open ring review
TRANSCRIPT
Review
The chemistry of selective ring-opening catalysts
Hongbin Du *, Craig Fairbridge, Hong Yang, Zbigniew Ring
National Center for Upgrading Technology, 1 Oil Patch Drive, Devon, Alta., Canada T9G 1A8
Received 6 June 2005; accepted 15 June 2005
Available online 30 August 2005
Abstract
Bitumen-derived crude and heavy oils require severe processing and produce middle distillate product with poor ignition quality. This
becomes a concern to refiners as tighter specifications on transportation fuel are promulgated. One process to address this issue is selective ring
opening of cycloparaffins to reduce the number of ring structures, while retaining the carbon number of a productmolecule. The process involves
bifunctional catalysts, both metal and acid sites, working in high-pressure, high-temperature reactor systems in the presence of hydrogen. The
acidic sites catalyze dehydrogenation, cracking, isomerization and dealkylation, while the metal sites promote hydrogenation, hydrogenolysis
and isomerization. Various compounds containing single, double and multiple rings have been used to model the ring-opening reactions and a
number of mechanisms have been proposed. The five-membered ring readily undergoes ring-opening reaction on either acid or metal catalysts
with the selectivity and activity dependent on the nature of the supported metal catalyst. The ring opening of six-membered ring compounds is
secondary, requiring an acidic function to isomerize a six-membered ring cycloparaffin to a five-membered ring. A balanced metallic-acidic
function catalyst is necessary to achieve optimal performance. A system dominated by acid function results in excess cracking, while a catalytic
systemwith high concentration of metals leads to mainly hydrogenation. Commercial hydrocracking catalysts using transitionmetal sulfides on
acidic supports usually require severe operating conditions due to their low activities of the metal sulfide compared to metal sites, leading to
extensive cracking of cycloparaffin side chains. Noble metals supported on acidic oxides are the most active catalysts for selective ring opening,
but these catalysts are very sensitive to poisoning by sulfur compounds in petroleum feedstocks. An understanding of the chemistry of selective
ring-opening catalysts, combined with theoretical studies of structure–activity relationships and high throughput experimentation methods,
provides opportunities in searching for new generations of selective ring-opening catalysts with high-performance and sulfur resistance.
Crown Copyright # 2005 Published by Elsevier B.V. All rights reserved.
Keywords: Ring opening; Catalysts; Model compound; Mechanism; Bifunctional catalysts
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2. Ring-opening catalysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.1. Cracking on solid acid catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2. Hydrocracking on bifunctional catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.3. Hydrogen spillover . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3. Selective ring opening of single-ring compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.1. Activity and selectivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.2. Proposed mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.3. Electronic effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.4. Geometric effect. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.5. Support effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
www.elsevier.com/locate/apcata
Applied Catalysis A: General 294 (2005) 1–21
* Corresponding author. Present address: State Key Laboratory of Coordination Chemistry, Nanjing University, Nanjing 210093, China.
Fax: +86 25 8331 4502.
E-mail address: [email protected] (H. Du).
0926-860X/$ – see front matter. Crown Copyright # 2005 Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2005.06.033
H. Du et al. / Applied Catalysis A: General 294 (2005) 1–212
4. Selective ring opening of multiple-ring compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
4.1. Indan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
4.2. Decalin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
4.3. Tetralin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4.4. Naphthalene. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
4.5. Phenanthrene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
4.6. Fluorene and fluoranthene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
5. Sulfur tolerance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
6. Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1. Introduction
Health concerns and local environmental awareness have
prompted worldwide regulatory actions on transportation
systems to limit emissions of criteria air contaminants
(hydrocarbons, carbon monoxide, nitrogen oxides and
particulate matter). In addition, concerns for global climate
change have highlighted the need to improve transportation
system efficiency to decrease emissions of carbon dioxide.
These regulations and concerns call for new technologies to
produce high quality liquid fuels and internal combustion
engines with maximum combustion effectiveness and
minimum emissions. The regulations primarily involve
reductions of sulfur levels, aromatics (particularly benzene),
olefins, volatility and the use of oxygenates in gasoline. The
sulfur compounds in fuels contribute to the emission of
sulfates with particulate matter, and have a deleterious effect
on the ability of future automobile engines to meet the more
stringent emissions standards. Aromatics in middle-dis-
tillate fuels produce particulates in the exhaust gases and, in
addition, have poor ignition properties, i.e. low cetane
number in diesel fuel and high smoke point in jet fuel.
Across the world, gasoline and diesel specifications are
becoming stricter, especially with regard to sulfur. In the US,
the Reformulated Gasoline Program Phase II, which took
effect on 1 January 2002, mandated a reduction in volatile
organic compounds (VOC) by 27%, nitrogen oxides (NOx)
by 7%, and toxic pollutants (aromatics, especially benzene)
by 22%. In 2004, the Environment Protection Agency
started to implement the Tier 2 Gasoline Sulfur Control
Program for cleaner gasoline, which limits the maximum
sulfur level to 300 ppm, and by January 2006 to 80 ppm. The
current maximum concentration of sulfur in on-road diesel is
limited to 500 ppm and 2000–5000 ppm for off-road diesel.
By June 2006, the new EPA regulations will limit sulfur
content to 15 ppm. Additional specifications for minimum
cetane index at 40, and maximum aromatics content at
35 vol% will also be imposed.
Canada has also launched programs for a staged sulfur
reductions in gasoline that introduced the specification of
150 ppm July 1, 2002, then to 30 ppm by January 1, 2005. A
maximum of 500 ppm sulfur in on-road diesel fuel was
mandated on January 1, 1998, with 15 ppm slated for June
2006.
In Europe, the European Commission mandated a
reduction in diesel sulfur levels from 500 to 350 wt ppm in
2000, and a further reduction to 50 wt ppm in 2005. The
regulations require higher average cetane number (currently
51 and 53 by 2005) than in the US and an even higher level of
55 is under consideration and will put additional pressure on
refiners. For gasoline, the regulations require reduction of
sulfur content from 150 to 50 ppm, and aromatics from 45 to
35 vol% by 2005. Future regulations will likely tighten the
limitation for the aromatic and toxics content and increase
fuel efficiency to reduce the resultant vehicle CO2 emissions.
This will probably promote light-duty, diesel-fuelled vehicle
production and the use of ultra-low sulfur diesel (ULSD).
In the last few years, refiners in the US and Canada have
invested in major refining expansion and revamping
construction projects because of the ultra-low sulfur fuel
requirements. At the same time, they are also facing
challenges of meeting other requirements including the
replacement of methyl tertiary butyl ether (MTBE) and
reduction of aromatics [1–3]. In addition, refiners have to cope
with increasingly heavier and poorer quality feedstocks.
According to the US Energy Information Administration, the
quality of crudes processed in the US declined between 1997
and 2001 at the approximate rates of �0.128 API/year and+0.057 wt% sulfur/year. This trend is expected to continue in
the next 5–10 years, forcing US refineries to import heavier
crudes due to the worldwide increase in crude demand,
decreasedproductionof light crudes frommature regions such
as the US and North Sea, and geopolitical factors. Increased
heavy oil supplies arose from Canada and Venezuela.
With 174 billion barrels of established bitumen reserves
(recoverable in situ and mineable), the Canadian oil sands
resource has been increasingly recognized as a strategic
source of North America energy supply. Alberta bitumen
production reached approximately 1 million b/d (barrels per
day) in 2003, accounting for 53% of Alberta’s total crude oil
and equivalent production. According to the Alberta Energy
Utilities Board (AEUB), total bitumen production will
increase to 3.5 million b/d by 2017, representing 2 million b/
d of synthetic crude oil (SCO) and 1.5 million b/d of
unprocessed crude bitumen. Compared to conventional
crude oil, SCO has the valued quality attributes of low sulfur
content and zero residues. However, SCO also has some
major disadvantages, largely related to its high aromaticity.
H. Du et al. / Applied Catalysis A: General 294 (2005) 1–21 3
Fig. 1. Cetane numbers of various compounds [76].
The heavy gas oil (HGO) bottom cut from SCO (about
38 vol%) contains more than 90% of cyclic hydrocarbons
which causes processing problems, particularly in fluid
catalytic cracking (FCC) units, which constitutes approxi-
mately 80% of the US conversion capacity. This has led to its
limited acceptance in a conventional FCC refinery intake. In
addition to the poor HGO quality, SCO also yields poor
quality middle distillate (potentially jet and diesel fuel
blending components). These factors have resulted in a price
differential between SCO and other light crudes. Similar
problems exist with other highly aromatic streams, such as
those from cokers and FCC units. These potential motor fuel
components have low cetane numbers and poor ignition
qualities. This will become a particular difficulty for refiners
if higher cetane specifications are introduced worldwide in
the near future. In order to accept larger volumes of bitumen-
derived crudes, both the producers and refiners will have to
address the quality issues and refining challenges. One way
to do this is to develop new catalytic processes that will
improve the quality of these streams. There has been a
growing demand for refining catalysts driven primarily by
refiners’ needs to meet the already legislated regulatory
requirements without capital infusion. It will require new
generation of high-performance hydroprocessing catalysts
to make higher quality fuels of the future.
Hydrogenation and hydrocracking are two commercially
proven refining technologies addressing the high aromaticity
issues [4]. The conventional hydrotreating catalysts help
saturate aromatic rings to naphthenes (cycloparaffins or
cycloalkanes) [5,6]. However, the increase of cetane number
via hydrogenation of aromatics is insufficient to significantly
increase thequality of themiddle distillates because thecetane
numbers of individual cycloalkanes, such as decalin are rather
low (Fig. 1). Early studies by Wilson et al. demonstrated that
the hydrogenation or a combination of hydrogenation and
aromatics extraction of bitumen-derived middle distillates
could only produce a distillate with 40 cetane number [7–9].
However, approximately97%of thearomatic carbonhad tobe
converted by severe hydroprocessing to achieve this.
Hydrocracking is, on the other hand, a process designed to
produce lighter and higher quality naphtha and middle
distillate fuel blending stocks by saturating aromatic rings and
reducing the number of ring structures via carbon–carbon
Fig. 2. b-Scission of linear and
bond scission in the presence of hydrogen [10,11]. However,
because of its extensive overcracking there is limited net
gain in production of desirable paraffins via this route, and
the improved distillate fuel quality results primarily from a
combination of hydrogenation of aromatics and a concentra-
tion of paraffins in a reduced volume of distillate product. The
ideal process would be a selective ring-opening catalyst that
can reduce the number of ring structures while retaining the
carbon number of a product molecule.
2. Ring-opening catalysis
2.1. Cracking on solid acid catalysts
Selective ring opening of naphthenes with minimum
cleavage of side chains is very complex and represents a
challenge to catalyst researchers. It is known that ring
opening can be catalyzed by the acid sites, e.g. the Bronsted
sites, via carbenium intermediates [12]. The reaction is
initiated by protolytic cracking, accompanied by protolytic
dehydrogenation, hydride transfer, skeletal isomerization,
b-scission and alkylation, as in the cases of cracking of
aliphatics over acid catalysts. The latter mechanisms have
been extensively studied [13]. The cracking of endocyclic
C–C bonds in cyclic hydrocarbons, however, is much slower
than those of aliphatics, presumably because of a higher
tendency of the alkenyl cation, formed by b-scission of a
cycloalkyl cation, to recyclize, or because of a lower b-
scission rate in cylcoalkylcarbon ions caused by an
unfavorable orientation of the p-orbital at the positively
charged carbon atom and the b-bond to be broken (Fig. 2)
cyclic hydrocarbons [14].
H. Du et al. / Applied Catalysis A: General 294 (2005) 1–214
[14]. As a result, the overall reaction is predominated by
isomerization and subsequent hydrocracking (b-scission) of
side chains of cyclic hydrocarbons, particularly of those
having substituents with more than five carbon atoms,
leading to significant dealkylation of pendant substituents on
the ring. The yields of desired high cetane ring-opened
products are usually low due to a high extent of consecutive
cracking reactions and a fast catalyst deactivation.
2.2. Hydrocracking on bifunctional catalysts
Starting in the 1950s, bifunctional catalysts consisting of
highly dispersed metal particles for hydrogenation or
dehydrogenation and an acidic support for cracking or
isomerization were introduced in refining processes, such as
catalytic reforming of naphtha, hydrotreating, hydrocrack-
ing and hydroisomerization [2]. These processes improved
fuel quality and at the same time avoided catalyst decay and
the formation of coke deposits. Early studies by Mills et al.
[15] and by Weisz and Swegler [16] described the reactions
over the bifunctional catalysts as proceeding over two
distinct catalytic sites: the reactants are first converted into
olefins on the metal site via hydrogenation/dehydrogenation
reactions (Eq. (1)); then the formed olefins are protonated at
the acid sites, leading to the formation of carbenium ions,
which subsequently undergo skeletal isomerization, crack-
ing or alkylation (Eq. (2)). The products are finally desorbed
from the acid sites as olefins, which are hydrogenated on the
metal sites in the presence of hydrogen (Eq. (3)). It was later
recognized that the reactions could also occur on one
reaction site, owing to activated hydrogen species, i.e.
spillover hydrogen [17]. The metal co-catalyst provides
spillover hydrogen, which migrates to the acid sites and
saturates the carbenium intermediates. In other words, the
acidic support cannot only initiate the formation of
carbenium ions pyrolytically or by the addition of protons
to olefins, but also hydrogenate carbenium ions with hydride
ions and promote their desorption as the saturated products.
In addition, the ring opening of naphthenes can also proceed
on certain metal sites via direct hydrogenolysis of an
endocyclic C–C bond [12], i.e. cleavage of a C–C bond with
the addition of hydrogen. Overall, the activity and selectivity
of ring opening on bifunctional catalysts are strongly
dependent on the properties of the metal and support as well
as on reaction conditions. These parameters include the type
of metal, metal particle size, acidity of the support, pore size
of the support, the interface length and strength balance
between the metal and acid sites, working conditions, such
as temperature, hydrogen pressure, etc. This has provided
opportunities to devise catalysts for highly selective ring
opening of naphthenic compounds in fuel without significant
dealkylation of any pendant substituents on the ring.
(1)
(2)
(3)
2.3. Hydrogen spillover
The classical model for bifunctional catalysis proposed
byMills et al. [15] and byWeisz and Swegler [16] envisaged
the reaction in three steps (Eqs. (1)–(3)), involving a gas
phase diffusion of olefinic intermediates between the two
catalytic sites. The model successfully explained a number
of experimental observations, but failed to account for the
role of hydrogen and the synergy between the two catalyst
components in controlling the selectivity and activity [17].
The extension of the classical model by incorporating the
hydrogen spillover concept allows a better interpretation of
experimental results, including those that did not fit into the
old theorem. The new model involves the formation of
mobile hydrogen surface species, i.e. spilt-over hydrogen
that allows the hydroconversion of the hydrocarbon at a
certain distance from the metal so that all reaction steps can
occur on one reaction site.
Spillover is now a well-known phenomenon in hetero-
geneous catalysis, involving the transport of active species
sorbed or formed on one surface onto another that does not
sorb or form the active species under the same conditions.
Several small molecules have been known to exhibit
spillover effects upon interactions with noble metals,
including hydrogen, oxygen, nitrogen, carbon monoxide
and organic species [18]. Hydrogen spillover plays an
important role in petroleum processes, e.g. in hydrotreating,
hydrocracking, hydrogenation and hydroisomerization.
A number of excellent reviews on this subject have been
published [17–21].
The nature of the spilt-over hydrogen species has been
discussed in the literature. Depending on the system studied,
different species have been claimed, including H atoms, H
ions, ion pairs, H3 species and bound species. Roland et al.
[18] proposed a model that describes the spilt-over species
as electron donors adsorbed on the surface, which
corresponds to H atoms (occupied, weakly chemisorbed)
or H+ ions (empty, strongly chemisorbed). Their ratio is
determined by the chemical properties (e.g. the presence of
Lewis and Bronsted acid sites) and the electronic properties
(e.g. electron density) of the solid. In bifunctional catalysts,
spilt-over hydrogen can donate an electron to the support to
form a proton on a Bronsted site, and a hydride ion may form
on a Lewis site in the process of charge balancing [13]. An
olefin formed on the metal by dehydrogenation can react
H. Du et al. / Applied Catalysis A: General 294 (2005) 1–21 5
with the proton on a Bronsted site to form a carbenium ion,
which may then combine with the migrated hydride ion from
a Lewis site to produce a saturated compound. Alternatively,
the carbenium ion formed by addition of a proton to an olefin
on a Bronsted site can migrate to a Lewis site where it reacts
with a hydride ion.
The creation of catalytic active sites by spilt-over
hydrogen has been recognized and used in supported metal
catalysts for hydrogenation (or dehydrogenation), hydro-
cracking and hydroisomerization. For instance, spillover of
hydrogen from a metal under certain conditions can make
the inert silica, or alumina, or carbon support active for
hydrogenation of olefin or benzene [22–27]. A number of
studies showed that mechanical mixtures of supported metal
catalyst and a support exhibited higher hydrogenation
activities than the supported metal catalyst alone. The
increased activities have been attributed to hydrogen
spillover. Without activation by hydrogen spillover, neither
silica nor alumina will adsorb hydrogen and both are inert
for hydrogenation catalysis.
Several studies have implied that hydrogen spillover can
give rise to Bronsted acid sites on oxide supports, including
zeolites. Fujimoto [26] found that mechanically mixed
catalyst Pt/SiO2 + H-ZSM-5 showed both high selectivity
and high conversion for benzene hydrogenation, equal to
those obtained over a Pt/ZSM-5 catalyst, while Pt/SiO2 or
H-ZSM-5 alone were not effective catalysts. He attributed
the activity of the mechanical mixture to the generation of
Bronsted acid sites by hydrogen spillover. Ohgoshi et al.
[27] demonstrated that the hybrid catalysts Pt/KA + NaYor
H-ZSM-5 showed high activity for isobutene hydrogena-
tion, while the Pt/KA catalyst and zeolite Y or H-ZSM-5
alone failed to hydrogenate isobutene because isobutene is
too big to fit into the pores of KA zeolite (3 A), and H2
cannot be activated on NaY or H-ZSM-5 zeolite. They
concluded that H2 was activated on Pt/KA. The resulting H
atoms spilt-over onto the NaYor H-ZSM-5 zeolite, and then
hydrogenated isobutene on the zeolite. Further study using a
H-ZSM-5 zeolite that contained almost no SiOH groups
mixed with Pt/KA yielded zero conversion of isobutene,
indicating the OH groups play an important role for
migration of spilt-over hydrogen. In the bifunctional CoMo/
zeolite hydrocracking catalysts, Co-Mo adsorbs H2 and
Fig. 3. Non-selective and selective hydr
supplies spilt-over hydrogen to acidic sites, where it forms
acidic hydroxyl groups and promotes acid-catalyzed
reactions [21].
3. Selective ring opening of single-ring compounds
3.1. Activity and selectivity
When adsorbed on metal surface at certain conditions,
hydrocarbon molecules undergo dehydrogenation/hydroge-
nation, skeletal isomerization and hydrogenolysis, including
ring opening/cyclization, and ring contraction/enlargement.
Certain noble metals, such as Pt, Pd, Ir, Ru and Rh have been
found to be selectively active for the ring opening for cyclic
hydrocarbons to the corresponding paraffins with the same
carbon number. The activity and selectivity depend mainly
on the metal catalysts, such as the type of metal, particle
size, crystal morphology, etc. The ring opening of cyclic
hydrocarbons on noble metals is highly sensitive to the
catalyst structure, particularly alkylcyclopentanes and
alkylcyclobutanes, which have been used as molecular
probes to characterize various types of catalysts. Several
excellent reviews have been published on this subject
[14,28–33].
The ring opening of methylcyclopentane (MCP) on
supported metal catalysts has been extensively studied
[30,34–37]. The reaction produces n-hexane (nHx), 2-
methylpentane (2MP) and 3-methylpentane (3MP). The
product distribution is dependent on the properties of the
metal. Over supported Pt catalysts with small particle sizes,
e.g. low loading and highly dispersed Pt, the ring opening of
MCP is non-selective (the rupture of endocyclic C–C bonds
is statistical, producing 2MP, 3MP and nHx in a ratio of
2MP:3MP:nHx = 40:20:40 related to the number of bond
types in the molecule) (Fig. 3). On the other hand, the ring
opening of MCP on large metal particles and metal surfaces
with low Miller index produces selectively 2MP and 3MP,
but no nHx. In some cases, the ring opening does not follow
the ‘selective’ or ‘non-selective’ mechanism, instead
producing unusually high amount of nHx or 3MP relative
to its statistical ratio. These ‘partially selective’ mechanisms
compete with the non-selective and selective mechanism,
ogenolysis of methylcyclopentane.
H. Du et al. / Applied Catalysis A: General 294 (2005) 1–216
occurring on high loaded 10% Pt/Al2O3 at high-temperature
[30], or Pt/zeolite [38–40].
The ring opening of MCP on Ir [12] and Rh [41], on the
other hand, is less sensitive to the particle size of the metals
in comparison with Pt catalysts. However, these metals
exhibit higher activity and selectivity to open the C5 ring in
bisecondary positions [36,42]. Other metal catalysts, such as
Co, Ni, Ru and Os, show extensive hydrogenolysis, yielding
a significant amount of fragments [41,42].
Generally, the ring-opening activities of alkylcyclopen-
tanes on metal catalysts decrease with increased number of
ring substitutents [34]. This is especially the case for Ir, and
to lesser extent Ru, Rh and Ni [36], all of which show a
preference for cleaving unsubstituted ring C–C bonds. A
significant decrease in activity was observed for cleaving
substituted ring C–C bonds over these metals. In compar-
ison, Pt is better able to break substituted ring C–C bonds
than Ir, Ru, Rh and Ni. However, the ring-opening rate over
Pt catalysts is sensitive to the cis-/trans-isomer ratio,
decreasing with increasing trans-isomer concentration. In
other words, Pt requires a cis-substitution of isomers when
breaking a tertiary–tertiary C–C ring bond.
The reaction conditions have also been found to affect the
product distribution. Changes in selectivity and activity of
MCP ring opening on supported Pt metal catalysts with
hydrogen pressure have been reported [34,43]. Higher
hydrogen pressures favor ring opening further from the
substituents. The increase in hydrogen partial pressure leads
to increased ring-opening activity on supported Pt catalysts at
relatively low temperatures (<300 8C). This reaches a
maximum and thereafter shows a steady decline, or a very
broad plateau with further increase in hydrogen partial
pressure. The hydrogen pressure at which the maximum
activity is obtained is a function of the support. However,
these volcano-type plots of ring-opening activity versus
hydrogen pressure are independent of the support and the
reduction temperature. The latter determines the particle size,
morphology and structure of themetal particle on the support,
and influences the activity. These resultswere attributed to the
competitive adsorption of hydrogen and hydrocarbon on the
metal surface. Increase in hydrogen pressure diminishes
dehydrogenation activity, leading to an increase in the overall
selectivity towards the ring-opening reaction.
Hydrogen pressure dependence on activity and selectivity
of ring opening has also been observed over other transition
metal catalysts [41,42,44–47]. The relationships between
the product formation rate and hydrogen pressure varies with
catalytic systems, reflecting differences in reaction mechan-
ism. Over Rh catalysts [45,46], the ring opening of 1,2-
dimethylcyclopropane goes through a maximum, and then a
minimum and then a continuous increase along with the
increase in hydrogen pressure. On Pd, Ni and Cu catalysts
[45], the formation rates of one product increases and levels
off with increasing hydrogen pressure, while for the other
product the curve passes through a maximum. These
observations were attributed to the changes of formation of
adsorbed intermediates at different hydrogen pressures. A
dissociative adsorption of hydrocarbon via the rupture of the
C–H bond(s) was assumed at low hydrogen pressure. An
increase in hydrogen pressure inhibits this C–H dissociation,
slowing down the reaction rate. In the case of Rh catalyst,
new adsorbed species were presumably formed via scission
of the ring C–C bond at high hydrogen pressure. These
species were desorbed to produce ring-opening compounds.
Compared to the five-membered ring cyclopentanes, the
ring opening of the four-, and six-membered ring
compounds are less extensively studied. Cyclobutanes are
more reactive towards ring opening than cyclopentanes due
to higher ring strain. The reactions can take place at low
temperature (<100 8C), exhibiting relatively low selectivity,
i.e. statistical product distribution [44,46]. On the other
hand, the ring opening of cyclohexanes over metal catalysts
[36,48,49] is much slower than five-membered rings, even
over the most active metal Ir [36]. Instead, cracking and
aromatization dominate the total reaction. For a significant
conversion of six-membered rings into alkanes without loss
of molecular weight, additional acid function is required to
convert the six-membered rings to the five-membered rings
via skeletal isomerization. The latter then readily undergo
selective ring opening on acid or metal sites.
3.2. Proposed mechanism
There have been several mechanisms proposed to account
for ring opening on metal surfaces, two of which are well
recognized: the multiplet mechanism and the dicarbene
mechanism [30,33]. The two mechanisms differ mainly in
how the reaction intermediate forms on the metal surface.
In the multiplet mechanism, cyclic hydrocarbons
physically adsorb either edgewise on two metal atoms, or
flat-lying on the metal surface. The former is called the
‘doublet’ mechanism, and the latter called the ‘sextet-
doublet’ mechanism. Both mechanisms compete with each
other in the reaction. The doublet mechanism is thought to
occur on small metal particles, and is used to explain the
selective hydrogenolysis of bisecondary C–C bonds.
According to this mechanism, cyclic hydrocarbon is
adsorbed perpendicular to the metal surface via a
bisecondary C–C bond, which then reacts with chemisorbed
hydrogen in a push–pull manner to produce ring-opening
products. Due to steric hindrance, the edge-wise adsorption
of the tertiary–secondary or tertiary–tertiary C–C bonds is
limited. In the ‘sextet-doublet’ mechanism, on the other
hand, the cyclic molecule is physically adsorbed flat-lying
on the metal surface, with the carbon atoms of the ring
located over the interstices of the metal plane, e.g. Pt(1 1 1).
For the five-membered ring cyclopentanes, one C–C bond
has to be stretched (Fig. 4), and this bond is readily attacked
by neighboring, adsorbed hydrogen, leading to hydrogeno-
lysis of the ring. The tertiary–secondary C–C bonds in
alkylcyclopentane could be ruptured via the five-atom
adsorbed mode. For cyclobutanes, all four C–C bonds are
H. Du et al. / Applied Catalysis A: General 294 (2005) 1–21 7
Fig. 4. Hydrogenolysis of methylcyclopentane via the multiplet mechanism [30].
stretched, resulting in higher reactivity but lower selectivity.
In contrast, there is no stretching when cyclohexanes and
paraffin adsorb on the metal surface, as all the carbon atoms
can fit the interstices of the Pt(1 1 1) plane. As a result,
cyclohexanes and paraffin are usually not hydrogenolysized,
but preferably dehydrogenated or isomerized.
Alternatively, the dicarbene mechanism involves the
chemisorption of the cyclic hydrocarbon molecules on the
metal surface after the rupture of several C–H bonds, forming
carbon–metal bonds or p-adsorbed olefins (Fig. 5). The non-
selective ring opening of MCP on metal catalysts can be
accounted for by thep-adsorbed olefinmechanism. Similar to
that in the sextet-doublet mechanism, MCP is adsorbed
parallel to the metal surface, but in a quasi-planar manner
involving only one metal atom. The selective ring opening of
MCP, on the other hand, involves 1,2-dicarbene complexes
that bond to several metal atoms and stand perpendicular to
the metal surface. Owing to steric hindrance, the hydro-
genolysis of the tertiary–secondaryC–Cbonds is retarded.On
some occasions, however, an exocyclic alkyl substituent
participates in the formation of a metallocyclobutane
intermediate, resulting in the selective breaking of tertiary–
tertiary and tertiary–secondary C–C bonds (Fig. 5b).
Fig. 5. Hydrogenolysis of methylcyclopenta
3.3. Electronic effect
As various mechanisms were proposed to account for
experimental facts such as selectivity, structural effect and
kinetic data, the nature of the catalyst–reactant interactions
remains unknown. In ring opening (cyclization), and
isomerization catalysis, metallocyclobutanes, metallocar-
benes and metallocarbynes were postulated as the possible
reaction intermediates, depending on the nature of the sites.
To generalize and simplify the reaction mechanisms, Garin
and Maire [29] proposed an agostic precursor as the first
species adsorbed on the metal surface, which consists of an
M H–C with a hydrogen atom bonded simultaneously to
both a carbon atom of a reactant and a transition metal atom
(Fig. 6). This species initiates the formation of a s-alkyl or
carbene species that further reacts to give rise to products of
ring opening (cyclization), hydrogenation (dehydrogena-
tion) and/or isomerization. It was proposed that an
electronic (or ligand) effect governs the relative contribu-
tion of s-alkyl or carbene precursor, while a geometric (or
ensemble size) effect determines the pathway to the bond
shift (isomerization), the cyclic, and the hydrogenolysis
reactions.
ne via the dicarbene mechanism [30].
H. Du et al. / Applied Catalysis A: General 294 (2005) 1–218
Fig. 6. Proposed reaction mechanism of n-C5H12 on a metal catalyst,
involving an agostic precursor species [29].
Fig. 7. Simplified reaction mechanism for the catalytic reforming of
hydrocarbons on transition metal surfaces [54].
In hydrocracking/isomerization of 2-methyl- and 3-
methylpentane on several group VIII metal catalysts [29],
studies have shown that Co and Ni tend to cleave multiple
bonds, leading to extensive cracking, while Pd, Pt and Ir
selectively rupture primary–secondary/tertiary, secondary–
secondary/tertiary (demethylation) and secondary–second-
ary C–C bonds, respectively. The differences in the
selectivity and activity of metals in hydrogenolysis and
isomerization may be related to their electronic structure
(density of states), as proposed by Saillard and Hoffmann
[50]. When a H2 or a hydrocarbon molecule (e.g. CH4)
approaches a metal surface, the electron transfer from d
orbitals of the metal M to a C–H s* antibond (M! s*)
dominates the early stages of the reaction, which weakens
the C–H s bond and forms the M–H bond. This is different
from transition metal complexes in which s!M electron
transfer leads the reaction because of the higher energy of
the occupied metal orbitals on the surface. For transition
metals at the left side of the periodic table, the surface is
positive relative to the bulk, while for metals at the right side
of the periodic table, the metal surface is negative. This is in
accordance with the fact that the heats of chemisorption of
hydrogen on metal surfaces decrease from left to right across
a periodic row [50,51]. These metals are more likely to form
metallocarbynes or carbenes, leading to multiple C–C
rupture. Within the same series of transition metals, e.g.
group VIII metal, the charge transfer from the bulk to the
surface decreases as d level occupancy for a metal surface
increases in traversing a periodic row [50,52]. Consistently,
the heats of chemisorption for hydrogen atoms on the metal
surface decrease in going from 3d to 4d to 5d (i.e. decrease
with increased atomic weight).
Based on the product distribution, hydrogenolysis
reaction on metal catalysts can be classified into two groups
[29]: the C2-unit mode, involving rupture of C–C bonds
between primary and secondary carbon atoms; and the iso-
unit mode, involving rupture of a C–C bond with a tertiary
carbon atom. In the C2-unit mode, metallocarbene species
may be formed upon removing at least two hydrogen atoms,
which lead to hydrogenolysis over isomerization. On the
other hand, in the iso-unit mode where only one H atom is
available on the tertiary carbon, s-alkyl species are formed,
leading to higher isomerization than hydrogenolysis. The
relative contribution of the two modes is determined by the
electronic effect, i.e. the nature of the metal catalyst.
The C2-unit mode is favored on monometallic Ru, Ir and
bimetallic Pt-Ru, Pt-Co, Pt-Ir and Pt-Mo catalysts. The C2-
unit mode-dominated reactions, e.g. catalyzed by Ir catalysts,
were thought to proceed via a dicarbene (or doublet)
mechanism. The reactant bonds perpendicularly to the metal
surface. The reactions are usually not sensitive to the particle
size of metals but to the number of substituents because of
steric hindrance. On the other hand, the iso-unit mode is
dominant onNi, Pt, Pd, Pt-Ni, Pt-WandPt-Pd catalysts. In this
case, the reactant is adsorbed parallel to themetal surface, and
the reaction proceeds via a multiplet mechanism. The particle
size of metal influences the selectivity of the reactions.
It is now widely accepted that the reactions of
hydrocarbon on metal are initiated by dissociative adsorp-
tion of the alkanes on the metal surface and followed by
subsequent hydride elimination from adsorbed alkyl
intermediates. The former is the rate-determining step in
the reaction, while the latter is thought to determine the
selectivity. Zaera [53–55] proposed that b-hydride elimina-
tion from the surface intermediate accounts for the
production of olefins, g-hydride elimination is responsible
for isomerization and cyclization, and a-hydride elimination
leads to hydrogenolysis products (Fig. 7). The relative rates
for a-, b- and g-dehydrogenations determine selectivity of
metal catalysts in hydrocarbon reforming. In general, the
reactivity for a-, b- and g-dehydrogenations increases with
early transition metals and with decreasing hydrocarbon
H. Du et al. / Applied Catalysis A: General 294 (2005) 1–21 9
chain lengths. The b-hydride elimination is the most
favorable. However, the relative rates for a-, b- and g-
dehydrogenations within the metal also change across the
periodic table, which results in the different catalytic
performance of different metals in hydrocarbon reforming.
For instance, Pt(1 1 1) displays comparable rates for a- and
g-dehydrogenations, while Ni(1 0 0) only shows a-dehy-
drogenation. These results explain the unique ability of Pt
for catalyzing hydrocarbon reforming as opposed to Ni,
which facilitates extensive cracking.
3.4. Geometric effect
The geometric effect on the catalytic properties is
exemplified by the particle size effects, which have been
extensively studied [28,30,31]. In hydroconversion of
hydrocarbons, large metal particle size promotes selective
hydrogenolysis, and small particle size favors non-selective
isomerization and hydrogenolysis. For instance, ring open-
ing on supported Pt metal catalysts exhibit pronounced
particle size effects on the selectivity. As illustrated by the
hydrogenolysis of MCP [56,57], catalysts with large Pt
particle size lead to selective ring opening, favoring 2MP,
and 3MP with the formation of nHx suppressed. Small Pt
particle size produces non-selectively nHx, 2MP and 3MP in
a statistical ratio. Other metals, e.g. Rh and Ir, favor the
selective hydrogenolysis with little effect of particle size.
The particle size has also affected the turnover rate
(activity) in the hydrogenolysis of cyclic hydrocarbons
[30,31]. On Pt/Al2O3, the turnover rate of the hydrogeno-
lysis of cyclopentane was shown to decrease with decreased
particle size. Similar results were observed on Rh/Al2O3. On
Ir/Al2O3, however, the hydrogenolysis of cyclopentane is
insensitive to the particle size. Pd/Al2O3, on the other hand,
presented a slight increase in the hydrogenolysis of
cyclopentane with decreasing particle size.
The origin of particle size effects is not clear. Some
possible explanations include electronic effects, morphol-
ogy, and support effects [28,30,31]. As the particle size is
reduced, the electronic bands of the metal particle become
distinct, and the electron energies increase. For most metal
particles, the electron energies begin to increase as the
particle size is reduced below 5 nm. For some metals, the
energies level off below a certain size, e.g. Pd at about 1 nm,
while for others, e.g. Pt, the increase in energies continue
with the decrease in particle size. As a result, the electronic
configurations of the surface atoms are different from those
of large particles, leading to changes in the catalytic
properties. On the other hand, the decrease in particle size
likely changes the morphology of the metal particle, which
could influence the catalytic properties. For instance, a large
fcc metal particle crystallizes in an octahedron geometry (8
faces), while most small fcc metal particles (d = 10 nm)
adopt cubooctahedral geometry (14 faces) as demonstrated
by quantum-mechanical calculations and experimental
studies. For even smaller particles a non-fcc arranged
icosahedron (20 faces) might become more stable [31]. As
the particle size decreases, the number of atoms of low
coordination at the edges or corners of the crystallites
increases, while the fraction of face atoms diminishes.
Therefore, small particles would favor catalysis by atoms of
low coordination, e.g. the non-selective, statistical hydro-
genolysis, giving higher rates. Large particles would favor
catalysis by face atoms of high coordination numbers, e.g.
selective hydrogenolysis. The effect of surface atom
coordination on the catalytic properties of the metal can
also account for the changes in the selectivity of the ring
opening of MCP over different metal surfaces. In controlled
tests using metal films with oriented faces [58,59] or
supported metal catalysts where the specific facets of the
metal particles were selectively prepared [41,60], studies
showed that the hydrogenolysis activity and product
selectivity varied from facet to facet of metal crystal. High
activities and non-selective hydrogenolysis were observed
on small polycrystalline particles with disordered surfaces,
or on flat, stepped, and kinked noble-metal single-crystal
surfaces of high Miller index containing high concentrations
of low-coordination sites. This particle size phenomenon
observed in heterogeneous catalysis is the basis of the
current interest in nanotechnology related to chemistry and
materials.
3.5. Support effect
Besides the above-mentioned electronic and geometric
effects, catalyst supports also play an important role in
heterogeneous catalysis. As demonstrated in many catalytic
processes, the supports are not inert in selective ring-
opening catalysis as previously thought, but alter the
catalytic properties of the metal particles through electronic
interactions, or by space confinements. The support effect
has attracted much attention and stimulated extensive
research worldwide. Excellent reviews have been published
on this subject [32,33,61,62].
A ‘partly’ selective mechanism for ring opening of MCP
was observed on noble metals supported on acidic oxides,
which produced a higher proportion of nHx than the
theoretical ratio, according to a non-selective mechanism.
The formation of nHx appeared to increase with increasing
support acidity [33,63,64]. Kramer and Zuegg [57] used the
adlineation model, coined by Schwab and Pietsch [65], to
account for the partial selectivity for nHx and 2MP. The
model assumed that ring opening occurs at the phase
boundary of metal and support. The support, i.e. an acidic
site, interacts with the tertiary C–H of MCP because this H
atom is most susceptible to attack by an acidic or cationic
site, while the neighboring carbon is attached to the metal
site. Ring opening of MCP adjacent to the methyl group
occurs, leading to the formation of additional nHx (Fig. 8).
This partly selective mechanism competes with a non-
selective mechanism occurring exclusively on the metal
surface. Alternatively, the partly selective ring opening of
H. Du et al. / Applied Catalysis A: General 294 (2005) 1–2110
Fig. 8. Partly selective ring opening of methylcyclopentane.
MCP may be explained by electronically asymmetric pairs
of metal atoms resulting from strong metal–support
interactions. The cationic sites of the support in bifunctional
catalysts exert an electric field on adjacent metal sites,
leading to the formation of partially charged metal atoms
that have been confirmed by various methods [33,66]. The
resulting Md�–Md+ exhibit functionalities similar to those of
the acid site–metal pair to selectively produce nHx and 2MP
but not 3MP (Fig. 8) [63].
Another ‘partly’ selective mechanism for ring opening of
MCP has been observed over zeolite-supported metal
catalysts, for example, on SAPO-11 [67], LTL [38,39,43]
and FAU [43,68]. Over these catalysts, significantly higher
selectivities to 3MP were observed compared to other
supported metal catalysts of similar particle size. These
enhanced selectivities were attributed to the constraint of the
one-dimension pores of zeolites that result in a preferred
orientation of the incoming MCPmolecule with its long axis
parallel to the direction of the pores. When the MCP
molecule reaches a metal particle inside a zeolite pore, ring
opening preferentially takes place through one of its ends,
resulting in a 3MP/2MP product ratio higher than the
statistical value of 0.5 (Fig. 8). Similarly, a higher 3MP/nHx
ratio than the statistical value of 0.5 can be explained by the
steric hindrance of the methyl group of MCP in zeolite pores
that restrict the rotation of the molecule when it approaches
the metal particle with its methyl end.
4. Selective ring opening of multiple-ring
compounds
Compared to the number of studies using mono C4, C5
and C6 naphthenic rings, there are fewer studies on ring
opening of more complex molecules containing multiple
rings. Recently, model compounds containing two fused
rings, such as indan, decalin, tetralin and naphthlene have
been used to model the upgrading of heavy oil fractions.
Unlike ring opening of alkylcyclopentanes, which can
readily occur at low temperature by hydrogenolysis on noble
metal catalysts, ring opening of fused C6 naphthenic rings
requires the addition of an acidic function to promote the
isomerization of the C6 rings into the more reactive C5 rings
for achieving reasonable ring-opening activity. The ring-
opening reactions are complex, usually accompanied by
hydrogenation, isomerization and cracking.
4.1. Indan
The indan molecule consists of a C5 ring fused to a
benzene ring, and represents a probable reaction intermediate
of reactions of heavy oil fractions on ring-opening catalysts.
The ring opening of indan was recently evaluated using noble
metal catalysts supported on boehmite by Nylen et al. [69].
They found that the hydrogen pressure plays a key role in the
reaction. At atmospheric pressure the ring opening of the C5
ring was dominant, leading to 2-ethyltoluene (60–70 mol%)
and minor amounts of n-propylbenzene (7–10%) (Eq. (4)).
The Ir and Pt-Ir catalysts showed superior catalytic activities
to the Pt catalyst. Pt was shown to selectively open the C5
naphthenic ring into 2-ethyltoluene with a selectivity
approximately seven times higher than for n-propylbenzene,
while Ir displays relatively high cracking activity. At high-
pressure (40 bar) and low-to-moderate temperature, hydro-
genation of indan into hexahydroindan was favored.
Increasing temperature resulted in increasing both the ring
opening and subsequent undesired cracking products.
(4)
4.2. Decalin
The six-member hydrocarbon rings are more stable than
five-member rings, with the ring strain energies of about
1 kcal/mol and 6–7 kcal/mol, respectively. McVicker et al.
[36] showed that the ease of ring opening of two-ring
naphthenes is directly related to the relative number of
saturated five-member rings in the molecule. Therefore, the
ring opening of decalin (4.4% conversion on 0.9% Ir/Al2O3)
with two saturated six-member rings is much slower than
perhydroindan (68%) with one six- and one five-member
ring, and bicyclo[3,3,0]octane of two five-member rings. An
acidic function in the catalyst has to be used to achieve a
significant amount of ring-opening products by promoting
the isomerization of six-member ring structures into the five-
member ones.
On acidic catalysts, the direct ring opening of decalin was
thought to proceed via cleavage of the corresponding
carbonium ion, which is initiated by the attack of a Bronsted
acid site on a C–C bond of decalin [70]. The alkylnaphthene
carbenium ion can further undergo b-scission and/or
isomerization to form lighter naphthenes or gaseous
molecules. Meanwhile, bimolecular reactions may occur,
H. Du et al. / Applied Catalysis A: General 294 (2005) 1–21 11
Fig. 9. Proposed mechanism for direct ring opening over acid catalyst [70]. PC, pyrolytic cracking; ISO, isomerization; TA, transalkylation; b, b-scission; HT,
hydride transfer.
such as hydrogen transfer, transalkylation or disproportio-
nation to produce various naphthenic species, as shown in
Fig. 9. Alternatively, ring opening of decalin can take place
as a result of isomerization (ring contraction), which
produces five-member ring hexahydroindan. In other words,
the ring-opening products of decalin originated from the
isomerization products. Indeed, stereo- and skeletal iso-
merizations of decalin were found to dominate the reaction
of decalin on proton-form zeolites including medium-pore
ZSM-5, MCM-22, large-pore HY, H-Beta, H-Mordenite,
UTD-1 and extra-large-pore H-MCM-41 [71,72]. The
decalin isomers were then consumed by consecutive
reactions of ring opening and cracking, yielding more than
200 products. Stereoisomerization of decalin was thought to
occur on a Bronsted site via a pentacoordinated carbocation
(Fig. 10b), or carbenium ion formed by protonic dehy-
drogenation, or proton exchange, or on Lewis acid sites by
hydride abstraction (Fig. 10c). The cis-decalin is much more
Fig. 10. Isomerization of cis-decalin to trans-decalin [71].
reactive than trans-decalin, and converts much more
selectively to ring-opening products than trans-decalin.
The latter mainly converts to cracking products. The skeletal
isomerization was initiated by ring contraction of decalin
carbocations or carbenium ions to methylbicyclo[4,3,0]no-
nanes, followed by their isomerization or further ring
contraction (Fig. 11). The isomers can undergo further
isomerization to yield a complicated mixture of isomers.
Among those isomers, methylbicyclo[4,3,0]nonanes,
dimethylbicyclo[3,2,1]octanes, dimethylbicyclo[3,3,0]oc-
tanes and methylbicyclo[3,3,1]nonanes are the most
abundant. The pore size of zeolites influences the isomer
product distribution, while the acidity and temperature do
not affect the selectivity of isomerization. Decalin hardly
penetrates the channels of the medium-pore zeolites (ZSM-
5, MCM-22); these zeolites are much less active than large-
pore zeolites as the reaction takes place only on external
surfaces [70]. In addition, the pore topology of zeolite has a
strong influence on diffusion and, consequently, on activity
and selectivity. As a result, dimethylbicyclooctanes were
Fig. 11. Proposed isomerization reaction network for decalin [71].
H. Du et al. / Applied Catalysis A: General 294 (2005) 1–2112
Fig. 12. Proposed mechanism for isomerization and ring opening of decalin over acidic catalyst [71]. PC, pyrolytic cracking; ISO, isomerization; DS,
desorption; b, b-scission; HT, hydride transfer.
found to be more abundant over H-Y than over H-Beta
zeolites, while more methylbicyclononanes are formed over
Beta zeolites [71]. The isomers are then converted into ring-
opening and cracking products. The ring strain and the
number of alkyl substituents in the isomers affect the ring
opening. Strain in the hydrocarbon ring results in ring
opening following the order C3 > C4� C5 � C7� C6.
The isomers with more alky substitutes are more prone to
ring opening by acid because these compounds consist of
more tertiary carbon atoms that more easily form carbenium
ions than the secondary carbon. As a result, the main ring-
opening products on large-pore zeolites HY, H-Beta, and H-
Mordenite [71] are propyl-ring-opened products (ROP)
(from methylbicyclo[4,3,0]nonanes), followed by ethyl-
ROP (from dimethyl bicycle[3,2,1]octane with 1 or 2 methyl
groups on the five member ring), methyl-ROP (from
dimethyl bicycle[3,2,1]octane with 2 methyl groups on
the six member ring), ethylpropyl-ROP and butyl-ROP. The
different isomer product distributions on different zeolites
are reflected in the ring-opening product (ROP) distribu-
tions, i.e. relatively more ethyl-ROPs and less propyl-ROPs
are observed over H-Y zeolites in comparison with H-Beta.
Fig. 13. Proposed mechanism for isomerization and ring opening of decalin over
desorption; b, b-scission; HT, hydride transfer; HYG, de-/hydrogenation; RO, ri
Amechanism for decalin isomerization and ring opening has
been proposed (Fig. 12), involving pyrolytic dehydrogena-
tion, pyrolytic cracking, hydride transfer, b-scission and
olefin desorption and adsorption.
Introduction of noble metal onto acidic zeolite catalysts
reduces the strength of Bronsted acid sites, and significantly
enhances isomerization and ring opening of decalin [72,73],
while the cracking is reduced. Stereosiomerization of cis-
decalin to trans-decalin proceeds on a metal site via the
dehydrogenation–hydrogenation mechanism (Fig. 10a), in
addition to the stereoisomerization facilitated solely by acid
sites (Fig. 10b and c). The skeletal isomerization occurs on
Bronsted acid sites as shown in Fig. 13, which is an essential
step for the ring opening of decalin. In the absence ofBronsted
acid sites, neither isomerization nor ring-opening reactions
occur. The main isomerization products are methylbicy-
clo[4,3,0]nonanes and dimethylbicyclo[3,2,1]octanes, simi-
lar to those on proton-form zeolites [70,71]. However, the
distributions of ring-opening products over Pt-modified
zeolites are different from those obtained on H-zeolites.
The dominant ring-opening products are methyl-cyclohex-
anes, followed by ethyl-cyclohexanes. Ring opening yields of
bifunctional catalyst [73]. PC, pyrolytic cracking; ISO, isomerization; DS,
ng opening.
H. Du et al. / Applied Catalysis A: General 294 (2005) 1–21 13
as high as 30 mol%were obtained; in comparison just 8 mol%
of ring-opening products were achieved on H-zeolites. The
differences were attributed to the addition of Pt. In addition to
ring opening initiated byBronsted acid sites as observedonH-
zeolites, Pt particles participated in the ring-opening reaction:
(1) to form olefin by dehydrogenation of decalin isomers or
ring-opening products, which were then protonated over
Bronsted acid sites and underwent isomerization and ring
opening; (2) to directly open rings of isomers and (3) to
provide spillover hydrogen to suppress the secondary
reactions (e.g. cracking) and to prevent the catalyst
deactivation.
4.3. Tetralin
Ring opening of tetralin via cracking on acidic supports
produced mainly C1 to C4 fractions, benzene, naphthalene
and C10 olefin/naphthene aromatics (mainly methylindanes
and naphthalene) [70,72]. Their relative selectivity varied
from one zeolite to another depending on pore topology and
size. Extensive cracking dominated on zeolites with medium
pore sizes (ZSM-5, MCM-22, ITQ-2) to form C3 and C4
fractions. Large pore zeolites (Beta, Y) were able to open the
naphthenic ring and minimize the dealkylation of the formed
alkylaromatics. Extra-large pore zeolites (UTD-1) and
mesoporous MCM-41 favored ring opening. A mechanism
was proposed to account for the reaction, involving the
attack of a Bronsted acid site either on the aromatic ring or
on the naphthenic ring (Fig. 14). The resulting carbenium
and carbonium ions underwent isomerization (ring contrac-
tion), b-scission and protolytic cracking to form ring-
opening products and isomerization products such as
Fig. 14. Proposed reaction networks for catalytic cracking of tetralin
methylindanes and methylindenes. Subsequently, dealkyla-
tion occurred to produce cracking products, while dis-
proportionation between carbenium and olefins led to heavy
products and coke. Fast hydride transfer between tetralin and
the adsorbed carbenium ions was attributed to the formation
of naphthalene, which was found to be one of the primary
products.
On bifunctional catalysts, ring opening of tetralin was
believed to occur exclusively on Bronsted acid sites, while
metal sites hydrogenated tetralin to decalin and promoted its
isomerization (ring contraction) which is a crucial step in the
ring opening of decalin as described in the previous section.
Hydrogenation of tetralin to decalin was found to dominate
on supported noble metal catalysts, especially at low
temperature [72,74–79]. The hydrogenation of tetralin was
reported to take place not only on metal centers but also on
acid sites by hydrogen spillover from the metal surface. The
formed decalin underwent skeletal isomerization, and then
subsequent ring-opening and cracking reactions. The rate of
isomerization, and hence the ring-opening rate increased
with the decrease in the distance between the metal and acid
sites and with the increase in metal loading. The ring-
opening product yields and product distribution were related
to the topology of zeolite structure, i.e. diffusional
limitations. The acidity, on the other hand, shifted the
optimal (maximum yield) reaction temperature for ring
opening. Mild acidity favored ring opening.
Supported noble metal catalysts are very sensitive to
sulfur compounds. Efforts have been made to improve the
sulfur resistance of noble metal-based catalysts for
hydrogenation. For instance, bimetallic catalysts, e.g. Pd-
Pt, have demonstrated moderate sulfur tolerance in
[70]. Iso, isomerization; PC, pyrolytic cracking; b, b scission.
H. Du et al. / Applied Catalysis A: General 294 (2005) 1–2114
hydrogenation [74,77,79]. However, the influence of sulfur
on the activity for ring opening of naphthenes is still unclear.
Rodriguez-Castellon et al. [74] found that the ring opening
and cracking of tetralin on Pt-Pd catalysts supported on
zirconium doped mesoporous silica were suppressed in the
presence of sulfur compounds, but not as much as
hydrogenation.
Ring opening of tetralin on transition metal catalysts, e.g.
NiMo and NiW supported on acidic materials, such as
silica–alumina or zeolites has been investigated [80–86].
These catalysts are able to catalyze the ring-opening reaction
of naphthalenes in the presence of sulfur compounds. Such
catalysts are usually used at severe operating conditions,
such as high hydrogen pressure and high-temperature, due to
their relatively low activities. Unlike noble metal catalysts,
the loading of NiW onto zeolites increases the amounts of
Bronsted acid sites [80,85]. However, the additional acid
sites themselves do not catalyze hydrocracking of tetralin.
The increased conversion of tetralin over pure zeolites can
be attributed to the bifunctionality. Sato et al. [80–82]
reported hydrocracking of tetralin over bifunctional NiW
sulfide catalysts supported on zeolites USY, HY and
mordenite (MOR) under typical hydrocracking conditions
at moderate temperature (623 K) and high hydrogen
pressure (6.1 MPa) with a low H2/feed ratio. They found
that the ring opening of tetralin required relatively strong
acid sites, and was related to the hydrogen transferability of
supports: NiW/USY > NiW/HY > NiW/MOR > NiW/
Al2O3. The NiW/Al2O3 catalyst functions mainly as a
hydrogenation catalyst, yielding decalin with small propor-
tions of heavy compounds, while zeolites and NiW/zeolite
catalysts produced a complex mixture of more than 300
Fig. 15. Proposed bimolecular mechanism for catalytic
compounds, consisting of products of cracking (alkane
gases, benzene), hydrogenation (decalin), isomerization
(decalin isomers), ring opening (alkylbenzene, monocyclo-
paraffins), dehydrogenation (naphthalene) and alkylation
(alkyltetralin, tricyclic compounds). NiW was found to
hydrogenate aromatic or olefinic compounds that were
produced in the reactions, and to supply spillover hydrogen
to the acid sites. Due to an insufficient supply of hydrogen
with NiW sulfide, hydrocracking of tetralin proceeded
mainly via a bimolecular pathway (Fig. 15), in addition to a
unimolecular mechanism, at the initial reaction stage,
leading to the formation of heavy compounds. In the long
run, the unimolecular process was dominant, and the heavy
compounds were gradually converted to light products. A
careful balance between the metal active sites of hydro-
genation and acid sites of cracking was necessary for
optimal performance for selective ring opening. Under
conditions with high hydrogen/tetralin ratios, on the other
hand, hydrocracking of tetralin proceeds via a monomole-
cular mechanism over transition metal or metal sulfide
catalysts [83–85], producing cracking compounds (volatiles,
benzene, toluene, xylene), ring opening (n-propylbenzene,
n-butylbenzene), isomerization (indan), hydrogenation
(decalin) and dehydrogenation (naphthalene) compounds.
No products heavier than decalins were observed [83–85].
4.4. Naphthalene
Hydrogenation of naphthalene to tetralin is dominant in
the hydroconversion of naphthalene over bifunctional
catalysts at low temperatures [40,87–95]. The hydrogena-
tion rate of naphthalene to tetralin is an order of magnitude
cracking of tetralin [81]. A, acid, DS, desorption.
H. Du et al. / Applied Catalysis A: General 294 (2005) 1–21 15
Fig. 16. Hydrogenation networks for naphthalene over bifunctional cata-
lysts [88]. M, metal sites; A, acid sites; Hsp, spilt-over hydrogen.
higher than that of tetralin to decalin, due to the strong
adsorption of naphthalene on the metal surface [87].
Because of a thermodynamic limitation, the hydrogenation
conversions decreased with the increase in reaction
temperature, while the isomerization and ring-opening
reaction increased [91]. In a bifunctional catalyst system,
the hydrogenation takes place via conventional metal-
catalyzed hydrogenation, or acid site induced hydrogena-
tion, involving migration of spillover hydrogen from metal
sites (Fig. 16). The latter was thought to account for the
increased hydrogenation activity of metal catalysts on acidic
supports [88,94].
Ring opening of naphthalene is a secondary reaction,
converting saturated and/or isomerized products on metal or
Bronsted sites. Albertazzi et al. [90] showed that Pd(4)-Pt(1)
supported on Mg/Al mixed oxide, which is a basic support
with no Bronsted acid sites, produced an appreciable amount
of ring opening (18 mol% yield), mainly alkycyclohexanes,
at moderate conditions around 300 8C. In the presence of
acid sites, the yield of ring opening increased up to 30 mol%
[91].
The ring opening increased with the increase in reaction
temperature. However, consequent cracking reactions of
ring-opening products are favored at high temperatures
(>375 8C), leading to lighter products. The optimal
Fig. 17. Hydroconversion networks for 1-methylnaphthalene over bifunctional ca
transfer; HYG, de-/hydrogenation.
operating temperature exists, e.g. at ca. 350 8C for ring
opening of 1-methylnaphthalene over Pt/USY [40].
The ring-opening product yields are also dependent on
the Bronsted acidity. Arribas and Martinez [40] studied the
influence of zeolite acidity for the coupled hydrogenation
and ring opening of 1-methylnaphthalene on Pt/USY
catalysts with varied aluminum content (i.e. acidity). They
found that selective ring opening was favored over zeolites
with a low density of Bronsted acid sites. Strong and high
density of Bronsted acid sites lead to extensive cracking and
dealkylation. They also found that the selective ring opening
of 1-methylnaphthalene increased with decreasing space
velocity and increasing total pressure with both cases
exhibiting a maximum for ring-opening conversion. The
unimolecular mechanism was proposed for the hydrocrack-
ing of 1-methylnaphthalene (Fig. 17).
4.5. Phenanthrene
Few studies on selective ring opening of three-ring cyclic
compounds have been reported. Several early studies were
carried out to elucidate the reaction networks for hydro-
conversion of three- or four-ring compounds on bifunctional
catalysts under hydrocracking conditions. Nonetheless,
these studies shed insights into understanding of the ring
opening of these model compounds. The studies are
summarized as follows.
Phenanthrenewas considered a suitable model compound
for hydrogenation and hydrocracking of liquefied coal over
bifunctional catalysts. Lemberton and Guisnet [96] carried
out a model test of hydroconversion of phenanthrene over
NiMo sulfide supported on alumina at 430 8C, 10 MPa of H2
in a stainless steel autoclave. They found that phenanthrene
hydroconversion proceeded through multi-step hydrogena-
tion, isomerization and cracking reactions. Phenanthrene
initially was hydrogenated into dihydrophenanthrene, which
in turn was hydrogenated into tetrahydro- and octahydro-
phenanthrenes. But further hydrogenation into perhydro-
talysts [40]. ISO, isomerization; DS, desorption; b, b-scission; HT, hydride
H. Du et al. / Applied Catalysis A: General 294 (2005) 1–2116
Fig. 18. Proposed mechanism for phenanthrene hydroconversion [96]. ISO, isomerization; DS, desorption; b, b-scission; HT, hydride transfer.
phenanthrene was not observed. On NiMo/Al2O3 with only
weak acidity, ring opening took place through the opening of
the saturated central ring of phenanthrene, yielding mainly
2-ethyl biphenyl, or the opening of a saturated terminal ring,
yielding mainly butyl naphthalene, methylpropyl naphtha-
lene and probably diethyl naphthalene. The former was
proposed to occur on an acid site, initiated by an acid attack
to a benzene ring (Fig. 18), via cleavage of aromatic C–
secondary C bond rather than cleavage of the 9–10 C–C
bond because of the formation of stable conjugated
carbocation (2-ethyl biphenyl). The ring opening of the
saturated terminal ring, on the other hand, was thought to
proceed via a bifunctional mechanism, involving isomer-
ization of the C6 ring into the C5 ring, followed by ring
opening of the isomers on acid sites. Cleavage of aromatic
C–secondary C bond was highly disfavored because of the
involvement of an unstable primary carbocation.
Similar results were obtained by Korre et al. [97] on a
presulfided NiW/USY catalyst at 350 8C, 68.1 atm of H2 in a
batch autoclave. They presented detailed kinetic studies on
the hydrocracking of phenanthrene. A reaction network
consisting of hydrogenation, isomerization, ring opening
and dealkylation was established (Fig. 19), which was then
used to obtain various reaction rates, equilibrium and
adsorption parameters. Di- and tetrahydrophenanthrenes
were found as the primary products. The latter was
hydrogenated further to octahydrophenanthrene, but with
no perhydrophenanthrene observed. Terminal ring saturated
tetra- and octahydrophenanthrenes then underwent isomer-
ization and subsequent ring opening, while the ring-opening
products (dimethyl- or ethylbiphenyls) of central ring
saturated dihydrophenanthrene were not observed. The
ring-opening conversion of the latter was thought to proceed
via reversible dehydrogenation/hydrogention to the terminal
ring saturated compounds. Kinetic studies showed that
hydrogenations of terminal aromatic rings were favored over
those of the central ring, and isomerization of sym-
octahydrophenanthrene was favored over those of asym-
octahydrophenanthrene. Isomerizations and ring opening
preferably occurred at position a to an aromatic ring or a
tertiary carbon because of the ability of the parent structure
with tertiary carbons or aromatic rings to stabilize a
carbocation intermediate. Therefore, isomerizations proceed
in the following orders: tetrahydrophenanthrene > sym-
octahydrophenanthrene > asym-octahydrophenanthrene >tetralin. Similar trends have been observed for ring opening
and dealkylation.
4.6. Fluorene and fluoranthene
The hydrogenation of fluorene initially occurred over
bifunctional catalysts [98]. The primary product hexahydro-
fluorene was then converted by isomerization, ring opening
and dealkylation. However, isomerization of hexahydrofluor-
ene took place in parallel with ring opening, and the ring
opening of hexahydrofluorene and its isomers occurred
through the central five-member ring (Fig. 20). In the presence
of a strong acid (e.g. on NiMo/zeolite Y), ring opening of
hexahydrofluorenes occurred preferably in parallel with
isomerization and dealkylation, yielding mainly cyclohexyl-
methylenebenzene and cyclohexyltoluene.
The hydroconversion of fluoranthene follows a similar
mechanism. Initial hydrogenation of fluoranthene on NiMo/
zeolite Y yields tetrahydrofluoranthene as the major primary
product, which is then converted by cracking and
isomerization to produce mainly 1-phenyltetralin, 2-phe-
nyltetralin, 2-phenlymethylindan, tetralin and benzene. Ring
opening of the five-member ring of tetrahydrofluoranthene
immediately followed the hydrogenation of fluoranthene
due to its high ring strain. The formed 1-phenyltetralin can
isomerize to the thermodynamically more stable isomer 2-
phenyltetralin. 1-Phenyltetralin can recombine with a proton
H. Du et al. / Applied Catalysis A: General 294 (2005) 1–21 17
Fig. 19. Phenanthrene hydroconversion networks over a presulfided NiW/
USY catalyst [97]. Relative pseudo-first order reaction constants are shown.
to form a carbenium ion, which rearranges to a more stable
benzylic carbocation by 1,5- and/or 1,3-hydride transfer
from the saturated ring. The latter is subject to a
hydrogenolytic cleavage of the C–C bond between two
aromatic rings to give tetralin and benzene. 2-Phenyltetralin,
on the other hand, does not form benzylic carbocation
because of unfavorable hydride transfer in this conforma-
tion. As a result, isomerization of 2-phenyltetralin took
place, yielding 2-phenylindan. The reaction networks for
hydrogenation and ring opening of fluoranthene are
summarized in Fig. 21.
5. Sulfur tolerance
The discussion thus far has considered ring-opening
reactions in model compounds where noble metal catalysts
are the most active catalysts for selective ring opening of
naphthenes. However, these catalysts are very sensitive to
sulfur compounds in real hydrocarbon feedstocks, which
normally need a severe hydrotreating pretreatment to reduce
the sulfur concentration to below a few ppm. The application
of these catalysts would thus be limited by such severe
pretreatment conditions, unless the sulfur tolerance pro-
blems could be solved. It is believed that the sulfur tolerance
of a noble metal catalyst is related to the electron density of
metal clusters. The sulfur poisoning is mainly due to the fact
that adsorption of H2S decreases the metal–support
interaction, which promotes platinum migration and leads
to a growth of platinum particle size [100]. Therefore, the
sulfur tolerance can be increased by decreasing Pt electron
density and/or enhancing metal–support interactions. By
choosing tetralin hydrogenation as a model reaction for
aromatics reduction, Chiou et al. [101] investigated the
effect of sulfur on the deactivation of g-Al2O3 supported Pt
catalysts. They found that the reversible reaction scheme is
unable to explain the decreased reactivation rate with the
severity of sulfur poisoning. TPR, fast FT-IR and electron
probe microanalysis (EPMA) suggested that the interactions
between CO and Pt, and H2 and Pt were weakened due to the
increase in sulfur poisoning. A sintering of platinum
particles occurred, promoted by the H2S adsorption at the
Pt–alumina interface, which may have decreased the Pt–
alumina interaction.
It is possible to improve sulfur resistance through the
addition of a second metal component to Pt or Pd catalyst,
i.e. via bimetallic interaction. There are well-documented
examples that alloying of metals can increase the activity,
selectivity and stability in catalytic reactions [102]. The
formation of the bimetallic phase could inhibit the Pt or Pd
migration. In addition, electron transfer from Pt or Pd to the
second metal would increase the sulfur tolerance of the
catalyst. Fujikawa et al. [103] performed screening tests of
various SiO2-Al2O3 noble metal catalysts (Pt, Pd, Re, Sn, Ir,
Ni, Mo and Ge) with hydrotreated light cycle oil (LCO)/
straight-run light gas oil (SRLGO) feedstocks containing
32–34 vol% aromatics and 172–474 ppm sulfur at a
temperature of 573 K, H2 pressure of 4.9 MPa, and LHSV
of 1.5 h�1. They found that the Pt-Pd/SiO2-Al2O3 catalyst
was the most highly active catalyst for aromatic hydro-
genation under the applied conditions. The co-existence of
Pt with Pd on SiO2-Al2O3 remarkably enhanced the catalytic
activity, which depended on the Pd/(Pt + Pd) weight ratio
and reached a maximum at about 0.7 Pd/(Pt + Pd) weight
ratio. TEM studies indicated that all active metals form Pt-
Pd bimetallic agglomerates on the catalyst. Long-term
stability tests demonstrated the excellent stability of the Pd-
Pt/SiO2-Al2O3 catalyst. Similar results have been reported
by Lin et al. [100]. The supported Pt catalysts on g-Al2O3
were modified by adding a secondmetal, such as Co,Mo, Ni,
Re, Ag and Pd. The results showed that Pd-Pt catalyst had
the highest sulfur resistance. Long-term stability tests over
Pt and Pd-Pt catalysts indicated the catalyst exhibited much
H. Du et al. / Applied Catalysis A: General 294 (2005) 1–2118
Fig. 20. Postulated mechanism for fluorene hydroconversion [98]. ISO, isomerization; DS, desorption; b, b-scission.
Fig. 21. Postulated mechanism for fluoranthene hydroconversion [99].
better catalytic performance for hydrogenation of diesel
feedstocks containing 28.4% aromatics, 369 ppm sulfur,
44.8 ppm nitrogen under the following commercial diesel
hydrotreating conditions: WHSV 3.0 h�1, H2 to oil mole
ratio 2.5, temperature 340 8C, pressure 580 psig. Pd-Pt
bimetallic interaction played a crucial role in improving
sulfur resistance. Fast FT-IR studies of CO adsorption
suggested that electrons are transferred from Pt to Pd by
bimetallic interaction, inhibiting the adsorption of H2S. In
addition, it was inferred that the Pd inhibited agglomeration
of Pt particles during hydrogen regeneration.
Another approach to enhance the sulfur resistance of
noble metal catalysts is to fine tune the metal–support
interaction. It has been reported that the sulfur resistance of
Pt catalyst can be increased by using acidic support, such as
SiO2-Al2O3 or zeolites. The interaction between the strong
acid site and the small cluster of noble metal results in the
electrons being withdrawn from the noble metal, thus
creating an electron-deficient metal particle [89,104–107].
This decreases the strength of the bonding interaction
between sulfur and metal and increases the sulfur tolerance
of the catalyst. Song and Schmitz [89] studied the
hydrogenation of naphthalene in n-tridecane at 200 8C in
the absence and presence of benzothiophene over mordenite
and zeolite Y-supported Pt and Pd, compared with Al2O3-
and TiO2-supported Pt and Pd catalysts. Both zeolite-
supported catalysts are substantially more active than the
Al2O3- and TiO2-supported catalysts. The presence of
benzothiophene decreases the activity of all the catalysts.
However, there are significant improvements in sulfur
tolerance of the noble metals when supported on zeolites.
Mordenite-supported Pd catalyst has shown the best activity
and the highest sulfur resistance [89].
6. Concluding remarks
Currently, there is increasingly interest in selective ring-
opening catalysis that can produce middle distillate with
high cetane number with bitumen-derived crude and heavy
H. Du et al. / Applied Catalysis A: General 294 (2005) 1–21 19
oils. Ring opening of cycloparaffins can be catalyzed via
carbenium intermediates by protolytic cracking on the
Bronsted sites. The cracking of endocyclic C–C bonds in
cyclic hydrocarbons is much slower than those of aliphatics.
As a result, the overall reaction is predominated by
isomerization and subsequent hydrocracking (b-scission)
of side chains of cyclic hydrocarbons, particularly of those
having substituents with more than five carbon atoms,
leading to significant dealkylation of pendant substituents on
the ring. The yields of desired high cetane ring-opened
products are usually low due to a high extent of consecutive
cracking reactions and a fast catalyst deactivation.
On the other hand, the ring opening of cycloparaffins can
proceed on certain metal catalysts via direct hydrogenolysis
of an endocyclic C–C bond, i.e. cleavage of a C–C bond with
the addition of hydrogen. The reaction is accompanied by
dehydrogenation/hydrogenation, skeletal and stereo isomer-
ization. The activity and selectivity depend mainly on the
type of metal, particle size, crystal morphology, etc. Certain
noble metals, such as Pt, Pd, Ir, Ru and Rh have been found
to be selectively active for the ring opening for cyclic
hydrocarbons to the corresponding paraffins with the same
carbon number. However, these metal catalysts lack activity
for ring opening of the six-member hydrocarbon rings.
The presence of the acid function is necessary for
selective ring opening of six-membered ring naphthenes in
feedstocks. The acid function promotes the formation of
intermediate carbenium ions, which isomerize into five-
membered rings that readily undergo subsequent ring
opening on acid or metal sites. The metal function supplies
spilt-over hydrogen to the acid sites to saturate the
intermediate carbenium ions and prevent the formation of
coke. The acid number, strength and distance between the
acid site and the metal site strongly influence the catalytic
performance. Insufficient supply of spilt-over hydrogen
leads to mainly acidic cracking as in the monofunctional
catalyst, and favors coke formation, whereas in a catalytic
system with spilt-over hydrogen in a sufficiently high
concentration, hydrogenation dominates, and admixing of a
metal-free component can further increase the conversion.
The optimal catalytic system requires a well-balanced
metallic-acidic function. In selective ring-opening catalysis,
the preferred acid function isomerizes six-membered ring
naphthenes to five-membered ring naphthenes with the
minimum number of ring substituents [36]. Such non-
branching ring contraction allows maximal ring-opening
rates and product selectivities and minimal undesired
hydrocracking and secondary acyclic paraffin isomerization
reactions.
The commercial process for selective ring opening
involves bifunctional catalysts, both metal and acid sites,
working together in high-pressure, high-temperature reactor
systems in the presence of hydrogen. The acidic sites
catalyze dehydrogenation, cracking, isomerization and
dealkylation, while the metal sites promote hydrogenation,
hydrogenolysis and isomerization. The widely available
catalysts using transition metal sulfides on acidic supports
usually require severe operating conditions due to their low
activities of the metal sulfide compared to the metal sites,
leading to extensive cracking of cycloparaffin side chains.
Noble metals supported on acidic oxides are highly active
catalysts for selective ring opening, but these catalysts are
very sensitive to poisoning by sulfur compounds in
petroleum feedstocks. There is a need to develop new
generations of selective ring-opening catalysts with high-
performance and sulfur resistance that could meet the future
increasing cetane specification.
The search for such catalysts through the conventional
trial-and-error method would be tedious. During the last
decade, rapid development of modern computing techniques
and computational chemistry have enabled one to run
theoretical calculations at a relatively low cost, and at the
same time allowed for the accurate descriptions of the
molecular structures, spectroscopy and predictions of
chemical reactivity. The application of computational
chemistry in heterogeneous catalysis has grown rapidly in
recent years, which has helped to better understand the
reaction mechanism, to explain the experimental data, and to
devise new catalysts. For instance, quantum theoretical
calculations of electron densities of alloy catalysts, aided by
modern characterization techniques, such as X-ray photo-
spectroscopy, scanning tunnel microscopy, XANES, TPD,
etc., have provided insights into understanding the origin of
alloying effects of the metal catalysts. These studies have
yielded qualitative relationships of the chemisorption bond
strength of the reactants on the metal catalysts and catalytic
activity [28,53,102,108]. Recently, Jacobsen et al. [109]
found that the catalytic activity for ammonia synthesis is a
function of nitrogen adsorption energy on the surfaces of the
various metals, which exhibits a typical volcano curve. The
ideal catalyst should possess an optimal binding energy; the
metals that bind nitrogen either too weakly or too strongly
are poor catalysts. Guided by the theoretical studies, the
groups found theoretically, and confirmed experimentally,
that the alloys Co–Mo, Fe–Ru, Fe–Co and Ni–Mo were
more active than pure Co, Mo, Ru or Fe. Similar strategies
have been applied to search for hydrodesulfurization
catalysts from sulfur binding energy–activity correlations
[110–117], and hydrogenation catalysts from hydrogen
dissociation energy–activity correlations [118]. It is
expected that such strategies will provide opportunities in
searching for new metal catalysts with high-performance
and sulfur resistance for selective ring-opening catalysis.
In comparison, hydrogenolysis of C–C bonds in a
hydrocarbon is much more complex, and is usually
accompanied by dehydrogenation, isomerization, cycliza-
tion, etc. Nonetheless, Zaera [53–55] discovered a relation-
ship between the selectivity of metal catalysts in
hydrocarbon reforming and the ability of metal to catalyze
a-, b- and g-dehydrogenations from chemisorbed surface
alkyl intermediates. The latter varies across the periodic
table, and can be correlated to temperature-programmed
H. Du et al. / Applied Catalysis A: General 294 (2005) 1–2120
desorption of alkyl halides on the metal surface. The
temperature-programmed desorption data, which corre-
spond to the temperature maxima for the desorption of
methane, ethylene or iso-butane via a-, b- or g-hydride
eliminations, respectively, indicate the degree of difficulty of
the dehydrogenation reactions. Metals with the ability to
promote a-hydride elimination facilitate hydrogenolysis in
hydrocarbon reforming. Such data, though scarce and
scattered in the literature, have yielded some useful insights
into the hydrocarbon reforming catalysis. Research into this
direction and establishment of quantitative structure–
activity relationships could provide opportunities to discover
new ring-opening catalysts that could meet future fuel
specifications.
High throughput experimentation methods, invented in
the pharmaceutical industry for fast discovery of new drugs
has attracted a lot of attention in the catalysis research field
[119]. Many research institutes and companies have adopted
this technology when searching for new catalysts or for
improving existing catalyst performance. This method
allows the synthesis and catalyst testing of tens and even
hundreds of materials at one time, greatly speeding up the
catalyst development process. Such techniques will provide
new opportunities in searching for bi- or multi-metallic
metal catalyst with high-performance and sulfur resistance
for selective ring-opening catalysis, and/or improving
hydrocracking catalysts with well-balanced acid and metal
functions. With the guidance of today’s advanced computa-
tional methods, the search will be even more markedly
shortened by zeroing in a limited number of promising
candidates.
Acknowledgements
Partial funding for the National Centre for Upgrading
Technology (NCUT) has been provided by the Canadian
Program for Energy Research and Development (PERD),
the Alberta Research Council (ARC) and the Alberta Energy
Research Institute (AERI). The authors thank Mr. Norman
Sacuta for proofreading the manuscript.
References
[1] T.G. Kaufmann, A. Kaldor, G.F. Stuntz, M.C. Kerby, L.L. Ansell,
Catal. Today 62 (2000) 77.
[2] C. Marcilly, J. Catal. 217 (2003) 47.
[3] S. Rossini, Catal. Today 77 (2003) 467.
[4] A. Nishijima, T. Kameoka, T. Sato, N. Matsubayashi, Y. Nishimura,
Catal. Today 45 (1998) 261.
[5] J. Barbier, E. Lamy-Pitara, P. Marecot, J.P. Boitiaux, J. Cosyns, F.
Verna, Adv. Catal. 37 (1990) 279.
[6] A. Stanislaus, B.H. Cooper, Catal. Rev. Sci. Eng. 36 (1994) 75.
[7] M.F. Wilson, J.F. Kriz, Fuel 63 (1984) 190.
[8] M.F.Wilson, I.P. Fisher, J.F. Kriz, Ind. Eng. Chem. Prod. Res. Dev. 25
(1986) 505.
[9] M.F. Wilson, I.P. Fisher, J.F. Kriz, Energy Fuels 1 (1987) 540.
[10] G. Valavarasu, M. Bhaskar, K.S. Balaraman, Pet. Sci. Technol. 21
(2003) 1185.
[11] J.W. Ward, Fuel Process. Technol. 35 (1993) 55.
[12] L.B. Galperin, J.C. Bricker, J.R. Holmgren, Appl. Catal. A 239
(2003) 297.
[13] K.A. Cumming, B.W. Wojciechowski, Catal. Rev. Sci. Eng. 38
(1996) 101.
[14] J.Weitkamp, S. Ernst, H.G. Karge, Erdol undKohle Erdgas 37 (1984)
457.
[15] G.A. Mills, H. Heinemann, T.H. Milliken, A.G. Oblad, Ind. Eng.
Chem. 45 (1953) 134.
[16] P.B. Weisz, E.W. Swegler, Science 126 (1957) 31.
[17] F. Roessner, U. Roland, J. Mol. Catal. A 112 (1996) 401.
[18] U. Roland, T. Braunschweig, F. Roessner, J. Mol. Catal. A 127 (1997)
61.
[19] W.C. Conner Jr., J.L. Falcone, Chem. Rev. 95 (1995) 759.
[20] B. Delmon, G.F. Froment, Catal. Rev. Sci. Eng. 38 (1996) 69.
[21] B. Delmon, Solid State Ionics 101–103 (1997) 655.
[22] K.M. Sancier, J. Catal. 20 (1971) 106.
[23] S.T. Srinivas, P. Kanta Rao, J. Catal. 148 (1994) 470.
[24] S. Ceckiewicz, B. Delmon, J. Catal. 108 (1987) 294.
[25] P. Antonucci, N.V. Truong, N. Giordano, R. Maggiore, J. Catal. 75
(1982) 140.
[26] K. Fujimoto, in: T. Inui, K. Fujimoto, T. Uchijima, M. Massi (Eds.),
Stud. Surf. Sci. Catal., vol. 77, Elsevier, Kyoto, 1993, p. 9.
[27] S. Ohgoshi, I. Nakamura, Y. Wakushima, in: T. Inui, K. Fujimoto, T.
Uchijima, M. Masai (Eds.), Stud. Surf. Sci. Catal., vol. 77, Elsevier,
Kyoto, Japan, 1993, p. 289.
[28] B. Coq, F. Figueras, Coord. Chem. Rev. 178–180 (1998) 1753.
[29] F. Garin, G. Maire, Acc. Chem. Res. 22 (1989) 100.
[30] F.G. Gault, Adv. Catal. 30 (1981) 1.
[31] M. Che, C.O. Bennett, Adv. Catal. 36 (1989) 55.
[32] G.L. Haller, D.E. Resasco, Adv. Catal. 36 (1989) 173.
[33] K. Hayek, R. Kramer, Z. Paal, Appl. Catal. A 162 (1997) 1.
[34] H. Zimmer, Z. Paal, J. Mol. Catal. 51 (1989) 261.
[35] Y. Zhuang, A. Frennet, Appl. Catal. A 134 (1996) 37.
[36] G.B. McVicker, M. Daage, M.S. Touvelle, C.W. Hudson, D.P. Klein,
W.C. Baird Jr., B.R. Cook, J.G. Chen, S. Hantzer, D.E.W. Vaughan,
E.S. Ellis, O.C. Feeley, J. Catal. 210 (2002) 137.
[37] M. Chow, S.H. Park, W.M.H. Sachtler, Appl. Catal. 19 (1985) 349.
[38] G. Jacobs, F. Ghadiali, A. Pisanu, A. Borgna, W.E. Alvarez, D.E.
Resasco, Appl. Catal. A 188 (1999) 79.
[39] W.E. Alvarez, D.E. Resasco, J. Catal. 164 (1996) 467.
[40] M.A. Arribas, A. Martinez, Appl. Catal. A 230 (2002) 203.
[41] D. Teschner, L. Pirault-Roy, D. Naud, M. Guerin, Z. Paal, Appl.
Catal. A 252 (2003) 421.
[42] D. Teschner, K. Matusek, Z. Paal, J. Catal. 192 (2000) 335.
[43] M. Vaarkamp, P. Dijkstra, J. van Grondelle, J.T. Miller, F.S. Modica,
D.C. Koningsberger, R.A. van Santen, J. Catal. 151 (1995) 330.
[44] B. Torok, M. Bartok, J. Catal. 151 (1995) 315.
[45] I. Palinko, J. Catal. 168 (1997) 543.
[46] B. Torok, I. Palinko, A. Molnar, M. Bartok, J. Catal. 159 (1996) 500.
[47] D. Teschner, Z. Paal, D. Duprez, Catal. Today 65 (2001) 185.
[48] F. Figueras, B. Coq, C. Walter, J.-Y. Carriat, J. Catal. 169 (1997) 103.
[49] G. Onyestyak, G. Pal-Borbely, H.K. Beyer, Appl. Catal. A 229 (2002)
65.
[50] J.-Y. Saillard, R. Hoffmann, J. Am. Chem. Soc. 106 (1984) 2006.
[51] E. Schustorovich, R.C. Baetzold, E.L. Muetterties, J. Phys. Chem. 87
(1983) 1100.
[52] E. Schustorovich, R.C. Baetzold, J. Am. Chem. Soc. 102 (1980)
5989.
[53] F. Zaera, Catal. Lett. 91 (2003) 1.
[54] F. Zaera, J. Phys. Chem. B 106 (2002) 4043.
[55] F. Zaera, Appl. Catal. A 229 (2002) 75.
[56] R. Kramer, H. Zuegg, J. Catal. 85 (1984) 530.
[57] R. Kramer, H. Zuegg, J. Catal. 80 (1983) 446.
[58] G. Rupprechter, K. Hayek, H. Hofmeister, J. Catal. 173 (1998) 409.
H. Du et al. / Applied Catalysis A: General 294 (2005) 1–21 21
[59] D. Kalakkad, S.L. Anderson, A.D. Logan, J. Pena, E.J. Braunsch-
weig, C.H.F. Peden, A.K. Datye, J. Phys. Chem. 97 (1993) 1437.
[60] L. Pirault-Roy, D. Teschner, Z. Paal, M. Guerin, Appl. Catal. A 245
(2003) 15.
[61] J.T. Miller, B.L. Mojet, D.E. Ramaker, D.C. Koningsberger, Catal.
Today 62 (2002) 101.
[62] M. Breysse, P. Afanasiev, C. Geantet, M. Vrinat, Catal. Today 86
(2003) 5.
[63] J.B.F. Anderson, R. Burch, J.A. Cairns, J. Catal. 107 (1987) 351.
[64] T.J. McCarthy, G.-D. Lei, W.M.H. Sachtler, J. Catal. 159 (1998) 90.
[65] G.M. Schwab, E. Pietsch, Z. Phys. Chem. B 1 (1928) 385.
[66] R. Kramer, M. Fischbacher, J. Mol. Catal. 51 (1989) 247.
[67] M. Hoffmeister, J.B. Butt, Appl. Catal. 82 (1992) 169.
[68] F. Garin, R. Girard, G. Maire, G. Lu, L. Guczi, Appl. Catal. A 152
(1997) 237.
[69] U. Nylen, J.F. Delgado, S. Jaras, M. Boutonnet, Appl. Catal. A 262
(2004) 189.
[70] A.Corma,V.Gonzalez-Alfaro,A.V.Orchillesy, J.Catal. 200 (2001)34.
[71] D. Kubicka, N. Kumar, P. Maki-Arvela, M. Tiitta, V. Niemi, T. Salmi,
D.Y. Murzin, J. Catal. 222 (2004) 65.
[72] M. Santikunaporn, J.E. Herrera, S. Jongpatiwut, D.E. Resasco, W.E.
Alvarez, E.L. Sughrue, J. Catal. 228 (2004) 100.
[73] D. Kubicka, N. Kumar, P. Maki-Arvela, M. Tiitta, V. Niemi, H.
Karhu, T. Salmi, D.Y. Murzin, J. Catal. 227 (2004) 313.
[74] E. Rodriguez-Castellon, J. Merida-Robles, L. Diaz, P. Maireles-
Torres, D.J. Jones, J. Roziere, A. Jimenez-Lopez, Appl. Catal. A
260 (2004) 9.
[75] M.A. Arribas, P. Concepcion, A.Martinez, Appl. Catal. A 267 (2004)
111.
[76] M.A. Arribas, A. Corma, M.J. Diaz-Cabanas, A. Martinez, Appl.
Catal. A 273 (2004) 277.
[77] H. Yasuda, Y. Yoshimura, Catal. Lett. 46 (1997) 43.
[78] J.-R. Chang, S.-L. Chang, J. Catal. 176 (1998) 42.
[79] H. Yasuda, T. Sato, Y. Yoshimura, Catal. Today 50 (1999) 63.
[80] K. Sato, Y. Iwata, T. Yoneda, A. Nishijima, Y. Miki, H. Shimada,
Catal. Today 45 (1998) 367.
[81] K. Sato, Y. Iwata, Y. Miki, H. Shimada, J. Catal. 186 (1999) 45.
[82] T. Sato, Y. Nishimura, K. Honna, N. Matsubayashi, H. Shimada, J.
Catal. 200 (2001) 288.
[83] R. Hernandez-Huesca, J. Merida-Robles, P. Maireles-Torres, E.
Rodriguez-Castellon, A. Jimenez-Lopez, J. Catal. 203 (2001) 122.
[84] D. Eliche-Quesada, J. Merida-Robles, P. Maireles-Torres, E. Rodri-
guez-Castellon, A. Jimenez-Lopez, Appl. Catal. A 262 (2004) 111.
[85] D. Li, A. Nishijima, D.E. Morrisz, J. Catal. 182 (1999) 339.
[86] E. Rodriguez-Castellon, L. Diaz, P. Braos-Garcia, J. Merida-Robles,
P. Maireles-Torres, A. Jimenez-Lopez, A. Vaccari, Appl. Catal. A
240 (2003) 83.
[87] K. Ito, Y. Kogasaka, H. Kurokawa, M. Ohshima, K. Sugiyama, H.
Miura, Fuel Process. Technol. 79 (2002) 77.
[88] K.C. Park, D.J. Yim, S.K. Ihm, Catal. Today 74 (2002) 281.
[89] C. Song, A.D. Schmitz, Energy Fuels 11 (1997) 656.
[90] S. Albertazzi, G. Busca, E. Finocchio, R. Glockler, A. Vaccari, J.
Catal. 223 (2004) 372.
[91] M. Jacquin, D.J. Jones, J. Roziere, S. Albertazzi, A. Vaccari, M.
Lenarda, L. Storaro, R. Ganzerla, Appl. Catal. A 251 (2003) 131.
[92] C. Petitto, G. Giordano, F. Fajula, C. Moreau, Catal. Commun. 3
(2002) 15.
[93] S. Albertazzi, R. Ganzerla, C. Gobbi, M. Lenarda, M. Mandreoli, E.
Salatelli, P. Savini, L. Storaro, A. Vaccari, J. Mol. Catal. A 200 (2003)
261.
[94] A. Corma, A. Martinez, V. Martinez-Soria, J. Catal. 169 (1997) 480.
[95] A.D. Schmitz, G. Bowers, C. Song, Catal. Today 31 (1996) 45.
[96] J.-L. Lemberton, M. Guisnet, Appl. Catal. 13 (1984) 181.
[97] S.C. Korre, M.T. Klein, R.J. Quann, Ind. Eng. Chem. Res. 36 (1997)
2041.
[98] A.T. Lapinas, M.T. Klein, B.C. Gates, A. Macris, J.E. Lyons, Ind.
Eng. Chem. Res. 30 (1991) 42.
[99] A.T. Lapinas, M.T. Klein, B.C. Gates, A. Macris, J.E. Lyons, Ind.
Eng. Chem. Res. 26 (1987) 1026.
[100] T.B. Lin, C.A. Jan, J.R. Chang, Ind. Eng. Chem. Res. 34 (1995) 4284.
[101] J.F. Chiou, Y.L. Huang, T.B. Lin, J.R. Chang, Ind. Eng. Chem. Res.
34 (1995) 4277.
[102] V. Ponec, Appl. Catal. A 222 (2001) 31.
[103] T. Fujikawa, K. Idei, T. Ebihara, H. Mizuguchi, K. Usui, Appl. Catal.
A 192 (2000) 253.
[104] J. Wang, L. Huang, Q. Li, Appl. Catal. A 175 (1998) 191.
[105] L.J. Simon, J.G. van Ommen, A. Jentys, J.A. Lercher, J. Catal. 201
(2001) 60.
[106] J. Zheng, M.J. Sprague, C. Song, Pet. Chem. Div. Prepr. 47 (2002)
100.
[107] J.T. Miller, D.C. Koningsberger, J. Catal. 162 (1996) 209.
[108] C&EN, November 29, 2004, p. 25.
[109] C.J.H. Jacobsen, S. Dahl, B.S. Clausen, S. Bahn, A. Logadottir, J.K.
Norskov, J. Am. Chem. Soc. 123 (2001) 8404.
[110] S. Harris, R.R. Chianelli, J. Catal. 86 (1984) 400.
[111] P. Raybaud, J. Hafner, G. Kresse, S. Kasztelan, H. Toulhoat, J. Catal.
190 (2000) 128.
[112] P. Raybaud, J. Hafner, G. Kresse, S. Kasztelan, H. Toulhoat, J. Catal.
189 (2000) 129.
[113] X. Ma, H.H. Schobert, J. Mol. Catal. A 160 (2000) 409.
[114] L.S. Byskov, J.K. Norskov, B.S. Clausen, H. Topsoe, J. Catal. 187
(1999) 109.
[115] I.I. Zakharov, A.N. Startsev, J. Phys. Chem. B 104 (2000) 9025.
[116] I.I. Zakharov, A.N. Startsev, G.M. Zhidomirov, V.N. Parmon, J. Mol.
Catal. A 137 (1999) 101.
[117] I.I. Zakharov, A.N. Startsev, G.M. Zhidomirov, J. Mol. Catal. A 119
(1997) 437.
[118] J. Greeley, M. Mavrikakis, Nat. Mater. 3 (2004) 810.
[119] R.J. Hendershot, C.M. Snively, J. Lauterbach, Chem. Eur. J. 11
(2005) 806.