Gallium Modified HUSY Zeolite as an Effective Co-support for NiMo Hydrodesulfurization Catalyst and the Catalyst's High Isomerization Selectivity

Article abstract of DOI:10.1002/chem.201701307

The effects of metal-modified acidic co-supports on the hydrodesulfurization (HDS) activity and isomerization selectivity of highly refractory organosulfur compounds such as 4,6-dimethyldibenzothiophene have been investigated. Y zeolite crystals with high Si/Al ratios and small crystallite sizes were successfully synthesized by a new hydrothermal synthesis approach. The synthesized Y zeolite crystals were ion-exchanged and stabilized. The prepared samples were then modified with different gallium contents using an impregnation method to adjust their acidity properties, and these modified samples were used as co-supports for NiMo sulfide HDS catalysts. The catalyst containing 10 wt.% zeolite Y modified by 2 wt.% gallium (NiMo/2GaY-ASA-A) exhibited the highest HDS activity, with 4,6-dimethyldibenzothiophene (4,6-DMDBT) conversion nearly double the rate of the catalyst without zeolite at 563 K, 4.0 MPa and liquid hourly space velocity (LHSV) of 40 h^?1. NiMo/2GaY-ASA-A also exhibited superior isomerization ability, with 3,4′-DMBP, 4,4′DMBP, and 3,6-DMDBT as the main products, indicating that the isomerization pathway was the main reaction route over NiMo/2GaY-ASA-A. The superior catalytic performance is related to the synergistic effect of the proper amount of medium and strong Br?nsted acid sites. The compounds 3,6-DMDBT and 3,7-DMDBT (isomers of 4,6-DMDBT) and 3,4,6-TMDBT and tetra-methyl-DBT (transmethyl products) were detected simultaneously in the HDS product of 4,6-DMDBT for the first time over NiMo/GaY-ASA-A catalysts. Finally, a new reaction network over NiMo/2GaY-ASA-A was proposed.

Full text of DOI:10.1002/chem.201701307

A Journal of
Accepted Article
Title: Gallium Modified HUSY Zeolite as an Effective Cosupport for
NiMo Hydrodesulfurization Catalyst and the Catalyst's High
Isomerization Selectivity
Authors: Wenwu Zhou, Yasong Zhou, Qiang Wei, Lin Du, Sijia Ding,
Shujiao Jiang, Yanan Zhang, and Qing Zhang
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Chem. Eur. J. 10.1002/chem.201701307
Link to VoR: http://dx.doi.org/10.1002/chem.201701307
Supported by

Chemistry - A European Journal
10.1002/chem.201701307
Manuscript
Gallium Modified HUSY Zeolite as an Effective Cosupport for NiMo Hydrodesulfurization
Catalyst and the Catalyst’s High Isomerization Selectivity
*
Wenwu Zhou, Yasong Zhou , Qiang Wei, Lin Du, Sijia Ding, Shujiao Jiang, Yanan Zhang and
Qing Zhang
State Key Laboratory of Heavy Oil Processing, China University of Petroleum, Beijing 102249,
PR China
*
Corresponding Author
Phone: 0086 010 89733501.
E-mail: zhouyasong2011@163.com.
1
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
Abstract:
The effects of metal-modified acidic cosupports on the HDS activity and isomerization
selectivity of highly refractory organosulfur compounds such as 4,6-dimethyldibenzothiophene
have been investigated. Y crystals with high Si/Al ratios and small crystallite sizes were
successfully synthesized by a new hydrothermal synthesis approach. The synthesized Y crystals
were ion-exchanged and stabilized. The prepared samples were then modified with different
gallium contents using a impregnation method to adjust their acidity properties, and these
modified samples were used as cosupports for NiMo sulfide hydrodesulfurization (HDS) catalysts.
The catalyst containing 10 wt.% Y zeolite modified by 2 wt.% gallium (NiMo/2GaY-ASA-A)
exhibited the highest HDS activity, with 4,6-dimethyldibenzothiophene (4,6-DMDBT) conversion
-
1
nearly double the rate of the catalyst without zeolite at 563 K, 4.0 MPa and LHSV of 40 h .
NiMo/2GaY-ASA-A also exhibited superior isomerization ability, with 3,4’-DMBP, 4,4’DMBP
and 3,6-DMDBT as the main products, indicating that the isomerization pathway was the main
reaction route over NiMo/2GaY-ASA-A. The superior catalytic performance is related to the
synergistic effect of the proper amount of medium and strong Brønsted acid sites. The compounds
3
,6-DMDBT and 3,7-DMDBT (isomers of 4,6-DMDBT) and 3,4,6-TMDBT and
tetra-methyl-DBT (transmethyl products) were detected simultaneously in the HDS product of
,6-DMDBT for the first time over NiMo/ GaY-ASA-A catalysts. Finally, a new reaction network
4
over NiMo/2GaY-ASA-A was proposed.
KEYWORDS: gallium-modified HUSY, 4,6-DMDBT hydrodesulfurization, isomerization,
transmethylation, selectivity
1
. Introduction
2
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
2
SO emitted by the combustion of organosulfur compounds in diesels has a significant negative
impact on the atmosphere, the environment and human health. The reduction of the sulfur content
in diesel fuels has received much attention in recent years due to stringent environmental
legislation worldwide. To meet this requirement, hydrodesulfurization (HDS) has high potential
for the removal of sulfur atoms from diesel fuels[1,2]. The greatest challenge in ultra-deep (with S
content below 10 ppm) HDS is the removal of highly refractory organosulfur compounds such as
4
,6-dimethyldibenzothiophene (4,6-DMDBT) from diesel fuels because the steric hindrance of the
methyl groups at the 4 and 6 positions strongly restricts desulfurization[2~5]. The catalysts
commonly employed in the petrochemical industry to remove sulfur atoms from transportation
2 2
fuels are cobalt- or nickel-promoted MoS or WS supported on γ-alumina[3~6]. Highly refractory
sulfur-containing compounds such as 4,6-DMDBT undergo HDS by direct desulfurization (DDS)
and hydrogenation desulfurization (HYD) over conventional catalysts[2,4,7~10]: DDS produces
3
,3’-dimethylbiphenyl (3,3’-DMBP) as the desulfurized product, and HYD produces tetrahydro-
and hexahydro- 4,6-DMDBT as intermediates, which can be further desulfurized to yield
,3’-dimethylcyclohexylbenzenes (3,3’-DMCHB) and 3,3’-dimethylbicyclohexyls (3,3’-DMBCH).
3
Due to the hindrance caused by the two methyl groups at the 4 and 6 positions, which strongly
prevent the adsorption of the sulfur atom at the active sites of the catalysts through
sigma-adsorption, sulfur removal from 4,6-DMDBT mainly proceeds through the HYD
pathway[4,11,12,13]. Thus, catalysts must possess both high hydrogenation activity to achieve the
hydrogenation of aromatic rings and high hydrogenolysis activity to achieve the removal of sulfur
atoms[14-16]. However, these commercial catalysts do not eliminate the hindrance caused by the
methyl groups at the 4 and 6 positions, and thus it remains difficult to remove sulfur atoms from
3
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
4
,6-DMDBT molecules. To overcome these drawbacks, catalysts should posess both high
hydrogenation activity to promote the saturation of aromatic rings and high methyl migration
activity to promote the isomerization of substituted DBTs. And acidic supports with high methyl
migration activity, especially zeolites, are widely used as cosupports for HDS catalysts[4,17,18].
Zhang et al.[19] observed that cyclohexenyl benzene, one of the desulfurized products of
dibenzothiophene (DBT), was isomerized to dimethyldecalin over an acidic Beta-KIT-6-supported
NiMo catalyst. Moreover, many nanocrystalline zeolites[20-24], such as Y, ZSM-5 and beta, show
higher catalytic activities and lower deactivation rates than conventional ones. Among all
investigated materials, Y zeolites are the most suitable acidic cosupports for high refractory
sulfur-containing compounds such as 4,6-DMDBT because their microporosity is comparable to
the size of 4,6-DMDBT molecules. HDS catalysts containing Y zeolites as a cosupport exhibit
relatively high isomerization ability, enabling the highly efficient HDS of highly refractory
organosulfur compounds such as 4,6-DMDBT in inferior diesel fuels[4,10,11]. However, the acid
strength, acid density, acid type and even textural properties of Y zeolites cannot satisfy the
demands of 4,6-DMDBT HDS. To improve the catalytic performance of Y zeolites, the
modification of Y zeolites using different metals is of great importance to adjust the acidic
properties and textural properties of Y zeolite[25,26]. This has a beneficial effect for either the
adsorption process or the reaction process in producing ultra clean diesels. Dapsens et al.[27]
modified Y zeolite with gallium and investigated the selective conversion of bio-based
dihydroxyacetone. They concluded that the post-synthetic alkaline-assisted galliation of FAU-type
zeolites generated outstanding acid catalysts. In addition, gallium-modified zeolites were highly
efficient for the isomerization of dihydroxyacetone. Moreover, E. Altamirano et al.[28,29]
4
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
employed gallium as an additive to investigate the effect of gallium over NiMo/Al
2
O
3
catalyst and
2 3
NiW/Al O catalyst. They concluded that small amount of gallium can promote the catalytic
activity of the corresponding catalysts for 4,6-DMDBT HDS due to the fact that tetrahedral
2 3
coordinated gallium species over Al O supports enhanced the sulfidability of Ni and Mo(W)
species and caused more Ni species participating in the decoration of NiMoS phase. Thus, we
modified HUSY zeolites with gallium as a co-support for a hydrotreating catalyst. After HUSY
zeolites are modified by gallium, some of the Si-O(H)-Al bonds can be substituted by Si-O(H)-Ga
bonds[30] because the Ga-O(H)-Si sites are more covalent than the Al-O(H)-Si sites. Moreover,
the average electronegativity value of 1.59 for Ga is lower than the value for Al (1.714)[31],
implying a lower acid strength of Brønsted protons for Ga-substituted zeolites. Previous studies
have indicated that gallium-containing zeolites can effectively decrease the production of lower
C1-C5 hydrocarbons and polyaromatic products in many petrochemical processes[32,33]. This
decrease occurs mainly because gallium species deposited on zeolites significantly improve the
hydrogenation activity of zeolites and lower the acid strength of the modified zeolites to avoid the
over-cracking of the hydrogenated hydrocarbons. Since the mechanistic studies of alkanes
aromatizing into aromatics indicates that gallium species deposited on zeolites have a significant
+
effect on the adsorption and activation of alkanes to produce activated C intermediates, we also
hypothesized that gallium-modified zeolites exhibit extraordinary activity for the isomerization of
hydrocarbons compared to conventional ones.
2 2 3
To test these hypotheses, we synthesized a nanocrystallite NaY zeolite with a high SiO /Al O
ratio and high crystallinity. The synthesized zeolites were hydrothermally treated and then
modified using different contents of gallium, and the corresponding Ni-Mo HDS catalysts were
5
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
synthesized. The effects of gallium modification on the acidity and textural properties of HUSY
zeolites were investigated, and the effects of gallium modification on the acidity, textural
properties and active-phase morphology were also investigated for the first time. Moreover, the
effects of medium and strong Brønsted acidity on catalytic performance and isomerization
selectivity were investigated. The highest activity for the HDS of 4,6-DMDBT was observed over
NiMo/2GaY-ASA-A, which exhibited the highest isomerization selectivity. All the catalysts’
supports are composed of 10 wt.% zeolite, 20 wt.% ASA and 70 wt.% Al2O3, the catalysts are
composed of 5 wt.% NiO, 15 wt.% MoO3 and 80 wt.% support. Finally, we proposed a reaction
network involving the isomerization and transmethylation of 4,6-DMDBT and related
2 3
intermediate products for the HDS of 4,6-DMDBT over NiMo/GaHUSY-ASA-γ-Al O catalyst.
To clarify the effect of gallium modification, the HDS reaction of 4,6-DMDBT was also studied
over a gallium-free zeolite-containing catalyst.
2
2
. Experimental Section
.1 Catalyst preparation
All NaY zeolites were synthesized via a new two-stage variable temperature crystallization
2
program. In a typical synthesis, 2.0 g of sodium hydroxide, 77 g of sodium silicate (28 wt.% SiO ,
9
2 2
.1 wt.% Na O, and 62.9 wt.% H O), 3.25 g of sodium aluminate, 3.5 g of aluminum sulfate, 20 g
of deionized water and 10 g of synthesized structural directing agent were mixed under vigorous
2 2 3 2 2
agitation, and the precursor gel composition was 7.5:1:16:240 Na O/Al O /SiO /H O. The mixed
gel was then aged at 315 K for 18 h and then pre-crystallized at 335 K in a Teflon autoclave.
Finally, the pre-crystallized system was hydrothermally crystallized at 368 K for another 24 h,
washed with deionized water, and calcined at 823 K for 5 h. The synthesized NaY zeolites were
6
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
4 3
ion-exchanged with 1.0 M NH NO aqueous solution 3 times at 363 K for 4 h and then calcined at
8
23 K for 5 h in a muffle furnace to obtain HY zeolites. The obtained HY zeolites were stabilized
by treating with citric acid (weight ratio of citric acid to HY zeolite = 1:12, Beijing Modern
Oriental Fine Chemical Co. Ltd., ≥99.8 %), and the product was denoted ultra-stable HY (HUSY)
zeolite. Gallium-modified HUSY zeolites with different Ga contents were prepared by incipient
wetness impregnation with a calculated aqueous solution of gallium nitrate (Aladdin, ≥99.9 %).
The obtained samples were dried at 393 K for at least 6 h after the evaporation of the solvent at
room temperature overnight, then calcined at 823 K for 5 h in a muffle furnace. The modified
2 3 2 3 2 3
zeolites containing 1 % Ga O , 2 % Ga O and 3 % Ga O were denoted 1GaHUSY, 2GaHUSY
and 3GaHUSY, respectively. Amorphous silica-alumina (ASA) was prepared as described
elsewhere in the literature [24], and the pseudo-boehmite was purchased from Shangdong
Aluminium Factory (Zibo, Shangdong). The gallium-modified HUSY, prepared ASA and
purchased pseudo-boehmite were mechanically mixed at a mass ratio of 1:2:7, extruded to form a
four leaf clover-cylindrical shape and crushed into 20 to 40 mesh. Ni-Mo catalysts were prepared
via an incipient wetness co-impregnation method with an aqueous solution of appropriate amounts
of nickel nitrate hexahydrate and ammonium heptamolybdate tetrahydrate, then dried at 393 K for
6
h after vaporization at room temperature overnight and calcined at 823 K for 5 h. The catalysts
were designated NiMo/Y-ASA-A, NiMo/1GaY-ASA-A, NiMo/2GaY-ASA-A and
NiMo/3GaY-ASA-A, where the number represents the mass content of gallium in the HUSY
zeolites. For all the synthesized catalysts, the total concentration of NiO in the catalysts was 5
3
wt.%, and the total MoO concentration was 15 wt.% in the catalysts. A reference catalyst,
NiMo/ASA-Awas also prepared by the same process except that the support consisted of 30 wt.%
7
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
2 3
ASA and 70 wt.% Al O .
2
.2 Material characterization
The powder X-ray diffraction (XRD) patterns of all synthesized and modified zeolites were
recorded on a PANalytical advance powder diffractometer with Cu Kα radiation (40 kV, 40 mA) in
the 2θ interval of 5~35°. The crystallite size of the synthesized zeolite was observed by field
emission scanning electron microscopy (SEM) on a Quanta 200F instrument, and the crystallite
size was statistically analyzed by counting at least 500 crystals. The Si, Al and Ga contents in the
synthesized zeolites were determined by inductively coupled plasma atomic emission
spectroscopy (ICP-AES) on a PE OPTIMA 5300DV instrument. Nitrogen adsorption-desorption
measurements were performed at 77 K using a Micromeritics ASAP 2010 analyzer on degassed
-
1
samples (10 mbar, 573 K, 4 h). The surface area of the examined samples was determined using
the Brunauer-Emmett-Teller (BET) equation, and the pore volume and pore diameter were
2 3
obtained using the Barrett-Joyner-Halenda (BJH) method with the N adsorption isotherm. NH
3
temperature-programmed desorption (NH -TPD) of the obtained materials was performed on a
Chembet 3000 chemical adsorption instrument. All measured materials were pretreated at 823 K
for 60 min in a mixture of Ar and air flow (v:v=3:1) and then cooled to 393 K and allowed to
3 3
adsorb NH for 30 min. Afterwards, the physically adsorbed NH was removed in an Ar flow for
9
0 min. The furnace temperature was then programmed to increase at a rate of 10 K/min from 393
K to 1023 K in 65 min. The amount of desorbed ammonia was detected by continuous effluent gas
monitoring equipped with a thermal conductivity detector (TCD). To assess the acidity of zeolites
and catalysts, pyridine-adsorbed Fourier transform infrared measurements (Py-FTIR) were
conducted on a Magna 560 FT-IR analyzer. The samples were dehydrated at 673 K for 3 h under a
8
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
-
2
vacuum of 10 Pa. Then, pure pyridine vapor was fed to the measured sample at room temperature
for 30 min, obtaining the saturated adsorption of py-FTIR spectra. The adsorbed pyridine was then
evacuated at 473 K, 553 K and 623 K for 60 min in turn, to obtain the desorption of py-FTIR
2 2
spectra at 473 K, 553 K and 623 K, respectively. H temperature-programmed reduction (H -TPR)
patterns of all the investigated catalysts were recorded on a TP-5000 multifunction adsorption
instrument. Before the TPR measurement, the catalyst was pretreated at 673 K for 60 min in air
2
flow. Then, the reducing gas (H /Ar = 1:19, v:v) was fed to the tested sample at a flow rate of 30
mL/min, and simultaneously, the furnace temperature was programmed to increase at a rate of 10
K/min from 393 K to 1273 K. The HRTEM images of all sulfide catalysts were taken on a Philips
Tecnai G2 F20 instrument with an operating electron voltage of 200 kV. At least 500 slabs taken
from different parts of each catalyst were counted to determine the size distribution of
molybdenum sulfide crystallites. We calculated the average length and stack number of MoS
2
slabs using equations reported elsewhere as follows[34,35]:
ꢀ
∑
n l
i=1 i i
Average slab length: L̅ =
⁄ ꢀ
(1)
(2)
∑
ni
i=1
ꢀ
i=1
∑
niNi
Average stack number: Nꢁ =
⁄
ꢀ
∑
ni
i=1
where l is the length of the slab, n is the number of slabs with length l , and N is the number
i
i
i
i
of layers in slab ꢂ. The MoS
2
dispersion 푓푀표 was also calculated using the following equation
reported in the literature[4,36]:
푡
푀표
푀표
∑
6(푛 −1)
푒푑푔푒
푖ꢅꢆ
푖
푓 ꢃ
ꢃ ∑푡
(3)
푀표
2
ꢇ3푛 −3푛 +1ꢈ
푡ꢄ푡푎푙
푖ꢅꢆ
푖
푖
where ꢉꢊꢋꢌꢍꢋ is the number of Mo atoms located on the edges of NiMoS slabs, ꢉꢊꢎ표ꢎꢏꢐ is the
total number of Mo atoms, ꢑ is the number of Mo atoms along one edge of a MoS
2
slab
ꢒ
9
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
i
determined from its length (L = 3.2(2n - 1)Å), and ꢓ is the total number of slabs determined by at
2
least 500 MoS slabs taken from HRTEM images of different catalysts. The positions of the
introduced gallium species were measured by X-ray photoelectron spectroscopy (XPS) on a VG
ESCA Lab 250 spectrometer using Al Kα radiation as the excitation light source. Prior to the test,
the C 1s peak with a binding energy of 284.6 eV and the Al 2p peak with a binding energy of 74.6
eV were used to calibrate the binding energy scale.
2
.3 Catalyst evaluation
HDS reactions of 4,6-DMDBT were performed on a fixed-bed reactor (8 mm inner diameter
and 500 mm in length) loaded with 2 g of catalyst with a particle size of 0.59–0.84 mm in
diameter diluted by 5 mL of quartz sand. Before the assessment, the catalyst was presulfided using
2 2
a solution of 6.0 mL CS dissolved in 500 mL of cyclohexane at 593 K and 4 MPa for 6 h in a H
-
1
-1
flow of 1200 mL min with a liquid hourly space velocity (LHSV) of 10 h . After that, the reactor
temperature was decreased to 563 K, and a mixture of 0.5 wt.% of 4,6-DMDBT dissolved in
cyclohexane was pumped to the reactor using an SZB-2 double-piston pump. The 4,6-DMDBT
HDS evaluations were conducted at temperatures of 558 K to 573 K, a total pressure of 4.0 MPa, a
-
1
-1
liquid hourly space velocity (LHSV) of 10.0 h to 150 h , and a hydrogen/oil ratio of 120 (v/v).
The liquid reaction products were carefully collected after a full stabilization period of 12 h. Then,
the collected products were immediately analyzed off-line on an Agilent 4890D gas
chromatograph equipped with a 60-m capillary Rtx-1 column (0.25 mm, RESREK) with a N
2
-
1
pressure of 0.3 MPa and flow velocity of 30 mL min as the column temperature increased from
-
1
3
23 K to 593 K with a heat rate of 15 K min after staying at 323 K for 1 min. To further identify
each of the compounds in the liquid products, a Finnigan Trace GC–MS consisting of a Trace
10
This article is protected by copyright. All rights reserved.

(4)
Chemistry - A European Journal
10.1002/chem.201701307
Ultral gas chromatograph (60 m - 0.25 mm - 0.5 μm) capillary column and an MS detector HP
973 was used to analyze the collected products.
The global rate constant (kHDS) of 4,6-DMDBT HDS reactions were calculated by employing
5
the pseudo-first order reaction according to the following two equations[4]:
퐹
푘퐻퐷푆 ꢃ ꢔ ꢕ ln
푚
-
1
-1
where 푘퐻퐷푆 stands for the reaction rate constant for the pseudo-first order reaction in mol g h ,
which is calculated by changing the reactant feeding rate, 퐹 is the reactant feed to the reactor in
-
1
mol h , m is the mass of the loaded catalyst in grams, and 푥 is the total conversion of
4
,6-DMDBT.
The activation energies of 4,6-DMDBT HDS reactions over all the mentioned catalysts at 563 K
were calculated by employing a pseudo-first order reaction according to equation 5[30,33] by
-
1
fixing the LHSV at 40 h and varied the reaction temperature in the range of 558 K to 573 K at an
interval of 5 K:
ꢘ1
푅푇ꢗꢑ( )
ꢘꢙ ⁄
Ea ꢃ
(5)
(
푇ꢖ ꢔ 푇ꢚ)
-
1
where Ea is the activation energy in kJ mol , 푥ꢖ is the 4,6-DMDBT conversion at higher
reaction temperature (푇ꢖ), 푥ꢚ is the 4,6-DMDBT conversion at lower reaction temperature (푇ꢚ).
3
3
. Results
.1 Textural properties
The XRD patterns of the synthesized NaY, HY, USY and HUSY zeolites and of HUSY
modified with different contents of gallium are shown in S1 and Fig. 1. All materials present the
typical diffraction features of FAU zeolites, and no diffraction peaks related to gallium species
were observed, indicating the formation of FAU zeolite cells and an intact FAU zeolite structure
11
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
after the modification process[5]. However, the lower intensities of the diffraction peaks imply
lower crystallinity after the ion-exchange process, stabilization process and modification process.
The crystallinity and framework SiO /Al O molar ratio of the synthesized zeolites were obtained
2
2
3
and are listed in Table 1. The framework SiO
2 2 3
/Al O molar ratio of the synthesized NaY zeolite
was 5.8. After ion-exchange and citric acid treatment of the NaY zeolite, the framework
2 2 3
SiO /Al O molar ratio increased to approximately 12.5 and 14.9 for the HY and USY zeolites,
2 2 3
respectively. The SiO /Al O molar ratio of the HUSY zeolite increased to approximately 23.5
(
2 2 3
S-Table 1). The increase in SiO /Al O molar ratio was clearly more pronounced during acid
treatment than the ion-exchange process, indicating greater removal of aluminum during the citric
acid treatment than during the ion-exchange process and stabilization of the Y zeolite due to
aluminum removal. Due to continuous dealumination, the HY zeolite first treated by citric acid
4 3 2 2 3
and then ion-exchanged in NH NO aqueous solution exhibited an even higher SiO /Al O molar
ratio than the other materials. The intensity of the diffraction peaks and crystallinity of the Y
zeolites decreased as the SiO
2
/Al
2
O
3
molar ratio increased, indicating that during the
dealumination, their crystallinity decreased. The intensities of the diffraction peaks of the different
gallium-modified HUSYzeolites were slightly lower than for HUSY, implying a slight decrease in
2 2 3
crystallinity during modification. In addition, the framework SiO /Al O molar ratio increased
slightly as the gallium content increased, indicating that a small amount of framework aluminum
was have been replaced by gallium species and some framework aluminum became
non-frameworkAl species during the modification.
Gallium was deposited onto the HUSY zeolite to obtain modified zeolites with detected
2 3
amounts of 1.26, 2.03 and 2.87 wt.% G O . The mesoporous surface area and pore volume of the
12
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
modified zeolites were substantially retained upon metal incorporation (Table 1), perhaps due to
the rather low loading. However, the microporous surface area and pore volumes of the zeolites
decreased as the incorporation of gallium increased. Although XRD analysis (Fig. 1) indicated a
2 3
slight decrease in crystallinity (Table 1), the absence of reflections specific to Ga O in the
patterns suggested the presence of a well-dispersed, nanostructured gallium phase. It is possible
that the pore structure of the USY zeolite underwent a small degree of destruction after
modification by gallium (Fig. 1) and that some portion of the micro pores were blocked by the
introduced gallium species and removed aluminum (Table 1). To further confirm the positions of
the introduced gallium species on HUSY zeolites, the XPS spectra of Ga 3d for each modified
zeolite were recorded (Fig. 2). The XPS spectra of Si 2p and Al 2p for 2GaHUSY were also
recorded (S2).
Based on the results presented in Fig. 2, we can conclude that there are four peaks at binding
2 3
energies of 21.4 eV, 24.7 eV, 26.0 eV and 27.2 eV that correspond to Ga O species, Ga-O(H)-Al
bonds, Ga-O(H)-Si bonds and Ga-OH species, respectively[31,37]. When the content of gallium is
relatively low, the main gallium species is Ga-O(H)-Al, and as the content of gallium increases,
2 3
the presence of Ga O and Ga-O(H)-Si increases. The result presented in S2A exhibits a peak at
the binding energy of 99.8 eV, which was attributed to the Si-O(H)-Ga bonds, and the result
presented in S2B exhibits a peak at the binding energy of 72.6 eV, which was attributed to the
Al-O(H)-Ga bonds[29,31,37]. All these results indicate that gallium species exhibit a preference to
coordinate with Al species or substitute the Al-OH groups in zeolites during the modification
process and the preference for Al diminishes for an increase of Ga loading.
3
.2 Acidity properties
13
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
Acidity plays an important role in the formation of the active phase and the HDS reaction[38],
and the acidities of all modified HUSY zeolites and the corresponding catalysts were investigated
using NH -TPD and Py-FTIR methods, respectively. Fig. 3 shows the NH -TPD profiles of the
3
3
3 3
modified HUSY zeolites and the catalysts. The NH -TPD profiles in Fig. 3A reveal two NH
desorption peaks for gallium-modified HUSY at approximately 510 K and 680 K, indicating the
3
presence of two types of acid sites with different acid strengths. In general, the NH desorption
3
peak at 510 K reveals the existence of weak acid sites, whereas the NH desorption peak at 680 K
indicates the existence of medium and strong acid sites. As the gallium content increased, the
lower-temperature desorption peak for gallium-modified HUSY shifted slightly to an even lower
temperature, suggesting that the acidity of the weak acid sites was weakened. The intensity of the
lower-temperature desorption peak also decreased, indicating that the amount of weak acid sites
decreased. The higher-temperature desorption peak also shifted to a lower temperature, but the
intensity of the higher-temperature desorption peak increased, indicating a decrease in acid
3
strength and an increase in the amount of medium and strong acid sites. The NH -TPD profiles in
Fig. 3B reveal more than two peaks for the catalysts at temperatures of 495 K and higher,
indicating the presence of complex acid types with different acid strengths. As the gallium content
3
increased, the lower NH desorption peak shifted to an even lower temperature, suggesting a
3
decrease in the acid strength of the weak acid sites[39]. However, the other NH desorption peak
shifted to higher temperatures, suggesting an increase in the acid strength of the stronger acid sites.
These results are similar to the observations of Meng et al.[40] over Y/ASA composite. They
stated that a particular conjunction of the tetrahedrons and thus a special skeleton structure might
be formed at the interface of the Y zeolite and ASA phases in the Y/ASA composite.
14
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
To observe the changes in Brønsted acid sites (BAS) and Lewis acid sites (LAS) on the
modified HUSYzeolites and the corresponding catalysts, Py-FTIR spectrum analyses and pyridine
desorption FTIR analyses were performed at different temperatures. The results are summarized in
Table 2 and S3A~S3F. Compared to the acidity properties of HUSY, the amount of weak BAS
decreased sharply as the gallium content increased, and the amount of weak LAS decreased
slightly (473 K). The amount of medium BAS increased obviously after gallium modification, and
2
GaHUSY contained the most medium BAS (553 K). However, the medium LAS decreased
slightly. The data in Table 2 demonstrate that both strong BAS and strong LAS (623 K) decreased
proportionally as the gallium content increased. In addition, the total amounts of BAS and LAS
decreased with increasing gallium content, and the B/L ratio increased. Similar trends were
observed for the corresponding catalysts. Moreover, there is an interesting result showing that the
-
1
total amounts of BAS decreased from 680 to 539 μmol g as the gallium content increased from 0
to 3 wt.%, and the trend became more pronounced. Regarding the total amounts of LAS, a slower
downtrend was first observed, and afterward, the trend reversed to an uptrend as the gallium
content increased from 2 wt.% to 3 wt.%.
3
2
.3 H -TPR measurement
2
The H -TPR technique can effectively characterize the reducibility of metal oxides loaded on a
2
support. Fig. 4 presents the H -TPR profiles for the investigated catalysts. Octahedrally
3
coordinated MoO species can be reduced at temperatures from approximately 600 K to 900 K,
when loaded on a support. Hexa-coordinated Mo(Ⅳ) species are formed that have strong
interactions with supports and thus are harder to reduce and often require higher temperatures,
2
from approximately 900 K to 1100 K [14,41,42]. The H -TPR profiles displayed in Fig. 4 revealed
15
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
2
two main H consumption peaks centered at temperatures of approximately 730 K and 1030 K.
2 3
The H consumption peak at 730 K can be attributed to the reduction of higher coordinated MoO
species to lower coordinated Mo(Ⅳ) species, and the H consumption peak shouldered at 1030 K
2
can be ascribed to the reduction of tetra-coordinated Mo(Ⅳ) species, which are difficult to reduce
due to their strong interaction with the supports. The reduction peaks of Ni species are probably
3 2
masked by the MoO reduction peaks. Compared to NiMo/Y-ASA-A, the H -TPR profiles of the
catalysts containing gallium-modified HUSY zeolites exhibited a significant change: the two
reduction peaks notably shifted to lower temperatures, indicating that the reducibility of Mo
species on these catalysts clearly improved. Moreover, the peak area at the lower temperature
increased, and the peak area at the higher temperature decreased. However, when the gallium
2
content was increased to 3 % in the modified zeolite, the H -TPR profile presented three reduction
2 3 2
peaks. The peak centered at 550 K is attributable to the reduction of Ga O to Ga O[31].
Compared with NiMo/2GaY-ASA-A, the Mo peak of NiMo/3GaY-ASA-A at lower temperature
shifted slightly to a higher temperature, and the Mo peak at higher temperature shifted to an even
higher temperature, suggesting that the reducibility of the Mo species on NiMo/3GaY-ASA-Awas
slightly deteriorated compared with NiMo/2GaY-ASA-A. These results can be explained by the
theory proposed in the literatures elsewhere[28,29] that the existence of gallium in tetrahedral
coordination over the aluminum sites, which results in the decrease of aluminum sites at the
surface of USY zeolites and leads to a decrease in the formation of Mo-O-Al bonds, which is
considered to be the real mode of the interaction between active metals and supports, thus the
interaction decreased.
3
.4 HRTEM results
16
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
HRTEM studies were performed to compare the sulfide metal dispersion degree of the
investigated catalysts. Fig. 5 displays the representative micrographs taken from different catalysts.
The statistical analyses of the average slab length and stacking layers of active MoS slabs over
2
sulfide catalysts were obtained by counting at least 500 slabs obtained from different parts of each
̅
ꢁ
catalyst. The average length 퐿 and stacking number 푁 were calculated according to Eq. 1 and
Eq. 2 and are listed in Table 3. Ahigh dispersion of the active phase is necessary for the formation
of large numbers of accessible active sites. Slabs with a shorter length and lower stacking number
are beneficial for the formation of more CUS sites, namely “type Ⅱ” active phase[29,42].
2
However, when the stacking height of MoS slabs is lower than the size of the aromatic ring, the
planar adsorption of reactant molecules will be strictly limited, and the HDS activity of highly
refractory sulfur-containing compounds will be constrained. Thus, we performed a more detailed
comparison by calculating the MoS
2
dispersion 푓푀표 according to Eq. 3, and the calculated results
are also listed in Table 3. The results demonstrate that both the average stacking number and the
2
average length of MoS slabs changed with the increased gallium content. For NiMo/ASA-A, the
2
average length of the MoS slabs was 5.3 nm, the average stacking number was 2.7, and there are
2
a large proportion of monolayer MoS slabs, which are less active for 4,6-DMDBT HDS reactions.
When HUSY was added as a cosupport, the average length decreased to 5.1 nm, and the average
stacking number increased to 3.5. More importantly, the amount of the less active monolayer
MoS
2
slabs decreased to a relatively small proportion. When the HUSY zeolites were modified by
different contents of gallium, the average length further decreased, the average stacking number
further increased with increasing gallium content, and the monolayer MoS
2
slabs nearly
disappeared from view. However, when the gallium content increased to 3wt.%, this trend was
17
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
2
reversed, especially for the amount of monolayer MoS slabs and the average length. The
2
dispersion and stacking of MoS slabs may have been improved on HUSY zeolites, and the
introduced gallium caused a clear decrease in the interaction of the active MoS species with the
2
supports. The highest 푓푀표 (0.29) was obtained over NiMo/2GaY-ASA-A with an average length
2
of 4.6 nm, an average stacking number of 3.8 and the smallest amounts of monolayer MoS slabs.
To visualize how Ga species influence the dispersion of Ni and Mo species over the sulfided
catalysts, Ga, Ni and Mo elemental scanning images of the NiMo/Y-ASA-A and the
NiMo/2GaY-ASA-A catalysts were obtained and the results are shown in S9 and S10. The results
imply that after gallium modification, there are more Ni species embedded in Mo species which is
beneficial for the formation of more NiMoS active phase. This is highly consistent with the results
observed from HRTEM images and the results obtained from XPS characterization.
3
.5 XPS results
To analyze the valence states of active metals and further investigate how the three metals (Ga,
Ni and Mo) interact in the sulfided catalyst, the XPS characterization of the sulfided catalysts were
carried out and the Mo 3d XPS spectra and the Ni 2p XPS spectra together with their
deconvolution results for the investigated sulfided catalysts are shown in S6 and S7 and the details
are summarized in Table 4. It can be seen that the sulfidation of Mo species(Mosulfidation) increase
in the order: NiMo/Y-ASA-A < NiMo/ASA-A < NiMo/1GaY-ASA-A < NiMo/2GaY-ASA-A <
NiMo/3GaY-ASA-A and the sulfidation of Ni species(Nisulfidation) increase in the order:
NiMo/Y-ASA-A
<
NiMo/ASA-A
<
NiMo/1GaY-ASA-A
<
NiMo/3GaY-ASA-A
<
NiMo/2GaY-ASA-A, more importantly, the proportion of NiMoS species increase in the same
2
order with the Nisulfidation. These results are consistent with the H -TPR of the oxide catalysts and
18
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
the HRTEM results of the sulfide catalysts. All these results are consistent with the results
observed over NiMo/Ga-Al2O3 catalyst by E. Altamirano et al.[28,43], and their theory is that
small amount of Ga causes a decrease in the interaction between support and active
2 4
metal(decrease in the amounts of Mo-O-Al bonds and NiAl O species). Due to its moderate
interaction between active metals and the supports, NiMo/2GaY-ASA-A exhibits the highest Mo
dispersion and the largest proportion of NiMoS species, which is considered as the real active
phase of the non-noble metal hydrotreating catalysts.
3
.6 4,6-DMDBT HDS results
,6-DMDBT is the most commonly representative highly refractory organosulfur compound in
4
inferior diesels. Thus, it is one of the most common probe molecules in the assessment of the HDS
performance of hydrotreating catalysts. Fig. 6 displays the 4,6-DMDBT conversions over different
-
1
-1
catalysts at different LHSVs. The LHSV was varied in the range of 10 h to 160 h with a
reaction temperature of 563 K, total pressure of 4 MPa and H /Oil ratio of 120:1. The results
2
showed that NiMo/2GaY-ASA-Aexhibited the highest 4,6-DMDBT conversion for all LHSVs. At
-
1
2
0 h , the most representative LHSV of our system, NiMo/2GaY-ASA-Aexhibited a 4,6-DMDBT
conversion of 57.6 %, approximately 2 times the value for NiMo/ASA-A and approximately 1.3
times the value for of NiMo/Y-ASA-A. Fig. 6 also indicates that the gallium-modified catalysts
exhibited higher 4,6-DMDBT conversion at all LHSVs.
To distinguish the different desulfurization routes and the contributions of each desulfurization
route to the overall HDS performance over each investigated catalyst, we varied the LHSVs to
control the total conversion of 4,6-DMDBT at approximately 50 % and analyzed the HDS
products by GC-MS. The results are displayed in Table 5. Twelve types of products were detected
19
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
over
catalysts
containing
Y
zeolites,
including
the
following:
and
4
4
,6-tetrahydro-dimethyldibenzothiophene
(4,6-THDMDBT)
,6-hexahydro-dimethyldibenzothiophene (4,6-HHDMDBT), which were identified as the
intermediates of the HYD pathway; 3,3’-dimethylcyclohexylbenzene (3,3’-DMCHB) and
,3’-dimethylbicyclohexyl (3,3’-DMBCH), which were identified as the desulfurized products of
3
the HYD pathway; 3,3’-DMBP, which was identified as the product of the DDS pathway. Some
other types of products, such as 3,4’-DMBP, which was observed for the first time in the HDS
reaction of 4,6-DMDBT, 4,4’-DMBP, and a small proportion of iso-methyl-isopropyltetralin
(
Iso-MIPT)[4], which were identified as the products of the isomerization (ISO) pathway, were
also detected over zeolite-containing catalysts (Fig. 7A). More importantly, 3,6-DMDBT and
,7-DMDBT, two isomers of 4,6-DMDBT, were also observed based on the GC peaks with
retention times of approximately 20.2 min and 20.6 min (Fig. 7B), which indicates that
,6-DMDBT first underwent methyl migration to form 3,6-DMDBT and 3,7-DMDBT before S
3
4
atom removal. Moreover, tri- and tetra-methyl-DBTs were also detected based on the two GC
peaks with retention times of 21.5 and 24.7 min (Fig. 7B and S5). These results indicate that the
transmethylation (TRM) of methyl groups between different 4,6-DMDBT molecules also occurred
over zeolite-containing catalysts before desulfurization. However, only five types of products,
4
,6-THDMDBT, 4,6-HHDMDBT, 3,3’-DMDBT, 3,3’-DMCHB and 3,3’-DMBCH, were observed
over NiMo/ASA-A, indicating that 4,6-DMDBT only underwent HYD and DDS over
NiMo/ASA-A during the hydrotreating process due to the lack of medium and strong Brønsted
acid sites. This result demonstrates that Brønsted acidity, especially medium and strong Brønsted
acidity from the acidic cosupports, plays a very important role in the HDS activity of 4,6-DMDBT,
20
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
especially in the selectivity of its HDS pathway. The main product of 4,6-DMDBT HDS over
NiMo/ASA-A was 3,3’-DMCHB, and the HYD pathway selectivity reached as high as 82 %,
indicating that 4,6-DMDBT was mainly transformed by the HYD pathway over nonacidic
catalysts, which shows good consistency with the results reported elsewhere in the
literature[4,9,11].
However, the main products of 4,6-DMDBT HDS over gallium-modified HUSY
zeolite-containing catalysts were 3,4’-DMBP and 4,4’-DMBP and ISO-MIPT, indicating that
4
,6-DMDBT was mainly transformed through the ISO pathway over acidic catalysts. Moreover,
the selectivity of 3,4’-DMBP and 4,4’-DMBP increased with the amount of medium Brønsted acid
sites and reached 45 % over NiMo/2GaY-ASA-A. The total isomerization selectivity also
increased with the amount of medium Brønsted acid sites and reached 59 % over
NiMo/2GaY-ASA-A. Furthermore, the selectivity of tri- and tetra-methyl-DBTs decreased with
the amount of strong Brønsted acid sites from 13 % over NiMo/Y-ASA-A to 4 % over
NiMo/2GaY-ASA-A and NiMo/3GaY-ASA-A. The total DDS reaction selectivity increased from
1
8 % over NiMo/ASA-A to 38 % over NiMo/Y-ASA-A, and then to 54 % over
NiMo/2GaY-ASA-A.
To further confirm each of the 4,6-DMDBT HDS pathways and the isomerization and the
transmethylation abilities of each catalysts, the global rate constant (kHDS) of the 4,6-DMDBT
HDS reactions were calculated using Eq. 4. The rate constants of each 4,6-DMDBT HDS pathway
were also calculated. The results are listed in Table 5. All the rate constants were calculated by
-
1
-1
varying the LHSV in the range of 10 h to 160 h with the other reaction conditions fixed. To
further compare the performance of each catalyst for the 4,6-DMDBT HDS reaction, the
21
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
activation energies of each catalyst at 563 Kwere calculated according to Eq. 6 by testing reaction
-1
temperatures of 558 K and 573 K with an LHSV of 40 h and the other conditions fixed as
described earlier. The results are also listed in Table 5.
The kinetic study results in Table 5 show that the overall activation energies of the 4,6-DMDBT
-
1
-1
HDS reaction decreased from 208 KJ mol to 159 KJ mol after the addition of 10 wt.% HUSY
zeolite. Moreover, the activation energies further dropped with increasing gallium modification
-
1
content to a minimum value of 134 KJ mol over NiMo/2GaY-ASA-A, and then this trend slightly
reversed. Otherwise, the global rate constant of the 4,6-DMDBT HDS reaction (kHDS) increased
-
4
-1 -1
-4
-1 -1
from 2.8*10 mol g h to 6.1*10 mol g h after the addition of acidic cosupport and further
-
4
-1 -1
-4
increased to 8.1*10 mol g h over NiMo/2GaY-ASA-A, then dropped slightly to 7.8*10 mol
-
1
-1
g h over NiMo/3GaY-ASA-A. These results indicated that catalysts containing HUSY zeolites
exhibited higher activity than NiMo/ASA-A, and NiMo/2GaY-ASA-A exhibited the highest
activity for the 4,6-DMDBT HDS reaction. The reaction rate constants of each 4,6-DMDBT HDS
pathway were also calculated at a 4,6-DMDBT conversion of approximately 50 %, and the results
are also shown in Table 6. The results showed that the rate constant of HYD pathway increased
slightly after the addition of acidic cosupport and then remained almost the same with increased
gallium content on HUSY zeolites; however, the total rate constant of the DDS pathway over the
zeolites containing catalysts was almost 4.5 times the level of NiMo/ASA-A, remained almost
unchanged before the acidic cosupports were modified by gallium, then increased to
approximately 9 times the value of NiMo/ASA-A over NiMo/2GaY-ASA-A. The most obvious
change was observed over the ISO pathway and TRM pathway, which were only observed over
the zeolite-containing catalysts, indicating that the acidity of the acidic cosupports plays an
22
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
important role in the reaction pathways of the 4,6-DMDBT HDS process. For the
-
4
-1
-1
isomerization(ISO) pathway, the rate constant increased from 2.3*10 mol g
h
over
over NiMo/2GaY-ASA-A, then decreased slightly to
.5*10 mol g h over NiMo/3GaY-ASA-A. For the transmethylation (TRM) pathway, the rate
-
4
-1
-1
NiMo/Y-ASA-A to 4.8*10 mol g
h
-
4
-1 -1
4
-
4
-1 -1
-4
constant decreased from 0.8*10 mol g h over NiMo/Y-ASA-A to approximately 0.3*10 mol
-
1
-1
g
h
over NiMo/2GaY-ASA-A and NiMo/3GaY-ASA-A. The ISO pathway selectivity
(
kISO/푘퐻퐷푆 ) values over different catalysts were also calculated, and the results showed an
increased from 0.37 over NiMo/Y-ASA-A to approximately 0.6 over NiMo/2GaY-ASA-A and
NiMo/3GaY-ASA-A.
4
4
. Discussion
.1 Effect of gallium modification on catalyst properties
The XRD patterns revealed that gallium modification caused a slight decrease in crystallinity
(
Fig. 1 and S-Table 1) due to the impregnation of gallium species, and from the XPS results of the
Ga 3d spectra for modified HUSY zeolites (Fig. 2), we can conclude that most of the gallium
species were introduced into the Yzeolite framework through the formation of Ga-O(H)-Al bonds
(
framework substitution at the zeolite surface) and Ga-O(H)-Si bonds (framework substitution at
2 3
the zeolite surface) [44-47]. Moreover, the XPS spectra also indicated that the Ga O species was
formed with the increase of gallium modification content (Fig. 2), thus decreasing the crystallinity.
The micropore surface area and pore volume of the modified zeolites decreased with increasing
gallium content, but the mesopore surface area was maintained (Table 1), indicating that the
Ga-species plugged some micropores. However, the decrease in pore diameter (Table 1) indicated
that most Ga-species were located on the mesopore surface of HUSYcrystals to form Ga-O(H)-Al
23
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
bonds. Moreover, Johnson et al.[48] stated that the acidic properties of the support were an
important factor in determining metal dispersion. The dispersion of the metal atoms on the support
increased with higher concentrations of acid sites, particularly the strong acid sites present in
steamed zeolite HY. In our case, the amounts of both strong BAS and strong LAS decreased with
increasing gallium content because the strong acid sites on the mesopore surface and external
surface or even in super cages of zeolites were covered by gallium species, as evidenced by the
fact that most of the introduced gallium species existed as Ga-O(H)-Al. The amount of medium
LAS decreased, while the amount of medium BAS increased with gallium content. The largest
amount of medium BAS and the smallest amount of medium LAS were obtained when the gallium
content was 2 wt.%. This result indicates that gallium modification can lead to LAS transfer into
BAS on HUSY zeolites at lower gallium content, but when the gallium content increases, the
opposite transformation occurs (for 3GaHUSY). The total amount of acid sites decreased slightly
before and after HUSY modification by a relative low content of gallium nitrite (1GaHUSY, Fig.
3
Aand Table 2), but with increasing gallium content, the amounts of both BAS and LAS declined,
indicating that not only did the introduced gallium species replace some aluminum species to form
Ga-O(H)-Al and Ga-O(H)-Si bonds on the mesopore surface and outer surface of HUSY, but
2 3
based on the emergence of Ga-O-Ga bonds, bulky Ga O also formed (Fig. 2). As mentioned
earlier, the dispersion of metal atoms on the support increased at a higher concentration of acid
sites, and gallium modification can cause variation of the acidity properties and textural properties
2 3
of zeolites. For the corresponding catalysts, the strongest acid centers were covered by Ga O or
substituted by Ga-O(H)-Si sites, and because Ga-O(H)-Si sites are more covalent than Al-O(H)-Si
sites[31,35,37], the interaction between the active phase and support was decreased. This decrease
24
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
can be explained by the fact that the average electronegativity value of 1.59 for Ga is lower than
the value for Al (1.71), implying a lower acid strength of both Brønsted protons and Lewis caves
for Ga-substituted zeolites. Thus, the reduction temperature of the catalysts decreased with the
2
increase in gallium modification content, as observed from the H -TPR profiles (Fig. 4), which
indicates that the interaction between the active metals and supports was weakened. The change in
the acidic properties of the support should influence the distribution of active metals on the surface
and thus affect the active phase in the catalyst grain. Moreover, the HRTEM results demonstrated
2
that after gallium modification, the average length of the MoS slabs decreased and the average
2
stacking number of the MoS slabs increased with increasing gallium content. The value of the Mo
dispersion also increased with increasing gallium content and reached its highest value over
NiMo/2GaY-ASA-A(Table 3). This phenomenon can be explained by the confinement effect from
2 3
the highly dispersed Ga O grains doped on the surface of HUSY zeolites. As the length of the
2
MoS slabs decreases and the stacking number increases, more highly active ‘type Ⅱ Ni-Mo-S
active phase’ will be formed, which will favor both the adsorption and the conversion of
4
4
,6-DMDBT.
.2 Effect of gallium modification on the catalyst performance of 4,6-DMDBT HDS
The zeolite-containing catalysts, especially the gallium-modified zeolite-containing catalysts,
exhibited much higher HDS activity than NiMo/ASA-A. Among all the prepared catalysts,
NiMo/2GaY-ASA-A exhibited the highest 4,6-DMDBT conversion at 563 K at all LHSVs. This
difference was mainly caused by the differences in the acidity properties of the supports and the
differences in active metal morphology caused by the acidity properties. As reported by several
researchers elsewhere, it has been confirmed that acidity, especially Brønsted acidity, plays an
25
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
important role in the activity of 4,6-DMDBT conversion[4,27]. Once protonized, the methyl
groups at the 4 and 6 positions of 4,6-DMDBT can be eliminated or shifted away by dealkylation
or isomerization, reducing the hindrance caused by these methyl groups reduced and rendering the
subsequent conversion much easier. Pérot[2] and Bataille et al.[49] used Yzeolite as a component
of their HDS catalyst and employed 4,6-DMDBT as a probe molecule in their study, observing
some isomerized intermediates. A similar conclusion was drawn by Sun and Prins et al.[11] when
they used another acidic zeolite, mesoporous ZSM-5, as support and 4,6-DMDBT HDS as a
reaction probe: they observed tri-methyl-cyclopentenyl benzene as one of the HDS products and
confirmed the existence of an isomerization pathway. Moreover, when they neutralized the acidity
of ZSM-5 zeolite by piperidine, tri-methyl-cyclopentenyl benzene disappeared in the final HDS
product, proving that the acidity of the supports determined the ISO pathway during the HDS of
4
4
,6-DMDBT. Isoda et al.[50] observed that the methyl groups at the 4 and 6 positions of
,6-DMDBT could migrate to the 3 and 7 positions on a Ni/Y catalyst, drastically decreasing the
activation energy and making the HDS reaction much easier. Recently, Wu and Duan et al.[4]
reported that 4,6-DMDBT can be transformed to 3,7-DMDBT over a NiMo hydrodesulfurization
catalyst supported on an acidic zeolite containing micro-mesoporous composite zeolite. Moreover,
they found that the isomerization selectivity corresponded to the quantity of Brønsted acid sites.
The kinetic study showed that after the methyl groups at the 4 and 6 positions of 4,6-DMDBT
were transferred to the 3 and 7 positions, the reactant molecule could be adsorbed on the active
sites through both planar adsorption and σ-adsorption rather than only through planar adsorption
for 4,6-DMDBT. Thus, the global reaction rate constant kHDS of the zeolite-containing catalysts
for the 4,6-DMDBT HDS reaction increased obviously, and the activation energies of the
26
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
4
,6-DMDBT HDS reaction dropped drastically after the addition of acidic zeolites (Table 6). As
described earlier, the cleavage of the C-S bond is favored by σ-adsorption, and thus it is much
easier for 3,6- and 3,7-DMDBTs to undergo DDS route than for 4,6-DMDBT over hydrotreating
catalysts. In addition, in this work, we observed that the ISO selectivity is closely related to the
total number of medium and strong Brønsted acid sites, while the TRM selectivity is only closely
related to the number of strong Brønsted acid sites. As shown in Fig. 8, the ISO selectivity and the
number of medium and strong Brønsted acid sites exhibit similar trends, as do the TRM selectivity
and the number of strong Brønsted acid sites. Based on this theory, NiMo/2GaY-ASA-A exhibits
the largest total number of medium and strong Brønsted acid sites (Table 2), and thus 4,6-DMDBT
can be easily transformed to 3,6- and 3,7-DMDBT, which can then easily produce 3,4’-DMBP and
4
,4’-DMBP as the main products via the DDS pathway(Table 4). On the other hand, because
NiMo/2GaY-ASA-A and NiMo/3GaY-ASA-A exhibit the lowest amounts of strong Brønsted acid
sites, the smallest amounts of trimethyl- and tetramethyl-DBTs are produced. Since trimethyl- and
tetramethyl-DBTs are even more refractory sulfur-containing compounds than 4,6-DMDBT and
are even harder to transform into other products over hydrodesulfurization catalysts, the formation
of these products should be avoided. NiMo/2GaY-ASA-A exhibits the highest 3,4’-DMDBT and
4
,4’DMDBT selectivity and the lowest trimethyl- and tetramethyl-DBTs selectivity. As mentioned
previously, considering the great similarity in chemical behavior between gallium and aluminum,
Ga-OH-Si sites formed at the mesopore surface of the HUSY zeolite, and some Al-O(H)-Si sites
were substituted by Ga-O(H)-Si/Al-O(H)-Ga sites (Fig. 2); Al-O(H)-Si sites were also formed at
the interface of different components of the catalyst supports[51,52]. Thus, some L acid sites,
together with some weak and strong B acid sites, were transformed into medium B acid sites
27
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
(
Table 2)[53]. Although the gallium-modified zeolites have lower acid strength than the original
zeolites, it is easier for gallium-modified zeolites to donate Brønsted protons. Therefore, gallium
modification can improve the reaction rate of the ISO route (highest kISO value and isomerization
selectivity in Table 6) and suppress the reaction rate of the TRM route (lowest kTRM in Table 6),
ultimately enhancing the HDS of 4,6-DMDBT.
The reaction network including the HYD and DDS pathways for 4,6-DMDBT over commercial
2 3
NiMo/Al O has been studied extensively[19,23,40]. It is widely accepted that in the HYD
pathway, 4,6-DMDBT first undergoes a hydrogenation process to produce 4,6-THDMDBT, which
is immediately transformed to 4,6-HHDMDBT; then, one S atom is removed to form
3
,3’-DMCHB; or the 4,6-HHDMDBT further hydrogenated to dodecahydro-DMDBT, which will
be hydrogenated to 3,3’-DMBCH. By contrast, in the DDS pathway, the S atom is directly
removed by breaking the C-S bond to produce 3,3’-DMBP. Due to its highly refractory nature and
adsorption manner, 4,6-DMDBT HDS reaction mainly undergoes the HYD pathway over
commercial NiMo/Al
NiMo/ASA-A is 3,3’-DMCHB, which was identified as the final product of the HYD pathway;
,3’-DMBP, which was identified as the final product of the DDS pathway; and relatively small
2 3
O catalysts. In our case, the main product of 4,6-DMDBT HDS over
3
amounts of 4,6-THDMDBT and 4,6-HHDMDBT (Table 5), which were identified as the
intermediates of the HYD pathway. This result is highly consistent with the results reported
elsewhere in the literature[54]. The absence of ISO and TRM products proves that weak Brønsted
acid sites do not contribute to the isomerization and transmethylation processes. However, the
product distribution and kinetic behavior of 4,6-DMDBT HDS over the zeolite-containing
catalysts was completely different, particularly over NiMo/2GaY-ASA-A; the main products were
28
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201701307
3
,4’- and 4,4’DMBP and 3,6- and 3,7-DMDBT, which were identified as the products and
intermediates of the ISO pathway rather than 3,3’-DMCHB and 3,3’-DMBP. The high
isomerization rate constant and ISO pathway selectivity (Table 6) over NiMo/2GaY-ASA-A
confirmed that NiMo/2GaY-ASA-Aexhibits a superior methyl group migration property. Based on
the differences in product distribution, kinetic behavior and catalyst properties, we conclude that
4
,6-DMDBT was first transformed to 3,6- and 3,7-DMDBT via an isomerization pathway over the
medium and strong B acid sites, followed by removal of the sulfur atom to form 3,4’- and
,4’-DMBP via the DDS pathway over the active sites of the catalysts. In the analysis of the HDS
4
products of 4,6-DMDBT, tri- and tetra-methyl-DBTs were identified simultaneously (Fig. 7 and
S5), and the amounts of these multi-methyl-DBTs were closely related to the number of strong B
acid sites (Fig. 8). We therefore conclude that methyl migration is closely related to the Brønsted
acidity of the hydrotreating catalysts and that the migration of methyl groups on the 4 and 6
positions of 4,6-DMDBT within one molecule (isomerization) is closely related to the amounts of
medium and strong BAS, whereas the migration of methyl groups from one molecule to another
(
transmethylation) is closely related to the quantity of strong BAS.
To further understand the 4,6-DMDBT HDS over our catalysts, an overall new reaction network
including the products and intermediates of the HYD, DDS, ISO and TRA pathways (Fig. 10) was
proposed based on the kinetic behavior and product distribution. The HYD and DDS pathways of
4
,6-DMDBT are the typical reaction pathways in the 4,6-DMDBT HDS over NiMo/ASA-A. The
ISO and TRM pathways of the 4,6-DMDBT HDS reaction only occur over the zeolite-containing
catalysts. In the DDS pathway, the S atom was directly removed by hydrogenolysis to produce
3
,3’-DMBP as the product. In the HYD pathway, 4,6-DMDBT was first hydrogenated to
29
This article is protected by copyright. All rights reserved.