Ring opening of decalin via hydrogenolysis on Ir/- and Pt/silica catalysts

Article abstract of DOI:10.1016/j.apcata.2012.03.010

The catalytic conversion of cis-decalin was studied at a hydrogen pressure of 5.2 MPa and temperatures of 250-410 °C on iridium and platinum supported on non-acidic silica. The absence of catalytically active Br?nsted acid sites was indicated by both FT-IR spectroscopy with pyridine as a probe and the selectivities in a catalytic test reaction, viz. the hydroconversion of n-octane. On iridium/silica, decalin hydroconversion starts at ca. 250-300 °C, and no skeletal isomerization occurs. The first step is rather hydrogenolytic opening of one six-membered ring to form the direct ring-opening products butylcyclohexane, 1-methyl-2-propylcyclohexane and 1,2- diethylcyclohexane. These show a consecutive hydrogenolysis, either of an endocyclic carboncarbon bond into open-chain decanes or of an exocyclic carboncarbon bond resulting primarily in methane and C₉ naphthenes. The latter can undergo a further endocyclic hydrogenolysis leading to open-chain nonanes. All individual C₁₀ and C₉ hydrocarbons predicted by this direct ring-opening mechanism were identified in the products generated on the iridium/silica catalysts. The carbon-number distributions of the hydrocracked products C_9^- show a peculiar shape resembling a hammock and could be readily predicted by simulation of the direct ring-opening mechanism. Platinum on silica was found to require temperatures around 350-400 °C at which relatively large amounts of tetralin and naphthalene are formed. The most abundant primary products on Pt/silica are spiro[4.5]decane and butylcyclohexane which can be readily accounted for by the well known platinum-induced mechanisms described in the literature for smaller model hydrocarbons, namely the bond-shift isomerization mechanism and hydrogenolysis of a secondary-tertiary carboncarbon bond in decalin.

Full text of DOI:10.1016/j.apcata.2012.03.010

Applied Catalysis A: General 425–426 (2012) 97–109
Contents lists available at SciVerse ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Ring opening of decalin via hydrogenolysis on Ir/- and Pt/silica catalysts
Andreas Haasa, Sandra Rabla, Marco Ferrarib, Vincenzo Calemmab, Jens Weitkampa,∗
a Institute of Chemical Technology, University of Stuttgart, 70550 Stuttgart, Germany
b Eni S.p.A., R&M Division, Via F. Maritano 26, 20097 San Donato Milanese, Italy
a r t i c l e i n f o
a b s t r a c t
Article history:
The catalytic conversion of cis-decalin was studied at a hydrogen pressure of 5.2 MPa and temperatures
of 250–410 ◦C on iridium and platinum supported on non-acidic silica. The absence of catalytically active
Brønsted acid sites was indicated by both FT-IR spectroscopy with pyridine as a probe and the selectivities
in a catalytic test reaction, viz. the hydroconversion of n-octane. On iridium/silica, decalin hydroconver-
sion starts at ca. 250–300 ◦C, and no skeletal isomerization occurs. The first step is rather hydrogenolytic
opening of one six-membered ring to form the direct ring-opening products butylcyclohexane, 1-methyl-
2-propylcyclohexane and 1,2-diethylcyclohexane. These show a consecutive hydrogenolysis, either of an
endocyclic carbon carbon bond into open-chain decanes or of an exocyclic carbon carbon bond result-
ing primarily in methane and C9 naphthenes. The latter can undergo a further endocyclic hydrogenolysis
leading to open-chain nonanes. All individual C10 and C9 hydrocarbons predicted by this “direct ring-
opening mechanism” were identified in the products generated on the iridium/silica catalysts. The
Received 26 January 2012
Received in revised form 1 March 2012
Accepted 3 March 2012
Available online 13 March 2012
Keywords:
Ring opening
Decalin
Iridium
Platinum
Hydrogenolysis
−
carbon-number distributions of the hydrocracked products C9 show a peculiar shape resembling a ham-
mock and could be readily predicted by simulation of the direct ring-opening mechanism. Platinum on
silica was found to require temperatures around 350–400 ◦C at which relatively large amounts of tetralin
and naphthalene are formed. The most abundant primary products on Pt/silica are spiro[4.5]decane
and butylcyclohexane which can be readily accounted for by the well known platinum-induced mecha-
nisms described in the literature for smaller model hydrocarbons, namely the bond-shift isomerization
mechanism and hydrogenolysis of a secondary-tertiary carbon carbon bond in decalin.
© 2012 Elsevier B.V. All rights reserved.
1. Introduction
metal catalysts. Since about 2000, a large number of publications
has appeared dealing with the hydroconversion of multi-ring, espe-
A promising route for upgrading low-quality middle-distillate
refinery streams with a high content of polynuclear aromatics
into valuable blending cuts for diesel fuel is their hydrogenation
to multi-ring naphthenes followed by selective ring opening to
alkyl naphthenes with one remaining ring or open-chain alkanes.
As the complete hydrogenation of polynuclear aromatics is well
developed [1–3], the selective hydrodecyclization of multi-ring
naphthenes without excessive hydrocracking into hydrocarbons
with less carbon atoms has recently become a major issue in het-
erogeneous catalysis.
cially bicyclic naphthenes, naphthenoaromatics and aromatics on
either bifunctional [4–14] or monofunctional acidic [4,5,9] cat-
alysts. By contrast, the literature on ring opening of multi-ring
naphthenes on monofunctional metal catalysts is rather scarce,
though there are papers available from two groups who studied
the field remarkably thoroughly and comprehensively [15–17].
Much more information is available concerning the metal-
catalyzed hydrocracking (often referred to as “hydrogenolysis”) of
monocyclic naphthenes, in particular cyclopentane, cyclohexane,
and their alkyl-substituted derivates. To a large extent, the nomen-
clature proposed by Maire et al. [18] for the various hydrogenolysis
mechanisms applicable to naphthenes has been widely adopted.
Basically, three mechanisms are usually discerned that were essen-
tially advanced by Gault and his school for metal-catalyzed ring
opening of naphthenes [19]: in the so-called non-selective mecha-
nism the likelihood of all endocyclic carbon carbon bonds for being
cleaved is considered to be equal. By contrast, the salient feature of
the selective mechanism is an exclusive cleavage of bonds between
Three principal families of solid catalysts are usually dis-
tinguished for the cleavage of carbon carbon bonds, viz. (i)
monofunctional acidic catalysts, (ii) bifunctional catalysts contain-
ing both Brønsted acid and metal sites, and (iii) monofunctional
∗
Corresponding author. Tel.: +49 711 685 64060; fax: +49 711 685 64065.
E-mail address: jens.weitkamp@itc.uni-stuttgart.de (J. Weitkamp).
0926-860X/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2012.03.010

98
A. Haas et al. / Applied Catalysis A: General 425–426 (2012) 97–109
secondary carbon atoms. Finally, the partially selective mechanism
describes the rupture of bonds between carbon atoms which are
substituted by at least one methyl group.
dropwise addition of a 1 molar aqueous ammonia solution. Subse-
quently, 206 mg of the iridium complex or 145 mg of the platinum
complex dissolved in 30 cm3 of demineralized water were added
dropwise under ongoing stirring. After stirring for another 18 h
the solid was filtered off, washed with 100 cm3 of demineralized
water and dried at 80 ◦C. Likewise, 2.59Ir/silica and 2.68Pt/silica
were prepared by suspending 4.0 g of silica in 50 cm3 of deminer-
alized water, the pH was again adjusted to 10, and a solution of
300 mg of the iridium complex or 215 mg of the platinum complex
in 20 cm3 of demineralized water were added dropwise. Thereafter,
the suspension was again stirred for 18 h, and the solid was filtered
off, washed with 50 cm3 of demineralized water and dried at 80 ◦C.
The resulting catalyst powder was pressed into tablets without a
binder at a pressure of 127 MPa, and the tablets were crushed and
sieved. The particle size fraction from 0.20 to 0.32 mm was trans-
ferred into the catalytic reactor. There, the adsorbed complex was
treated in an air flow at 0.1 MPa and 150 ◦C for 3 h in order to obtain
a high metal dispersion [27]. Finally, the noble metal was reduced
in a flow of hydrogen at 0.1 MPa and 400 ◦C for 2 h.
For elemental analysis of iridium and platinum, an optical emis-
sion spectrometer with an inductively coupled plasma (ICP-OES,
Varian Vista-MPX CCD) was used. The catalysts were dissolved in
a mixture of diluted hydrofluoric acid and aqua regia. Details con-
cerning the measurement of the water content are given in Ref.
[24].
The dispersion of the noble metal was determined by measuring
the amount of irreversibly adsorbed hydrogen in a Quantachrome
Autosorb-1-C instrument. For details of the experimental proce-
dure see Ref. [24]. An adsorption stoichiometry of nH/nmetal = 1 was
assumed. For the calculation of the metal particle size, a spheri-
cal shape and densities of, respectively, 22.42 and 21.45 kg dm−3
for iridium and platinum were assumed. The porosity of the cat-
alyst support silica was measured by nitrogen physisorption at
T = −196 ◦C in the same Quantachrome Autosorb-1-C instrument
after degassing the samples at T = 350 ◦C for 16 h. For evaluating
the BET specific surface area the range of p/p0 between 0.1 and 0.3
was used. For the N2 molecule a cross-sectional area of 1.62 nm2
was assumed.
With FT-IR spectroscopy and pyridine as a probe molecule,
information about the Brønsted acid sites of the silica support and,
exemplarily, of one metal-loaded catalyst was collected. A Bruker
Vector 22 spectrometer equipped with a high-vacuum sample cell
was employed. Self-supporting wafers were made from the pow-
dered samples of silica and of the precursor of 2.68Pt/silica after
drying at 80 ◦C. Subsequently, they were treated in an air flow in
the sample holder of the IR cell at 150 ◦C. After reduction of the
noble metal with hydrogen at T = 500 ◦C, pyridine was adsorbed at
T = 100 ◦C, and desorption took place at T = 200 ◦C. For experimental
details of the IR measurements see Ref. [25].
No evidence was obtained for the existence of Brønsted acid
sites on the silica carrier by FT-IR spectroscopy. This result was
corroborated by a catalytic test reaction, viz. the hydroconversion
of n-octane, which was carried out on the catalysts 0.77Ir/silica and
0.93Pt/silica. If Brønsted acid sites with a sufficient strength and in
a sufficient concentration had been present, n-octane would have
been converted into the typical products of bifunctional hydro-
carbon conversion, namely a mixture of iso-octanes at low and
moderate conversions and, at elevated conversions, hydrocracked
products with 3–5 carbon atoms, but without C1, C2, C6 and C7
hydrocarbons [28]. Conversely, in the absence of catalytically active
Brønsted acid sites, n-octane would be converted at the respec-
tive metal sites and, in particular, hydrogenolysis to light alkanes
including C1, C2, C6, and C7 hydrocarbons would occur. For the n-
octane test, a flow-type apparatus made of glass with a fixed-bed
reactor was used. The mass of catalyst, the hydrogen partial pres-
For example, when methylcyclopentane is hydroconverted on
iridium, the naphthenic ring is opened selectively at the bisec-
ondary positions [15]. Platinum shows a very similar ring opening
selectivity, however only, if the metal is poorly dispersed. On highly
dispersed platinum, by contrast, the ring is opened non-selectively
[18]. Upon introducing an increasing number of alkyl substituents
or when a six-membered instead of a five-membered ring is to
be opened, higher reaction temperatures have to be applied and
more than one ring-opening mechanism occur simultaneously
[15,20–22].
Five-membered rings are relatively easy to open on noble-metal
catalysts without making hydrocarbons with lower carbon num-
bers. Six-membered rings can be opened on this type of catalysts as
well, but larger amounts of by-products are usually formed. It has
been demonstrated in several publications that, on a non-acidic
support, iridium is the most active and selective metal for ring
opening of naphthenes [15,23].
In their detailed study of hydrogenolytic ring opening of
naphthenes on various metals, McVicker et al. [15] used both mono-
cyclic and bicyclic model hydrocarbons. In the hydroconversion of
bicyclo[3.3.0]octane on iridium/alumina, for example, both five-
membered rings were readily opened to give octane isomers. When
bicyclo[4.3.0]nonane, which contains both a five-membered and a
six-membered ring was hydroconverted on the same catalyst, the
five-membered rings were again rapidly opened, but open-chain
nonanes were formed in low yields only indicating that the six-
membered rings are remarkably recalcitrant to be opened. Decalin
(bicyclo[4.4.0]decane), which consists of two six-membered rings,
was also hydroconverted on the iridium/alumina catalyst and
turned out to react very slowly, with open-chain decanes being
formed in traces only.
Recently, we reported on the hydroconversion of decalin on
typical bifunctional catalysts, such as Ir/La-X and Pt/La-X zeo-
lites using improved methods for analyzing the very complex
product mixtures [24]. In another paper [25] a novel class of high-
performance ring-opening catalysts (HIPEROCs) was introduced
based on Ir/Na,H-Y and Pt/Na,H-Y zeolites with a low concentra-
tion of Brønsted acid sites. On these catalysts, decalin could be
hydroconverted to open-chain decanes with unprecedented selec-
tivities and yields up to ca. 40%. Preliminary results obtained in the
hydroconversion of decalin on non-acidic Ir/silica catalysts were
also published [26]. On these catalysts, a hitherto unknown carbon-
number distribution of the hydrocracked products was observed
for which the term “hammock-type curve” was coined. Here, we
present the results of a detailed investigation of decalin ring open-
ing on non-acidic Ir/silica and Pt/silica catalysts and interpret the
results in terms of hydrocarbon conversion mechanisms on the two
noble metals.
2. Experimental
2.1. Preparation and characterization of the catalysts
Four catalysts were prepared, viz. 0.77Ir/silica, 2.59Ir/silica,
0.93Pt/silica, and 2.68Pt/silica, by loading silica (Aerosil 380, Evonik
Degussa GmbH, average primary particle size according to the
manufacturer 7 nm) with [Ir(NH3)5Cl]Cl2 or [Pt(NH3)4]Cl2 via elec-
trostatic adsorption [27]. The numbers in the catalyst designations
indicate the metal content in wt.% on a dry basis. For the prepara-
tion of 0.77Ir/silica and 0.93Pt/silica, 8.0 g of silica were suspended
in 100 cm3 of demineralized water, this suspension was vigorously
stirred for 2 h, and during that time the pH was adjusted to 10 by
sure, the n˙ H /n˙ n-Oc ratio and the LHSV were 0.5 g, 0.1 MPa, 34 and
2

A. Haas et al. / Applied Catalysis A: General 425–426 (2012) 97–109
99
0.5 h−1, respectively. The reaction temperature was varied from
217 to 393 ◦C, and the products were analyzed on-line in a Hewlett-
Packard HP 5890 gas chromatograph equipped with a BP1 PONA
(50 m × 0.15 mm × 0.5 ␮m) capillary column and a flame ionization
detector (FID).
50
40
30
20
10
0
2.2. Hydroconversion of decalin
The Ir/- and Pt/silica catalysts were tested with cis-decalin
(Merck, purity >98%) as reactant in the gas phase. All catalytic
experiments were conducted in a high-pressure flow-type appara-
tus from stainless steel. Its main parts were a saturator filled with an
inert solid (Merck, Chromosorb P/AW, particle size 0.25–0.60 mm)
and the liquid reactant, in which the hydrogen stream was loaded
with vapors of cis-decalin in the desired concentration [29], and a
fixed-bed reactor. The mass of dry catalyst, the total pressure, the
partial pressure of cis-decalin and the LHSV amounted to 0.17 g,
5.2 MPa, 25 kPa and 0.4 h−1, respectively. The reaction temperature
was varied in the range from 250 to 410 ◦C. The reactor effluent
was depressurized in a needle valve, and the product stream was
then passed through a heated six-port gas sampling valve which
allowed an on-line introduction of full product samples into a gas
chromatograph (Agilent 6890N) equipped with a Supelco Petrocol
DH 150 (150 m × 0.25 mm × 1.0 ␮m) capillary column and a flame
ionization detector. The gas stream leaving the gas sampling valve
was routed through a cooling trap held at −10 ◦C where an integral
liquid hydrocarbon sample was collected over several hours.
The entire quantitative evaluation of the catalytic experiments
was based on the analysis of the full product samples sluiced into
the on-line gas chromatograph via the gas sampling valve. The
integral liquid product samples were used for qualitative peak
assignment via ancillary off-line GC/MS analyses. A second gas
chromatograph (Agilent 6890N) which was also equipped with a
Supelco Petrocol DH150 capillary column and coupled to a mass
spectrometer (Agilent 5876B inert XL MSD) was used for this pur-
pose. Peak assignment was further refined by (i) co-injection of
pure hydrocarbons whenever these were commercially available
and (ii) co-injection of a mixture of decane isomers of known
composition which was generated by catalytic isomerization of n-
decane [30]. A more comprehensive description of our methods for
peak assignment has been presented elsewhere [24].
200
250
300
350
400
Temperature / °C
Fig. 1. Conversion of n-octane on 0.77Ir/silica (ꢀ) and on 0.93Pt/silica (᭹) and yields
of hydrocracked products on 0.77Ir/silica (ꢁ) and on 0.93Pt/silica (ꢀ) at different
temperatures.
a group of hydrocarbons j formed divided by the molar amount of
decalin converted into C9− products:
(n˙ j)formed
(n˙ decalin
(m˙ )
Mdecalin
Mj
j
formed
Sj∗≡
=
·
(1)
)
(m˙
)
decalin
converted to C
−
converted to C
−
9
9
where n˙ , m˙ and M are, respectively, the molar flux, the mass flux,
and the molar mass. For the evaluation of Eq. (1) and the calculation
of the decalin conversion XDec, the yields of products Yj, and the
selectivities of products Sj see Ref. [24].
3. Results and discussion
3.1. Characterization of the catalysts
3.1.1. Physicochemical properties
For the silica support, a specific surface area of 391 m2 g−1
and a pore volume of 1.07 cm3 g−1 were determined. In Table 1,
the results of the physicochemical characterization of the metal
components are listed. The amount of noble metal taken up by sil-
ica in comparison with the amount of noble metal offered in the
solution during electrostatic adsorption was 60%, 69%, 92%, and
90% for 0.77Ir/silica, 2.59Ir/silica, 0.93Pt/silica, and 2.68Pt/silica,
respectively. For iridium, metal dispersions above 1 were formally
measured by hydrogen chemisorption, in agreement with reports
by others [31]. Such a finding is usually attributed to an adsorp-
tion stoichiometry nH/nIr > 1. The platinum particles appear to be
larger than the iridium particles with a similar metal loading. For a
given metal, the specific surface area referenced to the mass of the
overall catalyst, expectedly increases with increasing metal con-
tent. Somewhat surprisingly, the catalytic activity of 2.68Pt/silica
is not significantly higher than that of 0.93Pt/silica, even though
its content of platinum is higher by about a factor of 2.9. This may,
in part, be due to the much lower dispersion of 2.68Pt/silica, but
further unknown reasons might play a role as well.
For the subsequent discussion, it is convenient to classify
the product hydrocarbons from cis-decalin into the following six
groups:
1. tr-Dec: trans-decalin formed by a rapid stereoisomerization;
2. sk-Isos: skeletal isomers of decalin with two remaining naph-
thenic rings, a molar mass M = 138 g mol−1, and a molecular
formula C10H18
3. ROPs: ring-opening products with one remaining naphthenic
ring, M = 140 g mol−1, C10H20
4. OCDs: open-chain decanes, M = 142 g mol−1, C10H22
;
;
;
5. DHPs: dehydrogenation products, mainly tetralin and naphtha-
lene;
6. C9− : hydrocracked products with less than ten carbon atoms.
By FT-IR measurements with pyridine as a probe no absorp-
tion band at 1545 cm−1 was detected, neither on the silica support
without a noble metal, nor on 2.68Pt/silica. Hence, no Brønsted acid
sites of sufficient strength for the protonation of pyridine exist on
this catalyst. Since the other three noble-metal-containing catalysts
were prepared in an analogous manner, it is justified to assume that
all four catalysts are lacking Brønsted acid sites.
As will be shown, by far the fastest reaction of cis-decalin on the
noble metal/silica catalysts is its stereoisomerization into trans-
decalin. We found it convenient not to treat this compound as
a product but rather to lump it with cis-decalin to the pseudo-
reactant decalin (Dec).
3.1.2. n-Octane test
For the quantitative discussion of the hydrocracking reactions
to C9− hydrocarbons, the modified hydrocracking selectivities Sj*
will be used. Sj*is defined as the molar amount of a hydrocarbon or
In the hydroconversion of n-octane, catalyst 0.77Ir/silica was
considerably more active than 0.93Pt/silica, see Fig. 1. Due to cata-
lyst deactivation at the low hydrogen pressure of ca. 0.1 MPa, a more

100
A. Haas et al. / Applied Catalysis A: General 425–426 (2012) 97–109
Table 1
Physicochemical properties of the noble metals on the four catalysts (the specific surface area is referenced to the mass of the overall catalyst).
Noble metal
Catalyst
Content/wt.%
Dispersion
Particle size/nm
Specific surface area/m2 g−1
0.77Ir/silica
2.59Ir/silica
0.93Pt/silica
2.68Pt/silica
0.77
2.59
0.93
2.68
1.52
1.02
0.99
0.64
0.65
0.97
1.1
3.16
7.12
2.28
4.25
1.8
detailed interpretation of relative catalyst activities is not possible.
Hydrogenolysis was by far the predominating reaction on the irid-
ium catalyst with a selectivity of hydrocracked products beyond
93%, as indicated by the nearly overlapping curves of n-octane
conversion and yields of hydrocracked products. Other products
were mainly methylheptanes and ortho-xylene, formed by a very
minor extent of skeletal isomerization and aromatization, respec-
tively. By contrast, on 0.93Pt/silica, hydrocracking was much less
pronounced, with selectivities between 26 and 42%. Several other
reactions occurred on platinum, resulting in the diverging values
of conversion and yields of hydrocracked products in Fig. 1. These
were, in decreasing order of abundance, dehydrocyclization, mainly
to propylcyclopentane and 1-ethyl-2-methylcyclopentane, arom-
atization, mainly to ethylbenzene and ortho-xylene, and skeletal
isomerization, mainly to 2-, 3-, and 4-methylheptane. A simi-
lar product distribution in the hydroconversion of n-octane on
2.0Pt/alumina was interpreted in terms of pure metal catalysis [32].
In sharp contrast, the n-octane test led to completely differ-
ent results when bifunctional zeolite catalysts like 0.85Ir/La-X or
1.00Pt/La-X were used [24]: on Ir/La-X, skeletal isomerization via
carbocations occurred with selectivities of up to 58%. Likewise,
skeletal isomerization and consecutive hydrocracking were the sole
reactions on Pt/La-X, while the metal-catalyzed dehydrocyclization
and aromatization were virtually absent [24].
reasonable interpretation of the distribution curves in Fig. 2 are
a hydrogenolytic mechanism of carbon carbon bond cleavage on
the respective metal. It is straightforward to assume that hydrocar-
bon conversion on the silica-supported catalysts with higher metal
loadings occurs on the metal sites as well, without participation of
Brønsted acid sites.
3.2. Stereoisomerization of cis-decalin to trans-decalin
On all four catalysts the fastest reaction was stereoisomerization
of cis- to trans-decalin. The mole fraction ntr-Dec/(nc-Dec + ntr-Dec
)
assumed values from 0.83 to 0.92 at reaction temperatures from
250 to 410 ◦C and conversions in the range from 9 to 90%, as
found by Moraes et al. who converted a feed mixture of 40 mol%
cis- and 60 mol% trans-decalin on Ir/␥-Al2O3 with different irid-
ium loadings [17]. As discussed previously [24], the extent of
isomerization corresponds nicely to the values calculated for ther-
modynamic equilibrium. Since our catalysts lack Brønsted acid
sites, the rapid stereoisomerization is very likely to occur on the
metal sites via dehydrogenation to 1,9- or 9,10-octalins followed
by re-hydrogenation to trans-decalin, as proposed by Weitkamp
[33].
3.3. Hydroconversion of decalin on the four metal/silica catalysts
Typical carbon number distributions observed in the n-octane
test on 0.77Ir/silica and 0.93Pt/silica are shown in Fig. 2. These
curves deviate substantially from those obtained in the hydrocon-
version of n-octane on bifunctional catalysts [28]. While the latter
are bell- (or volcano-) shaped with pronounced maxima at C4 and
smaller but noticeable amounts of C3 and C5, significant amounts
of C1, C2, C6 and C7 are additionally formed on 0.77Ir/silica and
0.93Pt/silica. This is in agreement with the results obtained in n-
octane hydroconversion on non-acidic 2.0Pt/alumina [32]. The only
None of the catalysts showed any deactivation up to a time-on-
stream of 30 h under the conditions used in this study. Moreover,
hydrocarbons with more than ten carbon atoms were absent in the
product mixtures on all four catalysts.
3.3.1. Conversions and selectivities
The decalin conversions and product selectivities attained on
the four catalysts in dependence of the reaction temperature are
depicted in Fig. 3a–d. In line with a previous report [15] iridium
is found to be considerably more active than platinum in the ring
opening reaction. This may have its origin in both the differences in
their intrinsic catalytic activity and in the higher dispersion of irid-
ium (Table 1). To arrive at similar decalin conversions much lower
temperatures can be applied with Ir/silica than with Pt/silica. More-
over, large differences are observed in the main products made
on the iridium and the platinum catalysts: Ring opening is by far
the predominating reaction on the two iridium catalysts at low to
moderate conversions (Fig. 3a and b). It is furthermore seen that vir-
tually no skeletal isomerization of decalin occurs on iridium, again
in agreement with the work of McVicker et al. [15]. On both iridium
catalysts, some open-chain decanes are also formed, their selectiv-
ities passing through maxima with increasing temperature [26].
These maxima are due to an enhanced formation of OCDs by endo-
cyclic hydrogenolysis of ROPs, as the temperature is increased, and
a consecutive hydrocracking of OCDs into C9− hydrocarbons. The
maximum selectivities of OCDs amount to 16% for 0.77Ir/silica and
to 28% on 2.59Ir/silica which corresponds to OCDs yields of 11%
and 20%, respectively (Table 2). These selectivities and yields are
significantly lower than those reported recently for Ir/Na,H–Y zeo-
lite catalysts with a low concentration of Brønsted acid sites (SOCDs
up to 36% and YOCDs up to 31%, [25]). At 290 ◦C, i.e. under conditions
100
80
60
40
20
0
1
2
3
4
5
6
7
Number of C-atoms
Fig. 2. Modified hydrocracking selectivities Sj* in the hydroconversion of n-octane
on 0.77Ir/silica (ꢀ, Tr = 280 ◦C; Xn-Oc = 9%; YC = 8%; ꢀ Sj* = 263%) and 0.93Pt/silica
−
7
(᭹, Tr = 332 ◦C; Xn-Oc = 14%; YC = 4%; ꢀ Sj* = 224%).
−
7

A. Haas et al. / Applied Catalysis A: General 425–426 (2012) 97–109
101
100
80
60
40
20
0
100
80
60
40
20
0
100
80
60
40
20
0
100
80
60
40
20
0
0.77Ir/silica
2.59Ir/silica
0.93Pt/silica
2.68Pt/silica
b
c
d
a
300 320 340 360
Temperature / °C
260
280
300
360
380
400
360
380
400
Temperature / °C
Temperature / °C
Temperature / °C
Fig. 3. Conversion of decalin XDec (᭹) and selectivities of different groups of products (Ssk-Isos (ꢀ); SROPs (ꢀ); SOCDs (ꢁ); SC (ꢁ); SDHPs (ꢂ)) on the four metal/silica catalysts
−
9
at different temperatures.
of maximal OCDs selectivity, the following seven isomeric decanes
were formed on 2.59Ir/silica in the order of their abundance: an
unidentified OCD(1), 4-ethyloctane, 3,4-diethylhexane, n-decane,
an unidentified OCD(2), 5-methylnonane, and 4,5-dimethyloctane
(both diastereomers). Dehydrogenation of decalin to tetralin and
naphthalene (DHPs) is negligible on both Ir/silica catalysts with
SDHPs < 1%.
could still represent very valuable components in the products of
hydrodecyclization, with a boiling point falling within the range of
diesel fuel, a high cetane number and good cold-flow properties.
Likely pathways for the formation of OCNs in the ring opening of
decalin on Ir/silica will be discussed in Section 3.4.
The degree of branching in the alkanes formed by hydrodecy-
clization is decisive for both the cetane number and the cold-flow
properties of the diesel fuel produced. Whereas the cetane num-
ber generally decreases with increasing degree of branching [34], a
certain degree of branching is beneficial for the cold-flow proper-
ties [30]. Mono- and dibranched iso-alkanes are often considered
as a good compromise. In Fig. 4, the distribution of the differently
branched decane isomers formed from decalin on the two Ir/silica
and one Pt/silica catalysts are shown for the experiments leading to
maximal selectivities of OCDs (4-propylheptane was not formed on
any of the catalysts). On the two Ir/silica catalysts, very similar iso-
mer distributions with a clear preponderance of multi-branched
iso-decanes are found. On 2.68Pt/silica, considerably less multi-
branched iso-decanes and large amounts of methylnonanes occur.
The large amounts of multiply branched iso-decanes on the irid-
ium catalysts accounting for ca. 60% of all OCDs formed are not
unexpected. It is reasonable to assume that OCDs are formed from
decalin on the iridium catalysts by two successive hydrogenoly-
sis steps of endocyclic carbon carbon bonds. Since hydrogenolysis
on iridium is known to occur preferentially between unsubstituted
carbon carbon bonds [15], the branchings existing in decalin and
the intermediate ROPs with one remaining ring are expected to sur-
vive the two ring-opening steps, thereby leading to OCDs with at
least two branchings.
Pt/silica behaves differently from Ir/silica in many respects.
Apart from the fact that the platinum catalysts are much less active,
an almost threefold increase in the platinum content brings about
only a slight increase in the catalytic activity (compare Fig. 3c and
d). Perhaps the larger metal particles on 2.68Pt/silica (Table 1)
contain only slightly more crystallographic sites that are active in
the decalin hydroconversion. Another main difference compared
to Ir/silica is the ability of both Pt/silica catalysts to promote the
skeletal isomerization of decalin. Mechanistic aspects of decalin
isomerization on platinum catalysts without Brønsted acid sites
will be discussed below, see Section 3.6. Fig. 3c and d reveals
that, at low conversions, the selectivities of skeletal isomeriza-
tion and ring opening are approximately equal and amount to ca.
40% each. With increasing temperature, both selectivities decrease.
Throughout the entire temperature range measured the selectivi-
ties of OCDs and C9− hydrocarbons are very low. Instead, decalin is
dehydrogenated to tetralin and naphthalene to a significant extent.
Roughly, the amount of dehydrogenated products coincided with
the one calculated for equilibrium at a hydrogen partial pressure
of 5.2 MPa and temperatures of 350–410 ◦C. Ancillary experiments
with an empty and a silica-filled reactor showed that tetralin and
naphthalene were formed with similar yields as in the catalytic
experiments, indicating that the reactor walls are capable to pro-
mote the dehydrogenation of decalin at these high temperatures.
Another principal difference between the performance of the
Ir/silica and Pt/silica catalysts is seen from Table 2, second-last
column: Noticeable amounts of open-chain nonanes (OCNs) are
formed in the hydroconversion of decalin on the two Ir/silica
catalysts, but they do not form at all on Pt/silica. While in our model-
compound study with decalin as reactant, these OCNs appear in
the group of hydrocracked products (C9− ), such alkanes formed
by abstraction of methane from hydrocarbons in a real feedstock
3.3.2. Formation of hydrocracked products with less than ten
carbon atoms
Generally, hydrocracking of decalin, its skeletal isomers, ring-
opening products, and open-chain decanes into hydrocarbons with
less than ten carbon atoms is undesirable, because it is associ-
ated with an increased consumption of valuable hydrogen and
because such lighter hydrocarbons will, at least in part, be outside
the boiling range of diesel fuel. It has been shown, however, that
a detailed look at the distribution of these lighter hydrocarbons
Table 2
Maximum yields of open-chain decanes (OCDs) and selectivities of open-chain nonanes (OCNs) and hydrocracked products obtained on the four catalysts.
Catalyst
Tr/%
XDec/%
SOCDs/%
YOCDs,max./%
SOCNs/%
SC − /%
9
0.77Ir/silica
2.59Ir/silica
0.93Pt/silica
2.68Pt/silica
350
290
410
410
71
73
43
53
16
28
2
11
20
1
7
8
0
0
48
35
6
4
2
13

102
A. Haas et al. / Applied Catalysis A: General 425–426 (2012) 97–109
100
80
60
40
20
0
100
0.77Ir/silica
2.59Ir/silica
80
60
40
20
0
n-C
M-C
E-C
MBDe
n-C
M-C
E-C
8
MBDe
10
9
8
10
9
100
80
60
40
20
0
2.68Pt/silica
n-C
M-C
E-C
8
MBDe
10
9
Fig. 4. Breakdown of differently branched open-chain decanes observed on three catalysts in the ring-opening experiments leading to maximal selectivities of OCDs (MBDe:
multiply branched decanes).
with 1–9 carbon atoms is worthwhile, since it can furnish valu-
able information concerning the mechanism of hydrocracking that
is operative on the catalyst under investigation [24,26]. For exam-
ple, hydrogenolysis of decalin on iridium supported on a non-acidic
support has been shown to result in carbon number distributions
with large amounts of methane and C9 hydrocarbons, smaller but
still appreciable amounts of ethane plus C8 and propane plus C7,
and very low amounts of C4–C6 hydrocarbons [26]. In view of their
characteristic shape, these distribution curves were referred to as
“hammock-type curves”. By contrast, distinctly different M-shaped
carbon-number distribution curves were observed when decalin
was hydrocracked on bifunctional catalysts comprising metallic
and Brønsted acid sites [24].
In Fig. 5 the carbon number distributions of the hydrocracked
products measured on the four silica-supported catalysts at compa-
rable decalin conversions are depicted. Typical hammock-shaped
distributions are obtained on the two iridium catalysts (Fig. 5a).
The curves are slightly unsymmetrical indicating some secondary
hydrogenolysis of the largest hydrocarbons, especially C9, even at
the relatively low yields of hydrocracked products. Entirely differ-
ent distribution curves are found for the two platinum catalysts.
Again, methane predominates, but relatively large amounts of C2,
C3, C4, and C6 hydrocarbons are also formed. The large amounts
of C1 and C2 rule out a cracking mechanism via carbocations, in
line with the lack of Brønsted acid sites of these catalysts. In fact,
we believe that the patterns of hydrocracked products shown in
Fig. 5b are characteristic for hydrogenolysis of decalin, the skeletal
isomers and ring-opening products formed from it around 400 ◦C
on platinum.
methylcyclohexane are found in the C6 and C7 fractions, respec-
tively, formed on the two Pt/silica catalysts.
3.4. Direct ring-opening mechanism on the iridium/silica
catalysts
As shown in Fig. 3a and b, virtually no skeletal isomeriza-
tion of decalin takes place on Ir/silica, but the metal is very
active in the ring-opening reaction. There are five possible direct
ring-opening products (direct ROPs) from decalin, namely butylcy-
clohexane (B-CHx), cis- and trans-1-methyl-2-propylcyclohexane
(1-M-2-P-CHx), and cis- and trans-1,2-diethylcyclohexane (1,2-
DE-CHx). Indeed, these direct ROPs are strongly dominating at
low conversions, as the relevant section of a pertinent gas chro-
matogram shows (Fig. 6). It should be noted that these results are
fully in-line with the selectivities of decalin ring opening reported
very recently by Moraes et al. [17] for Ir/alumina catalysts with
metal loadings ranging from 0.6 to 1.8 wt.%. Cyclodecane, another
potential direct ROP, which would be formed from decalin, if its
bridging carbon carbon bond were cleaved by hydrogenolysis, was
never present in the products obtained on the two iridium catalysts,
as ascertained by co-injection of cyclodecane and a liquid product
from decalin conversion on 2.59Ir/silica.
The section of the gas chromatogram presented in Fig. 6 further-
more shows that, beside the direct ROPs, a few other important
product hydrocarbons appear at low decalin conversion, even
though their concentrations are lower than those of the direct ROPs.
Two of these peaks were identified as cis- and trans-1-ethyl-2-
methylcyclohexane (1-E-2-M-CHx) which are predicted to form
by a consecutive abstraction of methane from direct ROPs. We
refer to these C9 hydrocarbons as “direct C9 ROPs”. The signal of a
third direct C9 ROP, viz. propylcyclohexane (P-CHx) unfortunately
overlaps with the signal of an unidentified open-chain decane des-
ignated as OCD(1). By co-injection of propylcyclohexane with a
liquid product mixture from decalin it was verified that the signal
in question stems mainly from the OCD, so that the relatively small
More detailed information concerning the compositions of the
C4, C6 and C7 fractions formed on the four catalysts is given in
Table 3. Relatively little iso-butane is found in the C4 fraction on
all catalysts, and the molar ratio of methylcyclopentane and iso-
butane is far from unity. These features are independent evidence
for the absence of a carbocationic cracking mechanism via the
paring reaction [24]. Relatively large amounts of cyclohexane and

A. Haas et al. / Applied Catalysis A: General 425–426 (2012) 97–109
103
a
b
100
80
60
40
20
0
100
80
60
40
20
0
0.77Ir/silica
2.59Ir/silica
0.93Pt/silica
2.68Pt/silica
1
2
3
4
5
6
7
8
9
1
2
3
4
5
6
7
8
9
Number of C-atoms
Number of C-atoms
Fig. 5. Modified hydrocracking selectivities Sj* in the hydroconversion of decalin (0.77Ir/silica: Tr = 330 ◦C; XDec = 47%; YC = 16%; ꢀ Sj* = 246%. 2.59Ir/silica: Tr = 270 ◦C;
−
9
XDec = 37%; YC = 7%; ꢀ Sj* = 217%. 0.93Pt/silica: Tr = 410 ◦C; XDec = 43%; YC = 3%; ꢀ Sj* = 274%. 2.68Pt/silica: Tr = 410 ◦C; XDec = 53%; YC = 7%; ꢀ Sj* = 294%).
−
−
−
9
9
9
Table 3
Details on the C4, C6 and C7 fractions of the products formed in the hydroconversion of decalin (i-Bu: iso-butane; M-CPn: methylcyclopentane; n-Hx: n-hexane; 2-M-Pn:
2-methylpentane; 3-M-Pn: 3-methylpentane; CHx: cyclohexane; M-CHx: methylcyclohexane).
n
n
n
+n
+n
3−M-Pn
n
n
n
n
n
n
M-CHx
n
C
7
n-Hx
2−M-Pn
Tr/◦C
XDec/%
YC − /%
i-Bu
M-CPn
n
C
6
M-CPn
CHx
C
6
Catalyst
n
C
6
9
C
i−Bu
4
0.77Ir/silica
2.59Ir/silica
0.93Pt/silica
2.68Pt/silica
330
270
410
410
47
37
43
53
16
0.26
0.12
0.36
0.14
0.03
0.00
0.28
0.05
0.77
0.80
0.23
0.35
0.10
0.00
0.72
0.27
0.16
0.16
0.38
0.53
0.60
0.58
0.75
0.81
7
3
7
amounts of propylcyclohexane could not be reliably determined in
a quantitative manner.
feature of hydrogenolysis on iridium to preferentially cleave bonds
between two secondary carbon atoms over bonds between a sec-
ondary and a tertiary carbon atom [15,19]. The direct ROPs undergo
consecutive hydrogenolysis on the iridium, either in the remaining
six-membered ring or in an alkyl side chain. In the former case,
one out of seven decane isomers listed in the bottom-left box of
Fig. 7 is predicted to form. We refer to these isomers as “direct
open-chain decanes” or “direct OCDs”. Side-chain hydrogenoly-
sis of the direct ROPs gives a light alkane, typically methane, and
an alkylated cyclohexane with less than ten, typically nine car-
bon atoms. As shown in the bottom-center box of Fig. 7, methane
abstraction from the direct ROPs results in the formation of either
1-ethyl-2-methylcyclohexane or propylcyclohexane. We call these
“direct C9 ROPs”. In the group of direct C9 ROPs, cis- and trans-
1-ethyl-2-methylcyclohexane are strongly prevailing, while little
propylcyclohexane is formed (see Fig. 6). Endocyclic hydrogenoly-
sis of the direct C9 ROPs is expected to give one out of seven nonane
isomers listed in the bottom-right box of Fig. 7. We call these
nonane isomers “direct open-chain nonanes” or “direct OCNs”. Of
the seven direct OCDs listed in Fig. 7, the five first-mentioned
ones were safely identified in the product obtained on 2.59Ir/silica
under mild reaction conditions (Tr = 290 ◦C, XDec = 73%). The two
remaining ones, denoted as unidentified OCD(1) and OCD(2) in
Section 3.3.1, were found to be present in the same product with
a high degree of certainty. At more severe conditions and ele-
vated conversions direct OCDs, direct C9 ROPs, and direct OCNs
all undergo consecutive hydrogenolysis on iridium to give lighter
hydrocarbons, and their quantitative distributions are described by
hammock-type curves as shown in Fig. 5a.
The results discussed so far can be accounted for by the reaction
scheme in Fig. 7 for decalin hydroconversion on iridium on a sup-
port lacking Brønsted acid sites. No skeletal isomerization occurs,
because neither the support nor the metal is active for this reaction.
Instead, direct opening of one six-membered ring occurs through
hydrogenolysis resulting in the five direct ROPs (three are shown
of which two form as cis- and trans-isomers). As seen from Fig. 6,
the most abundant direct ROPs are 1-methyl-2-propylcyclohexane
followed by 1,2-diethylcyclohexane, while significantly less butyl-
cyclohexane is formed. This selectivity is consistent with the known
In Fig. 8 the selectivities of direct ROPs, direct OCDs, direct C9
ROPs, and direct OCNs formed from decalin on the 2.59Ir/silica cat-
alyst are plotted versus the reaction temperature. (Note that the
values plotted in Fig. 8 pertain to the direct ROPs, the direct OCDs,
the direct C9 ROPs, and the direct OCNs as defined in the four boxes
of Fig. 7 and that these values differ from those plotted in Fig. 3b.)
As implied by the scheme in Fig. 7, direct ROPs are indeed primary
products from which direct OCDs, direct C9 ROPs, and direct OCNs
form in consecutive reactions.
Fig. 6. Section of the gas chromatogram of the product of decalin hydroconver-
sion on the 2.59Ir/silica catalyst at 250 ◦C and XDec = 17%. The direct ring-opening
products (direct ROPs) are: butylcyclohexane (B-CHx), cis- and trans-1-methyl-
2-propylcyclohexane (1-M-2-P-CHx), and cis- and trans-1,2-diethylcyclohexane
(1,2-DE-CHx). The direct C9 ROPs are cis- and trans-1-ethyl-2-methylcyclohexane
(1-E-2-M-CHx) and propylcyclohexane (P-CHx). a and b denote the two stereoiso-
mers.

104
A. Haas et al. / Applied Catalysis A: General 425–426 (2012) 97–109
Fig. 7. The direct ring-opening mechanism for hydrogenolytic hydrodecyclization of decalin and consecutive hydrogenolysis reactions over iridium on non-acidic supports
(note that decalin, 1-methyl-2-propylcyclohexane, 1,2-diethylcyclohexane, and 1-ethyl-2-methylcyclohexane occur as cis- and trans-isomers).
As column 4 of Table 4 shows, on the two iridium/silica catalysts
at low decalin conversion (12 and 17%, respectively), the content
of direct ROPs in the total ring-opening products is as high as 96%.
Also given in Table 4 are the predicted distributions of individual
direct ROPs for the two limiting cases of a purely non-selective ring
opening (equal probability of cleavage of all endocyclic bonds) and
a selective ring opening (no cleavage of endocyclic bonds between
secondary and tertiary carbon atoms; equal rates of cleavage of
endocyclic bonds between two secondary carbon atoms). Based
on the fact that no cyclodecane was formed on any iridium cata-
lyst, cleavage of the bridging bond between the two tertiary carbon
atoms in decalin is viewed to be absent on the two catalysts con-
taining iridium. As the data in Table 4 reveal, the experimental
distributions of the direct ROPs observed on the two iridium/silica
catalysts can neither be understood in terms of a purely selective
nor in terms of a purely non-selective hydrogenolysis mechanism.
Rather, the measured distributions of the direct ROPs contain fea-
tures of both mechanisms, and they may both be operative on the
iridium/silica catalysts, in agreement with what has been claimed
by McVicker et al. [15] for the hydrogenolytic ring opening of
methylcyclohexane and by Moraes et al. [17] for ring opening of
decalin on iridium/alumina catalysts. The distributions of the direct
ROPs listed in Table 4 for the platinum/silica catalysts will be dis-
cussed in Section 3.6.
As already shown in the second last column of Table 2, it is a
salient feature of decalin hydroconversion on the iridium/silica cat-
alysts that noticeable amounts of open-chain nonanes (OCNs) are
formed. It is worthwhile to throw a closer look at their compo-
sition, especially at low decalin conversions. Table 5 reveals that
100% of the nonane isomers found under these conditions belong
to the group of direct OCNs, as listed in the bottom-right box of
Fig. 7. Particularly abundant isomers are 3-ethyl-4-methylhexane,
3-ethyl-2-methylhexane, 4-methyloctane, 3,4-dimethylheptane,
3-ethylheptane, and n-nonane. The scheme depicted in Fig. 7
implies that the open-chain nonanes are formed by endocyclic
hydrogenolysis of the direct C9 ROPs propylcyclohexane and
1-ethyl-2-methylcyclohexane. Of these, propylcyclohexane can
probably be neglected as a precursor for the formation of OCNs,
since it is present in a much lower concentration than the
two stereoisomers of 1-ethyl-2-methylcyclohexane, see Fig. 6.
Table 4
Composition of the direct ring-opening products obtained on the four non-acidic catalysts at low decalin conversions.
S
S
S
S
S
Tr/◦C
XDec/%
direct ROPs /%
/%
1-M-2-P-CHx /%
1,2−DE-CHx /%
S
direct ROPs
B-CHx
direct ROPs
Catalyst
S
S
ROPs
direct ROPs
0.77Ir/silica
2.59Ir/silica
0.93Pt/silica
2.68Pt/silica
290
250
370
370
–
12
17
15
20
–
96
96
65
79
100
100
13
13
54
64
40
0
52
51
34
28
40
67
35
36
12
8
20
33
Predicted for non-selective ring opening of decalin
Predicted for selective ring opening of decalin
–
–

A. Haas et al. / Applied Catalysis A: General 425–426 (2012) 97–109
105
Table 5
Detailed composition of the open-chain nonanes (OCNs) formed in the decalin hydroconversion on the iridium/silica catalysts at low conversion.
Catalyst
0.77Ir/silica
2.59Ir/silica
Predictedbyringopeningof1-ethyl-2-methylcyclohexane
Non-selective
Selective
Tr/◦C
XDec/%
Sdirect OCNs/SOCNs (%)
Sn-No/Sdirect OCNs (%)
290
12
100
8
250
17
100
8
16 2/3
16 2/3
16 2/3
0
0
0
0
S4-M-Oc/Sdirect
S3-E-Hp/Sdirect
S4-E-Hp/Sdirect
(%)
(%)
(%)
14
11
0
12
9
3
OCNs
OCNs
OCNs
0
S3,4-DM-Hp/Sdirect
S3-E-2-M-Hx/Sdirect
S3-E-4-M-Hx/Sdirect
(%)
(%)
(%)
9
22
36
13
20
35
16 2/3
16 2/3
16 2/3
33 1/3
33 1/3
33 1/3
OCNs
OCNs
OCNs
It appears hence justified to consider the latter direct C9 ROPs
as the predominant precursors for OCN formation. Also given
in Table 5 are the predicted OCN distributions expected for
a purely non-selective and a purely selective ring opening of
1-ethyl-2-methylcyclohexane. When comparing the experimen-
tally measured distributions of OCNs with those predicted on
the basis of the two ring-opening mechanisms, an explanation
for the absence (0.77Ir/silica) or near-absence (2.59Ir/silica) of
4-ethylheptane in the product group of direct OCNs is readily
obtained. This nonane isomer cannot form by ring opening of 1-
ethyl-2-methylcyclohexane, neither via the non-selective nor via
the selective mechanism. In quantitative terms, neither the non-
selective mechanism alone nor the selective mechanism alone
can account for the measured distributions of OCNs. Rather, the
real ring-opening mechanism seems to feature elements of both
established mechanisms. It should be mentioned that McVicker
et al. arrived at similar conclusions for ring opening of 1,2,4-
trimethylcyclohexane on iridium/alumina [15].
carbon-number distribution of the hydrocracked products (see
Fig. 5a). The predominance of methane and C9 fragments plus
ethane and C8 fragments can only be rationalized by invoking
a hydrogenolytic mechanism for carbon carbon bond cleavage
(rather than by a bifunctional hydrocracking via carbocations). One
such mechanism which accounts, in a qualitative manner, for the
products observed at low decalin conversion, has been presented
in Fig. 7. It implies that no skeletal isomerization of hydrocarbons
happens in this catalytic system, neither on metallic iridium nor
on the non-acidic support. To gain more quantitative insight into
the catalytic chemistry on iridium/silica, an attempt was made to
simulate the hydrogenolytic pathways and to compare the car-
bon number distributions thereby predicted with those measured
experimentally on 2.59Ir/silica.
The underlying reaction network for the simulation is depicted
in Fig. 9 which is a simplified version of Fig. 7. The relevant infor-
mation from Fig. 9 is that hydrocracked products (C9− ) are formed
from decalin on iridium/silica via two principal pathways: Both
start with the hydrogenolytic opening of one six-membered ring
to the five direct ROPs. The latter are key intermediates on the way
3.5. Simulation of the hammock-type distribution curves for the
hydrocracked products formed on the iridium/silica catalysts
to hydrocarbons with nine or less carbon atoms: in path 1, C9
−
products are readily made from direct ROPs by hydrogenolysis in
an alkyl side chain. Alternatively, in path 2 direct OCDs are first
formed by endocyclic hydrogenolysis of direct ROPs, and the direct
OCDs undergo a consecutive hydrogenolysis to C9− hydrocarbons.
For translating the network in Fig. 9 into quantitative carbon-
As a characteristic feature of decalin ring opening over irid-
ium on a non-acidic support, hammock-type curves result for the
−
number distributions of the C9 products, assumptions have to
be introduced as to the relative probabilities of cleavage of the
various carbon carbon bonds in the respective precursor hydro-
carbons. In the simplest version of the simulation, it is assumed
that all carbon carbon bonds in decalin (except the bridging bond,
vide infra), all exocyclic carbon carbon bonds in the direct ROPs,
all endocyclic carbon carbon bonds in the direct ROPs, and all
carbon carbon bonds in the direct OCDs undergo hydrogenolysis
at equal probabilities. The bridging carbon carbon bond in decalin
is assumed not to be cleaved at all, as no cyclodecane was formed
(cf. Section 3.4.). Furthermore, the assumptions are made that the
rates of formation of C9− hydrocarbons via routes 1 and 2 in Fig. 9
approximately equal each other and that no further hydrogenolysis
of C9− to smaller moieties happens.
2.59Ir/silica
60
40
20
0
The curve in Fig. 10 with the open circles shows the
result of this simulation. A comparison with the experimen-
tal distribution curves in Fig. 5a reveals that the princi-
pal feature of the hammock-type distribution curves, i.e.
a
strongly decreasing contribution of hydrogenolysis leading to
(C1 + C9) > (C2 + C8) > (C3 + C7) > (C4 + C6) > (C5 + C5) is already cor-
rectly predicted in qualitative terms, but in the experiment, the
preponderance of hydrocracking reactions leading to C1 + C9 is even
more pronounced than in the simulation. In an attempt to obtain a
better match between the result of the simulation and the exper-
imental curves, the simulation was repeated starting from the
260
280
300
Temperature / °C
Fig. 8. Selectivities of direct ROPs (ꢀ), direct OCDs (ꢁ), direct C9 ROPs (ꢀ), and direct
OCNs (ꢁ) in the hydroconversion of decalin on 2.59Ir/silica.

106
A. Haas et al. / Applied Catalysis A: General 425–426 (2012) 97–109
Fig. 9. Underlying network for the simulation of hydrocracking decalin to hydrocarbons with less than ten carbon atoms via hydrogenolysis on non-acidic iridium catalysts.
distribution of the individual direct ROPs measured on 2.59Ir/silica
at Tr = 250 ◦C and low conversion of XDec = 17% (see Table 4), instead
of from decalin. This simulation led to a slightly improved fit with
the experiment (Fig. 10, full circles), but the quantitative agreement
continued to be poor.
A more rigorous simulation takes into consideration the well
established fact that, in the hydrogenolysis of hydrocarbons on
iridium, cleavage of carbon carbon bonds between two unsubsti-
tuted (secondary or primary) carbon atoms is faster than cleavage of
carbon carbon bonds between carbon atoms of which at least one
is substituted (tertiary or quaternary). To account for this regiose-
lectivity in a quantitative manner, the factors ˛unsubst. or ˛subst. are,
respectively, defined:
hydrogenolysis of direct OCDs (route 2 in Fig. 9) allows one to calcu-
late a simulated carbon number distribution of the hydrocracked
products. It is depicted in Fig. 11 along with the experimentally
determined distribution for decalin hydrogenolysis on 2.59Ir/silica
at Tr = 250 ◦C and XDec = 17%. Obviously, this simulation results in
a carbon-number distribution that matches the experimentally
determined one significantly better. The grey and black bars at each
carbon number in Fig. 11 indicate to what extent the hydrocarbons
were, respectively, formed by exocyclic hydrogenolysis of direct
ROPs (route 1 in Fig. 9) or via direct OCDs (route 2 in Fig. 9). The
results of the simulation based on the underlying assumptions sug-
gest that slightly more than 50% (black bars) of the hydrocracked
products are formed via route 2, i.e. via direct OCDs. The relatively
good agreement between the experimental and the simulated dis-
tribution of the hydrocracked products are also considered to be
strong support for the mechanistic pathways suggested in Fig. 7 for
decalin hydroconversion on non-acidic iridium/silica catalysts.
In the course of this simulation, a predicted distribution of
the formed direct OCDs was also obtained (Fig. 12). Although the
mechanistic input for the simulation was taken from the catalytic
experiment at Tr = 250 ◦C, whereas the measured OCD distribution
on 2.59Ir/silica shown in Fig. 4 holds for Tr = 290 ◦C, both graphs
are qualitatively similar. Furthermore, in the experiment 95% of all
OCDs were direct OCDs at Tr = 290 ◦C. Even though 76% of all OCDs
are predicted to be multi-branched by the simulation, whereas a
lower value of 63% was obtained in the experiment, we believe
that the qualitative similarity of both distributions can be looked
upon as supporting the underlying mechanistic assumptions.
number of hydrogenolyzed C C bonds between unsubstituted C−atoms
˛
≡
unsubst.
total number of C C bonds hydrogenolyzed
(2)
number of hydrogenolyzed C C bonds involving substituted C − atoms
˛
subst.
≡
total number of C C bonds hydrogenolyzed
(3)
and
˛
= 1 − ˛subst.
(4)
unsubst.
From the distribution of the direct ROPs formed from decalin on
2.59Ir/silica at low conversion (XDec = 17%) at Tr = 250 ◦C (Table 4),
˛subst. = SB–CHx/Sdirect ROPs = 0.13, and ˛unsubst. = 0.87 are obtained.
Starting from the measured distribution of the individual direct
ROPs and applying the applicable value of ˛ for the exocyclic
hydrogenolysis of direct ROPs (route 1 in Fig. 9), the endo-
cyclic hydrogenolysis of direct ROPs (route 2 in Fig. 9), and the
80
Experiment
Simulation
100
Starting from decalin
Starting from real mixture
of direct ROPs
Hydrogenolysis of ROPs
60
40
20
0
Hydrogenolysis of OCDs
80
60
40
20
0
1
2
3
4
5
6
7
8
9
Number of C-atoms
1
2
3
4
5
6
7
8
9
Fig. 11. Predicted carbon-number distribution of the hydrocracked products in the
hydroconversion of decalin on iridium/silica using the assumptions of a preferred
cleavage of unsubstituted carbon carbon bonds (filled circles). The experimen-
Number of C-atoms
tally determined curve for decalin hydroconversion on 2.59Ir/silica at Tr = 250 ◦
C
Fig. 10. Predicted carbon-number distributions of the hydrocracked products in
the hydroconversion of decalin on iridium/silica using the assumption of equal
probabilities of carbon carbon bond cleavage in the precursors of C9− .
and XDec = 17% is also shown (empty circles). For the significance of black and grey
bars see text.

A. Haas et al. / Applied Catalysis A: General 425–426 (2012) 97–109
107
30
100
80
60
40
20
0
Simulated
20
10
MBDe
n-C
M-C
E-C
8
10
9
0
0
10
20
30
40
50
60
Fig. 12. Breakdown of the simulated selectivities of OCDs.
Conversion / %
Fig. 14. Selectivities of butylcyclohexane on 0.93Pt/silica (ꢂ) and 2.68Pt/silica (ꢁ)
and of spiro[4.5]decane on 0.93Pt/silica (ꢃ) and 2.68Pt/silica (ꢁ) in dependence of
the decalin conversion.
3.6. Details of the reactions on the platinum/silica catalysts
On both platinum/silica catalysts, the most abundant prod-
uct hydrocarbons at all conversions were butylcyclohexane and
spiro[4.5]decane (beside tetralin at elevated temperatures). For
example, on 2.68Pt/silica at the lowest conversion of 9% (Fig. 13),
ca. 50% of all ROPs formed consisted of butylcyclohexane, and
spiro[4.5]decane accounted for 62% of all sk-Isos. In fact, Fig. 14
strongly suggests that butylcyclohexane and spiro[4.5]decane are
primary products in decalin conversion on the two non-acidic plat-
inum/silica catalysts.
The formation of butylcyclohexane from decalin is readily
understood in terms of a hydrogenolytic rupture of the endocyclic
secondary-tertiary carbon carbon bond, in much the same man-
ner as n-heptane has previously been reported to be the prevailing
product in the hydrogenolysis of methylcyclohexane on Pt/alumina
[15]. Cleavage of the bonds between two secondary carbon atoms
leading to the ROPs cis- and trans-1-methyl-2-propylcyclohexane
and cis- and trans-1,2-diethylcyclohexane occurs to a considerably
lesser extent on the platinum/silica catalysts than on iridium/silica
(see Table 4). It is remarkable that for the iridium catalysts the
distribution of direct ROPs is practically independent of the metal
content, whereas for the platinum catalysts the selectivities depend
on the metal content as reported by others [18,19]. Interestingly,
even the bridging C C bond in decalin, i.e. the bond between the
two tertiary carbon atoms, is broken to a small extent on the two
platinum/silica catalysts, at variance to hydrogenolysis on iridium.
On 2.68Pt/silica, for example, cyclodecane was formed from decalin
with a maximal selectivity of 0.5% at a reaction temperature of
410 ◦C.
As demonstrated very recently [24], spiro[4.5]decane is read-
ily formed from decalin on bifunctional catalysts, such as Pt/La-X
zeolite, probably at the level of carbocations at the Brønsted acid
sites via a rapid type A rearrangement [35]. However, the platinum
catalysts used in this study are lacking Brønsted acid sites, and no
spiro[4.5]decane at all was formed on the iridium/silica catalysts.
We hence prefer to invoke a purely platinum-catalyzed mechanism
for isomerizing decalin into spiro[4.5]decane. In fact, the capability
of platinum metal to induce skeletal isomerization of hydrocar-
bons is well established in the catalytic chemistry of light alkanes
and referred to as the bond-shift mechanism [36,37]. It is generally
believed to proceed via cyclopropanoid species and was observed
to be the predominant type of metal-catalyzed isomerization of
n-octane when the hydrogen pressure was increased from 0.5 to
2.0 MPa [32]. As shown in Scheme 1, applying the bond-shift mech-
anism to decalin leads one to predict spiro[4.5]decane as a primary
product.
It should be mentioned that a rigorous application of the
bond-shift mechanism to decalin results in the prediction of
ten additional isomers, viz. the cis- and trans-isomers of bicy-
clo[5.3.0]decane, 1-, 2- and 3a-methylbicyclo[4.3.0]nonane and
of bicyclo[4.3.1]decane, beside spiro[4.5]decane. A pattern of ten
prominent peaks with a size comparable to that of spiro[4.5]decane
was not observed. The two stereoisomers of bicyclo[5.3.0]decane
were formed with a maximal total selectivity of only 2%. As
an alternative pathway, one could consider the formation of
spiro[4.5]decane by a platinum-catalyzed 1,5-dehydrocyclization
[19,38,39] of butylcyclohexane formed by ring opening of decalin
(Scheme 2).
The problem with such a mechanism is that spiro[4.5]decane
would be expected to be a secondary product formed from the pri-
mary product butylcyclohexane, whereas Fig. 14 seems to suggest
that spiro[4.5]decane is rather a primary product.
Fig. 13. Section of the gas chromatogram of the product of decalin hydroconver-
sion on the 2.68Pt/silica catalyst at 350 ◦C and XDec = 9%. a and b denote the two
stereoisomers.

108
A. Haas et al. / Applied Catalysis A: General 425–426 (2012) 97–109
Scheme 1. Possible platinum-catalyzed formation of spiro[4.5]decane from decalin via the bond-shift mechanism.
Scheme 2. Hypothetical platinum-catalyzed formation of spiro[4.5]decane from decalin via ring opening to butylcyclohexane followed by 1,5-dehydrocyclization.
4. Conclusions
and butylcyclohexane. The latter is readily accounted for by a
hydrogenolytic rupture of a secondary-tertiary carbon carbon
bond in decalin. Platinum on non-acidic silica appears to catalyze
skeletal isomerization beside hydrogenolysis, and the so-called
bond-shift mechanism [37] via cyclopropanoid species could be
responsible for the direct formation of spiro[4.5]decane.
Two iridium/silica and two platinum/silica catalysts with
different metal contents were prepared and tested in the hydrode-
cyclization of cis-decalin. Two characterization techniques, viz.
FT-IR with pyridine as probe and measuring the selectivities
of n-octane hydroconversion, indicated that catalytically active
Brønsted acid sites were lacking on the metal/silica catalysts. This
enabled a detailed study of cis-decalin conversion on the respective
metal without overly complex product distributions or an interfer-
ence of bifunctional catalysis via carbocationic intermediates.
Stereoisomerization of cis- to trans-decalin was by far the fastest
reaction: the cis/trans-equilibrium corresponding to a mole frac-
tion ntr-Dec/(nc-Dec + ntr-Dec) ≈ 0.8–0.9 was attained at all reaction
temperatures from 250 to 410 ◦C over the whole range of decalin
conversions. This rapid stereoisomerization must be concluded to
occur on the respective metal via 1,9- or 9,10-octalins, as proposed
in the literature [33]. It is likely that this result has a considerable
bearing on decalin hydroconversion over metal catalysts which do
possess acid sites [24,25].
Acknowledgment
We gratefully acknowledge Dipl.-Chem. Dominic Santi for the
catalyst characterization by FT-IR. Moreover, technical assistance
by Dipl.-Chem. Tobias Holl in the preparation of this paper is highly
appreciated.
References
[1] A. Stanislaus, B.H. Cooper, Catal. Rev.: Sci. Eng. 36 (1994) 75–123.
[2] B.H. Cooper, B.B.L. Donnis, Appl. Catal. A: Gen. 137 (1996) 203–223.
[3] M. Sanati, B. Harrysson, M. Faghihi, B. Gevert, S. Järås, in: J.J. Spivey (Ed.), Cat-
alytic Hydrodearomatization, Catalysis, vol. 16, The Royal Society of Chemistry,
Cambridge, 2002, pp. 1–42.
On the non-acidic catalysts used in this investigation, ring open-
ing of decalin occurs by hydrogenolysis on the metal. There are
pronounced differences in hydrodecyclization on the two met-
als used: Iridium is much more active than platinum in the ring
opening reaction, but iridium supported on non-acidic silica has
practically no activity for skeletal isomerization of hydrocarbons,
at variance to platinum. As a consequence, decalin hydroconver-
sion on Ir/silica starts with the direct opening of one six-membered
ring resulting in a small number of direct ring-opening products
(direct ROPs). These alkylated cyclohexanes undergo consecutive
hydrogenolysis reactions, either of an endocyclic carbon carbon
bond to give one out of seven open-chain decanes (direct OCDs) or
of an exocyclic carbon carbon bond to give a small alkane, typically
methane, and one out of two C9 naphthenes (direct C9 ROPs). The
latter can undergo another endocyclic hydrogenolysis leading to
one of seven open-chain nonanes (direct OCNs). All individual C10
and C9 hydrocarbons thereby predicted could be identified in the
products generated on iridium/silica at low conversions. The direct
ring-opening mechanism could be simulated in a quantitative man-
[4] D. Kubicˇka, N. Kumar, P. Mäki-Arvela, M. Tiitta, V. Niemi, H. Karhu, T. Salmi, D.Y.
Murzin, J. Catal. 227 (2004) 313–327.
[5] M. Santikunaporn, J.E. Herrera, S. Jongpatiwut, D.E. Resasco, W.E. Alvarez, E.L.
Sughrue, J. Catal. 228 (2004) 100–113.
[6] M.A. Arribas, P. Concepción, A. Martínez, Appl. Catal. A: Gen. 267 (2004)
111–119.
[7] K.C. Mouli, V. Sundaramurthy, A.K. Dalai, J. Mol. Catal. A: Chem. 304 (2009)
77–84.
[8] M.A. Arribas, A. Corma, M.J. Díaz-Caban˜as, A. Martínez, Appl. Catal. A: Gen. 273
(2004) 277–286.
[9] H. Ma, X. Yang, G. Wen, G. Tian, L. Wang, Y. Xu, B. Wang, Z. Tian, L. Lin, Catal.
Lett. 116 (2007) 149–154.
[10] M.A. Arribas, J.J. Mahiques, A. Martínez, in: A. Galarneau, F. Di Renzo, F. Fajula,
J. Védrine (Eds.), Zeolites and Mesoporous Materials at the Dawn of the 21st
century, Studies in surface science catalysis, vol. 135, Elsevier, Amsterdam,
2001, p. 303.
[11] M.A. Arribas, A. Martínez, G. Sastre, in: R. Aiello, G. Giordano, F. Testa (Eds.),
Impact of Zeolites and Other Porous Materials on the New Technologies at the
Beginning of the New Millennium, Studies in surface science catalysis, vol. 142,
Part B, Elsevier, Amsterdam, 2002, pp. 1015–1022.
[12] H. Liu, X. Meng, D. Zhao, Y. Li, Chem. Eng. J. 140 (2008) 424–431.
[13] K.C. Mouli, V. Sundaramurthy, A.K. Dalai, Z. Ring, Appl. Catal. A: Gen. 321 (2007)
17–26.
[14] S. Nassreddine, L. Massin, M. Aouine, C. Geantet, L. Piccolo, J. Catal. 278 (2011)
253–265.
[15] G.B. McVicker, M. Daage, M.S. Touvelle, C.W. Hudson, D.P. Klein, W.C. Baird, B.R.
Cook Jr., J.G. Chen, S. Hantzer, D.E.W. Vaughan, E.S. Ellis, O.C. Feeley, J. Catal. 210
(2002) 137–148.
ner whereby the peculiar carbon-number distributions of the C9
−
hydrocarbons referred to as hammock-type curves could be readily
predicted.
Platinum/silica catalysts require high reaction temperatures in
the vicinity of 400 ◦C. This brings about the formation of relatively
large amounts of tetralin and naphthalene. The dominating primary
products found on Pt/silica at low conversions are spiro[4.5]decane
[16] M. Daage, G.B. McVicker, M.S. Touvelle, C.W. Hudson, D.P. Klein, B.R. Cook, J.G.
Chen, S. Hantzer, D.E.W. Vaughan, E.S. Ellis, in: A. Galarneau, F. Fajula, F. Di
Renzo, J. Védrine (Eds.), Zeolites and Mesoporous Materials at the Dawn of
the 21st Century, Studies in Surface Science and Catalysis, vol. 135, Elsevier,
Amsterdam, 2001, p. 159.

A. Haas et al. / Applied Catalysis A: General 425–426 (2012) 97–109
109
[17] R. Moraes, K. Thomas, S. Thomas, S. Van Donk, G. Grasso, J.-P. Gilson, M. Houalla,
J. Catal. 286 (2012) 62–77.
[18] G. Maire, G. Plouidy, J.C. Prudhomme, F.G. Gault, J. Catal. 4 (1965) 556–569.
[19] F.G. Gault, in: D.D. Eley, H. Pines, P.B. Weisz (Eds.), Advances in Catalysis, vol.
30, Academic Press, New York, 1981, pp. 1–95.
[20] S.L. González-Cortés, S. Dorkjampa, P.T. Do, Z. Li, J.M. Ramallo-López, F.G.
Requejo, Chem. Eng. J. 139 (2008) 147–156.
[21] S. Dokjampa, T. Rirksomboon, D.T.M. Phuong, D.E. Resasco, J. Mol. Catal. A:
Chem. 274 (2007) 231–240.
[28] J. Weitkamp, in: J.W. Ward, S.A. Qader (Eds.), Hydrocracking and Hydrotreating,
Am. Chem. Soc. Symp. Ser., vol. 20, American Chemical Society, Washington, D.
C., 1975, pp. 1–27.
[29] J. Weitkamp, R. Gläser, in: G. Ertl, H. Knözinger, F. Schüth, J. Weitkamp (Eds.),
Handbook of Heterogeneous Catalysis, 2nd edn., vol. 4, Wiley-VCH, Weinheim,
2008, pp. 2045–2053.
[30] J. Weitkamp, Ind. Eng. Chem. Prod. Res. Dev. 21 (1982) 550–558.
[31] G.B. McVicker, R.T. Baker, R.L. Garten, E.L. Kugler, J. Catal. 65 (1980) 207–220.
[32] J. Weitkamp, Proceedings of the 9th Iberoamerican Symposium on Catalysis
Sociedad Iberoamericana de Catálise, vol. 2, Lisbon, 1984, pp. 1332–1341.
[33] A.W. Weitkamp, in: D.D. Eley, H. Pines, P.B. Weisz (Eds.), Advances in Catalysis,
vol. 18, Academic Press, New York, 1968, pp. 1–110.
[22] P.T. Do, W.E. Alvarez, D.E. Resasco, J. Catal. 238 (2006) 477–488.
[23] R.J. Chimentão, G.P. Valenc¸ a, F. Medina, J. Pérez-Ramírez, Appl. Surf. Sci. 253
(2007) 5888–5893.
[24] S. Rabl, A. Haas, D. Santi, C. Flego, M. Ferrari, V. Calemma, J. Weitkamp, Appl.
Catal. A: Gen. 400 (2011) 131–141.
[25] S. Rabl, D. Santi, A. Haas, M. Ferrari, V. Calemma, G. Bellussi, J. Weitkamp,
Micropor. Mesopor. Mater. 146 (2011) 190–200.
[26] (a) J. Weitkamp, S. Rabl, A. Haas, D. Santi, M. Ferrari, V. Calemma, in: S. Ernst,
A. Lercher, J. Lichtscheidl, M. Marchionna, F. Nees, E. Santacesaria (Eds.), The
Future Role of Hydrogen in Petrochemistry and Energy Supply, DGMK Tagungs-
bericht 2010-3, DGMK, Hamburg, 2010, pp. 77–86;
[34] R.C. Santana, P.T. Do, M. Santikunaporn, W.E. Alvarez, J.D. Taylor, E.L. Sughrue,
D.E. Resasco, Fuel 85 (2006) 643–656.
[35] J. Weitkamp, M. Hunger, in: J. Cejka, H. van Bekkum, A. Corma, F. Schüth (Eds.),
Introduction to Zeolite Science and Practice, Studies in surface science and
catalysis, vol. 168, Elsevier, Amsterdam, 2007, pp. 787–835.
[36] J.R. Anderson, N.R. Avery, J. Catal. 2 (1963) 542–544.
[37] G.C. Bond, in: M.V. Twigg, M.S. Spencer (Eds.), Fundamental and Applied Catal-
ysis, Springer, New York, 2005, pp. 625–628.
ˇ
(b) Oil Gas Eur. Mag. 37 (2011) 94–98.
[27] J.T. Miller, M. Schreier, A.J. Kropf, J.R. Regalbuto, J. Catal. 225 (2004) 203–212.
[38] Y. Barron, G. Maire, J.M. Muller, F.G. Gault, J. Catal. 5 (1966) 428–445.
[39] P. Légaré, F. Garin, J. Catal. 214 (2003) 336–340.