You are on page 1of 30

Applied Catalysis A: General 278 (2005) 143–172

www.elsevier.com/locate/apcata

Review

On the hydrodesulfurization of FCC gasoline: a review


Sylvette Bruneta, Damien Meya, Guy Pérota,*, Christophe Bouchyb, Fabrice Diehlb
a
UMR CNRS 6503, Catalyse en Chimie Organique, Université de Poitiers, Faculté des Sciences
Fondamentales et Appliquées 40, avenue du Recteur Pineau, 86022 Poitiers Cedex, France
b
Institut Français du Pétrole, IFP-Lyon, BP 3, 69390 Vernaison, France
Received 20 July 2004; received in revised form 11 October 2004; accepted 13 October 2004
Available online 23 November 2004

Abstract

The possible origins of sulfur impurities in FCC gasoline are reviewed and discussed. Their mechanism of formation during the FCC
process as well as their mechanism of transformation on hydrotreating catalysts are also examined.
The article focuses on the desulfurization of FCC gasoline by means of catalytic processes considering the fact that deep desulfurization
must be achieved (in accordance with new regulations) while preserving octane rating of the fraction. The various parameters (presence of a
promoter, nature and modification of the support, additives and poisons) which may influence the selectivity in hydrodesulfurization (HDS)
versus olefin hydrogenation are also discussed. Existing and potential processes for the HDS of FCC gasoline with octane preservation are
described.
# 2004 Elsevier B.V. All rights reserved.

Keywords: FCC gasoline; Selective hydrodesulfurization; Sulfur impurities

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144

2. Sulfur impurities and olefins in FCC gasoline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144


2.1. Nature of the sulfur impurities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
2.2. The olefins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145

3. Origin of sulfur impurities in FCC gasoline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145


3.1. Main sulfur impurities in FCC feedstocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
3.2. Formation of sulfur compounds in FCC gasoline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
3.2.1. The various possibilities (general scheme) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
3.2.2. Reactivity of sulfur compounds under FCC conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
3.2.3. Recombination of H2S with olefins and diolefins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
3.2.4. Reaction scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151

4. The HDS of FCC gasoline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 152


4.1. Studies regarding reaction mechanisms and catalytic centers involved in the hydrogenation of olefins and HDS of
thiophene and related compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
4.1.1. The HDS of thiophene and related compounds: reaction products and mechanism . . . . . . . . . . . . . . . . . . . 152
4.1.2. Reaction of thiophene and hydrogenation of olefins: kinetics and poisoning studies . . . . . . . . . . . . . . . . . . 153
4.1.3. Hydrogenation of olefins over sulfides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154

* Corresponding author. Tel.: +33 549 453 674; fax: +33 549 453 899.
E-mail address: guy.perot@univ-poitiers.fr (G. Pérot).

0926-860X/$ – see front matter # 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2004.10.012
144 S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172

4.2. The HDS of real feeds and of model FCC gasoline factors influencing the HDS/HYDO selectivity . . . . . . . . . . . . . 154
4.2.1. Designing a test reaction to evaluate catalysts for the HDS of FCC gasoline . . . . . . . . . . . . . . . . . . . . . . . 154
4.2.2. Effect of the promoter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
4.2.3. Effect of the nature of the support and of various additives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
4.3. Existing desulfurization processes and technologies—practical aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
4.3.1. Selective HDS processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
4.3.2. HDS processes with octane recovery or compensation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
4.3.3. Other processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169

5. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170

1. Introduction options being the hydroprocessing of the feed and reducing


the sulfur content during the FCC process [7,15]). Another
The exhaust gases from motor vehicles contribute to a possibility is the fractionation of the FCC gasoline in order
large extent to air pollution through their content in NOx and to blend the heavier fraction which contains most of the
SOx. Moreover, sulfur is a well-known poison for catalytic sulfur and less olefins into the gasoil fraction [7 and
converters [1]. This led the governments of numerous references therein]. The aim of these operations is to reduce
countries to adopt new regulations which aim at a drastic the amount of the sulfur compounds in the gasoline and/or to
reduction of sulfur emissions by imposing a very low change their nature in order to facilitate their subsequent
concentration of this element in fuels (50 ppm or less by elimination. However, FCC gasoline contains also a great
2005; 10 ppm or less by 2009) [2–5 and references therein]. quantity of olefins (20–40 wt.%), which provides it with a
Commercial gasoline is made up of different fractions fairly good octane number [19]. Therefore the challenge is
coming from reforming, isomerization and fluid catalytic to eliminate a maximum of the sulfur impurities with a
cracking (FCC) units. Those coming from the reforming and minimum olefin saturation.
isomerization units are produced from distillation cuts, and In the following sections, we will examine successively:
consequently contain little or no sulfur because (i) the sulfur-
containing compounds present in crude petroleum have  the nature of the olefins present in FCC gasoline and of the
generally high boiling points [6] and (ii) the feedstocks used sulfur impurities;
in the isomerization and reforming units are generally  the possible origin of the latter;
hydrotreated. On the opposite, the atmospheric residues or  the possibility of designing a simple test reaction to
the vacuum distillates which constitute FCC feedstocks evaluate the ‘‘HDS/olefin hydrogenation’’ (HDS/HYDO)
contain significant amounts of sulfur, 0.5–1.5 wt.% gen- selectivity of hydrotreating catalysts;
erally. Consequently FCC gasoline, which represents 30–  studies on the HDS of FCC gasoline;
40% of the total gasoline pool, is by far the most important  the existing desulfurization processes.
sulfur contributor in gasoline, up to 85–95% [2–5,7 and
references therein]. In the recent years a large number of Reaction mechanisms of both the formation of sulfur
articles dealing with the sulfur distribution in FCC effluents impurities during FCC and of the transformation of these
have been published [1,8–15]. The most important class of compounds and of olefins on HDS catalysts will also be
sulfur compounds present in FCC gasoline is made of discussed.
thiophene and its light alkyl derivatives in addition to
benzothiophene [16–18]. Most of these compounds are not
present in FCC feedstocks. They could either result from the 2. Sulfur impurities and olefins in FCC gasoline
direct transformation of the sulfur compounds present in
the feedstock or come from the recombination of FCC 2.1. Nature of the sulfur impurities
products. Whatever their origin, the new regulations make it
necessary to remove these sulfur organic compounds from The main sulfur components of FCC gasoline are thiols,
the FCC gasoline almost completely and as we will see a sulfides, thiophene and alkylthiophenes, tetrahydrothio-
great deal of research has also been devoted recently to this phene, thiophenols and benzothiophene [11,17 and refer-
important issue. ences therein, 18]. Alkylthiophenes which are typically in
In this review we will focus on gasoline hydrodesulfur- the boiling range of gasoline include three and four carbon
ization which is, together with non-catalytic sulfur atoms-substituted thiophenes (C3- and C4-thiophenes) but it
elimination processes, one of the three main options to be is difficult to tell whether these are polymethylated or longer
considered in order to produce pure gasoline (the other two chain-substituted thiophenes. Recent data reported by Xia and
S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172 145

Table 1 Table 2
Gas chromatographic analysis of thiophene derivatives in FCC naphtha [21] Distribution of sulfur impurities in three different FCC feedstocks (data
Compounds Percent ratio of Sulfur content from Ref. [17])
total thiophenes in naphtha (mg/g) Feed A Feed B Feed C
(sulfur %) Mercaptans and sulfides 7 12 26
Thiophene 6.3 69.2 Thiophenes 59 56 42
2-Methylthiophene 10.0 109.8 Benzo and multiring thiophenes 30 17 29
3-Methylthiophene 13.8 152.6 Oxidic sulfur 4 15 3
Dimethylthiophenes 35.4 389.8
Unknown thiophene 4.0 43.9
Iso-propylthiophene 2.4 26.4 FCC gasoline should have an internal di-, tri- or
Methyl ethylthiophene 4.2 46.1
tetrasubstituted double bond. Actually, the analysis of
Trimethylthiophene 5.6 61.5
Unknown thiophene 1.6 17.6 FCC fractions shows that terminal olefins represent about
Trimethylthiophene 4.1 45.0 25 wt.% only of the total olefin fraction, most of them
4-Carbons-alkylthiophene 10.5 116.4 having a branched skeleton [11,19].
Unknown thiophene 2.1 23.0

coworkers [20,21] confirm that thiophene sulfur represents a 3. Origin of sulfur impurities in FCC gasoline
large fraction of the total sulfur content in FCC gasoline
(60 wt.% and over). By using gas chromatography they 3.1. Main sulfur impurities in FCC feedstocks
detected more than 20 different kinds of thiophenes among
which a certain number (di- and trimethyl-, ethyl-, In order to identify the origin of the sulfur impurities in
ethylmethyl-, di- and triethyl-, iso-propyl-, tertiobutyl-) could FCC gasoline, it is essential to know what are the sulfur
be identified by GC/MS analysis (Table 1). compounds present in FCC feedstocks. The potential sulfur
components in these feedstocks are thiophene, benzothio-
2.2. The olefins phene, dibenzothiophene, polyaromaticthiophenes and their
alkylated derivatives as well as thiols and sulfides. By
The main reaction involved in catalytic cracking is the combining the results obtained from various methods which
transformation of long chain paraffins through b-scission of were available in the literature, including XPS and mass
intermediate carbenium ions (Scheme 1). The consequence spectral analysis, Cheng et al. [17] reported a semiquanti-
of this is that catalytic cracking should produce equal tative distribution of sulfur species in various feedstocks
amounts of olefins and of paraffins. This is not the case (Table 2). From this study, it appears that the main
because of H-transfer phenomena which take place during sulfur components are thiophenes. This is not in accordance
the process and which convert olefins and for instance with earlier studies by Wollaston et al. [23] who reported
cycloparaffins into paraffins and aromatics [22]. On the that the main sulfur components in various feedstocks
other hand, the reaction takes place mainly via the most were dibenzothiophenes and heavier compounds. In fact,
stable tertiary carbenium ions which result from a FCC feedstocks contain generally very small amounts of
rearrangement of the normal paraffins contained in the feed thiophene and thiophene derivatives [24] although in certain
so that the primary cracking products and the olefins in cases polyalkylated thiophenes or long alkyl chain
particular are expected to have a branched skeleton. thiophenes were mentioned as possible constituents of
Moreover, under the conditions of catalytic cracking, olefins FCC feedstocks [23]. However if we except these
are expected to isomerize very readily and since internal compounds, thiophene or thiophene derivatives are rather
olefins are thermodynamically more stable than terminal in the gasoline than in the gas oil or residue boiling
(methylenic) olefins, a large fraction of the olefins present in ranges [17].

Scheme 1. Simplified reaction path for the catalytic cracking of n-hexadecane through isomerization and b-scission of an intermediate carbenium ion.
146 S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172

3.2. Formation of sulfur compounds in FCC gasoline the addition of H2S to the olefins or the diolefins produced
through hydrocarbon cracking followed by a cyclization into
3.2.1. The various possibilities (general scheme) tetrahydrothiophenic compounds which can in turn dehy-
As already mentioned, most of the sulfur impurities drogenate into thiophenic compounds (Scheme 2b).
present in FCC gasoline (mainly thiophene and short alkyl
chain thiophenes) are not present in the FCC feedstocks. 3.2.2. Reactivity of sulfur compounds under FCC
Therefore, several possibilities (in fact two main routes a conditions
priori) have to be considered to account for the formation Quite a number of studies have dealt with the trans-
of the ‘‘organic sulfur compounds’’ of the gasoline boiling formation of thiophene on FCC type catalysts [26–29].
range during the FCC process [25]. However, as will be shown below, thiophene itself is rather
They could be produced through the cracking of heavier unreactive under the conditions of FCC and the mechanism
alkyl thiophenes or of other heavy sulfur-containing of its decomposition on pure acidic materials is still not well
molecules which are present in the feedstock (Scheme 2a). understood. In fact most of the recent relevant studies deal
However, as indicated by Cheng et al. [17], 35–45% of the with other sulfur compounds like alkylthiophenes, tetra-
feed sulfur is converted into H2S so that another possibility is hydrothiophene, benzothiophene [15,17,25,30].

Scheme 2. Possible ways of formation of sulfur-containing compounds present in FCC gasoline: transformation of heavy sulfur compounds contained in the
feed (a); reaction of H2S (produced by desulfurization of sulfur impurities of the feed) with olefins or diolefins resulting from the catalytic cracking of the
hydrocarbons of the feed (b); cyclization of long chain alkylthiophenes formed during the process (c).
S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172 147

Scheme 3. Mechanism of C–S bond cleavage in thiophene and desorption of the intermediate enethiol over a pure acidic zeolite in the absence of hydrogen [29].

3.2.2.1. Thiophene. Early studies on the reactivity of character?) which leads to the formation of a triple bond.
thiophene on acid catalysts were in fact related to This overall process repeated twice affords butadiyne and
hydrocracking [26–28]. They showed that the reactivity H2S as proposed earlier by Welters et al. [28]. We may notice
of thiophene was indeed very low on the pure acidic that each of the C–S bond cleavages could as well occur
materials used as supports for hydrocracking catalysts. directly in one step through a b-elimination process
Nevertheless the results indicated that thiophene could be involving the b-hydrogen and the protonated sulfur atom.
decomposed on these materials although deactivation was The second mechanism was considered by the authors as a
very fast. In their study on the adsorption and reaction of mechanism with prehydrogenation, which is not really the
thiophene on ZSM-5 zeolites, Garcia and Lercher [31] case. Actually according to this mechanism which leads to
confirmed that C–S bonds of thiophene could be broken on butadiene and H2S, both C–S bonds in thiophene are broken
pure acidic zeolites, which led to unsaturated thiols as through the same process which again involves two steps, in
primary products. The latter were supposed to transform fact two consecutive nucleophilic displacement reactions.
subsequently into aromatic sulfur compounds and even- The first step is actually the same as in the previous
tually into coke. Later on, Welters et al. [28] found also that mechanism (Scheme 3) and leads to a vinylic enethiol
pure acidic supports had a substantial activity in thiophene adsorbed at a surface oxygen atom. The second step involves
desulfurization and that the activity increased linearly with the displacement of the oxygen by a hydride ion provided by
the increasing acidity of the materials. They concluded that dihydrogen adsorbed on another adjacent oxygen with a
the reactivity of thiophene on pure zeolites could not be basic character (Scheme 4). This process repeated twice
explained on the basis of the generally accepted mechanisms makes it possible to desorb butadiene and H2S. Again we
involved in the HDS of thiophene on sulfide catalysts. They may notice that the elimination process could as well
suggested a direct desulfurization mechanism involving H+ occur directly through the displacement of the protonated
and leading to butadiyne (which was supposed to sulfur atom by the hydride, despite the fact that this can be
polymerize) plus H2S through b-elimination. This inter- regarded as rather difficult (as is the case with any
pretation was supported by a subsequent theoretical study displacement occurring at a sp2 carbon atom). Nevertheless,
from the same group [29]. Actually, in their study involving the calculations show that the process involving hydrogen is
DFT calculations the authors considered two possible easier. This is quite sensible and expected. However, the
mechanisms in order to account for the desulfurization of scope of this interesting study is to a certain extent limited
thiophene on acid catalysts. The first mechanism (Scheme 3) since the authors did not examine the possibility of the
is a process which does not involve hydrogen. Each of the decomposition of thiophene via tetrahydrothiophene,
two C–S bond cleavages in adsorbed thiophene leading to which is considered as the main route over FCC catalysts
desulfurization occur in two steps. The first step consists in [15,17,25,30].
the displacement of the sulfur by a nucleophilic surface Several other reports pointed out the importance of
oxygen atom; the second is a b-elimination process hydrogen sources for the transformation of thiophene over
involving another surface oxygen anion (with a basic acid catalysts. Yu et al. [32–34] observed that the conversion

Scheme 4. Mechanism of C–S bond cleavage in thiophene and desorption of the intermediate enethiol over a pure acidic zeolite in the presence of hydrogen
[29].
148 S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172

of tetrahydrothiophene was the key step in the decomposi-


tion of thiophene and that cyclohexane was the best hydr-
ogen-donor hydrocarbon compared to n-octane and iso-
octane in the presence of HY zeolite.
Scheme 5. Decomposition of thiophene through disproportionation [32].
3.2.2.2. Comparison of the reactivities under FCC condi-
of thiophene into H2S was multiplied by about 10 when tions of various sulfur impurities. Several reports [15,17,
propane (about 20 mol/mol) was added to the feed. Their 25,30] account for the reactivities of the main sulfur
interpretation was that desulfurization occurred through impurities of FCC gasoline under conditions similar to those
disproportionation according to Scheme 5. However, of the process. They agree on most of the observations and
they observed an excess in H2S with respect to benzothio- interpretations. Only a few conclusions differ from one work
phene so that they supposed that an alternate path for the to the other.
desulfurization of thiophene such as the one leading to Harding et al. [30] studied the kinetics and reaction
butadiene existed [29]. Butadiene could then transform into scheme of the transformation of various sulfur compounds
aromatic hydrocarbons, which is known to occur readily on (thiophene, tetrahydrothiophene, 2-ethylthiophene, 2-n-
HZSM-5 [35]. As we will see another possibility which hexylthiophene, 3-n-hexylthiophene) over a USY and over
would also explain the beneficial role of a hydrogen source a REY-zeolite-based catalyst in a FCC unit. They compared
would be the transformation of thiophene into butadiene via the reactivities of these various sulfur compounds by using
tetrahydrothiophene [12,15,18,25,30]. binary mixtures of each of them with n-hexadecane, the
This hypothesis is supported by a more recent study by reactivity of which was taken as a standard. By this
Zhu et al. [36] who also pointed out that the decomposition technique, they found that in desulfurization thiophene was
of thiophene on acidic catalysts was greatly favored by the about 10 times less reactive than n-hexadecane; 2-
use of a H-donating solvent such as tetraline. Zhu et al. found ethylthiophene had approximately the same reactivity as
that on FCC catalysts with strong acid sites, thiophene was n-hexadecane while tetrahydrothiophene was more than 10
apparently converted directly into hydrocarbons plus H2S times as reactive as n-hexadecane, which means that the
but was also transformed through three other parallel reactivity of tetrahydrothiophene was about two orders of
reactions: magnitude greater than that of thiophene under the same
conditions. Therefore, they concluded that the main
 the hydrogenation into tetrahydrothiophene; mechanism of removal of sulfur from thiophene-type
 alkylation via electrophilic substitution; molecules present in gasoline was their transformation into
 polymerization. tetrahydrothiophene with subsequent ring opening to form
H2S. This is in accordance with other recent studies
On this type of catalyst the conversion of thiophene into H2S [12,15,18,25]. Regarding 2- and 3-n-hexylthiophenes, the
was enhanced in the presence of a H-donating solvent. A- authors encountered some difficulties caused by the
gain, as will be shown in the next section, this suggests that isomerization of these two reactants into each other.
thiophene was probably converted in a first step through Nevertheless they found that 2- and 3-n-hexylthiophenes
hydrogen transfer to give tetrahydrothiophene which is e- were about 4 times as reactive as n-hexadecane (and
xpected to be highly reactive under FCC conditions and can consequently as 2-ethylthiophene). They found also that the
decompose rapidly through an acid-catalyzed process. This reactivity in cracking of these two compounds was
is also in accordance with the fact that the catalysts which approximately the same and concluded that it did not
have a low H-transfer activity yield more sulfur in gasoline depend on the position of the alkyl chain with respect to the
than the catalysts which have a high H-transfer activity [18]. S-atom. In fact, this is not necessarily true. Actually as
When the acidity of the catalyst was lower, the conversion of shown in a recent report [25] these two compounds
thiophene was also lower but the general trends and reaction isomerize so easily into each other that it is their equilibrium
scheme remained apparently the same. However, not much mixture which decomposes. It was therefore impossible to
detail was given regarding the hydrocarbons which were reach the actual reactivity of each of them under the
formed initially as well as concerning the desulfurization experimental conditions which were used. On the other
mechanism. Desulfurization of thiophene was supposed to hand, Harding et al. [30] conclude that the H-transfer
occur directly through proton catalysis. On the other hand, capacity of the catalyst did not affect the reactivity of the
the formation of tetrahydrothiophene was supposed to in- sulfur compounds, which is apparently in disagreement with
volve the addition of protons of the zeolite and the partic- other studies [18].
ipation of metal active centers which would be oxidized. The Corma et al. [14,15] did more or less the same study by
authors did not consider the possibility of hydride transfer using a different technique. They used an automated MAT unit
which is often put forward regarding catalytic cracking r- to study the transformation of a heavy vacuum gas oil
eactions. Actually, Fu et al. [12] carried out a similar study in (HVGO) as the base feed which was spiked with various
which they concluded that the formation and decomposition sulfur compounds representative of those found in FCC
S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172 149

gasoline (dibutylsulfide, ethylthiophene, 2-pentylthiophene, reaction schemes and mechanisms. In accordance with other
benzothiophene and 3-butylbenzothiophene) over an equili- studies, they found that thiophene was nearly unreactive
brated commercial FCC catalyst. They found that mercaptans (less than 0.2% conversion under the chosen operating
were highly reactive and produced H2S and olefins which conditions). The presence of molecular hydrogen or
were in turn converted into alkanes through H-transfer and did methylcyclohexane did not improve its reactivity which
not contribute to coke formation. They paid a particular remained very low and only traces of desulfurization
attention to the reactivity of thiophene and alkylthiophenes products were found. Under the same conditions 2-methyl-
which are considered as the main sulfur impurities in gasoline. and 3-methylthiophene were quite reactive but essentially
For this purpose, they studied pure thiophene and 2- through isomerization into each other and through dis-
methylthiophene and found that these two compounds which proportionation into equimolar amounts of thiophene and
had a rather low reactivity (thiophene especially) were mainly dimethylthiophene. 2-Ethylthiophene was slightly more
converted into coke. In accordance with the work of Harding reactive than the methylthiophenes; the main product was
et al. [30] who found that thiophene compounds were first the 3-ethylthiophene isomer, the other products were
converted into tetrahydrothiophene before being cracked, thiophene and diethylthiophene (the disproportionation
they explained the higher reactivity of 2-methylthiophene products) ethenylthiophene (a dehydrogenation product
compared to that of thiophene by supposing that its saturation resulting from H-transfer), traces of di- and tetrahydrothio-
by H-transfer was easier because of the possibility of forming phene (also resulting from intermolecular H-transfer
a tertiary carbenium ion. This led them to the conclusion that between the reactant molecules) and a few mol% of C6
under FCC conditions, H-transfer was the rate-limiting step of hydrocarbons (the desulfurization products). In fact, the
the decomposition of thiophene and of short chain reaction scheme was rather complex due to numerous
alkylthiophenes. This is in accordance with the observation possibilities of intermolecular reactions such as H-transfer
of Alkemade and Dougan [18] that hydrogen transfer was and disproportionation.
particularly important for the conversion of ‘‘small’’ The hexylthiophenes gave interesting results. 2-n-Hexyl-
alkylthiophenes. The transformation of HVGO spiked with and 3-n-hexylthiophenes were much more reactive than 2-
both ethylthiophene and tetraline confirmed them in their ethylthiophene, at least four times after stabilization of the
interpretation that H-transfer was playing a major role in the catalyst (if we consider the overall conversion) and about 60
reaction since the conversion of ethylthiophene was higher in times (if we consider only desulfurization). Moreover, as
the presence of tetraline than in its absence. On the other hand reported by Harding et al. [30], the reactivities of both
benzothiophene was found to have a very low reactivity when compounds in desulfurization were approximately the same.
added to HVGO. The authors found that a small portion only However, as indicated above, it is probably not because the
was cracked. Under the same conditions 3-butylbenzothio- position of the alkyl group with respect to the sulfur atom has
phene gave methylbenzothiophene and propane through no influence on the reactivity, it is rather because the
cracking of the alkyl chain in b-position with respect to the isomerization of the two reactants into each other is so rapid
ring and contributed also to coke formation significantly. On that it is impossible to differentiate their individual
the basis of their results and assumptions, Corma et al. [15] reactivities. For each reactant, the main desulfurization
proposed a set of interconnected reactions for the transforma- products were cracking product (C1–C6) and C10 aromatics
tion of sulfur compounds on FCC catalysts (Scheme 6). such as diethylbenzene. The sulfur-containing products were
Leflaive et al. [25] compared the reactivities of various the isomer, thiophene, various alkylthiophenes resulting
individual sulfur compounds diluted in nitrogen at 500 8C, from the cracking of the alkyl chain in different positions,
under atmospheric pressure and paid a particular attention to tetrahydrothiophenes and ethylbenzothiophene which was

Scheme 6. Interconnected reaction scheme for the formation of sulfur impurities under FCC conditions [15].
150 S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172

indeed the more abundant. This is very important in regard However, the chain has to be at least five carbon atoms
of the comprehension of the origin of sulfur impurities in long in order to provide a stable carbenium ion for cyclization
FCC gasoline. Corma et al. [15] made a similar observation (Scheme 7). With butylthiophene a primary carbenium
with n-pentylthiophene which led to methylbenzothiophene. ion would be involved, which would make the reaction
It was proposed [15,25] that this compound could result very slow.
from the intramolecular Friedel–Crafts alkylation of the Tetrahydrothiophene was found very reactive, about 50
long chain alkylthiophene followed by dehydrogenation times more reactive in desulfurization than thiophene under
(Scheme 7). This reaction can indeed explain why, as standard conditions after stabilization of the catalyst. This is
reported by Cheng et al. [17], the concentration of in accordance with the results reported by Harding et al.
benzothiophene in FCC gasoline has a tendency to increase [30]. As generally reported [14,15,38], it gave dehydro-
with increasing conversion. This is generally interpreted by genation products (thiophene, about 55 mol% and dihy-
supposing that benzothiophene is formed by dealkylation of drothiophene, 5 mol% at 10% conversion); desulfurized
high-molecular weight sulfur compounds present in the products (but-1-ene, but-2-enes, but-1,3-diene, n- and iso-
feed. The results of Corma et al. [15] and of Leflaive et al. butanes, 30%); cracking products (C1–C3 hydrocarbons,
[25] as well as very recent results of Lappas et al. [37] mainly propene, 11%); traces (3%) of various products, such
suggest that benzothiophene and its alkylated derivatives can as alkylthiophenes, benzothiophene, benzene, toluene and
also be formed during catalytic cracking by cyclization of xylenes. Dehydrogenation was clearly a primary reaction.
long chain alkylthiophenes and dehydrogenation. This can No effect of hydrogen when used as the carrier gas instead of
provide an important source of benzothiophene derivatives nitrogen, nor of added Ni or V to a fresh catalyst was
since long chain thiophenes can be formed in situ through observed on the selectivity of the reaction. Therefore it was
alkylation of light thiophene compounds by olefins produced concluded that the reaction was exclusively acid-catalyzed.
by cracking of hydrocarbons during the FCC process. Concerning the desulfurization reaction, it seemed that but-

Scheme 7. Transformation of n-hexylthiophene through cyclization in the presence of a FCC catalyst.


S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172 151

1,3-diene was the only product formed initially. Taking into 3.2.4. Reaction scheme
account thermodynamics and in particular the fact that under Except for benzothiophene which can be formed directly
the conditions of FCC, tetrahydrothiophene can dehydro- from heavier benzothiophene derivatives and also for some
genate into thiophene, it is clear that if tetrahydrothiophene of the alkylthiophenes which can be formed from long chain
can be formed, for instance by condensation of H2S with alkylthiophenes if present in the feed, the other sulfur
olefins or diolefins and cyclization (the reverse process of its compounds found in FCC gasoline result from a more
decomposition), thiophene compounds which were not complex process of formation. This process includes the
originally present in the FCC feed can appear in FCC recombination of compounds issuing both from the
gasoline. decomposition of the sulfur compounds present in the feed
(H2S in particular) and from the cracking of the
3.2.3. Recombination of H2S with olefins and diolefins hydrocarbons (such as olefins and diolefins). Therefore it
Corma et al. [15] studied the effect of the addition of both is clear that the two main options presented in Scheme 2 are
1-nonene and ethylthiophene to a HVGO. They did observe plausible. However as proposed by Corma et al. [15], quite a
a positive effect of 1-nonene on the cracking conversion of number of other reactions must be taken into consideration
the gas oil but surprisingly, they did not observe any direct (e.g. alkylation reactions) in order to give a more complete
interaction between 1-nonene and ethylthiophene and very representation of the fate of sulfur under FCC conditions.
little effect on the sulfur level. Leflaive et al. [25] studied the According to the results obtained by Corma et al. [15], if
transformation of hex-1-ene and of dienes (penta-1,3-diene long chain alkylbenzothiophenes are present in the feed,
and hexa-1,5-diene) in the presence of H2S over an they can easily transform either through dealkylation or
equilibrated FCC catalyst. As could be expected, hex-1- cracking of the side chain into benzothiophene and short
ene was very reactive (100% conversion initially) and led chain alkylbenzothiophenes which will end up in the
principally to its alkene isomers and to cracking products gasoline range. As indicated by the authors, these
(90%), but also to alkylthiophenes (about 10%) and to traces compounds can also undergo alkylation into heavier
of hexanethiols. Under similar conditions which were compounds and eventually into coke. Moreover, if long
defined as standard, penta-1,3-diene led to 3% conversion chain alkylthiophenes are present in the feed (which
(not considering the isomers). The products were hydro- however is seldom the case [23]), they can transform both
carbons (75%), methylthiophene (24%) and pentanethiols into benzothiophene derivatives through cyclization and
(about 1%). Hexa-1,5-diene gave slightly different results. dehydrogenation and into thiophene and short chain
The main products were also hydrocarbons (98%), propene alkylthiophenes through dealkylation or cracking of the
principally, and traces of alkylthiophenes and tetrahy- side chain [15,25].
drothiophene derivatives; hexanethiols were not detected. On the other hand, most of the sulfur compounds present
Tetrahydrothiophene and thiophene derivatives were sup- in FCC gasoline are formed indirectly from the compounds
posed to be formed through the acid-catalyzed addition of present in the feed. For instance, long chain alkylthiophenes
H2S to the alkene or diene followed by cyclization as shown can result from the alkylation of preexisting thiophene or
in Scheme 8 in the case of hex-1-ene. thiophene derivatives [12,15,25] formed in a first step and

Scheme 8. Formation of 2,5-dimethylthiophene through the addition of H2S to hex-1-ene (the scheme shows only the carbenium ions involved in the
cyclization) [25].
152 S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172

they can lead in turn to the compounds mentioned in the regarding the hydrodesulfurization versus hydrogenation
foregoing discussion (Scheme 2c). The other important selectivity.
possible route is the addition of H2S to olefins or diolefins
resulting from the cracking of hydrocarbons [14,15,25]. The 4.1. Studies regarding reaction mechanisms and catalytic
intermediate thiol compound could cyclize into tetrahy- centers involved in the hydrogenation of olefins and HDS
drothiophenic compounds which would dehydrogenate of thiophene and related compounds
readily into thiophenic compounds [38]. By considering
the mechanism of this process, the reaction is expected Given the importance of the subject in relation with
to be particularly easy in the case of diolefins. However, petroleum refining, numerous studies of the next decades
this reaction is reversible and it was also reported were dealing with the HDS of thiophene and related
[12,15,18,25,30] that, under FCC conditions, thiophene compounds with the objective of understanding the reaction
transformed into tetrahydrothiophene which decomposed mechanism mainly through the identification of the reaction
very rapidly into H2S and olefins. products and intermediates [41–46]. On the other hand,
Nevertheless, short alkyl chain thiophenes (methyl- and because thiophene was found to afford butadiene and/or
ethylthiophenes) undergo mainly isomerization and dis- butenes as primary products [41,42,44–46], a number of
proportionation reactions and thiophene is quite unreactive. authors addressed more specifically the question of whether
No significant desulfurization was observed with these the sites active in HDS were the same as those active in
compounds which can therefore be considered as fairly hydrogenation or not [41,47–56]. As mentioned already, this
stable under FCC conditions [18,25]. issue has gained a renewal of interest in recent years with the
All the reactions which were observed are typical acid- need of deep desulfurization coupled to the necessity of
catalyzed reactions (formation of thiols, cyclization, saving a maximum of octane rating in the fractions
hydrogen transfer, isomerization, disproportionation, crack- composing commercial gasoline.
ing), and consequently can be expected to occur readily in a
FCC unit. 4.1.1. The HDS of thiophene and related compounds:
The relative contributions of the possible pathways very reaction products and mechanism
likely depend on the composition of the feed, but also on the Kolboe [43] compared the rates of desulfurization of
reaction conditions, in particular on the contact time. thiophene, tetrahydrothiophene and n-butanethiol over
Considering the possible mechanisms of formation of the various catalysts including alumina-supported cobalt–
sulfur impurities, it can be concluded that their concentra- molybdenum. He found that the reactivities of thiophene
tions depend both on thermodynamics (intermediates and tetrahydrothiophene were similar while that of n-
resulting from the addition of H2S to olefins or diolefins) butanethiol was much higher (15–40 times). He found also
and on kinetics (products resulting from consecutive steps). that thiophene and tetrahydrothiophene gave qualitatively
similar (although not equal) product distributions. Never-
theless he concluded that both reactants did not give the
4. The HDS of FCC gasoline same primary desulfurization products, in other words
that tetrahydrothiophene was not an intermediate in the
As early as in the mid-1950s, in a study aiming at desulfurization of thiophene. Thiophene and tetrahydrothio-
improving the lead susceptibility of cracked gasoline phene were supposed to give adsorbed diacetylene and
fractions by sulfur reduction, Casagrande et al. [39] pointed butadiene respectively and in each case the reaction was
out that it was possible to adjust the operating conditions of a supposed to involve two C–S bond cleavages through a b-
process using a nickel-tungsten sulfide catalyst so as to elimination process. Accordingly, this meant that no
perform efficient hydrodesulfurization while maintaining a hydrogenation of thiophene was required prior to its
good octane rating in the fraction. This was made possible desulfurization. The author explained that the primary
by taking advantage of the acidic properties of the catalyst in adsorbed diacetylene was not likely to appear among the
order to promote skeletal isomerization of the olefins and gaseous products because it is known to be strongly
hence compensate their saturation. However, Meerbott and adsorbed. The role of hydrogen was then to clean the surface
Hinds [40] found that under the conditions required for from the acetylenic residue.
substantial skeletal isomerization to take place, extensive However, certain studies regarding the HDS of thiophene
saturation of olefins resulted. This enlightens the necessity over molybdenum sulfide catalysts concluded that butadiene
of finding means to limit olefin hydrogenation, which is of was the sole primary product [41,45] but it was also shown
course even more crucial in the present context (where lead that this product was so reactive under the conditions of most
additives are totally prohibited and the content in aromatics of the experiments that it could not be detected [41,42]. In
as well as the use of oxygen additives are limited) than in the fact, several authors found that both butadiene and butenes
fifties. Meerbott and Hinds [40] reported also qualitative were primary products [44,46] assuming parallel reaction
evidence that normal olefins saturate faster than branched pathways. It seems actually that the most likely reaction
olefins, which as we will see could be of some importance scheme and mechanism is the one proposed by Kwart et al.
S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172 153

cleavage itself although a concerted elimination process


means in principle two-points adsorption in the transition
state as shown by Kwart et al. [44].
What precedes (in particular the fact that the HDS of
thiophene requires most probably its preliminary total or
partial hydrogenation) could obviously lead to predict the
actual difficulties encountered today to deeply desulfurize
FCC gasoline without saturation of the olefins. Nevertheless
as we will see in the next section, the hope was placed in part
on the fact that one could reasonably envisage that the two
reactions occurred on distinct centers.
Scheme 9. Mechanism of C–S bond cleavage in dihydrogenated thiophene
over sulfided CoMo/Al2O3 and related catalysts [44]. 4.1.2. Reaction of thiophene and hydrogenation
of olefins: kinetics and poisoning studies
[44]. The authors proposed a mechanism in which thiophene Several authors [41,49–51,55,56] studied the effect of
is first partially hydrogenated, which makes the first C–S coreactants (H2S, organic sulfur compounds) and poisons
bond cleavage possible through b-elimination (Scheme 9). (ammonia, pyridine, etc.) on the HDS of thiophene and
The intermediate could then add a second molecule of related compounds (alkylthiophenes, tetrahydrothiophene)
hydrogen before the second C–S bond cleavage occurred and on the hydrogenation of olefins.
through the same mechanism as the first one so that Most of them [41,49,51,55,56] found that the two
butadiene could be formed. However, it was also proposed reactions had not the same sensitivity to the coreactants or
[44] that once the first C–S bond would be opened, the poisons and concluded that they occurred on distinct centers.
double bond the most remote from the C–S bond in the For instance, by using pyridine as poison, Desikan and
butadienyle sulfide intermediate (Scheme 9) could be Amberg [41] concluded that the sites involved in each of the
reduced before the vinylic sulfide double bond. The result two reactions differed in acid strength, the strongest being
would be the desorption of but-1-ene instead of butadiene. responsible for olefin hydrogenation. Another typical
Consequently it is not surprising that butenes appeared as investigation was reported by Satterfield and Roberts [49]
primary products in most cases and were found to transform who made a kinetic study of thiophene hydrodesulfurization
eventually into butane [41,42,44,46,51–53]. Another pro- and of the hydrogenation of the butene intermediate. Their
posal made by Sauer et al. [57] on the basis of a kinetic study Langmuir–Hinshelwood kinetic modeling of both reactions
carried out with model catalysts was the formation of and particularly of the inhibiting effect of H2S on these
butadiene with 2,5-dihydrothiophene as intermediate. One reactions led them to conclude that butene was not
could also suppose that thiophene is first totally hydro- hydrogenated on the desulfurization sites. They acknowl-
genated into tetrahydrothiophene which then can undergo edge however that it is often difficult to draw definite
twice a C–S bond cleavage step to produce butadiene as conclusions regarding reaction mechanisms from kinetic
proposed by Hargreaves and Ross [53] or to produce either equations. Nevertheless, numerous other kinetic studies [63–
butadiene or butene as can be deduced from the work of 66] were in favor of the existence of two types of active sites,
Moser et al. [46] who suggested surface butene thiolates as one for hydrogenolysis and one for hydrogenation.
intermediates. Without entering into details which are In fact several studies [51–53] concluded that if the sites
beyond the scope of this review, we may add that the involved in hydrogenation and desulfurization are different,
elimination is also in accordance with 35S [58–60] and there must be some connection or relationship between the
deuterium tracer experiments [45,61] of the hydrodesulfur- two. For instance, Lee and Butt [51] who also studied the
ization of thiophene showing that the S-atom of the kinetics of the HDS of thiophene and hydrogenation of
produced H2S comes from the catalyst rather than from butene do conclude from their rate correlations that the
the organic substrate while part of the hydrogen atoms at catalytic centers for the two reactions are different but that
least comes from the latter and not exclusively from the they have similar sensitivities to pyridine. On the other hand
hydrogen of the gas phase. A detailed description of the in their study on Mo and CoMo-on-alumina catalysts,
various steps of the overall hydrodesulfurization process Hargreaves and Ross [53] proposed that hydrogenation and
accounting for these observations was reported recently desulfurization occurred on distinct but interrelated centers.
[62]. Both were found to be promoted by Co although the former
The debate about one-point versus two-point or multi- were less affected than the latter.
point mechanisms [43,44,49,50] is in a way difficult matter. Other authors even come to the conclusion that both
Actually, it is clear that if at least partial hydrogenation of reactions occur on the same centers [50,54]. Lipsch and
the substrate is required prior to C–S bond cleavage, which Schuit [50] found in particular that pyridine had the same
seems very likely, then it is sensible to assume a multipoint effect on butene and thiophene adsorption, which they
adsorption. However, this is not necessarily so for C–S bond considered as an indication that both adsorb on the same
154 S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172

sites. However, their CoMo catalysts were activated by some differences in the sequence of reactivities as compared
reduction with hydrogen and they found that sulfidation did to transition metals. In particular they did not find the same
not improve the activity, which limits the significance of the decrease in reaction rates with increasing number and size of
results with respect to practical hydrodesulfurization condi- substituents on the double bond, which they assumed was
tions. Moreover, this is in complete disagreement with the due to the fact that on their catalyst, the rate-limiting step of
work of Okamoto et al. [54] who found that the activity of their the reaction was the addition of the second hydrogen atom
Mo-on-alumina catalyst in the transformation of thiophene instead of the first one on transition metals.
depended exclusively on the sulfidation degree of molybde-
num. Nevertheless, Okamoto et al. found that the relative 4.2. The HDS of real feeds and of model FCC gasoline
activity for the hydrogenation of butenes with respect to factors influencing the HDS/HYDO selectivity
desulfurization was constant, which led them also to conclude
that both reactions occurred on the same centers [54]. The studies undertaken on the desulfurization of gasoline
However, as we will see in a subsequent section, more in the present context of drastic limitation of sulfur emissions
recently Hatanaka and coworkers found that the olefins aim at understanding (or identifying) the factors which govern
inhibited the HDS of the sulfur impurities contained in FCC the selectivity in desulfurization with respect to the
gasoline [11]. A similar observation was made previously by hydrogenation of olefins (HDS/HYDO) in order to improve
Meerbott and Hinds [40] although these authors mentioned this selectivity. For this purpose, certain groups have used real
that this was true only under certain conditions. To account feeds or spiked gasoline [7,11,19] while others have used
for this phenomenon and for the fact that H2S had not the model or synthetic gasoline [19,67,68,75–77].
same effect on the hydrogenation of normal and iso-olefins,
Hatanaka et al. [19,67,68] suggested the existence of three 4.2.1. Designing a test reaction to evaluate catalysts for
categories of active centers. One category of centers was the HDS of FCC gasoline
supposed to be active in HDS and to adsorb n- and iso- 4.2.1.1. Measuring the ‘‘HDS/olefin hydrogenation’’ (HDS/
olefins, the second category of centers was supposed to be HYDO) selectivity. Several methods can be used to measure
active in n-olefin hydrogenation and the third to be active in the HDS/HYDO selectivity of hydrotreating catalysts. In
iso-olefin hydrogenation. numerous studies using thiophene as reactant the butane
Obviously, the question of the number and nature of the concentration among the desulfurization products was
catalytic centers involved in hydrogenation and hydro- considered as representative of the hydrogenation activity
desulfurization is not yet definitely settled. It can also be of the catalyst and compared to its HDS activity was used to
considered that the surface is not stable in the presence of the estimate its selectivity. However, since we are dealing with
reactants [69] and that the interconversion of sites can occur consecutive reactions it is difficult to obtain precise
depending on gas phase composition [65,70–72]. For quantitative values with this method. Another possibility
instance, the interconversion of hydrogenation and hydro- consists in measuring in separate experiments the activity of
genolysis centers was supposed to take place through the catalyst in the hydrogenation of one given olefin on the
reaction with H2S and/or with spillover hydrogen [71,72 and one hand and in the HDS of a sulfur compound such as
references therein] (see Section 4.2.2), the consequence of thiophene or benzothiophene on the other. This, of course,
which is that the distribution of these sites depend very much is not at all representative of actual situations in refining
on the H2/H2S molar ratio. This hypothesis made it possible and in particular does not account for possible competi-
to account for the activity of sulfide catalysts in olefin tion phenomena between olefins and sulfur compounds. In
hydrogenation and in hydrodesulfurization [70–72]. principle, the most reliable method is to measure
simultaneously both activities by using mixtures of chosen
4.1.3. Hydrogenation of olefins over sulfides sulfur compounds and olefins (a synthetic gasoline) or to
As we have seen, there is in the literature a large number measure the HDS and hydrogenation activities by using a
of studies on the hydrogenation of butenes either in real FCC gasoline feed. The former is certainly the most
combination with the HDS of thiophene or separately. precise although it is not necessarily representative of what
However, there is only a very limited number of studies would be obtained with real feedstocks, the latter is of course
dealing with the hydrogenation of olefins of different the most representative of real practical conditions but is not
structure over sulfide catalysts. In their study on the always precise and the results not always extendable to
isomerization of olefins Meerbott and Hinds [40] noticed all feedstocks. Another possibility could be the kinetic
however that n-olefins seemed to saturate faster than modeling [71,72] of the hydrodesulfurization of FCC
branched olefins, which is in agreement with what is found gasoline using the lumping technique or the more recent
on metals [73]. Uchytil et al. [74] studied the hydrogenation approach based on structural contributions [78–80].
of a series of olefins with various structures over a Co-on-
alumina catalyst. However the samples were reduced and not 4.2.1.2. Components of a ‘‘synthetic gasoline’’. The com-
sulfided, which again reduces the interest of the study in the ponents of a model or synthetic FCC gasoline were chosen
context of hydrodesulfurization. Nevertheless they found so as to represent at best the compounds present in real feeds
S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172 155

Scheme 10. Main products of transformation of alkenes on hydrotreating catalysts.

(see Section 2). As we have seen, FCC gasoline contains products as described in Scheme 11, compounds like 2-
mainly thiophene and alkylated thiophenes as well as small methylthiophene can undergo reactions such as methyl
amounts of benzothiophene as sulfur compounds. In migration and disproportionation (Scheme 12). However,
addition to saturated hydrocarbons and aromatics it contains these reactions are seldom observed with conventional
also olefins mostly branched and internal rather than normal hydrotreating catalysts.
and methylenic or terminal olefins [11,19]. Therefore certain
authors when using only one representative compound of 4.2.1.5. Side reactions. We must also consider the possi-
each category have chosen for instance methylthiophene and bility of reactions between model olefins or olefins issuing
one branched olefin in solution in n-heptane [75–77]. Others from desulfurization and H2S to produce thiols (recombina-
have used several different olefins (for instance, iso-octene, tion or recombinant mercaptans). In fact, as early as in 1975,
oct-1-ene, cyclohexene) and several thiophene derivatives Satchell and Crynes [81] pointed out that the deep
[19,67,68]. desulfurization of cracked (olefinic) naphtha could be
limited by both the olefin and hydrogen sulfide partial
4.2.1.3. Selection and reactivity of the olefin. Depending pressures in the HDS reactor. Olefins, hydrogen sulfide and
on the reaction conditions, on the nature of the hydrotreating hydrogen could react according to Scheme 13.
catalyst as well as on their own structure, olefins may
undergo various reactions. However, the catalysts which we
have to consider in this review are either weakly or non-
acidic. Consequently, olefins are expected to undergo only
isomerization and hydrogenation reactions. For instance, a
terminal olefin such as oct-1-ene is expected to undergo
principally double bond migration and hydrogenation into
n-octane (Scheme 10a). Similarly di-iso-butylene or iso-
octene (2,4,4-trimethylpent-2-ene) and 2,3-dimethylbut-2-
ene give essentially double bond migration and hydrogena-
tion into iso-octane and 2,3-dimethylbutane, respectively
(Scheme 10b), while cyclohexene gives almost exclusively
cyclohexane. Given the nature of the olefins present in FCC Scheme 11. Possible pathways for the transformation of 2-methylthiophene
gasoline, an olefin with a branched skeleton and internal over sulfided CoMo catalysts.
double bond which can give rise to a rather limited number
of isomers such as 2,3-dimethylbut-2-ene seems to be quite
appropriate as a model molecule.

4.2.1.4. Selection and reactivity of the sulfur compoun-


d. Alkylthiophenes can be considered as representative of
the major sulfur impurities of FCC gasoline. In addition to
desulfurization and consecutive reactions of the primary Scheme 12. Isomerization and disproportionation of 2-methylthiophene.
156 S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172

question arises of whether the promoter could have an effect


on the HDS/HYDO selectivity for FCC gasoline or not.
Indeed, one could wonder if, as certain results in the
literature seemed to indicate, the optimum in synergy for
HDS and olefin hydrogenation could occur for different
Scheme 13. Reactions between olefins, hydrogen sulfide and hydrogen concentrations in promoter [71,99], or if the effect could be
(adapted from Ref. [81]). different in magnitude for the two reactions [101].
Hatanaka et al. [68] studied the effect of Co in CoMo-on-
Considering that the recombination of H2S with olefins is alumina catalysts on the main reactions involved in the
at the thermodynamic equilibrium (recombination being hydrotreating of FCC gasoline: the HDS of thiophenic
much faster than hydrogenation), the following relationship compounds and the hydrogenation and isomerization of
between thiol, hydrogen sulfide and olefin partial pressures, olefins. For this purpose, they used mixtures made of
Pthiol ¼ ½Polefins ½PH2 S ðk1 =k2 Þ; thiophene and an olefin (di-iso-butylene or oct-1-ene)
dissolved in toluene. As expected they found that Co had a
should apply at a given temperature. This means that the promoting effect on the HDS of thiophene but surprisingly
amount of thiols should be proportional to the partial they found also that it had an inhibiting effect, although
pressures of H2S and of olefins. In fact, for high desulfur- moderate, on the hydrogenation of iso-olefins and a
ization levels (i.e. when hydrogen sulfide partial pressure is pronounced inhibiting effect on the hydrogenation of n-
practically constant), the amount of thiols should be propor- olefins. This was for them a confirmation of the existence of
tional to the amount of the olefins in the treated gasoline. three different categories of centers as mentioned above (see
This was indeed observed. Obviously in the specific case of Section 4.1.2).
the selective HDS of FCC gasoline, where a maximum of the In their study on the selective hydrodesulfurization of
olefins should be preserved from hydrogenation, this phe- FCC naphtha, Miller et al. [7] examined various factors
nomenon can be of great importance. including the effect of Co on the properties of MoS2
catalysts. As we will see below, they found that the only
4.2.2. Effect of the promoter factor which led to lower olefin saturation, in other words to
The promotion or synergy effect of Co or Ni on the an improved HDS/hydrogenation selectivity was the
activity in HDS of Mo or W sulfide catalysts has been known promotion by Co. At sulfur concentrations in the gasoline
for many years and various theories or hypotheses were greater than about 300 ppm at least, the olefin concentration
proposed to explain it [82–84 and references therein]. (hence the RON) decreased much less than the sulfur
Mainly two of them have been the subject of a large debate concentration (Fig. 1), which was not the case with the
in the literature: (i) the remote control theory [84,85] in unpromoted catalysts. Below about 100 ppm however olefin
which it is supposed that the phase made of the sulfided saturation occurred significantly.
promoter is necessary to produce spillover hydrogen which Recently, Choi et al. [75] reported results on the effect of
then migrates to the MoS2 or WS2 phase where it can react Co and Sn introduced by surface organometallic chemistry
with the organic substrate or create new catalytic centers; (ii) (SOMC) on the activity and selectivity of molybdenum-
the model based on the so-called mixed CoMoS or NiMoS based catalysts in the transformation of a model FCC
phases discovered by Topsøe and coworkers [83 and gasoline containing 3-methylthiophene and 2,3-dimethyl-
references therein, 86–90]. A combination of the two but-2-ene. They found that Co enhanced HDS more than
models was also proposed to explain the gain in activity olefin hydrogenation. Their catalyst modified by SOMC was
obtained by mixing for instance Co9S8 with a NiMoS sample nearly as active as a conventional commercial CoMo
[91]. The mixed phase model was interpreted as an catalyst but was not more selective. On the other hand tin
illustration of the Sabatier principle by supposing that the introduced by the same method had an effect which was
promoter donates electrons to the main constituent and leads opposite to the desired effect: it inhibited the HDS activity
to a weakening of the metal–sulfur bond down to the more than the hydrogenation activity.
optimum required for the HDS activity [92 and references Apart from the inhibiting effect of Co on the
therein, 93]. This interpretation was recently updated on the hydrogenation of olefins reported by Hatanaka et al. [68]
basis of ab initio calculations using the density functional the results are in general agreement with recent and
theory and a new metal–sulfur bond energy model was more ancient studies which indicate that the promoter
proposed [94]. However, the promoter has also an effect on enhances the rate of C–S bond cleavage steps more than
many other reactions catalyzed by sulfides [71,95 and the rate of hydrogenation steps. This was found with
references therein, 96–99] including the hydrogenation of compounds representative of diesel fuel sulfur impurities
olefins [71,95 and references therein, 99]. It was also shown such as dibenzothiophene or 4,6-dimethyldibenzothiophene
that the synergetic maximum (corresponding generally to a [62,98,102,103] as well as with thiophene [52,53,88,
promoter/(promoter + Mo (or W)) atomic ratio of about 0.3) 101,104,105,106 and references therein]. Actually, if one
could depend on reaction conditions [100]. Then, the excepts the work of Hagenbach et al. [107] in which the
S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172 157

primary butenes [88,101,104,105,106 and references


therein] found also that HDS was enhanced to a greater
extent by the promoters than hydrogenation but they did not
come exactly to the same conclusion as Hargreaves and Ross
[52,53]. For instance, Massoth and Chung [104] found that
the promotion by cobalt of the secondary hydrogenation of
butenes was appreciably lower than that of the HDS
reaction. However, they pointed out that since hydrogena-
tion is consecutive to desulfurization it is rather difficult to
appreciate the extent of the promotion of the former. They
noticed in particular that at the same thiophene conversion
the hydrogenation selectivities of all the promoted catalysts
were quite similar. They concluded accordingly that the
same sites were involved in both reactions.
Moreover it seems, according to the results reported by
Candia et al. [101], that the promoting effect of nickel on the
hydrogenation of olefins is more significant than the effect of
cobalt while the difference is not so pronounced in the HDS
of thiophene. This is in accordance with results reported
recently showing that cobalt had nearly no effect and nickel
had clearly a promoting effect on the hydrogenation of
cyclopentene (Fig. 2) [108] while both had similar effects on
the HDS of dibenzothiophene [98]. This may explain why
cobalt-promoted catalysts are more efficient than nickel-
promoted catalysts in the selective HDS of FCC gasoline.
Fig. 1. Relationship between sulfur concentration (ppm sulfur remaining in
the gasoline) and olefin concentration (a) or RON (b). (x) FCC naphtha feed;
(*) CoMo/alumina; (*) CoMo/2% K-alumina (data from Ref. [7]).
4.2.3. Effect of the nature of the support and of various
additives
It has been known since many years that the nature of the
authors found similar effects on the HDS of thiophene and support can have a critical effect on the catalytic properties
on the hydrogenation of olefins, most of the studies which of sulfides [109,110]. The subject was updated very recently
are relevant to the issue of selective HDS of FCC gasoline in a series of review articles [111] which pointed out that the
have indeed shown that the promoter had a more significant effect of the support can be very complex as it can affect in
effect on HDS than on hydrogenation [52,53,88,101,104, an interconnected manner the dispersion as well as the
105,106 and references therein]. In fact, the catalysts used geometric and electronic properties of the active component
by Hagenbach et al. [107] were obtained by sulfiding a blend [112]. In this article we shall focus on the studies related to
of the cobalt and molybdenum oxides and although the the HDS of FCC gasoline.
authors considered that the promoting effects they measured In this respect, it was shown that acidic supports such as
were significant (30–40% enhancement in activity), these zeolites can promote the hydrogenation activity of metals
effects were much weaker than those reported in other [113,114] as well as of sulfides [115–117]. The enhancement
studies made with commercial catalysts [104] or with
catalysts where at least both sulfide precursors were
deposited on the same support [52,53,104,105,106 and
references therein].
By studying the HDS of thiophene and the hydrogenation
of but-1-ene separately, Hargreaves and Ross [52,53] found
that their most active promoted catalyst was about 10 times
more active in the HDS of thiophene than the unpromoted
catalyst whereas it was only three times more active in the
hydrogenation of but-1-ene. This added to kinetic con-
siderations led them to conclude that C–S bond cleavage and
hydrogenation occurred on distinct although interrelated
centers, the hydrogenation centers being less affected by the
Fig. 2. Hydrogenation of cyclopentene over sulfided NiMo/alumina and
promoter. CoMo/alumina catalysts (150 8C, atmospheric pressure). Effect of the
Several other groups who studied simultaneously the promoter/(promoter + molybdenum) atomic ratio (Pr/(Pr + Mo)) on the
HDS of thiophene and the consecutive hydrogenation of the activity. Open symbols: commercial catalysts [108].
158 S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172

of the activity was attributed, in part at least, to electronic


transfer phenomena from the metal or sulfide to acceptor
sites in the zeolite lattice. Such transfers would favor the
adsorption of the substrate and therefore the hydrogenation
activity of the metal or sulfide [116]. However, this kind of
effect could also be interpreted in terms of a modification of
the intrinsic hydrogenation properties of the material.
Evidence for the electronic transfer was found in FT-IR
of adsorbed CO which showed that the absorption bands
were shifted towards high frequencies indicating an electron
deficiency of the catalytic centers [116]. Conversely it can be
expected that increasing the basicity of the support should
decrease the hydrogenation activity. This is indeed what was Fig. 3. Effect of SiO2 content in the support on the HDS/hydrogenation
observed with Pt-zeolites [118,119] in the case of benzene selectivity of CoMo/SiO2–Al2O3 catalysts (data from Ref. [123]).
hydrogenation and with molybdenum sulfide in the case of
1-methylnaphthalene hydrogenation [120]. Moreover, it was HDS/hydrogenation selectivity increased (Fig. 3). This was
also expected that basic supports could keep deactivation by particularly clear when taking into consideration pure SiO2.
coke deposition to a low level [121,122]. Several attempts Otherwise (when pure SiO2 was not taken into considera-
were made either by changing the nature of the support or by tion), the effect was not very significant and, as indicated
modifying the hydrotreating catalysts or their alumina above, it was actually the opposite of what would be
support with various additives. Actually, we will see that expected from changing the acidity. Incidentally, the authors
there are some discrepancies regarding the effect of the concluded that the change in HDS/hydrogenation selectivity
acid–base properties of the support or catalyst on the provided support to the existence of different catalytic
hydrogenation activity. centers for HDS and hydrogenation.
Okamoto and coworkers [125,126] examined also the
4.2.3.1. Effect of the nature of the support. A number of effect of the nature of the support (Al2O3, TiO2, ZrO2 and
studies have dealt with the effect of the support on the HDS SiO2) on the activity in the HDS of thiophene and in the
and hydrogenation activities of hydrotreating catalysts hydrogenation of butadiene of supported Mo sulfide and of
although they did not address the question of the HDS of supported CoMo sulfide model catalysts. The CoMo
FCC gasoline specifically [123–126]. catalysts were prepared by chemical vapor deposition
For instance, Muralidhar et al. [123] studied separately (CVD) of Co(CO)3NO on the supported Mo sulfide. They
the HDS of thiophene and the hydrogenation of hex-1-ene as found that the Co species deposited in this way were located
well as the hydrocracking of 2,4,4-trimethylpentene over at the edges of the MoS2 particles. Significant support effects
various supported CoMo catalysts. They compared CoMo were found with the unpromoted catalysts (Fig. 4). In HDS,
supported on different types of alumina to CoMo supported the SiO2-supported MoS2 catalyst was the most active and
on SiO2–Al2O3, SiO2–MgO and TiO2. They found that the the Al2O3-supported MoS2 catalyst was the least active
CoMo–Al2O3 catalysts were more active than the others, while in the hydrogenation of butadiene the ZrO2-supported
both in HDS and in hydrogenation, which they attributed to a MoS2 catalyst was by far the most active and the SiO2-
better dispersion of the Mo phase on Al2O3. However, the supported MoS2 catalyst the least active. The interpretation
Al2O3-supported catalysts were less active in hydrocracking,
which is an indication that they were less acidic than the
other catalysts. They found also that increasing the SiO2
content in SiO2–Al2O3 decreased both the HDS and
hydrogenation activities but increased the hydrocracking
activity. Since it is known that increasing the SiO2 content in
SiO2–Al2O3 increases its acidity [127] the effect on
hydrocracking was indeed expected. However, the results
regarding the hydrogenation of hex-1-ene are not in line with
the expectation that increasing the acidity of the support
should increase the hydrogenation activity (and vice versa).
A possible explanation of the results of Muralidhar et al.
[123] could be an increase in deactivation by coking due to
the increase in acidity. Nevertheless according to the results Fig. 4. Effect of the nature of the support on the activities of MoS2 catalysts
in thiophene HDS (open bars) and in butadiene hydrogenation (black bars).
of the authors, the hydrogenation activity decreased more The turnover frequencies were calculated on the basis of the number of sites
than the HDS activity as the SiO2 content increased, determined by NO adsorption (data from Ref. [125]; hydrogenation activity
particularly in the case of the CoMo catalysts, so that the of the MoS2/ZrO2 catalyst was divided by 10).
S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172 159

of the authors was that the HDS/hydrogenation selectivity of


these unpromoted catalysts was in agreement with the rim-
edge theory of Daage and Chianelli [128] since a more
significant stacking of the layers in the MoS2 crystallites was
observed with SiO2. However, the authors did not exclude
the possibility of geometric or electronic effects. The same
effects were not found with the Co promoted catalysts. The
promotion effect itself depended on the support and in fact
when the CoMoS phase was formed the specific activity of
the Co species at the edge sites of MoS2 particles became
equalized. Consequently, the thiophene HDS activity was
proportional to the amount of Co whatever the support,
except for SiO2, which the authors interpreted by supposing Fig. 5. Relationship between the number of acid sites (measured by NH3
that the interactions between the CoMoS phase and the adsorption) and the HDS and hydrogenation activities of sulfided CoMo
support were weaker than those between the MoS2 edge sites catalysts (data from Ref. [124]).
and the support for the unpromoted catalysts. Therefore,
they concluded that no real support effect existed for the Klimova et al. [135] used Mg–Al mixed oxides
reaction regarding at least the Al2O3, TiO2 and ZrO2 (x = MgO/(MgO + Al2O3) = 0–1) prepared by the sol–gel
supports. The higher activity of the SiO2-supported catalyst method as supports for Mo and NiMo catalysts. They used
was attributed to the formation in this specific case of the thiophene HDS as the test reaction and estimated the
CoMoS(II) phase [89] which would not exist on the other hydrogenation activity through the production of butane
supports. On the other hand the Al2O3-supported catalyst among the HDS products. The results indicated that even
showed the best activity in the hydrogenation of butadiene with the incorporation of small amounts of magnesia to
while the catalysts supported on TiO2, ZrO2 and SiO2 alumina the hydrogenation function of the unpromoted Mo
showed the same activity. Obviously HDS and hydrogena- catalyst was substantially reduced while the thiophene
tion were affected differently by the support. From the conversion was much less affected (Fig. 6a). The gain in
results obtained in both HDS and hydrogenation one could HDS/hydrogenation selectivity resulting from the addition
conclude that the best catalyst regarding the HDS/ of magnesia was attributed to the fact that the size of the
hydrogenation selectivity and therefore the best catalyst MoS2 crystallites increased with increasing magnesia
for the HDS of FCC gasoline would be CoMo/SiO2. content as compared to their size on pure alumina. This
Another recent study by Flego et al. [124] aimed at interpretation implies that hydrogenation takes place at
improving the activity of CoMo catalysts in view of the deep corner sites, the addition of magnesia leading to an increased
desulfurization of fuels and more particularly of gasoline. proportion of edge versus corner sites. However, with the Ni-
They prepared CoMo catalysts with various supports promoted catalysts with a Ni/(Ni + Mo) ratio of 0.3 there
containing metal (Zr, Ga, Si, and B) oxides having different was a continuous and substantial drop in HDS activity with
acid–base properties and measured their activity in the HDS increasing magnesia loading which paralleled the decrease
of thiophene. They estimated also the hydrogenation activity in hydrogenation activity (Fig. 6b). Consequently, there was
of the catalysts from the butane selectivity in the HDS no substantial gain in HDS/hydrogenation selectivity with
products. Both HDS and hydrogenation activities increased these catalysts. The decrease in activity with increasing
with increasing acid site density although again, the HDS magnesia content was attributed by the authors to the
activity increased more than the hydrogenation activity, the formation of magnesium molybdate and of a NiO–MgO
latter passing through a maximum (Fig. 5). The consequence solid solution which impeded the incorporation of Ni at the
was that the support had only a very small effect on the HDS/ edges of the MoS2 particles and therefore the promotion
hydrogenation selectivity. The authors interpreted the effect. The consequence was that it was necessary to use Ni/
relative decrease in hydrogenation activity of the samples (Ni + Mo) atomic ratios up to 0.6 to recover the promotion
containing boron oxide by supposing that the olefins were effect of Ni.
too strongly adsorbed on the acid sites to migrate to the In recent contributions, Zdrazil and coworker [121,140]
sulfide phase. reported detailed studies on the preparation and catalytic
Among the various supports which were used in this properties of MgO-supported catalysts. As pointed out
context, MgO and other materials containing MgO or other previously [135,141] the main difficulty with MgO supports
basic compounds received particular attention. Several is to avoid their deterioration by contact with water during
patents recommended the use of MgO-supported catalysts the introduction of the active phase precursor. Actually, a
for the selective HDS of FCC gasoline [129–134] and a key point is the tendency of MgO to hydrate into Mg(OH)2
series of recent academic studies on MgO and hydrotalcite- and to form solid solutions with CoO and NiO in the
derived oxides as supports addressed also this issue more presence of water, which makes the preparation of
specifically [121,135–140]. Co(Ni)Mo catalysts with this support quite delicate. For
160 S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172

Table 3
Effect of hydrotalcite (HT) and of potassium on the HDS and hydrogenation
activities of CoMo catalysts [136]
Support Additive HDS (%) HYD (%) Selectivity RON
factor
Al2O3a – 97.8 98.7 0.88 –
Al2O3a K 87.8 70.9 1.70 –
HT/Al2O3 (4/1, w/w)a 51.8 17.8 3.72 –
HTa 50.3 15.9 4.03 –
HTa K 46.9 11.5 5.18 –
Al2O3b 99.0 80.6 2.8 9.5
HTb 69.1 7.0 16.2 1.8
a
HDS of model compounds.
b
HDS of FCC gasoline.

350 8C. They attributed the high synergistic effect they


observed to the basicity of the support which according to
them favors high dispersion and therefore increases the ratio
of surface edge planes/basal planes and hence the number of
potential sites for promotion.
Zhao and coworkers [136–139] reported a series of
studies on CoMo catalysts with oxides from hydrotalcite and
related materials as supports for the selective HDS of FCC
gasoline. They used model reactions (the HDS of thiophene
and the hydrogenation of hex-1-ene) as well as the HDS of
Fig. 6. Effect of the molar fraction of MgO (x) in Al2O3 supports on the FCC gasoline. They found [136] that the catalysts with
HDS and hydrogenation activities of non-promoted MoS2 catalysts (a) and mixed hydrotalcite–alumina supports containing a high
NiMo catalysts (b) (350 8C, atmospheric pressure; data from Ref. [135]).
concentration of hydrotalcite or the catalysts with pure
hydrotalcite as support had a better HDS/hydrogenation
this reason Zdrazil and coworker developed a non-aqueous selectivity both for model compounds and for FCC gasoline
method for the deposition of MoO3 on MgO [142–144]. (Table 3). The consequence was that the loss in octane
More recently [121,140] they reported results on the number was much less significant with hydrotalcite than
introduction of Co(Ni) promoters on a MoO3/MgO starting with alumina as support. However, under the same
material. However, they did not address explicitly the conditions, the HDS conversion was much lower with
question of the HDS/hydrogenation selectivity but were hydrotalcite than with alumina as support. Nevertheless, a
essentially interested in improving and preserving the very recent study [139] confirmed that CoMo catalysts with
dispersion of the MoO3 precursor and in limiting the coking supports containing hydrotalcite were more efficient than a
process which occurs during HDS with other supports. They traditional CoMo/Al2O3 catalyst for the selective HDS of
found that the texture of the starting MoO3/MgO was indeed FCC gasoline. The CoMo/Al2O3 catalyst was the most
maintained during its impregnation with methanol solutions active in HDS but also in the hydrogenation of olefins while
of Co or Ni nitrates and that the Co(Ni)Mo catalysts the CoMo catalyst with a mixed Zn–Al oxide support was
prepared in this manner were 1.5–2.3 times more active than the least active for both reactions but was, by far, the
their commercial Al2O3-supported counterparts in the HDS most efficient regarding the HDS/hydrogenation selectivity
of benzothiophene [140]. They also pointed out that with this (Fig. 7).
kind of support, the activity of the catalysts and in particular The same group [138] compared also a traditional
the promotion effect of Co(Ni) on MoS2 depended on the CoMo/Al2O3 catalyst to a CoMo deposited on MgO–Al2O3
calcination temperature very much. According to them this oxide obtained from hydrotalcite and to a CoMo catalyst
may explain for instance why they observed a much higher with MgO dispersed on Al2O3 as support. The supported
promotion effect than Klimova et al. [135]. Their catalysts MgO solid base was obtained by the so-called monolayer
were calcined at 350 8C for less than one hour while those of dispersion method which consisted in mixing Mg(NO3)2-
Klimova et al. were calcined at 500 8C for 24 h. The 6H2O with Al2O3 at a concentration (about 12 wt.%) such as
conclusion of Klicpera and Zdrazil [140] was that for a good a monolayer was obtained. The authors showed that the
promotion to occur two conditions must be fulfilled: (i) the catalyst obtained by this method exhibited the best HDS/
Co(Ni) promoter must be deposited by a method (non- hydrogenation selectivity in the HDS of FCC gasoline
aqueous) such that the dissolution of Mo species as (Fig. 8).
MgMoO4 and the hydration of MgO be avoided and (ii) They reported [137] also that the addition of HZSM-5 to
the promoted material should not be calcined above 300– the alumina support of a CoMo catalyst increased slightly
S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172 161

However, several authors [7,123] who studied also the


effect of alkali metals and alkaline earth elements on the
properties of hydrotreating catalysts did not find any
improvement regarding the HDS/hydrogenation selectivity.
For instance, Miller et al. [7] found no effect at all of the
addition of cesium and potassium to MoS2 and of potassium
to CoMo catalysts regarding both their activities and their
HDS/hydrogenation selectivity or octane loss in a range of
sulfur contents between zero and 800 ppm (Fig. 1).
Muralidhar et al. [123] found that sodium and potassium in
particular decreased both the HDS and hydrogenation
activities of CoMo catalysts markedly but that the latter was
Fig. 7. Hydrodesulfurization of FCC gasoline. HDS (open bars) and
less affected than the former. This of course is not the effect
hydrogenation (hatched bars) activities (% conversion) of alumina- and
hydrotalcite (HT)-supported catalysts (data from Ref. [139]). which would be desirable in the case of FCC gasoline.
On the opposite, Zhao et al. [136,137] obtained a
beneficial effect of alkali metals. By using their synthetic
gasoline made of thiophene and hex-1-ene in cyclohexane,
they found that potassium could indeed improve the HDS/
hydrogenation selectivity of CoMo catalysts with both
Al2O3 and hydrotalcite as supports (Table 3) [136]. A similar
effect although rather limited was obtained with barium in
the HDS of FCC gasoline [137]. Mey et al. [77,150,151]
observed also a positive effect of potassium on the HDS/
hydrogenation selectivity of a commercial CoMo catalyst
but this was at a relatively low HDS conversion. By using
also a model FCC gasoline (2-methylthiophene and 2,3-
dimethylbut-2-ene in solution in orthoxylene and n-heptane)
Fig. 8. Hydrodesulfurization of FCC gasoline. HDS (open bars) and olefin
they were able to measure the HDS and the olefin
hydrogenation (hatched bars) activities of sulfided CoMo catalysts. Cat-1: hydrogenation activities at the same time but through two
commercial CoMo/Al2O3 catalyst; Cat-2: CoMo supported on MgO–Al2O3 distinct reactions (see Section 4.2.1). The authors found that
obtained from hydrotalcite; Cat-3: CoMo supported on MgO–Al2O3 potassium inhibited HDS (Fig. 9a) less than olefin
obtained by the so-called monolayer dispersion method (data from Ref. hydrogenation (Fig. 9b), the result was an improvement
[138]).
of HDS/hydrogenation selectivity (Fig. 9c). They found also
that the isomerization of 2,3-dimethylbut-2-ene into of 2,3-
the RON (because of increasing hydroisomerization of the dimethylbut-1-ene was slower with the catalyst containing
paraffins) but decreased the HDS conversion of the gasoline. potassium (Fig. 9d).
However up to now, Zhao and coworkers [136–139] did As can be seen from the foregoing, there are noticeable
not propose any general explanation regarding the effect differences among the results which were reported. The
of the nature of the support on the selectivity of their reasons for such discrepancies are not clear yet. We may
hydrotreating catalysts. notice however that Muralidhar et al. [123] measured the
A complementary approach in line with the idea that less HDS and hydrogenation activities separately while Zhao et
acidic catalysts should be less active in hydrogenation was al. [136] and Mey et al. [77,150,151] made competitive
to modify either the support or the catalyst with various experiments. Moreover, the results may also depend on the
additives such as alkali metals, alkaline earth elements, conversion which is not always known precisely. Actually, it
carbon or nitrogen poisons. is sensible to expect that when approaching 100% HDS
conversion it becomes more and more difficult, even with
selective catalysts, to avoid olefin hydrogenation since the
4.2.3.2. Alkali metals and alkaline earth elements. The use inhibition of the latter by sulfur compounds which is quite
of such additives was recommended in several patents [145– significant at low HDS conversion [76] is disappearing.
149]. In particular, Hatanaka et al. [148,149] showed that the Nevertheless, several proposals were made to account
presence of potassium (either predeposited on the support for the beneficial effect of alkali metals and alkaline earth
or introduced at the same time as the Co and Mo active elements. As we shall see, some of them are hard to
components) in a HDS catalyst inhibited coke deposition conciliate. Hatanaka et al. [148,149] supposed that
and therefore improved the HDS stability of the catalyst over potassium inhibited coke deposition through olefin poly-
a large period of time. They claimed also that it could merization and therefore improved the HDS stability of the
minimize octane loss [149]. catalyst. However, they observed also that octane loss was
162 S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172

Fig. 9. Transformation of a model FCC gasoline (T = 200 8C, P = 20 bar). Formation of the HDS products from 2-methylthiophene vs. residence time (a);
formation of the hydrogenation products (HYDO) from 2,3-dimethylbut-2-ene vs. residence time (b); HDS/HYDO selectivity vs. HDS yield (c); formation of
2,3-dimethylbut-1-ene from 2,3-dimethylbut-2-ene vs. HDS yield (d). Effect of the presence of potassium [150,151].

limited presumably because olefin hydrogenation was Another possible explanation proposed by Mey et al.
also inhibited. Consequently, it must be assumed that both [77,150,151] is that the decrease in olefin hydrogenation
reactions occur on discrete centers otherwise olefin activity of the potassium-containing catalyst compared to
hydrogenation should be equally stabilized. However, the the non-modified catalyst would be the consequence of the
catalysts were not compared exactly under the same con- inhibition of double bond isomerization due to the
ditions, which makes the conclusions questionable. More- neutralization by the alkali metal of the acid sites of the
over, as we will see in the next section coke predeposition support. Actually, with the unmodified catalyst, it was found
was also found to improve the HDS/hydrogenation that the isomerization of 2,3-dimethylbut-2-ene into 2,3-
selectivity [19,67]. Therefore we can wonder how both dimethylbut-1-ene was very fast [76] and that the latter (a
coke deposition and its inhibition could produce the same terminal olefin) hydrogenated much faster than the former
effect. (an olefin with a tetrasubstituted double bond) [77,151]. This
In line with the interpretation of support effects it can may also explain at least in part the effect of potassium on
also be supposed that alkali metals and alkaline earth the HDS/hydrogenation selectivity. With the unmodified
elements would induce electronic effects which would catalyst, the isomerization was very fast so that 2,3-
limit hydrogenation in the same manner as basic supports dimethylbut-1-ene could be regenerated in concentrations
[118–120]. According to Mey et al. [150,151] such close to the equilibrium as soon as it was consumed through
electronic effects existed with their potassium-containing hydrogenation, which was apparently no longer the case
catalyst as shown by FT-IR spectroscopy of adsorbed CO with the catalyst containing potassium. This explanation
and could explain in part at least the decrease in may also account for the effect of potassium reported by
hydrogenation activity they observed. However, the fact Hatanaka et al. [149] since catalytically cracked gasoline
that HDS is supposed to involve hydrogenation steps prior contains mostly internal olefins ( 75 vol.%) which are less
to C–S bond cleavage makes this explanation question- reactive than 1-olefins. It may account for the effect of
able unless we suppose again, as several authors did coking they reported as well (see next section) [19,67].
[19,41,49,51,53,68], that the reactions occur on different They found that coking limited iso-olefin (di-iso-butylene)
catalytic centers and that potassium inhibits those involved hydrogenation more than terminal olefin (oct-1-ene)
in olefin hydrogenation mainly. One may notice that this hydrogenation, which they attributed to the deactivation
constitutes an easy way out. by coking of catalytic centers which would hydrogenate iso-
S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172 163

Fig. 11. Hydrodesulfurization of FCC gasoline over a CoMo/Al2O3 catalyst


Fig. 10. Effect of coke deposition on the activities of a commercial CoMo/ aged with straight run naphtha (300 8C, 0.4 MPa). Effect of time-on-
Al2O3 catalyst in the HDS of thiophene (*) and in the hydrogenation of stream on the HDS (*) and hydrogenation (*) activities (data from
cyclohexene (*) (data from Ref. [155]). Ref. [19]).

olefins specifically. An alternative explanation would be that deactivation in HDS, if any was made undetectable under the
coking deactivates the acid sites which are the most efficient conditions of the experiment. Their interpretation was that
in isomerizing iso-olefins into terminal olefins. coke formation resulting from olefin oligomerization deac-
tivated the olefin hydrogenation sites selectively. Moreover,
4.2.3.3. Carbon predeposition—deactivated or spent cata- the authors observed that part of the adsorbed oligomers
lysts. Deactivated hydrotreating catalysts [152] or spent could be washed out from the catalyst by toluene while the
resid upgrading catalysts [153] were claimed to achieve rest was irreversibly deposited on it. This added to the fact
extensive and selective HDS of FCC naphtha, that is HDS that they observed two periods of deactivation regarding
with minimum octane loss. Of course, the reason for the olefin hydrogenation (fast at the beginning then slow and
improvement of the selectivity of these catalysts as steady) led them to conclude in accordance with their
compared to the selectivity of regular (fresh) hydrotreating previous proposal [68] that there were two different active
catalysts is not clear because, in addition to coke, the sites for olefin hydrogenation, one which is easily deac-
deactivated or spent catalysts can contain other deposits such tivated through olefin oligomerization and the other hardly.
as metals (Fe, Ni, V, etc.) as well as other contaminants, for As indicated in the previous section another possible
instance heavy sulfur- or nitrogen-containing compounds. explanation is that the catalytic centers (presumably on the
Arteaga et al. reported a study on the deactivation and support) active in olefin isomerization are rapidly deacti-
regeneration of an industrial CoMo/Al2O3 catalyst vated through coke deposition so that after a short period on
[154,155]. The catalyst ageing was obtained by coke stream the isomerization of internal olefins into terminal
deposition from either buta-1,3-diene or methylcyclopen- olefins becomes very slow and, from then on, the fraction
tane. The HDS and hydrogenation activities were measured of olefins which can be the most rapidly hydrogenated
by using a mixture containing thiophene and cyclohexene corresponds only to the terminal olefins originally present in
dissolved in cyclohexane. They found that coke deposition the feed. Consequently, since internal olefins are much less
inhibited both HDS and olefin hydrogenation (Fig. 10). reactive than terminal olefins, the extent of hydrogenation
However, the HDS/hydrogenation selectivity was apparently becomes less significant.
little affected by coke deposition [154]. Whatever the explanation, the effect of coke deposition
More recently, Hatanaka et al. [19,67] studied the effect on the HDS/hydrogenation selectivity was patently obvious
of coke deposition on the HDS of catalytically cracked from the results of Hatanaka and coworkers and it gave them
gasoline. The authors examined the HDS of FCC gasoline on the idea that it could be possible to selectively predeactivate
a presulfided CoMo catalyst aged with a straight run naphtha the hydrogenation sites by a coking pretreatment. In the
[19]. They found that the HDS/hydrogenation selectivity continuation of their study, they compared fresh catalysts to
changed with time-on-stream: the HDS activity decreased the same catalysts which underwent various pretreatments:
gradually from the beginning while the olefin hydrogenation simple ageing with straight run naphtha after sulfiding;
activity decreased drastically during the first 30 h on stream sulfiding and ageing followed by coking with a mixture of
and stabilized afterwards (Fig. 11). According to them, this cyclohexene and 1-methylnaphthalene. They found [19,67]
confirms their previous assumption [68] that HDS on the one that at equal HDS conversions of a FCC gasoline, the olefin
hand and part at least of olefin hydrogenation on the other conversion was much lower with the catalyst on which coke
occur on different catalytic centers (see Sections 4.1.2 had been deposited than with the fresh sulfided and aged
and 4.2.2). However, it should be mentioned that HDS catalyst (Fig. 12). In other words the precoked catalyst had a
conversion started from 100% (Fig. 11) so that initial better HDS/hydrogenation selectivity than the fresh sulfided
164 S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172

Fig. 12. Hydrodesulfurization of FCC gasoline over CoMo/Al2O3 catalysts Fig. 13. Effect of various catalyst pretreatments on the HDS of thiophene in
(300 8C, 0.4 MPa). Effect of various treatments on HDS and olefin hydro- the presence of di-iso-butylene (CoMo/Al2O3; 175 8C; 1.3 MPa; data from
genation (data from Ref. [19]): (*) sulfided and aged fresh catalyst; (*) Ref. [19]): (*) sulfided catalyst; (*) sulfided and aged catalyst; (&)
sulfided, aged and coked fresh catalyst; (&) coked, sulfided and aged fresh sulfided, aged and coked catalyst.
catalyst; (^) sulfided spent catalyst; (&) regenerated, sulfided and aged
spent catalyst. decrease the reactivity (and its inhibition increase it). This
was actually what was observed in the work of Hatanaka et
and aged catalyst. They found also, in accordance with the al. [19].
Texaco patents [152,153], that a refinery spent catalyst
containing 8.8 wt.% coke had a good HDS/hydrogenation 4.2.3.4. Nitrogen. In a recent patent, Hatanaka et al. [156]
selectivity and noticed that the effect disappeared when the reported that the modification with a basic organic nitrogen
spent catalyst was regenerated. On the other hand, coking compound of a presulfided CoMo or NiMo hydrotreating
before sulfiding had no effect on the selectivity. Again they catalyst led to an increase of its HDS/hydrogenation
concluded on the basis of the assumption of three different selectivity. Starting from the observation made in a previous
catalytic centers (for thiophene HDS (type 1), for n-olefin study that H2S poisoned HDS and the hydrogenation of n-
hydrogenation (type 2) and for iso-olefin hydrogenation olefins but activated the hydrogenation of isoolefins which
(type 3)) that the coking pretreatment deactivated type 3 are found in large amounts in FCC gasoline [67,68], they
centers selectively, which again can be questioned. Actually, thought that a basic poison should have the reverse effect.
their main argument was that their FCC gasoline contained Actually, they found that after treating the presulfided
mostly iso-olefins. However, as indicated above it is clear catalyst with a basic compound such as pyridine for instance
that under the HDS conditions internal olefins (n-olefins as at 150 8C, then returning to a nitrogen-free feedstock, the
well as iso-olefins) can isomerize into terminal olefins, catalyst could recover nearly 100% of its original HDS
especially on fresh catalysts. Moreover, one can wonder activity but 50% only of its olefin hydrogenation activity.
whether it is really sensible to call for three different types of Hence, the HDS/hydrogenation selectivity of the catalyst
sites. In fact the authors acknowledge that the way that coke was increased significantly and the decrease in octane
could affect the three categories of sites is unclear. To number prevented to a large extent. The effect of the
ascertain their interpretation they used again three different nitrogen compound was confirmed by studying the HDS of
catalysts (sulfided, sulfided and aged, sulfided aged and both synthetic gasolines containing thiophene, benzothio-
coked) in the transformation of model reaction mixtures
made of thiophene and di-iso-butylene on the one hand, and
of thiophene and oct-1-ene on the other [19]. The results
confirmed that the hydrogenation of iso-olefins (here di-iso-
butylene) was markedly decreased by coking with respect to
the HDS of thiophene (Fig. 13) while (surprisingly) oct-1-
ene hydrogenation was less inhibited than thiophene HDS
(Fig. 14). Again, this can perfectly be interpreted in terms of
isomerization of the olefins and of its inhibition. Actually,
the isomerization of the olefins (and consequently its
inhibition) is expected to have opposite effects on the
hydrogenation depending on whether we are starting from
an internal or a terminal olefin: in the former case the
isomerization should increase the reactivity since terminal
Fig. 14. Effect of various catalyst pretreatments on the HDS of thiophene in
olefins are more reactive than internal olefins (and the presence of oct-1-ene (CoMo/Al2O3; 175 8C; 1.3 MPa; data from Ref.
conversely its inhibition should decrease it) while, for the [19]): (*) sulfided catalyst; (*) sulfided and aged catalyst; (&) sulfided,
same reason, in the latter case, the isomerization should aged and coked catalyst.
S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172 165

phene and olefins and FCC gasoline in the presence of a


presulfided CoMo/Al2O3 catalyst treated with pyridine at
150 8C [157]. It was found that after partial desorption of the
latter at 300 8C, the HDS of thiophene was little affected and
that of benzothiophene was slightly inhibited while, on the
other hand, olefin hydrogenation and thiol formation were
strongly inhibited, which means that the HDS/hydrogena-
tion selectivity of the catalyst was improved.

4.3. Existing desulfurization processes and


technologies—practical aspects

Several recent review articles were dedicated to new


technologies and processes for deep desulfurization of motor Fig. 15. The Prime G+1 process (IFP-AXENS).
fuels [2,4,5,158,159]. In this section we will focus on the
desulfurization of FCC gasoline and more specifically on the
selective catalytic processes. Other processes regarding the mercaptans. Under the mild conditions of the process
purification of FCC gasoline will be reviewed briefly. (relatively low temperature and medium pressure), this
As pointed out by Rhodes [160], hydrotreating FCC reaction which is thermodynamically controlled can afford
gasoline through a conventional process would lead to a relatively high concentrations of these ‘‘recombination
significant octane loss. Because of this, existing catalytic mercaptans’’ and can therefore constitute a serious obstacle
processes for the desulfurization of FCC gasoline are the to achieve the goal of ultra-low content of sulfur imposed by
result of two different strategies: selective HDS preserving the regulation (10–15 wt ppm S) for the near future.
octane number (Prime G+1, SCANfining1 and catalytic
distillation (CDHydro/CDHDS1)) and deep desulfurization 4.3.1.1. The ‘‘Prime G+1’’ process (Axens-IFP). The
associated to octane recovery through alkane isomerization Prime G+1 process [161–163] is licensed worldwide and
(Oct-Gain and Isal). However, up to now, the selective HDS today more than 80 units are under design, construction or
processes are by far the ones which are the most generally operation. Typically, the full Prime G+1 process scheme
selected by the refiners. comprises a selective hydrogenation unit (SHU), a splitter
(which separates the light from medium and heavy naphtha)
4.3.1. Selective HDS processes and a selective HDS unit for the treatment of the medium and
FCC gasoline can be separated into three main fractions heavy cracked naphtha (Fig. 15).
(light, medium and heavy cracked naphtha) through splitting Three main reactions are taking place in the SHU (Prime
or distillation [158]. The light fraction (LCN) contains a G+1 first step): (i) the selective hydrogenation of the
large part of the olefins with mercaptans as sulfur impurities diolefins (typically the diolefin content in the full range FCC
which can be withdrawn through caustic treatment. The naphtha goes from 0.5 to 2 wt.%); (ii) the double bond
medium or intermediate fraction (MCN or ICN) has a low isomerization of the olefins; (iii) the conversion of light
octane rating and contains essentially thiophene and light mercaptans and light sulfides into heavier sulfur compounds.
alkylthiophenes as sulfur impurities. The HCN fraction The first reaction ensures the protection of the catalyst used
contains a relatively low concentration of olefins but most in the selective HDS unit from deactivation. The specific
of the sulfur impurities including benzothiophene and tailoring of the catalyst (HR845) and the choice of
alkylbenzothiophenes [158 and references therein]. appropriate process conditions allow the total hydrogenation
Several processes (Prime G+1, SCANfining1, CDHydro/ of the diolefins with a very limited amount of olefin
CDHDS1) combine this possibility of fractionating the saturation. The double bond isomerization reaction trans-
gasoline with catalytic HDS in order to optimize the overall forms the olefins present in the FCC naphtha into their
HDS/hydrogenation selectivity and to minimize octane loss. thermodynamic equilibrium at the temperature of the SHU
Moreover, as we will see, selective HDS processes generally reactor. In fact, this reduces the amount of terminal olefins
include a selective hydrogenation unit ahead of the HDS and produces higher octane olefins (internal olefins have a
section in order to ensure the required duration under better RON than terminal olefins [164]). As we have seen,
operation of the HDS catalyst (typically 3–5 years). The this could also contribute to the limitation of olefin
main purpose of this unit is to selectively hydrogenate the saturation in the HDS reactor since terminal olefins are
diolefins contained in the FCC gasoline so as to prevent the more reactive in hydrogenation than internal olefins. The
deactivation of the catalyst. As mentioned previously conversion of light mercaptans and light sulfides to heavier
(Section 4.2.1.5), the difficulty to reach high levels of sulfur compounds through reaction with olefinic compounds
sulfur removal in the hydrotreated gasoline is partly due to makes it possible to transfer them to the medium and heavy
the possible recombination of H2S with olefins to form fractions which will be hydrotreated in the HDS unit.
166 S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172

Fig. 16. Selectivity of the Prime G+1 process compared to conventional


hydrotreating. Olefin saturation vs. extent of HDS. Fig. 17. The Prime G+1/OATS1 integration [166].

The combination of the splitter with the SHU (Prime thiophenic sulfur, see Section 4.3.3) process from BP in
G+1 first step) leads to the production of a mercaptan-free, conjunction with the Prime G+1 process [165,166].
low-sulfur LCN fraction (C5, 65 8C) which does not need to Similarly to the SHU of Prime G+1, the OATS1 process
be hydroprocessed. Hence, the most valuable and reactive converts the light sulfur species (sulfides, mercaptans) into
olefins in the feed (typically C5-olefins) are preserved from heavier sulfur compounds, but in addition it transforms also
hydrogenation and recombination reactions. It has to be most of the thiophene into alkylthiophenes through its
mentioned that the end point of the LCN stream (65 8C) is reaction with light olefins. Therefore, the end-point of the
imposed by the presence of thiophene in the feed. Actually, stream coming out of this unit can be as high as 100 8C.
a great challenge of the fractionating step is to avoid Actually, when integrated in the Prime G+1 process [166],
the distilling of thiophene into the light fraction. This is the OATS1 unit is fed with the MCN (65–120 8C) coming
because thiophenic sulfur cannot be easily removed by the from the splitter next to the SHU (Fig. 17). This MCN
conventional or related extractive processes and can fraction contains olefins, thiophene and methylthiophenes.
therefore contribute significantly to the presence of sulfur The OATS1 unit makes it possible to switch these
in the gasoline pool if it is left in the light fraction. Another thiophenic compounds to the HCN fraction which goes to
advantage of the presence of the splitter is that it makes it the Prime G+1 selective HDS unit. The resulting benefit of
possible to reduce the size of the HDS unit by decreasing the the Prime G+1/OATS1 association is (i) the production of
volume of the gasoline to be treated. ultra-low sulfur, mercaptan free LCN (C5, 65 8C) with no
The Prime G+1 second unit (selective HDS) achieves the octane loss thank to the Prime G+1 first step, (ii) the
deep HDS of the MCN and HCN fractions with (i) a minimum production of ultra-low sulfur MCN (65–100 8C) with no
of olefin saturation (and no aromatic loss) leading to a octane loss thanks to the OATS1 reactor and (iii) the
maximum octane retention (Fig. 16), (ii) a minimized amount production of ultra-low sulfur HCN (100 8C+) with minimal
of recombination mercaptans (production of a ‘‘sweet’’ octane loss and a minimized amount of mercaptans, thanks
naphtha) and (iii) no cracking reactions, which means nearly to the Prime G+1 second step
100% liquid yield (no increase in light product formation and
consequently no Reid vapor pressure increase). All this can be 4.3.1.2. The ‘‘SCANfining1’’ process (ExxonMobil). This
achieved thanks to a dual catalyst reactor system operated process was invented by Exxon Research and Engineering
under optimal conditions. The first HDS catalyst (HR806) was and Akzo Nobel who developed jointly a selective HDS
tailored so as to have a minimum hydrogenation activity catalyst called RT-225 for this purpose [2 and references
towards olefins while having still sufficient hydrodesulfuriza- therein, 167]. At the end of 2003, 32 refineries had selected
tion activity to convert the most refractory sulfur compounds the SCANfining1 process [168]. The process is based both
encountered in cracked naphtha. The polishing catalyst on the development of a selective HDS catalyst and on the
(HR841) exhibits almost no olefin hydrogenation activity at optimization of process conditions [167]. Fig. 18 shows that
all, but is able to reduce the amount of recombination with such a selective catalyst when used in SCANfining1,
mercaptans and to complete the desulfurization of the final the extent of olefin saturation is much lower than with a
product in order to meet the legal requirements. Moreover, regular HDS catalyst used in a conventional HDS process
hydrogen consumption of the overall process is also limited as (20% instead of 80%) and therefore the octane loss is also
is the case with all selective processes. minimized. The process (Fig. 19) has some similarities with
The first two commercial units were started in September the Axens Prime G+1 process.
2001 in Germany. They produced a 10 ppm sulfur gasoline The naphtha feed is first pretreated with hydrogen in a
with a very limited octane loss (1–1.5 points only). saturator designed for the conversion of the diolefins which
Recently Axens-IFP and BP formed an alliance which can cause fouling in the heat exchangers as well as in the
enables Axens to offer the OATS1 (olefin alkylation of HDS reactor. The effluent of the pretreating section is heated
S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172 167

Fig. 18. Selectivity of the SCANfining process (A) compared to conventional hydrotreating (B) [2].

Fig. 19. Schematic diagram of the SCANfining process (ExxonMobil) [158].

and passes through the HDS reactor (containing a CoMo The authors [2] claim that the SCANfining1 selective
catalyst (RT225)) where it is desulfurized with a minimum catalyst developed by ExxonMobil and Akzo Nobel could
of olefin saturation. The desulfurized naphtha is then cooled still be improved by poisoning selectively hydrogenation (or
and fractionated. As for the Axens Prime G+1 process, rim) sites according to the concept proposed by Daage et al.
fractionation is essential and the design of the splitter as well [128,170]
as the choice of the cut points are particularly critical [169].
Because of the very low sulfur content objectives it is 4.3.1.3. Catalytic distillation (CDTech). Catalytic distilla-
necessary to remove the sulfur from the LCN which contains tion (CD) combines separation of FCC gasoline in several
the valuable C5–C6 hydrocarbons. This can be done with fractions through distillation and catalytic HDS in a same
caustic extraction for example, without affecting the olefins. elaborate process using in its simplest fitting up a single
However, as for the Prime G+1 process, it is very important reactor or vessel, in fact a distillation column packed with a
to prevent thiophene from distilling into the light fraction. desulfurization catalyst [5,158 and references therein]. As
Other options of this process are also available [169]. we have seen, the concentration of olefins as well as the
SCANfining1 can be associated with Sweeten Plus1 pre- concentration and nature of the organic sulfur compounds in
treating which involves the oxidation of the mercaptans to the gasoline depend very much on the boiling range of the
higher boiling point disulfides using a conventional caustic latter. The olefins concentrate in the low-boiling fraction
sweetener. The high-boiling point disulfides are known to be while most of the sulfur compounds are found in the high-
stable enough to be separated from the LCN fraction and boiling fraction. Moreover, the differences in nature of the
sent to the HDS reactor together with the intermediate and sulfur compounds require different levels of severity in order
heavy cracked naphtha. The post-treatment of the effluent to achieve complete or nearly complete elimination of these
with the EXOMER1 process (extraction of mercaptans, see compounds. Catalytic distillation makes it possible to treat
Section 4.3.3) is also possible. separately various fractions of FCC gasoline under the most
168 S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172

Fig. 20. Schematic diagram of the CDHDS process (CDTech) [5].

appropriate conditions for each of them in a single operation while the rest of the diolefins is hydrogenated into olefins.
and allows at the same time to preserve to a large extent The bottoms (medium plus heavy fractions) enter the
olefins from saturation. In fact, the severity of the CDHDS1 section where they undergo desulfurization under
desulfurization conditions is controlled by the boiling such conditions that a maximum sulfur removal is achieved
temperature range of the fraction itself. The lighter fractions and a minimum of saturation of the olefins is occurring
containing the olefins and the most reactive sulfur thanks to the fact that the latter concentrate at the top of the
compounds will be in contact with the desulfurization column where the temperature is lower.
catalyst at a relatively low temperature at the top of the
column. The heavier fractions containing more refractory
4.3.2. HDS processes with octane recovery or
sulfur impurities will undergo desulfurization at higher
compensation
temperatures at the bottom of the column (Fig. 20).
The process can be made even more efficient by using 4.3.2.1. Oct-Gain1 (ExxonMobil) and ISAL1 (UOP-IN-
two columns instead of one (Fig. 21). In this design, two TEVEP) processes. These processes do not resort to
distillation columns (CDHydro1 for the light fractions and selective HDS. They associate conventional deep HDS to
CDHDS1 for the heavy fractions) are packed with subsequent octane recovery through isomerization of the
desulfurization catalysts. CDHydro1 can be used separately paraffins and alkylation essentially. These two processes are
to produce various hydrogenated fractions while in quite similar but use different catalysts and operating
combination with CDHDS1 it is used to produce low conditions [158 and references therein]. Both use a fixed bed
sulfur gasoline with a minimum octane loss. In this case the reactor with a first catalyst bed over which the organic sulfur
mercaptans and part of the diolefins contained in the lighter compounds of the feed are converted into H2S and
fraction combine together to give heavy sulfur compounds hydrocarbons and the olefins almost completely hydro-
genated. The second bed containing a different catalyst leads
to octane recovery through cracking, isomerization and
alkylation of the paraffins, which compensates a significant
olefin saturation [171,172]. The ISAL1 process uses a
CoMo–P/Al2O3 associated to a Ga–Cr/HZSM-5 zeolite
[5,158 and references therein, 173]. Anyhow the second
catalyst, if not a pure acidic catalyst and especially when
containing precious metals must be designed to be sulfur
tolerant [174,175]. One of the drawbacks of these processes
can be some yield loss in gasoline due to cracking into light
products. This however could be improved with new
catalysts used in the ISAL1 process [172]. Another
disadvantage with respect to selective processes is the
hydrogen consumption needed for the quasi total olefin
Fig. 21. Combined CDHydro and CDHDS technologies (CDTech) [158]. saturation.
S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172 169

4.3.2.2. Prearomatizing and HDS (Phillips Petroleum is apparently not very attractive regarding selectivity and
Co.). Another option is to prearomatize the gasoline before efficiency and the choice of the solvent seems to be really
regular HDS [176]. Although olefin saturation is almost critical [5].
complete, octane is boosted because of the increase in OATS1 (olefin alkylation of thiophenic sulfur) from BP
aromatic concentration. However, the present and future [183,184] is an acid-catalyzed process which makes it
regulations limiting aromatic concentration in gasoline possible to shift the sulfur compounds from light to heavier
make this process much less attractive than Oct-Gain1 or fractions of the gasoline by increasing their molecular
ISAL1 [5]. weight. This can be obtained by alkylation of thiophene with
the olefins present in the gasoline. As we have seen, this
4.3.3. Other processes process can be advantageously associated with other
Numerous other processes (mainly non-catalytic) can be processes such as Prime G+1 for instance. Other processes
considered in order to achieve deep desulfurization of FCC are also proposed in view of the separation of sulfur
gasoline (for details, see for instance, [5,158]). Existing or compounds from the gasoline such as S-Brane1 (Grace)
promising technologies involve selective or reactive which uses a polymer membrane [185] and Exomer1
adsorption as well as alkylation of sulfur compounds. (Merichem, ExxonMobil) involving caustic extraction
Selective oxidation or extraction seem up to now less [186]. Processes aiming at the extraction of sulfur by using
appropriate for FCC gasoline and will not be considered ionic liquids [187,188] or by photochemical oxidation [189]
here. are also considered.
S-Zorb1, a process from Phillips Petroleum Co. [177]
uses a solid sorbent (a metal that reacts with sulfur to give a
sulfide) which is continuously withdrawn from a fluid (or 5. Conclusion
bubbling) bed reactor. The sorbent is able to extract the
sulfur from the organic molecule and to release the There are several possibilities regarding the origin of
hydrocarbon part of it in the medium. The process operates sulfur impurities in FCC gasoline.
in gas phase to avoid retention of the oil in the porosity. If long chain alkylthiophenes are present in the feed, they
Hydrogen is needed to limit coking as well as to help the can easily transform into thiophenic compounds during the
extraction of the sulfur from the organic molecules and its FCC process as shown in the case of n-hexylthiophene for
capture by the sorbent. The sulfided sorbent is sent to a instance. However, long chain alkylthiophenes can also
regenerator where the sulfur is burnt. The sorbent is then result from the alkylation of thiophene by olefins issuing
reduced before recycling. As shown by Tawara et al. [178– from the cracking of the paraffins and lead to short chain
180], zinc oxide associated to nickel oxide can play such a alkylthiophenes (methylthiophenes in particular) through
role of sorbent through a mechanism proposed by Babich cracking of the side chains. The thiophenic compounds
and Moulijn [5]. Actually, the sorbent consists in silica (20– present in FCC gasoline can also result from a series of
60 wt.%), alumina (5–15 wt.%), zinc oxide (15–60 wt.%) consecutive reactions including the addition of H2S to the
and nickel and/or cobalt (a few wt.%). Hydrogen olefins or diolefins present in the reactor, cyclization into
consumption is as low as 4.4–13 l/l of gasoline, the pressure tetrahydrothiophene derivatives, dehydrogenation of the
is between 7 and 21 bar, the temperature between 340 and latter through H-transfer.
415 8C and the LHSV between 4 and 10 h1. This process is Part of the benzothiophene compounds present in the
really innovating as the chemistry it involves is inventive high boiling fraction of the gasoline can result from the
compared with conventional HDS processes. It really limits transformation of heavier substituted benzothiophene
the octane number loss but is very costly regarding inves- derivatives present in the feed. However, they can also
tments, operating costs and SOx treatment after regenera- result from the cyclization of long chain (containing at least
tion. This added to some apprehension of the refining five carbon atoms) alkylthiophenes.
industry to use non-fixed bed technologies explains probably Thiols and sulfides result mainly from the addition of H2S
why a very few units have been built up to now. to olefins. They can also be the result of the partial
The IRVAD1 process developed by Black and Veatch decomposition of other sulfur impurities.
Pritchard [181] which uses a promoted alumina-based Numerous studies are in favor of distinct centers for HDS
sorbent and the TReND1 process, another reactive and olefin hydrogenation. This encouraged numerous groups
adsorption process by the Research Triangle Institute to modify conventional catalysts or to design new catalysts
[182] are similar in their principle to S-Zorb1 but are still in view of favoring HDS with respect to the hydrogenation of
at the pilot or exploratory stage. Other selective adsorption olefins.
processes such as SARS (selective adsorption for removing The selectivity of hydrotreating catalysts used in the HDS
sulfur) from Pennsylvania State University [158 and of FCC gasoline can be estimated at best by using model
references therein] remain at the laboratory scale. synthetic gasoline or real FCC gasoline. It is clear that the
GT-DeSulf1 (GTC) separates the organosulfur com- presence of a promoter (Co in particular) favors the HDS/
pounds and aromatics by extractive distillation. This process olefin hydrogenation selectivity of the catalyst. In general it
170 S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172

was found that the promoter enhanced HDS or C–S bond [12] J. Fu, P. Wang, M.-Y. He, Am. Chem. Soc. Prepr. Div. Pet. Chem. 45
cleavage more than hydrogenation steps. (4) (2000) 697.
[13] T. Myrstad, B. Seljestokken, H. Engan, E. Rytter, Appl. Catal. A:
The nature of the support has also an influence on the Gen. 192 (2000) 299.
selectivity. Basic supports such as magnesium oxide and [14] A. Corma, C. Martinez, P. Gullbrand, Am. Chem. Soc. Prepr. Div.
hydrotalcite in particular lead to improved HDS/olefin Pet. Chem. 44 (4) (1999) 490.
hydrogenation selectivity. Another approach was to modify [15] A. Corma, C. Martinez, G. Ketley, G. Blair, Appl. Catal. A: Gen. 208
(2001) 135.
conventional catalysts with alkali metals or alkaline earth
[16] T.G. Albro, P. Dreifuss, R.F. Wormsbecher, J. High Res. Chromatogr.
elements. It was indeed found that these elements can 16 (1993) 13.
definitely improve the selectivity although there exist some [17] W.C. Cheng, G. Kim, A.W. Peters, X. Zhao, K. Rajagopalan,
discrepancies among the results. In some instances no effect M.S. Ziebarth, C.J. Pereira, Catal. Rev.-Sci. Eng. 40 (1/2) (1998)
at all was detected. These discrepancies can be attributed 39.
mainly to differences in HDS conversion. It is clear that [18] U. Alkemade, T.J. Dougan, in: M. Absi-Halabi, J. Beshara, H.
Qabazard, A. Stanislaus (Eds.), Catalysts in Petroleum Refining
when approaching 100% HDS conversion, the experimental and Petrochemical Industries 1995, Elsevier, 1996, p. 303.
or process conditions have to be adjusted carefully otherwise [19] S. Hatanaka, M. Yamada, O. Sadakane, Ind. Eng. Chem. Res. 37
the effect of any modifier on the HDS/olefin hydrogenation (1998) 1748.
selectivity is bound to disappear. [20] C. Yin, G. Zhu, D. Xia, Am. Chem. Soc. Prepr. Div. Pet. Chem. 47 (4)
Preadsorbed basic poisons such as pyridine or carbon (2002) 391.
[21] C. Yin, G. Zhu, D. Xia, Am. Chem. Soc. Prepr. Div. Pet. Chem. 47 (4)
predeposits were also found to improve the selectivity. (2002) 398.
Various interpretations can be put forward such as [22] B.C. Gates, Catalytic Chemistry, Wiley, New York, 1992, 254.
electronic effects due to a decrease in acidity of the support [23] E.G. Wollaston, W.L. Forsythe, I.A. Vasalos, Oil Gas J. 69 (1971) 64.
which would decrease the hydrogenation activity of the [24] D.D. Whitehurst, T. Isoda, I. Mochida, Adv. Catal. 42 (1998) 345.
[25] P. Leflaive, J.L. Lemberton, G. Pérot, C. Mirgain, J.Y. Carriat, J.M.
catalyst or selective poisoning of the hydrogenation centers.
Colin, Appl. Catal. A: Gen. 227 (2002) 201.
Another explanation could be the neutralization or poison- [26] A. Lopez Agudo, A. Benitez, J.L.G. Fierro, J.M. Palacios, J. Neira, R.
ing of the sites (presumably on the support) which can Cid, J. Chem. Soc., Faraday Trans. 88 (3) (1992) 385.
catalyze the isomerization of the internal olefins into more [27] R. Cid, J. Neira, J. Godoy, J.M. Palacios, S. Mendioroz, A. Lopez
reactive terminal olefins. Agudo, J. Catal. 141 (1993) 206.
Among the catalytic desulfurization processes, selective [28] W.J.J. Welters, V.H.J. de Beer, R.A. van Santen, Appl. Catal. A: Gen.
119 (1994) 253.
HDS processes like SCANfining1 and Prime G+1 seem to [29] X. Saintigny, R.A. van Santen, S. Clémendot, F. Hutschka, J. Catal.
be very efficient and are already in practice world-wide. 183 (1999) 107.
Indeed, both of these processes which are based on [30] R.H. Harding, R.R. Gatte, J.A. Whitecavage, R.F. Wormsbecher, Am.
conventional catalytic fixed-bed technology and need low Chem. Soc. Symp. Ser. 552 (1994) 286.
[31] C.L. Garcia, J.A. Lercher, J. Phys. Chem. 96 (1992) 2669.
cost investments, are by far the most attractive for the
[32] S.Y. Yu, W. Li, E. Iglesia, J. Catal. 187 (1999) 257.
industry. Catalytic distillation seems also promising. Other [33] S.Y. Yu, W. Li, E. Iglesia, Am. Chem. Soc. Prepr. Div. Pet. Chem. 45
possibilities are to restore octane after conventional HDS (2) (2000) 354.
(with processes such as Oct-Gain1 or ISAL1) or to anti- [34] S.Y. Yu, W. Li, E. Iglesia, in: A. Corma, F.V. Melo, S. Mendioroz,
cipate octane loss by prearomatizing the feed. J.L.G. Fierro (Eds.), Proceedings of the 12th International Congress
on Catalysis, Elsevier, Stud. Surf. Sci. Catal. 130 (2000) 899.
[35] C.D. Chang, A.G. Silvestri, J. Catal. 47 (1977) 249.
[36] G. Zhu, D. Xia, G. Que, Am. Chem. Soc. Prepr. Div. Pet. Chem. 46
(4) (2001) 329.
References [37] A.A. Lappas, J. Valla, I.A. Vasalos, C.W. Kuehler, J. Francis, P.
O’Connor, N.J. Gudde, Am. Chem. Soc. Prepr. Div. Pet. Chem. 47 (1)
[1] R. Gatte, R. Harding, T. Albro, D. Chin, R.F. Wormsbecher, Am. (2002) 50.
Chem. Soc. Prepr. Div. Fuel Chem. 37 (1) (1992) 33. [38] A. Drahoradova, M. Zdrazil, React. Kinet. Catal. Lett. 33 (2) (1987)
[2] T.G. Kaufmann, A. Kaldor, G.F. Stuntz, M.C. Kerby, L.L. Ansell, 459.
Catal. Today 62 (2000) 77. [39] R.M. Casagrande, W.K. Meerbott, A.F. Sartor, R.P. Trainer, Ind. Eng.
[3] C. Marcilly, Stud. Surf. Sci. Catal. 135 (2001) 37. Chem. 47 (4) (1955) 744.
[4] C. Song, X. Ma, Appl. Catal. B: Environ. 41 (2003) 207. [40] W.K. Meerbott, G.P. Hinds, Ind. Eng. Chem. 47 (4) (1955) 749.
[5] I.V. Babich, J.A. Moulijn, Fuel 82 (2003) 607. [41] P. Desikan, C.H. Amberg, Can. J. Chem. 42 (1964) 843.
[6] R.L. Martin, J.A. Grant, Anal. Chem. 37 (6) (1965) 649. [42] P. Kieran, C. Kemball, J. Catal. 4 (1965) 394.
[7] J.T. Miller, W.J. Reagan, J.A. Kaduk, C.L. Marshall, A.J. Kropf, J. [43] S. Kolboe, Can. J. Chem. 47 (1969) 352.
Catal. 193 (2000) 123. [44] H. Kwart, G.C.A. Schuit, B.C. Gates, J. Catal. 61 (1980) 128.
[8] W.M. Kreucher, Am. Chem. Soc. Prepr. Div. Pet. Chem. 39 (4) (1994) [45] K.F. McCarty, G.L. Schrader, J. Catal. 103 (1987) 261.
507. [46] W.R. Moser, G.A. Rossetti Jr., J.T. Gleaves, J.R. Ebner, J. Catal. 127
[9] M. Kulakowski, Am. Chem. Soc. Prepr. Div. Pet. Chem. 39 (4) (1994) (1991) 190.
494. [47] P.J. Owens, C.H. Amberg, Adv. Chem. Ser. 33 (1961) 182.
[10] Y. Jacquin, T.D. Chan, M.G. Coronado, J. Cosyns, French Patent [48] P. Desikan, C.H. Amberg, Can. J. Chem. 41 (1963) 1966.
2476118 (1980). [49] C.N. Satterfield, G.W. Roberts, AIChE J. 14 (1) (1968) 159.
[11] S. Hatanaka, M. Yamada, O. Sadakane, Ind. Eng. Chem. Res. 36 [50] J.M.J.G. Lipsch, G.C.A. Schuit, J. Catal. 15 (1969) 179.
(1997) 1519. [51] H.C. Lee, J.B. Butt, J. Catal. 49 (1977) 320.
S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172 171

[52] A.E. Hargreaves, J.R.H. Ross, in: G.C. Bond, P.B. Wells, F.C. [91] M. Karroua, H. Matralis, P. Grange, B. Delmon, J. Catal. 139 (1993)
Tompkins (Eds.), in: Proceedings of the Sixth International Congress 371.
on Catalysis, London, 1976, The Chemical Society, London, 1977, [92] R.R. Chianelli, Catal. Rev.-Sci. Eng. 26 (1984) 361.
p. 937. [93] S. Harris, R.R. Chianelli, J. Catal. 98 (1986) 17.
[53] A.E. Hargreaves, J.R.H. Ross, J. Catal. 56 (1979) 363. [94] R.R. Chianelli, G. Berhault, P. Raybaud, S. Kasztelan, J. Hafner, H.
[54] Y. Okamoto, H. Tomioka, T. Imanaka, S. Teranishi, J. Catal. 66 Toulhoat, Appl. Catal. A: Gen. 227 (2002) 83.
(1980) 93. [95] J.F. Le Page, Catalyse de Contact, Technip, Paris, 1978, 441.
[55] M.E. Bussell, G.A. Somorjai, J. Catal. 106 (1987) 93. [96] M. Ternan, J. Catal. 104 (1987) 256.
[56] S.M.A.M. Bouwens, J.P.R. Vissers, V.H.J. De Beer, R. Prins, J. Catal. [97] G. Pérot, S. Brunet, N. Hamzé, in: M.J. Phillips, M. Ternan (Eds.), in:
112 (1988) 401. Proceedings of the Ninth International Congress on Catalysis, vol. 1,
[57] N.N. Sauer, E.J. Markel, G.L. Schrader, R.J. Angelici, J. Catal. 117 Calgary, Institute of Canada, Ottawa, 1988, p. 19.
(1989) 295. [98] J. Mijoin, V. Thevenin, N. Garcia Aguirre, H. Huze, J. Wang, W.Z. Li,
[58] C.G. Gachet, E. Dhainaut, L. de Mourgues, J.P. Candy, P. Fouilloux, G. Pérot, J.L. Lemberton, Appl. Catal. A: Gen. 180 (1999) 95.
Bull. Soc. Chim. Belg. 90 (1981) 1279. [99] D. Pirotte, J.M. Zabala, P. Grange, B. Delmon, Bull. Soc. Chim. Belg.
[59] G.V. Isagulyants, A.A. Greish, V.M. Kogan, in: M.J. Phillips, M. 90 (1981) 1239.
Ternan (Eds.), in: Proceedings of the Ninth International Congress on [100] I.A. Van Parijs, G.F. Froment, Appl. Catal. 21 (1986) 273.
Catalysis, vol. 1, Chemical Institute of Canada, Ottawa, 1988, p. 35. [101] R. Candia, B.S. Clausen, J. Bartholdy, N.-Y. Topsøe, B. Lengeler, H.
[60] T. Kabe, W. Qian, W. Wang, A. Ishihara, Catal. Today 29 (1996). Topsøe (Eds.), in: Proceedings of the Eighth International Congress
[61] R.J. Mikovsky, A.J. Silvestri, H. Heinemann, J. Catal. 34 (1974) 324. on Catalysis, vol. 2, Berlin, Verlag-Chemie, Weinheim, 1984, p. 375.
[62] J. Mijoin, G. Pérot, F. Bataille, J.L. Lemberton, M. Breysse, S. [102] T. Kabe, A. Ishihara, W. Qian, Hydrodesulfurization and Hydro-
Kasztelan, Catal. Lett. 71 (2001) 139. denitrogenation Chemistry and Engineering, Kodansha/Wiley–VCH,
[63] B.C. Gates, J.R. Katzer, G.C.A. Schuit, Chemistry of Catalytic Tokyo/Weinheim, 1999, 133.
Processes, McGraw-Hill, New York, 1979, 390. [103] F. Bataille, J.L. Lemberton, P. Michaud, G. Pérot, M. Vrinat, M.
[64] M. Vrinat, Appl. Catal. 6 (1983) 137. Lemaire, E. Schulz, M. Breysse, S. Kasztelan, J. Catal. 191 (2) (2000)
[65] I.A. Van Parijs, G.F. Froment, Ind. Eng. Chem. Prod. Res. Dev. 25 409.
(1986) 431. [104] F.E. Massoth, K.S. Chung, in: T. Seiyama, K. Tanabe (Eds.), in:
[66] M.J. Girgis, B.C. Gates, Ind. Eng. Chem. Res. 30 (1991) 2021. Proceedings of the Seventh International Congress on Catalysis,
[67] S. Hatanaka, M. Yamada, O. Sadakane, Am. Chem. Soc. Prepr. Div. Tokyo, 1980, Kodansha/Elsevier, Tokyo/Amsterdam, 1981, p. 629.
Pet. Chem. 42 (3) (1997) 558. [105] J.C. Duchet, E.M. van Oers, V.H.J. de Beer, R. Prins, J. Catal. 80
[68] S. Hatanaka, M. Yamada, O. Sadakane, Ind. Eng. Chem. Res. 36 (1983) 386.
(1997) 5110. [106] M. Zdrazil, M. Kraus, Stud. Surf. Sci. Catal. 27 (1986) 257.
[69] S. Eijsbouts, Appl. Catal. A: Gen. 158 (1997) 53. [107] G. Hagenbach, P. Courty, B. Delmon, J. Catal. 23 (1971) 295.
[70] I.A. Van Parijs, G.F. Froment, B. Delmon, Bull. Soc. Chim. Belg. 93 [108] G. Pérot, Am. Chem. Soc. Prepr. Fuel Chem. 48 (1) (2003) 127;
(1984) 823. M. Bremaud, L. Vivier, G. Pérot, V. Harlé, C. Bouchy, in press.
[71] B. Delmon, G.F. Froment, Catal. Rev.-Sci. Eng. 38 (1996) 69. [109] F. Luck, Bull. Soc. Chim. Belg. 100 (1991) 781.
[72] Y.W. Li, B. Delmon, J. Mol. Catal. A: Chem. 127 (1997) 163. [110] M. Breysse, J.L. Portefaix, M. Vrinat, Catal. Today 10 (1991) 489.
[73] M. Kraus, Adv. Catal. 29 (1980) 151. [111] Y. Okamoto, M. Breysse, G.M. Dhar, C. Song (Eds.), Effect of
[74] J. Uchytil, E. Jakubickova, M. Kraus, J. Catal. 64 (1980) 143. support in hydrotreating catalysis for ultra clean fuels, Catal. Today
[75] J.-S. Choi, C. Petit-Clair, D. Uzio, Stud. Surf. Sci. Catal. 143 (2002) 86 (2003) (special issue).
585. [112] Y. Okamoto, M. Breysse, G.M. Dhar, C. Song, Catal. Today 86
[76] D. Mey, S. Brunet, G. Pérot, F. Diehl, S. Kasztelan, Am. Chem. Soc. (2003) 1.
Prepr. Div. Pet. Chem. 47 (1) (2002) 69. [113] P. Gallezot, Catal. Rev.-Sci. Eng. 20 (1979) 121.
[77] D. Mey, S. Brunet, G. Pérot, F. Diehl, Am. Chem. Soc. Prepr. Fuel [114] W.M.H. Sachtler, A.Yu. Sthakheev, Catal. Today 12 (1992) 283.
Chem. 48 (1) (2003) 44. [115] B. Moraweck, B. Bergeret, M. Cattenot, V. Kougionas, C. Geantet,
[78] G.F. Froment, G.A. Depauw, V. Vanrysselberghe, Ind. Eng. Chem. J.L. Portefaix, J.L. Zotin, M. Breysse, J. Catal. 165 (1997) 45.
Res. 33 (1994) 2975. [116] M. Breysse, M. Cattenot, V. Kougionas, J.C. Lavalley, F. Maugé, J.L.
[79] G.F. Froment, G.A. Depauw, V. Vanrysselberghe, Stud. Surf. Sci. Portefaix, J.L. Zotin, J. Catal. 168 (1997) 143.
Catal. 106 (1997) 83. [117] P. Leyrit, T. Cseri, N. Marchal, J. Lynch, S. Kasztelan, Catal. Today
[80] V. Vanrysselberghe, G.F. Froment, Ind. Eng. Chem. Res. 37 (1998) 65 (2001) 249.
4231. [118] A. de Mallmann, D. Barthomeuf, J. Chim. Phys. 87 (1990) 535.
[81] D.P. Satchell, B.L. Crynes, Oil Gas J. (1975) 123. [119] D. Barthomeuf, Catal. Rev.-Sci. Eng. 38 (1996) 521.
[82] R.R. Chianelli, M. Daage, M.J. Ledoux, Adv. Catal. 40 (1994) 177. [120] H. Shimada, T. Sato, Y. Yoshimura, J. Hiraishi, A. Nishijima, J. Catal.
[83] H. Topsøe, B.S. Clausen, F.E. Massoth, Hydrotreating Catalysis— 110 (1988) 275.
Science and Technology, Springer-Verlag, Berlin, 1996. [121] M. Zdrazil, Catal. Today 86 (2003) 151.
[84] B. Delmon, Catal. Lett. 22 (1993) 1. [122] M. Breysse, P. Afanasiev, C. Geantet, M. Vrinat, Catal. Today 86
[85] B. Delmon, in: H.F. Barry, P.C.H. Mitchell (Eds.), in: Proceedings of (2003) 5.
the Third International Conference on Climax Molybdenum, Ann [123] G. Muralidhar, F.E. Massoth, J. Shabtaı̈, J. Catal. 85 (1984) 44.
Arbor, MI, 1979, p. 73. [124] C. Flego, V. Arrigoni, M. Ferrari, R. Riva, L. Zanibelli, Catal. Today
[86] H. Topsøe, B.S. Clausen, R. Candia, C. Wivel, S. Mørup, J. Catal. 68 65 (2001) 265.
(1981) 433. [125] Y. Okamoto, K. Ochiai, M. Kawano, K. Kobayashi, T. Kubota, Appl.
[87] C. Wivel, R. Candia, B.S. Clausen, S. Mørup, H. Topsøe, J. Catal. 68 Catal. A: Gen. 226 (2002) 115.
(1981) 453. [126] Y. Okamoto, T. Kubota, Catal. Today 86 (2003) 31.
[88] H. Topsøe, R. Candia, N.-Y. Topsøe, B.S. Clausen, Bull. Soc. Chim. [127] K. Tanabe, Solid Acids and Bases and their Catalytic Properties,
Belg. 93 (1984) 783. Academic Press, New York, 1970.
[89] R. Candia, O. Sørensen, J. Villadsen, N.-Y. Topsøe, B.S. Clausen, H. [128] M. Daage, R.R. Chianelli, J. Catal. 149 (1994) 414.
Topsøe, Bull. Soc. Chim. Belg. 93 (1984) 763. [129] A.P. Yu, E.C. Myers, US Patent 4,132,632 (1979).
[90] H. Topsøe, B.S. Clausen, Catal. Rev.-Sci. Eng. 26 (1984) 395. [130] R.J. Bertolacini, T.A. Sue-A-Quan, US Patent 4,140,626 (1979).
172 S. Brunet et al. / Applied Catalysis A: General 278 (2005) 143–172

[131] R.J. Bertolacini, US Patent 4,203,829 (1980). Refinery Processing, AIChE National Meeting, New Orleans,
[132] E.P. Dai, D.E. Sherwood, R.H. Petty, US Patent 5,340,466 (1994). 2002, p. 180.
[133] E.P. Dai, D.E. Sherwood, B.R. Martin, R.H. Petty, US Patent [164] J.C. Guibet, Carburants et Moteurs, Technip, Paris, 1997, 236.
5,441,630 (1995). [165] Q. Debuisschert, J.L. Nocca, J. Esquier, in: Proceedings of the Japan
[134] C. Sudhakar, M.R. Cesar, R.A. Heinrich, US Patent 5,525,211 Petroleum Institute Petroleum Refining Conference, Tokyo, October,
(1996). 2002.
[135] T. Klimova, D.S. Casados, J. Ramirez, Catal. Today 43 (1998) [166] J. Magné-Drisch, F. Picard, A. Pucci, Q. Debuisschert, J.L. Nocca, P.
135. Burnett, in: Proceedings of the AIChE Spring Meeting, New Orleans,
[136] R. Zhao, C. Yin, C. Liu, Am. Chem. Soc. Prepr. Div. Pet. Chem. 46 April, 2003 (Paper 49a).
(1) (2001) 30. [167] J.P. Greeley, S. Zaczepinski, T.R. Halbert, G.B. Brignac, A.R. Gentry,
[137] R. Zhao, C. Yin, H. Zhao, X. Dong, C. Liu, Am. Chem. Soc. Prepr. R.L. Kraus, S. Mayo, Am. Chem. Soc. Prepr. Div. Pet. Chem. 45
Div. Pet. Chem. 47 (1) (2002) 60. (2000) 357.
[138] C. Yin, R. Zhao, C. Liu, Am. Chem. Soc. Prepr. Div. Pet. Chem. 47 [168] J.S. Rosvoll, K. Grante, K. Moljord, P.A. Skjolsvik, B. Thorvaldsen,
(1) (2002) 63. S. Johansen, S. Kuhnle, P.A. Sorum, A.J. Sapre, E.M. Roundtree, in:
[139] R. Zhao, C. Yin, H. Zhao, C. Liu, Am. Chem. Soc. Prepr. Div. Pet. Proceedings of the European Refining Technical Conference, ERTC
Chem. 47 (3) (2002) 309. Annual Meeting, London, November, 2003.
[140] T. Klicpera, M. Zdrazil, J. Catal. 206 (2002) 314. [169] B.B. Garland, T.R. Halbert, J.P. Greeley, R.A. Demmin, T.J. Davis,
[141] E. Hillerova, Z. Vit, M. Zdrazil, Appl. Catal. 118 (1994) 111. World Refin. 10 (7) (2000) 14.
[142] T. Klicpera, M. Zdrazil, Catal. Lett. 58 (1999) 47. [170] M. Daage, R.R. Chianelli, A.F. Ruppert, Stud. Surf. Sci. Catal. 75
[143] T. Klicpera, M. Zdrazil, J. Mater. Chem. 10 (2000) 1603. (1993) 571.
[144] T. Klicpera, M. Zdrazil, Appl. Catal. A: Gen. 216 (2001) 41. [171] G.J. Antos, R.B. Solari, R. Monque, Stud. Surf. Sci. Catal. 106 (1997)
[145] E. Dai, D.E. Sherwood, US Patent 5,358,633 (1994). 27.
[146] C. Sudhakar, G.G. Sandford, P.L. Dahlstrom, M.S. Patel, E.L. [172] N.P. Martinez, J.A. Salazar, J. Tejada, G.J. Antos, M. Anand, E.J.
Patmore, US Patent 5,423,976 (1995). Houde, Vision Tecnologica 7 (2000) 77.
[147] C. Sudhakar, US Patent 5,770,046 (1998). [173] J.A. Salazar, L.M. Cabrera, E. Palmisano, W.J. Garcia, R.B. Solari,
[148] S. Hatanaka, T. Miyama, H. Seki, S. Hikita, EP 0736589 A1 (1996). US Patent 5,770,047 (1998).
[149] S. Hatanaka, O. Sadakane, S. Hikita, T. Miyama, US Patent [174] R.M. Jao, I.J. Leu, J.R. Chang, Appl. Catal. A: Gen. 135 (1996) 301.
5,853,570 (1998). [175] J.K. Lee, H.K. Rhee, J. Catal. 177 (1998) 208.
[150] D. Mey, S. Brunet, C. Canaff, G. Pérot, C. Bouchy, F. Diehl, F. [176] C.A. Drake, S.D. Love, US Patent 6,083,379 (2000).
Maugé, J. Catal. 227 (2004) 436. [177] J. Gislason, Oil Gas J. 99 (2002) 74.
[151] D. Mey, Ph.D. Thesis, The University of Poitiers, France, 2002. [178] K. Tawara, J. Imai, H. Iwanami, Sekiyu Gakkaishi 43 (2000) 105.
[152] C. Sudhakar, G.G. Sandford, US Patent 5,286,373 (1994). [179] K. Tawara, T. Nishimura, H. Iwanami, Sekiyu Gakkaishi 43 (2000)
[153] C. Sudhakar, G.G. Sandford, A.K. Bhattacharya, US Patent 114.
5,423,975 (1995). [180] K. Tawara, T. Nishimura, H. Iwanami, T. Nishimoto, T. Hasuike, Ind.
[154] A. Arteaga, J.L.G. Fierro, F. Delannay, B. Delmon, Appl. Catal. 26 Eng. Chem. Res. 40 (2001) 2367.
(1986) 227. [181] R.L. Irvine, US Patent 5,730,860 (1998).
[155] A. Arteaga, J.L.G. Fierro, P. Grange, B. Delmon, Stud. Surf. Sci. [182] B.S. Turk, R.P. Gupta, Am. Chem. Soc. Prepr. Div. Pet. Chem. 46
Catal. 34 (1987) 59. (2001) 392.
[156] S. Hatanaka, O. Sadakane, US Patent 6,120,679 (2000). [183] J. Wiltshire, BP Technol. Mag. 32 (2000) 10.
[157] S. Hatanaka, O. Sadakane, H. Okazaki, Sekiyu Gakkaishi 44 (2001) [184] B.D. Alexander, G.A. Huff, V.R. Pradhan, W.J. Reagan, R.H. Cayton,
36. US Patent 6,024,865 (2000).
[158] C. Song, Catal. Today 86 (2003) 211. [185] L.S. White, M. Lesemann, Am. Chem. Soc. Prepr. Div. Pet. Chem. 47
[159] F.L. Plantenga, R.G. Leliveld, Appl. Catal. A: Gen. 248 (2003) 1. (2002) 45.
[160] A.K. Rhodes, Oil Gas J. (1994) 16. [186] C. Fredrick, Hydrocarb. Process. 81 (2) (2002) 45.
[161] J.-L. Nocca, J. Cosyns, Q. Debuisschert, B. Didillon, in: Proceedings [187] A. Bösmann, L. Datsevich, A. Jess, A. Lauter, C. Schmitz, P.
of the NPRA Annual Meeting, San Antonio, TX, 2000. Wasserscheid, Chem. Commun. (2001) 2494.
[162] Anon., Hydrocarb. Process. 79-11 (2000) 87. [188] S.G. Zhang, Z.C. Zhang, Green Chem. 4 (2002) 376.
[163] F. Baco, Q. Debuisschert, N. Marchal, J.-L. Nocca, F. Picard, D. [189] Y. Shiraishi, Y. Taki, T. Hirai, I. Komasawa, Ind. Eng. Chem. Res. 38
Uzio, in: Proceedings of the Fifth International Conference on (1999) 4538.

You might also like