impregnation, have ber of edge plane atoms and the edge dispersion are differ-
not been reported, even though Ram ´ı rez et al. speculated ent. The information on the basal dispersion was obtained
that B3 is predominantly tetrahedrally coordinated in the by XPS measurements from which the fraction of surface
Co–Mo/� -Al2O3–B(x) catalysts they investigated (31). The atoms relative to the total number of atoms was calculated.
C
B K-edge XANES spectra of our NMABx catalysts in- On the other hand, EXAFS has been proven to be a pow-
dicated that boron is trigonally coordinated with oxygen. erful technique for studying the edge dispersion or lateral
Thus, boron may not enter the tetrahedral positions in the dimension of active metal phases in the Ni–Mo/� -Al2O3
lattice of � -Al2O3. The boron species probably form mono- catalysts (18, 39).
layers on the surface of � -Al2O3 particles when the con-
For calcined Ni–Mo/� -Al2O3–B2O3 catalysts, the basal
tent of B2O3 is low (Fig. 9). With further increases in the dispersion of Ni and Mo species increases markedly when
B2O3 loadings, the boron species may form multiple layers even a small amount of boron was added to � -Al2O3, as ev-
and the interaction between � -Al2O3 and the boron species idenced by the higher concentrations of Ni and Mo on the
becomes weaker. However, the boron species are well dis- surface of catalysts (Figs. 5 and 6). However, the further
persed until the B2O3 loading reaches about 14 mol% at increase in the loading of B2O3 reduces the basal disper-
which point an crystalline B2O3 phase appears, as shown in sion of the Ni and Mo species and, thus, the atomic con-
the XRD patterns (Fig. 2). The boron species is most likely centrations of Ni and Mo on the surface of catalysts de-
to be B2O3, even though the borate species containing [
3]
B
crease. These results are in good agreement with the XRD
only are not excluded. In addition, the NH3 temperature- patterns of these catalysts (Fig. 2), which showed that a
programmed desorption of our NMABx catalysts showed crystallite MoO3 phase first appears at the B2O3 loading
that the addition of boron to alumina only slightly improves of about 5 mol% , and its diffraction intensity increases

364
LI ET AL.
with further loadings of B2O3. The sulfidation did not cause cursors. When the content of B2O3 is low, the boron species
significant change in the basal dispersion of Mo and Ni, forms monolayers on the surface of Al2O3 particles, which
since the atomic concentrations of Ni and Mo in both the is speculated to prevent Ni and, perhaps, Mo atoms from
calcined and the sulfided catalysts are similar in profiles entering into the lattice of � -Al2O3, thus increasing the pro-
(
Figs. 5 and 6). Of course, the different loadings of B may portion of octahedral Ni species and favoring the formation
affect the chemical states of Ni and Mo in the catalysts. For of Ni–Mo–S phases in the sulfided catalysts. At the B2O3
example, B species may prevent Ni and Mo from entering loading of about 1 mol% , Mo and Ni oxide phases form
the lattice of alumina (27), which may result in the change well-dispersed monolayers on the surface of � -Al2O3 par-
in the atomic concentrations calculated by XPS.
ticles, and the lateral dimension of the Ni and Mo species is
The lateral size or edge dispersion of the MoS2 particles relatively large. However, increasing loadings of B2O3 re-
is an important parameter for hydrotreating Ni(Co)–Mo duce the surface area and pore size of the support (31), pro-
sulfide catalysts. As shown in Fig. 8A, the FTM of both mote the formation of the multiple-layered boron species
Mo–S and Mo–Mo shells increase markedly with the addi- and weaken the interaction of both Mo and Ni oxide precur-
tion of boron and then slightly decrease with further load- sors with the � -Al2O3–B2O3 supports. Thus, the Mo and Ni
ings of B2O3. These results show that the average coordi- oxide phases tend to form multiple layers with limited lat-
nation numbers of S (N(S)) and Mo (N(Mo)) around Mo eral dimensions. The Mo oxide phase may agglomerate and
also increase markedly with the addition of boron and then form crystallites. The sulfidation extension of the Ni–Mo/
slightly decrease with the increasing content of B2O3 if it is � -Al2O3–B2O3 catalysts also decreases with additional
assumed that the order–disorder states of the absorbing Mo loadings of B2O3.
atoms are similar for these samples (47). In fact, we noted
The HDS activity of DBT over the NMABx catalysts de-
that the disorder of these catalysts may be slightly different, pends both on the conversion rates to diphenyl (by hydroc-
for example, the Mo–Mo disorder (see Fig. 7) may partially racking) and to cyclohexylbenzene (byhydrogenation). The
contribute to the FTM(Mo–Mo), because the FTM is also incorporation of boron into � -Al2O3 has a beneficial effect
inversely proportional to the power of the square of the dis- on the HDS of DBT over the conventional Ni–Mo/� -Al2O3
order. However, we could not make a thorough quantitative catalysts. At a B2O3 loading of about 1 mol% , the acidity
EXAFS analysis to distinguish the contributions of coor- and thus the hydrocracking ability of the catalysts have a
dination number and the structural disorder to the FTM more marked increase, and meanwhile the basal dispersion
using our current EXAFS software. Thus, only for a quali- of the active metal phases reaches an optimum. Thus, the
tative comparison, the lateral size appears to increase, and total HDS activity of DBT over the Ni–Mo/� -Al2O3– B2O3
the edge dispersion decreases with the addition of a small catalysts reaches a maximum. However, the HDS of DBT
amount of boron to � -Al2O3, and the further loadings of is more dependent on the high basal dispersion of metal
boron may slightly increase the edge dispersion.
species and the conversion rate to cyclohexylbenzene. The
maximum conversion rate to cyclohexylbenzene may be re-
lated to the synchronized tendencies of the atomic concen-
trations or basal dispersion of the active Mo and Ni phases
on the surface of the catalysts. With additional loadings of
B2O3, the interaction of both Mo and Ni oxide precursors
with the supports becomes weaker. The Mo oxide precursor
may also agglomerate to form a crystalline phase. Thus, the
hydrogenation ability of the catalysts and, thus, the total
HDS activity of DBT over these catalysts decrease. Our re-
sults on the effects of boron on the basal dispersion of active
Ni and Mo species and, thus, on the HDS activity of DBT
are in good agreement with those of phosphorus in Ni–Mo/
Effects of Boron on the Dispersion of Active Metal
Species and HDS Activity of DBT
Previous studies on the effects of additives to the con-
ventional Co(Ni)–Mo(W)/� -Al2O3 focused on catalytic ac-
tivity, selectivity, and catalyst lifetime. Less work has been
done on the effects of additives on the structure, nature,
and dispersion of active metal species in the catalysts. Phos-
phorus is one of the most commonly used additives in the
molybdenum-based hydrotreating catalysts supported on
alumina. A great deal of research was done to understand
the effects of phosphorus on the chemical states and disper-
sion of active species and HDN and HDS catalytic activities.
However, the real functions of phosphorus as an additive in
the Ni–Mo/� -Al2O3 are still unclear (48). The few studies
�
-Al2O3 catalysts (48), even though the exact functions of
addition of boron to these catalysts require elucidation.
CONCLUSIONS
on boron as an additive suggest that it has effects similar to
those of phosphorus (29–31). However, the effects of boron
XRD, XPS, and EXAFS were used to characterize the
on the chemical states and dispersion of active species and dispersion of active Ni and Mo species in the Ni–Mo/
the catalytic activity of the catalysts are even more contro- � -Al2O3–B2O3 catalysts. These three techniques provide
versial.
complementary information on the dispersion of the active
As discussed above, the addition of boron to the � -Al2O3 species: XRD is used to reveal the distribution and specia-
support causes the redistribution of Mo and Ni oxide pre- tion of the Ni and Mo phases in the bulk catalysts; XPS

HDS ACTIVITY OF DBT OVER Ni–Mo/� -Al2O3–B2O3 CATALYSTS
365
determines the surface atomic concentrations and basal 12. Lipsch, J. M. J. G., and Schuit, G. C. A., J. Catal. 15, 179 (1969).
1
1
1
1
3. Voorhoeve, R. J. H., and Stuiver, J. C. M., J. Catal. 23, 228 (1971).
dispersion of the active species; and Mo K-edge EXAFS
is a powerful technique for studying the number of edge
atoms and the edge dispersion. The combination of these
three techniques provide a complete picture for character-
4. Hagenbach, G., Courty, Ph., and Delmon, B., J. Catal. 31, 264 (1973).
5. Grange, P., and Delmon, B., J. Less-Common Met. 36, 353 (1974).
6. Topsøe, H., Clausen, B. S., Candia, R., Wivel, C., and Mørup, S.,
J. Catal. 68, 433 (1981).
izing the dispersion of the active metal species and impor- 17. Wivel, C., Candia, R., Clausen, B. S., Mørup, S., and Topsøe, H.,
J. Catal. 68, 453 (1981).
tant information needed to understand the catalytic activity
and selectivity of hydrotreating Ni(Co)–Mo(W) sulfide cat-
alysts.
1
1
8. Clausen, B. S., Topsøe, H., Candia, R., Villadsen, J., Lengeler, B.,
Als-Nielsen, J., and Christensen, F., J. Phys. Chem. 85, 3868 (1981).
9. Bouwens, S. M. A. M., Koningsberger, D. C., de Beer, V. H. J., Louwers,
S. P. A., and Prins, R., Catal. Lett. 5, 273 (1990).
Adding boron to � -Al2O3 generally increases the acidity
of the � -Al2O3–B2O3 supports and slightly improves the 20. Louwers, S. P. A., and Prins, R., J. Catal. 133, 94 (1992).
hydrocracking ability of Ni–Mo/� -Al2O3–B2O3. Thus, the
2
1. Topsøe, H., Clausen, B. S., Topsøe, N. Y., Hyldtoft, J., and Nørskov,
J. K., ACS Petrol. Div. Prepr. 38, 638 (1993).
conversion rate of DBT to diphenyl increases slightly with
additional loadings of B2O3. The conversion rate of DBT to
cyclohexylbenzene first increases with increasing amounts
of B2O3, up to about 1 mol% of B2O3, and then decreases
with increasing amounts of B2O3. This is in good agreement
with the dispersion of active Mo and Ni phases and the
hydrogenation ability of Ni–Mo/� -Al2O3–B2O3 catalysts.
As a result, the total conversion rate of DB over Ni–Mo/
22. Kabe, T., Qian, W. H., and Ishihara, A., J. Catal. 149, 171 (1994).
23. Peil, K. P., Galya, L. G., and Marcelin, G., J. Catal. 115, 441 (1989).
24. Sato, S., Hasebe, S., Sakurai, H., Urabe, K., and Izumi, Y., Appl. Catal.
29, 107 (1987).
2
2
5. Curtin, T., McMonagle, J. B., and Hodnett, B. K., Appl. Catal. A: Gen.
9
3, 91 (1992).
6. Lafitau, H., Neel, E., and Clement, J. C., in “Preparation of Catalysts”
B. Delmon, P. Jacobs, and G. Poncelet, Eds.), p. 393. Elsevier, Ams-
terdam, 1976.
(
�
-Al2O3–B2O3 reaches a maximum at a B2O3 content of 27. Stranick, M. A., Houalla, M., and Hercules, D. M., J. Catal. 104, 396
(
1987).
about 1 mol% and then decreases with further increases
in B2O3, which is more dependent on the dispersion of the
active metal phases and the hydrogenation ability of the
Ni–Mo/� -Al2O3–B2O3 catalysts.
2
2
8. Morishige, H., and Akai, Y., Bull. Soc. Chim. Belg. 104, 253 (1995).
9. Tsai, M. C., Chen, Y. W., Kang, B. C., Wu, J. C., and Leu, L. J., Ind.
Eng. Chem. Res. 30, 1801 (1991).
3
0. Li, C. P., Chen, Y. W., Yang, S. J., and Wu, J. C., Ind. Eng. Chem. Res.
32, 1573 (1993).
3
1. Ram ´ı rez, J., Castillo, P., Cede n˜ o, L., Cuevas, R., Castillo, M., Palacios,
J., and L o´ pez-Agudo, A., Appl. Catal. A: Gen. 132, 317 (1995).
2. Portela, L., Grange, P., and Delmon, B., J. Catal. 156, 243 (1995).
3. Kerkhof, F. P. J. M., and Moulijn, J. A., J. Phys. Chem. 83, 1612 (1979).
4. Matsubayashi, N., Tamayama, M., Shimada, H., Sato, T., Yoshimura,
Y., Mori, Y., Kawamata, H., Abe, M., Ogino, K., and Nishijima, A.,
Sekiyu Gakkaishi 37, 594 (1994).
ACKNOWLEDGMENTS
3
3
3
DL acknowledges the STA fellowship from the Science and Technol-
ogy Agency (STA) of Japan. Our thanks go to K. Masuda and Y. Kanda,
Catalysts & Chemicals Industry Ltd., Japan, for their technical assistance;
K. Sato, Petroleum Energy Center, Japan, for measuring BET surface ar-
eas; M. Kasrai, University of Western Ontario, Canada, for his help in
measuring some of the B K-edge XANES spectra. We thank the staff at
the Synchrotron Radiation Center (SRC), the University of Wisconsin,
Madison, WI, for their technical assistance; the National Science Foun-
dation for its support of the SRC; and the advisory committee of Photon
Factory for accepting this project CT (95G231).
3
3
5. Bancroft, G. M., Can. Chem. News 44, 15 (1992).
6. Matsubayashi, N., Kojima, I., Kurahashi, M., Nishijima, A., Itoh, A.,
and Utaka, A., Rev. Sci. Instrum. 60, 2533 (1989).
3
3
7. Li, Dien, Bancroft, G. M., Fleet, M. E., Hess, P. C., and Yin, Z. Y., Am.
Mineral. 80, 873 (1995).
8. Mensch, C. T. J., van Veen, J. A. R., van Wingerden, B., and van Dijk,
M. P., J. Phys. Chem. 92, 4961 (1988).
REFERENCES
39. Shimada, H., Matsubayashi, N., Sato, T., Yoshimura, Y., Imamura, M.,
Kameoka, T., and Nishijima, A., Catal. Lett. 20, 81 (1993).
40. Scofield, J. H., J. Electron Spectrosc. Related Phenom. 8, 129 (1976).
41. Penn, D. R., J. Electron Spectrosc. Related Phenom. 9, 29 (1976).
42. Zmierczak, W., Quader, Q., and Massoth, F. E., J. Catal. 106, 65
(1987).
43. Li, C. P., and Hercules, D. M., J. Phys. Chem. 88, 456 (1984).
44. Grimblot, J., Dufresne, P., Gengembre, L., and Bonnelle, J. P., Bull.
Soc. Chim. Belg. 90, 1261 (1981).
1
2
3
4
5
6
7
. Grange, P., Catal. Rev. Sci. Eng. 21, 135 (1980).
. Furimsky, A., Catal. Rev. Sci. Eng. 22, 371 (1980).
. Topsøe, H., and Clausen, B. S., Catal. Rev. Sci. Eng. 26, 395 (1984).
. Chianelli, R. R., Catal. Rev. Sci. Eng. 26, 361 (1984).
. Delannay, F., Appl. Catal. 16, 135 (1985).
. Topsøe, H., and Clausen, B. S., Appl. Catal. 25, 273 (1986).
. Topsøe, H., Clausen, B. S., Topsøe, N. Y., and Pedersen, E., Ind. Eng.
Chem. Fundam. 25, 25 (1986).
45. Topsøe, N. Y., J. Catal. 64, 235 (1980).
8
. Prins, R., de Beer, V. H. J., and Somorjai, G. A., Catal. Rev. Sci. Eng. 46. Eijsbouts, S., Heinerman, J. J. L., and Elzerman, H. J. W., Appl. Catal.
3
1, 1 (1989).
A: Gen. 105, 53 (1993).
9
. Startsev, A., Catal. Rev. Sci. Eng. 37, 353 (1995).
47. Teo, B. K., “EXAFS: Basic Principles and Data Analysis,” p. 349.
Springer-Verlag, Berlin, 1986.
1
1
0. Delmon, B., Bull. Soc. Chim. Belg. 104, 173 (1995).
1. Topsøe, H., Clausen, B. S., and Massoth, F. E., “Hydrotreating cata- 48. Atanasova, P., Halachev, T., Uchytil, J., and Kraus, M., Appl. Catal. 38,
lysis: Science and Technology,” p. 310. Springer-Verlag, Berlin, 1996. 235 (1988).