Vous êtes sur la page 1sur 95

583

Chapter 13

REACTIONS OF ALKANES AND REFORMING OF NAPHTHA

13.1

Fundamentals

13.1.1 Adsorption of hydrocarbons under reaction conditions


Some of the spectroscopic and other surface science techniques, such as UV-VIS or
FT-IR spectroscopies, are well suited to monitor adsorption in situ, while catalytic
reactions are running. However, many other techniques such as EELS, LEED, XPS and
UPS, requiring high vacuum, are not applicable. Moreover almost all spectroscopic
techniques which can be used with metals and alloys suffer from the inherent drawback
that the most observable and easily visible species are usually those which are only of
indirect importance for the reaction studies, being for example poisons arising in situ from
the reaction itself. It is extremely difficult to detect species which can be considered with
confidence as reactive intermediates. Even the name indicates that their steady state
concentration on the catalyst surface will be low. Thus, regardless of the ever continuing
accumulation of the most valuable information provided by various spectroscopic
techniques, new attempts are always being made to identify the reactive intermediates by
chemical methods; some catalytic reactions indicate quite clearly what these intermediates
must be.
Very important information on the reactivity and adsorption modes of hydrocarbons
on metals and alloys has been obtained by the hydrocarbon-deuterium equilibration
reactions. Kemball and his associates were responsible for much of progress in this field,
by developing the theoretical basis of these reactions [1-3]. The literature offers several
reviews on this subject [3-5], the matter is also discussed in chapter 10.2.
Exchange reactions
Exchange reactions have revealed several important aspects of the behaviour of
alkanes on catalysts, as we have already seen in chapter 10.
Let us briefly summarize the points (i)-(v) which we have to keep in mind when
discussing reforming reactions on metals and alloys.
(i) Although the bond strength (dissociation-energy) of C-H bonds is higher in
many molecules than the dissociation energy of the C-C bond, the C-H groups react first
[1-4] (see also chapter 10). All alkanes can be bonded to the active site through a multiple

584

chapter 13

bond on one carbon-atom [4,5]


(ii) Formation of metal-carbon bonds (see the estimates in [6-10]) can make the
dissociations in alkanes thermodynamically feasible.
(iii) Methane is considerably less active than higher alkanes and needs higher
temperatures for its exchange [4] (see also chapter 10).
Recent molecular beam and vibrational spectroscopy investigations revealed why just the
adsorption of methane is so difficult. To initiate the C-H dissociation, methane molecule
must be pressed against the surface [11 ] to achieve a better contact of both C and H with
the surface atoms, as shown schematically by scheme I:
Scheme I

Methane is a rather small molecule, its Van der Waals interaction with the surface is weak
and therefore, either a high temperature of the whole system or at least a high kinetic
energy of

CH 4

such as in supersonic beams, is necessary to press it against the surface for

it to be dissociatively adsorbed. Higher alkanes interact physically with the surface in a


way which is strong enough to allow good sticking and subsequent dissociation, without
any activation energy or with only a small one. Figure 1 shows an appropriate LennardJones diagram, describing the dissociative adsorption of alkanes and alkenes. One can
expect: the higher the molecular weight, the stronger the physical interaction, the deeper
the minimum on the curves 1 and 2 and the lower the activation energy of dissociative
adsorption.

Epot

If,\

I 2~~~----~olko nes
/ ~lkenes3
v~-. diss. adsorbed

Io'c-H'
distclnce
from the

figure 1
Lennard-Jones potential energy diagram showing schematically the energy
changes along the transitions:
alkanes ~ dissociative adsorption (1,3)
alkenes ---) dissociative adsorption (2,3).

surface

As mentioned on other places in this book, this diagram does not show exactly
how a molecule reaches its final state of adsorption. A molecule can bounce against the
walls of potential wells, the exact movement depending on details of energy transfer
between the molecule and the metal or alloy surface, and on other things which cannot be
shown by such a simple diagram. However, by using such diagrams we can easily indicate

Reactions of alkanes and reforming of naphtha

585

the energy of the system at different distances of a molecule from the surface, regardless
of how it gets in the given state.
The extent of multiple exchange on different metals can be quantitatively characterized by parameter M (see chapter 10), defined as
m

M - E
i=l

where di stands for concentrations (in %) of the product containing i deuterium atoms. For
methane, m equals 4. While palladium and platinum show a low extent of multiple
exchange with methane, ruthenium, cobalt, nickel and rhodium give much more. One can
conclude that the first two metals form multiple bonds such as

/~

CH

CH2

/1\

Scheme II

II

reluctantly, but the latter metals do it easily. When we consider a C-C bond splitting in its
simplest possible form, viz.

c-

-~

-~

I
\
/
-X-

/
C

+
\

/
~ ~

Scheme III

II

/
C

\
"4

We have to assume that it can occur either by radical-forming, with an activation energy
of 340 kJ/mol or higher, or when metals help (as indicated in scheme III) by forming
multiple bonds dissociation can presumably occur with a lower activation energy.
Experimentally found activation energies of C-C hydrogenolysis are always lower than
about 200 kJ/mol, so that a mechanism using metal-carbon multiple bonds is very likely.
Therefore, Kemball suggested and verified [3c] a sympathetic correlation between the
propensities for multiple exchange on the one hand and hydrogenolysis on the other.
Methane is a very good diagnostic molecule for establishing the multiple bond
formation, but it is not ideal in all respects. The multiple bonds between the molecule and
the metal surface can be formed

only to one and the same carbon atom. For example a

higher alkane such as ethane can form multiple bonds and induce multiple exchange by
forming offS-bound species (see chapter 10). Therefore, it was necessary to check whether

586

chapter 13

Kemball's correlation between multiple methane exchange and hydrogenolysis of a higher


alkane also holds when multiple exchange running through the ac~ species is established
with the same molecule for which the hydrogenolytic results are considered. A molecule
which would allow such correlation to be studied appeared to be cyclopentane. Before we
discuss the results, a few words must be said about the exchange reaction of this molecule.
Cyclopentane exchange gives as initial products CsH9D by reversal of its adsorption
as a cyclopentyl radical, and molecules having from two to five deuterium atoms formed
by multiple 0tl3...c~...0~13 exchange on one side of the molecule. Somehow or other [4,5] the
molecule may 'roll over' so that hydrogen atoms on the other side can exchange; CsD10
will then be the major product.
A temperature may be chosen such that the a13 mechanism gives mainly CsHsD5
while the ~o~-adsorbed species are still present, giving CsH8D 2. Thus the low ratio ds/d 2
under these conditions indicates whether a metal readily forms the ota species: the results
of this study [5b] are shown in figure 2, which compares this ratio with activity at a high
temperature for cyclopentane hydrogenolysis.

figure 2
The relation of the activity in
cyclopentane hydrogenolysis with the
propensity to form multiple bonds.
Activity is characterized by the
temperature region (bars) at which
the conversion increases from the
first traces to 3% while the ds/d2
ratio is measured at the conversion
in exchange of o~=15%.

573

/
/

T(K}

PV
/

/..73

./I/RI h
/

~_ _. __ ~
373 -

"~ -'Ni

Co

Ir

Ru
273

dS/d 2

Obviously there is a sympathetic correlation between the hydrogenolytic activity and the
propensity to form metal-carbon multiple bonds. Hydrogenolytic activity has also been
plotted as a function of the multiple exchange parameter M, established with methanedeuterium exchange on the same metals. The results are shown in figure 3.

Reactions of alkanes and reforming of naphtha

587

673
Pd

T(K)
573 -

.~.... . . . .

-.\
\

473 -

\ \ ~Rh

373 -

273
0

N
1523 K)

figure 3
The relation of the activity in cyclopentane
hydrogenolysis with the parameter M, characterizing the formation of multiply bound species of methane. Multiplicity M of the methane/D e exchange reaction is determined at 523K
and 10% overall exchange of methane. Activity
in cyclopentane hydrogenolysis, as in figure 2.
[5b1.

We can now quite safely conclude that the hydrogenolytic activity and the propensity in
mutiple bond formation (c~a) follow - as suggested by Kemball [3] - the same sequence:
Ni, Co, Ru > Ir > Rh >> Pt, Pd

(2)

It is quite unfortunate that, up to now, there is no theoretical explanation available of this


order in activities. The smaller particles of all metals are less efficient in forming metalcarbon multiple bonds, and are also (except platinum) less active in hydrogenolysis
(5a,b,c]. However they can survive selfpoisoning and keep their activity longer. This has
to be kept in mind in the discussion of the results obtained with e.g. ruthenium-copper
alloys (see below).
(iv) Higher alkanes can induce multiple exchange by c~g-bound species. There was
some doubt [1-3] whether with these molecules c~-bound species are of any importance
at all. However, there seems to be no reason for such doubts. Figure 4 shows at its left
side the product distribution of ethane/D 2 exchange on Pt(II) complexes. The high
contribution by d 3- and d6-products indicates that with the mono-nuclear Pt(II) complexes
and ac~ multiple bond is indeed possible. On the right side of figure 4 is the product
distribution obtained with a platinum film [2,3]. We can immediately see that the
distribution also shows a high d 3 peak. It is impossible to explain it by one single
exchange mechanism yielding dl-, d 2- and d3-products as well as the d6-product. Thus two
or more different multiple exchange c~,13 mechanisms had to be postulated. Alternatively,
one can assume that, next to the ag multiple exchange leading to d6-product, there is a
parallel multiple c~a exchange. The second possibility seems to be more likely. Kemball et
al. [1-5] have established that ag adsorption of alkanes is not only easier than the c~a
adsorption, but also easier than the ct7 and other similar adsorptions (see also chapter 10).

588

chapter 13

Pt 2

%d i

Pt-metol

100~

134~

40

20-

I 2

I 2

Z. 5 6

D-exchonged

figure 4
Initial product distribution of the exchange of ethane with deuterium (platinum) and with

D20 (Pte+) [2,3,5c,d].

(v) When interpreting the results obtained by exchange reactions at higher

temperatures or when planning such experiments one has to be aware of the following
complication. The large difference in the reactivities of the C-H and the C-C bonds means
that the results of exchange reactions can be analysed in a straightforward manner usually
only in the region of temperature where C-C bonds do not react. When the temperature is
raised to induce skeletal reactions, the molecule undergoes many consecutive reactions on
the C-H bonds before a reaction on one of the C-C bond is accomplished. By this
circumstance, it is generally impossible to derive from the results on exchange reactions
information on the binding to the surface of intermediates in simultaneous reactions. In
some favourable cases this is easily possible and very valuable information can be then
obtained. For example, in the exchange reaction of adamantane there is a simultaneous
isomerizarion [12] and since the only product of exchange was that with one deuterium
atom it was concluded that skeletal isomerization must be in principle possible from the
state in which the intermediate species is bound by a single M-C bond to the surface.
They suggest, for example, for isomerization of neopentane (2,2-dimethylpropane) an
intermediate shown in figure 5; the reactive intermediate here has the form of a pseudo
metal-bound carbenium ion. This approach [12] is actually a combination of the exchange
reaction with the use of 'archetype' or 'diagnostic' molecules, an approach which is
treated in the next section.

589

Reactions of alkanes and reforming of naphtha

/
CH

i 2

CH

CH
," ,3
,," ' , , / C H

C'---CH ~

CH

~M~

C"

/CH

,i,
~M~

"'"D CH CH

CH 3

I\ c

H3

~M~

figure 5
Pseudo-carbenium-ion mechanism of 3C-skeletal isomerization [12] schematically.

Diagnostic or archetype molecules


Another possible approach in the identification of reactive intermediates is to use
the so-called diagnostic or archetype molecules. J.R.Anderson called them 'archetype'
molecules, because their structure makes them suited to form just one reactive intermediate. This approach can be most easily explained with the molecule neohexane (2,2dimethylbutane).
i

ct5'

f C
C

I
C--- C
C 4 ~C ~"

C4

C--_C

~C
I

\
c--dc~c
c~c
'

[somerization :
2.3-dimethylbutane
2- methylpentane

3-methylpentane

Hydrogenolys~s:
C1 . 2-methyl butane

CI .2-methylbutane

C1. neopentane

C2.2-methylpropane

figure 6
Adsorption modes of neohexane and the products arising from these modes.
C1 = methane; C2 = ethane.

Figure 6 shows two o~7-bonded and one o~6-bonded species, with a list of products
underneath which can be associated with each of these intermediates. When the reaction is
monitored at a low conversion and the system does not show too much consecutive

590

chapter 13

reaction, one can derive from the product distribution which intermediates are the most
favoured on the surface of the metal or alloy studied. It appeared from studies using
neohexane that only platinum and palladium favour the aT modes at low temperature, and
they also show appreciable isomerization. All other metals show a strong preference for
the o~g-adsorption mode, and hydrogenolysis [13,14]. The propensity to o~7 adsorption and
isomerization follows the order: Pt>__Pd>Ir>Ni > other metals. The order is somewhat
similar to (but the reverse) of that seen in the foregoing section, concerning multiple bond
formation.
The role of consecutive reactions increases as Pt_<Pd<Ir<Ni< other metals. The
consequences of consecutive reactions are demonstrated by figure 7, showing results
obtained with iridium:

%
CONC.

CONC.
5O

615K

639K

662K

-10

615K

662 K

639K

30-

_-6

%
CONC.
50-

IT

-2

10-

CONC.
622 K

C1C2iB 2MP
N P 3MP

639K

661K

-10

622K

639K

661K

Prop iP nHX CHX


Bu nP McP

figure 7
Product distributions obtained with reactions of neohexane on iridium catalysts; small
(upper part) and large (lower part) particles are compared Cl-methane, C2-ethane, iBisobutane, NP-neopentane, 2MP-2methylpentane, 3MP-3methylpentane, prop-propane, iPisopentane, nHX-n-hexane, CHx-cyclohexane.

Small particles (the upper part of figure 7) which are known to be less influenced than
large particles (lower part) by carbonaceous layers formed during the reaction (chapter 6),
show even at the lowest temperature a lower molar concentration of neopentane than of
methane. It means that the ~B-mode, which induces the production of the same amounts of

Reactions of alkanes and reforming of naphtha

591

neopentane and methane, leads simultaneously to consecutive reactions of the still


adsorbed isopentyl species. This and other similar consecutive reactions cannot be
suppressed by shortening the contact time, since the consecutive reactions occur during
one sojourn of the reactant on the surface. Large particles, covered by carbonaceous layers
to a higher extent, show more of isomerization and adsorption by the ~y-modes. Obviously, selfpoisoning favours a small extent of consecutive reactions and stimulates formation
of o~y'-mode [15]. It seems that three phenomena are sympathetically correlated: (i) the
activity in the formation of the multiple metal-carbon bonds (see chapter 5) (ii) the activity
in hydrogenolysis of alkanes, and (iii) the extent of consecutive catalytic reactions in the
adsorbed state.
The influence of several consecutive steps occuring upon one adsorption sojourn on
the surface on the product distribution must not be underestimated. We meet this problem
not only with exchange reactions and with reactions of diagnostic molecules discussed
here, but also in reactions with 13C or 14C labelled molecules.
Another example of the use of diagnostic molecules concerns molecules which can
form carbon-metal multiple bonds but which due to branching in the molecule cannot form
a system of two or three conjugated C=C bonds. An example of such a molecule is 2,2dimethylpentane: this molecule does not undergo cyclization at moderate hydrogen
pressures although it should if forming of metal-C multiple bonds were a sufficient
condition for it [16,17]. The conclusion therefore is that formation of a conjugated
multiply-unsaturated structure (which cannot be formed here) is a pre-requisite for
cyclization.
Zimmer et al. [18] reported that, when the hydrogen pressure is decreased and the
surface coverage by hydrogen is thereby suppressed, new routes (probably through
multiple consecutive reactions in the adsorbed state) become available which lead to an
extended cyclization of 2,2-dimethylpentane and similar molecules. It seems that, when |
is low, the metal-carbon multiple bonds are formed more easily and they are also more
favoured at equilibrium. At higher pressures of hydrogen, i.e. higher OH' s, the temperature
must be increased if such multiple bonds are to be formed, and the prevailing intermediates contain carbon-carbon double bonds, i.e. polyenic species.
While neohexane can be used to establish whether a given metal or alloy is able to
form t~B-, t~y- or t~y'-bonds, the 2,2,3,3-tetramethylbutane molecule can be used to test the
reactivity of adsorbed species for internal C-C bond fission. With Pt/AI203 the propensity
for this splitting is high as witnessed by high concentration of isobutane in the products
[ 19]. This could be an indication of the occurence of t~8- or t~t~88- or similar bonding.
The use of the following molecules belongs to the same group of experiments. For
example, from 2,2,4-trimethylpentane, various xylenes can be formed. However, this is
only possible after 1,1,3-trimethyl-cyclopentane has been formed. Since the mechanism of
Finnlayson et al. [16,17] cannot operate because of the quartenary carbon-atom, we can

592

chapter 13

say that the intermediate for aromatization is a 5C-ring, which must be 'standing' and not
'lying'. It is most likely that the intermediate on the way from 1,1,3-trimethylcyclopentane
to xylenes is a 3C-ring comprising the carbon bearing the two methyl groups. Obviously,
by using a proper iso-octane molecule, the existence of two intermediates can be proven
[20]. The fastest way to make benzene from an alkane such as hexane is by a sequence of
dehydrogenations, viz. hexane-hexene-hexadiene-hexatriene-benzene. This has been proven
by using ~4C labelled alkanes and alkadiene [21-26]. This way is so much preferred at high
temperatures that some aromatics are formed from substituted cyclopentanes by the
sequence: Cs-ring opening - dehydrogenation - C6-ring closure.
When

platinum

and palladium

alloys

are investigated

as y-A1203-supported

catalysts, it is sometimes more difficult to gain clear information on intermediates


adsorbed on the metal, because of the occurrence of simultaneous reactions on acidic
centres [20,21,27], leading to the same products.
13.1.2 Kinetics of skeletal reactions
It is mainly hydrogenolysis which has been studied by analyzing kinetics and for
which investigators have tried to relate the parameters of the kinetic equations to the
composition and the structure of the intermediates participating in the rate-determining
step of the reaction. However, the same principles can be applied to derive equations for
e.g. isomerization, and in some cases this has been done, too [28].
The first kinetic equation rationally derived for hydrogenolysis of ethane is
probably that by Cimino et al. [29]. It was assumed that ethane is adsorbed dissociatively
and this adsorption is in equilibrium with the gas phase ethane when the state of C2Hx has
been achieved. This is expressed by the equations:
C2H 6 = C2H x + a H 2

(3)

a- 6-x

(4)

The following and rate-determining step is then the C-C bond rupture involving gaseous or
weakly adsorbed di-hydrogen:
The rest of the reactions leading to C H 4 formation are fast. The equation:
k reac

C2Hx + H 2

. . . . . . ) CHy + CH z

(5)

Reactions of alkanes and reforming of naphtha

593

(6)

k~Pc~n6.( 1 -O q n ) = kr,,~, "Oc~vl,P ~


can be rewritten as

oc: m =

=A

with n_<l. The last expression would lead to


-- /

(1-via)

(8)

rate = kreacOCzHx.PH2 = g Pc2n~.PiI~


which is easily verified [28,29]. It appeared that the exponent m which is given by:

m = 1 -na = 1 - 6 n + 3x
2

(9)

increases with increasing temperature. However n is almost independent of temperature


and that would mean that x should increase with temperature: x indicates the content of
hydrogen in the adsorbed species, but this should probably rather decrease than increase as
the temperature is raised. There are more logical contradictions in this treatment, and
therefore, Sinfelt and Taylor [30] suggested a different sequence of steps:
k1
C2H 6 ~

C2H 5 + Haas (equil)

(10)

k2

K2
CzH 5 + Had s

-----> CzH x + ~ H 2

CzH x + H 2

------) CH 2 + CH x

(11)

A steady state is then assumed for |

(12)
and, with a = 2 and b = ki/k K 2, the final

equation was found in the form:

rate= klPc2a~ ~ pc2n6.p~


1 +bpH 2

(13)

594

chapter 13

The above treatment is not without problems and modifications of it followed later [31].
These equations could explain the negative order in hydrogen pressure found in a certain
temperature region

(m < 0), but failed to explain why with some metals (e.g. iron and

rhenium) m exceeded unity. Further it remained unexplained why, in the formal power
rate law r = k pH2m pHcn, the parameter m is larger than unity, for

CnH2n+2with n greater

than 2.
Some workers therefore proposed that reactive desorption, in which hydrogen
participates, is the rate-determining step [32]. However, there are also suggestions that the
rate of adsorption is the rate-determining step [33].
Let us take now two examples to show how the variations with temperature of the
hydrocarbon and hydrogen formal orders can be rationalized by full kinetic equations
derived with the assumption that the C-C bond rupture is the rate-determining step.
(i) It is assumed that the hydrocarbon fragment involved in the C-C bond rupture retains at
different temperatures and pressures, or with different hydrocarbons, a different number of
H atoms: a low number at low hydrogen pressure and high temperature and vice versa
[34]. In what follows, two situations are analyzed: the C-C bond rupture occurs from (a)
CzH 5

and (b) from C2H2. The adsorption site is indicated by Z, all species are assumed to

be bound to a single site. a) C2H 5 Z + Z - C H 3 Z + C H 2 Z; b) CzH 2 Z + Z - 2CH Z


It is assumed that, with an excess of adsorbed hydrogen on the surface, the rate r of
ethane hydrogenolysis is in case a):

r= klPc:~ Pv~
(Pe2H6+k2PH2)2

(14)

For case b), with C2H 2 undergoing the C-C bond rupture:

r=

k3Pc2HeP~2
--

(Pc2H~+lr

(15)

2.5.2

At a constant hydrogen pressure, the rate can show a maximum as a function of the
hydrocarbon pressure (figure 8). At a constant hydrocarbon pressure and in the higher
range of hydrogen pressures a negative order (in the approximation of the power rate law)
in hydrogen is predicted. Moreover, for case b) a region of hydrogen pressure is predicted
to exist, in which the rate of reaction increases with increasing hydrogen pressure, that is,
hydrogen has here in the approximate power rate law a positive order.
In principle the idea of shifting the rate-determining step amongst the various
elementary steps of the surface reactions, i.e. going for example, from C2H5 to C2H2 as the
key intermediate, can explain the change in the hydrogen order when going from, for
instance, ethane to heptane. One has just to assume then that the H/C ratio on the C-C
group undergoing fission is lower with heptane than with ethane, which is not unreasona-

Reactions of alkanes and reforming of naphtha

595

ble.
~Qtel

3.010f

figure 8
Hydrogenolysis of ethane at constant hydrogen
pressure (1.2 kPa); the rate as a function of
ethane pressure. Catalyst: 1.1% wt Pt, 0.7% wt
Fe, on silica. Curve 1)is plotted according to

;,00~.

eq. 14, curve 2 according to eq. 15. [34].

0,002

ii) The second approach represented at the time of its publication [35] a substantial
innovation in the routine of writing down the kinetic equations according the procedure
worked out by Hougen and Watson and others (see chapter 6) a long time ago. In this
approach [35] it is properly acknowledged that there are always side reactions accompanying the target hydrocarbon reaction, viz. reactions of deposition and removal of the
carbonaceous layer. The latter reaction leads to a steady state coverage of the surface by
carbonaceous deposits and this is a function of the hydrogen pressure and temperature. In
other words, the approximate power rate law should be written as:
r =kNwork P~cP~

(16)

where Nwork stands for the number of sites on the working surface uncovered by the
carbonaceous layer. It can reasonably be assumed that:
r

N~o~k = const.pn 2

(17)

When the mechanism of the reaction leads to equations such as 14 or 15, which can be
approximated by 16, with m less than zero, the total order in hydrogen pressure, r + m,
can be positive.
This short treatment probably suffices to illustrate the problems one has with using
and determining kinetics. For any practical application, knowledge of the kinetics is
essential, since it forms the basis of reactor and process design. Further, mechanisms
suggested on other than a kinetic basis should be checked by formulating and verifying the
kinetic equation. However conclusions from kinetics on mechanism and key intermediates
should always be made cautiously. Since very different mechanisms can lead to very
similar equations, kinetics does not supply in this respect more than a hint, which
nevertheless, can be very valuable (see also chapter 6).

596

chapter 13

13.1.3 Model reactions of alkanes on metal catalysts


With regard to skeletal reactions of hydrocarbons, let us now consider the segment
of the periodic table of elements which is most relevant for metal catalysis. We can
conclude on basis of the literature, with some extrapolation, that the metals of the Groups
3-6 show strong tendencies to be converted into stable hydrides and carbides upon contact
with hydrogen and hydrocarbons respectively. The non-stoichiometric carbides can show
sometimes interesting catalytic properties [36-38], but we do not intend to discuss this
matter in detail. Formation of hydrides and carbides is suppressed in many alloys. Not
much is known about Group 7, but the properties of elements in Groups 8-12 are known
quite well. Since Groups 11 and 12 show a very low or zero activity in skeletal reactions
of hydrocarbons, we focus just on the iron triad and platinum group metals (table 1).
table 1

10

Fe

Co

Ni

Ru

Rh

Pd

Os

Ir

Pt
(VIII)

All these metals are active in hydrogenolysis, this reaction always being strongly prevailing, except with palladium and platinum. However, as we shall see below, the manner of
hydrogenolysis is not the same on all metals. Platinum in a pure state as a film [39] or
mounted on an inert carrier such as carbon or silica [40] can induce isomerization and

dehydrocyclization and at higher temperatures also aromatization providing the molecule


contains sufficient carbon atoms for these processes to occur. Palladium deactivates very
quickly but it shows a high selectivity to dehydrocyclization [41]. A comparison of metals
is illustrated by two tables. Selectivity parameter S(i~ indicates how much of the reactant is
converted in each of the three groups of reactions: isomerisation, cyclisation and hydrogenolysis (cracking).
table 2

Selectivity of supported metals in n-hexane reactions [42,43]


Metal T(K)

&so._~__% S_Scycl_~__S,:rack

type of cracking

Pd

678

17.5

42.7

39.8

terminal

Pt
Fe

568
523

Ni

523

0
0
0

0
0
0

100
100
100

multiple
multiple
multiple

Reactions of alkanes and reforming of naphtha

table 3

597

Selectivity of supported metals in heptane reactions [43,44]


metal T(K)

SScracl~%

Pd

573

Pt

548

Rh

386

Ru

361

Ir

398

table 4

Sis~

S_Scycl

90.5

6.2

3.1

37.4

46.8

15.8

93.0

7.0

92.5

7.5

87

13

Product distribution in n-pentane reactions with

H2

% molar
Catalyst

T,K

]~.I-t2

~-pent C1

Pt/SiO 2
(16wt%)

585

0.9

0.1

619

0.9

0.1

Ni/SiO 2

623

2.5

( 19wt %)

623
623

c_c_.2

17

15

20

18

0.5

85.9

6.6

4.7

8.1

5.0

0.5

77.0

8.5

8.3

5.6

5.0

2.0

52.0

0.5

3.0

4.4

i-Cs

c-Cs

S_aso

52

67

43

59

~ycl

iso-C 5 = isopentane; c-C5 = cyclopentane;

Siso + Scycl-~- Scracking = 1 (cracking is hydrogenolysis)


Experiments performed in an open flow apparatus under the conditions indicated.

Hydrogenolysis

For several reactions the order of activities in hydrogenolysis of the Group 8-10
metals has been already established. For hydrogenolysis of cyclopentane under mild
conditions and minimal selfpoisoning [45]:
Ru > Ni,Co > Ir > Rh >> Pt,Pd
For ethane hydrogenolysis according to Sinfelt [46]:
Os>Ru>Ni>Rh>Ir>Re>Co>Fe>Cu>Pt,
or according to Sarkany and Tetenyi [47], also for ethane,
Ru > Rh > Ir > Co > Ni > P d > Pt

Pd

(T=478K)
(T = 525K)

The position of some metals is quite clear: ruthenium always stands among the metals of
greatest activity and platinum and palladium are always amongst the least active. The

598

chapter 13

metals nickel, cobalt, rhodium and iridium do not differ very much in their activity, which
is always quite high, and a small variation in the measuring conditions, leading to
variations in selfpoisoning, can easily change the order of activity and cause discrepancies
between various groups of workers.
Hydrogenolysis can proceed in several manners,

as we shall schematically

demonstrate by the example of n-hexane:


a) terminal fission
CCCCCC ~ CCCCC + C ~ CCCC + 2C ~ CCC + 3C etc.
b) internal fission

CCCCCC ~ 2CCC
CCCCCC ~ CCCC + CC

c) random fission, when the probabilities of splitting at any position are equal
d) multiple fission

CCCCCC ~ 6C; CCCCCC ---->2CC + 2C, etc.

To characterize the fission quantitatively, several parameters have been suggested. The
multiple fission parameter M e appeared to be very useful and sensitive [48]"
n-1

(n-i)c,

Mr---

(18)

i=2

C1

wherein n is the number of carbon atoms in the initial molecule, and

Ci

the concentration

of a molecule with i carbon atoms. Upon terminal fission each step releases just one
molecule of methane (C1) so that Mf should be one. If the fission is multiple, M e is less
than 1 and if internal splitting prevails, M e exceeds 1. Hungarian workers who have
produced a great wealth of information on hydrocarbon reactions prefer to use the socalled depth of hydrogenolysis, characterized by the fragmentation parameter ~ [49]:
n-1

~=

(19)

n-1

~ i Cin1

Montarnal and Martino have characterized terminal splitting in relation to internal and
multiple fissure, by using the parameter Pf [50]"
C1

(20)

c_:
which increases when multiple splitting is more important and decreases when the internal
fission becomes prevailing. Whatever parameter is used, the conclusions are usually
similar. The application of M e leads to the conclusion that for example nickel, cobalt and

Reactions of alkanes and reforming of naphtha

599

ruthenium show a very high propensity for multiple splitting and platinum the lowest.
Similarly, those who use the parameter ~ derived from their results that ruthenium,
osmium and nickel prefer deep hydrogenolysis, while the opposite is true for platinum,
palladium, iridium and rhodium. Exactly the same is found for multiple fission when using
parameter Pf: it increases in the order platinum, ruthenium, nickel, cobalt.
It has already been indicated elsewhere that the o~13 adsorption mode can be
considered as a typical intermediate of fast hydrogenolysis. However, the reactions of
neohexane reveal that ~7 adsorption can also induce hydrogenolysis. Where possible, this
reaction will probably use several surface atoms as the active site(s), but with platinum it
is also likely that metathesis-like intermediates (metallo-cyclobutane) operate. Metallocyclobutane is a well-established species in organometallic chemistry; the situation is
summarized in figure 9.

\c_c /

ot[3.

/I

\/

I x.

c
Jl
~

\/

figure 9
Hydrogenolysis induced by o~fi
or c~7 adsorption modes, respec-

c
II
N

tively.

ct'y

The

~-bound

species

can be adsorbed on sites with

C
\/(~
/ \ / ~,

N
/\
/CiC

\/
+ C

more than one metal atom


(see scheme 3, above).

II
N

I~

\/

/
\
N

--

-~

-~

The reactivity in hydrogenolytic splitting of different positions in hydrocarbon


molecules has been systematically studied by Leclercq et al. [51], with Pt-A1203 catalysts.
They defined a reactivity parameter:
60

observed rate of reaction of a particular bond


rate expected, statistical random rupture

The statistical rate is obtained as the total rate of hydrogenolysis divided by the number of
bonds which can be broken (4 in pentane, 5 in hexane, 6 in methylcyclopentane, etc.) The
results of such evaluation are shown in table 5.

600

chapter 13

table 5
Rates of hydrogenolysis of C-C bonds at 573K (PH2 = 0.9 atm.; P

Hydrocarbons
1 2

C-C-C-C

Broken
bonds
1

Reactivity
factor: co
0.9
1.2

Hydrocarbons
123
C-C-C-C-C-C-C

hydrocarbon ----

Broken
bonds

0.1 atm.)

Reactivity
factor: co
1.1
0.6
1.3

123
C-C-C-C

08
0.5
1.9

C
1123

C-C-C-C
I

0.8
2.2
0.35

C
1 2 3 4

0.85

3
4

0.9
1.2
1.15

C-C-C-C-C

1 2

C-C-C-C-C

1 2

C-C-C-C
I

1.8
0.4
0.6

0.7
1.65

C
112 3 4
C-C-C-C-C
I

1.45

C
C
112 3 4
C-C-C-C-C
I

C
112 3

C-C-C-C

CC

0.25
3.7
0.09

I I

0.35
3.8
0.09
1.05

0.7
2.7
0.6

CC
1 2

C-C-C-C-C
I

1.15
0.45

CC
112 I

0.45
4.3

C-C-C-C
I I

CC
1234

C-C-C-C-C
I I

CC

1
2
3
4
5

0.95
0.8
0.85
1.75
0.7

1 2

C-C-C-C-C
III

CCC
3

1.1
0.5
1.6

Reactions of alkanes and reforming of naphtha

601

We can see from the results obtained with 2,2,3,3-tetramethylbutane that with platinum the
~x8 adsorption mode can also induce hydrogenolysis. When the aB, o~y, ~8 modes of
binding are all possible, we probably cannot exclude formation of the 1,5 or 1,6 species
also. The latter ones could be some of those intermediates which can lead to dehydrocyclization and aromatization. Establishing similar reactivity tables, such as table 3, is
more difficult with other metals because of the larger extent of multiple (in the adsorbed
state) and consecutive (via the gas phase) reactions.
We have learned above several very well-established facts. (i) There is a correlation of activity in hydrogenolysis with the propensity to form multiple bonds. (ii) With all
species so far postulated as necessary intermediates of hydrogenolysis we expect formation
of multiple bonds. Therefore, one can expect that metals which, under the reaction
conditions, form species with a lower H/C ratio would be more active in hydrogenolysis
than the metals which give a higher H/C ratio. This can be documented by two references
from the literature [52,53]; table 6 shows the H/C ratio of species formed from ethene
[53]. The lower index indicates the temperature at which the result shown is obtained.
H/C ratio of species formed from ethene

table 6
cat.

(I--I/C)a28

(I--I/C)483

Co

0.82

0.33

Ni

1.12

0.25

Rh

1.2

1.34

Pd

2.1

1.83

Pt

2.5

2.05

The higher H/C ratios correlate with the lower hydrogenolytic activity (see above). The
other very similar example is from cyclopropane hydrogenation. Cyclopropane reacts at
quite low temperatures according to two reactions:
c - C 3 H 6 -I- H 2 ----) C3H 8 (hydrogenation)

(21)

c-C3H 6 -!- 2 H 2 ---> C2H 6 + C H 4 (hydrogenolysis)

(22)

Metals have been found to be more selective for propane (reaction 22), the higher the H/C
ratio in the adsorbed layer formed by adsorption of cyclopropane at low pressures on
metal films at room temperature [52]. Hydrogenolysis of the C-C bond requires vice versa
a low H/C ratio. Notice that due to the electronic structure of cyclopropane, reaction (22)

602

chapter 13

reminds us of the hydrogenation of alkenes more than the C-C bond splitting in saturated
molecules (see also chapter 11).

Dehydrocyclization
There is a large variety of intermediates to be considered here: di<y-adsorbed 5C
intermediates (i.e. 1,5-adsorbed)bonded to one or more atoms, multiply-bonded carbenic
and carbynic 5C-intermediates and rt-complexed intermediates. Also a rt-complexed, 8bonded species has been suggested [5,54]. This is all summarized by figure 10.
Di-(~-adsorbed

5C-intermediates

c/C--/c
~

c/Cc/c

I_cS

/ /c /

(o)
Carbenic

and

5C-intermediates

C--C

/Cx
C
C

C --C

C--C

C--C

Olefinic

5C-intermediates

II

II

(d)

/C~
C
C

111

ill
Nm~

~mN

(e}

(schematically)

/C

/C----,C\

I
.

II

/C
C--C

l
N

(c}

(b)

(o)

(schematically)

/Cx
C
C

/C~
c
c

/Cx
C
C
C~C

carbynic

(b)

c<-.. c

C--C

--,\

C\c /

,,

~" " ~ c c~C~


/,- .,\
C'--

figure 10
Possible cyclic intermediates of isomerization. The same intermediates operate in dehydrocyclization and aromatization.

None of these species has been ever directly seen by any spectroscopy under reaction
conditions, but for each of them there are some, more or less strong, reasons to assume

Reactions of alkanes and reforming of naphtha

603

their existence. With metals in their pure state unmodified by e.g. sulfur or other modifier,
operating as monofunctional catalyst, the formation of intermediates such as these depicted
in figure 10 can to an appreciable extent be expected only with platinum and palladium.
However, small amounts of cyclic and aromatic hydrocarbons are formed from alkanes
with other metals of Groups 8-10.
We have already mentioned in this chapter that 1,6-aromatization can also occur by
successive dehydrogenation accompanied probably by rt-complexing of intermediates; this
holds in particular at higher temperatures. It seems that diminishing of the particle size of
platinum stimulates this way of aromatization or the reverse way of ring opening.

Isomerization
This group of reactions can proceed through two classes of intermediates.
(i)
Intermediates involving three carbon atoms (3C-intermediates) seen clearly in
isomerization via (xT'-intermediates with neohexane, and in isomerization of butane.
These two molecules cannot form any other intermediate leading to isomerization.
(ii)
5C-intermediates involving five carbon atoms of hydrocarbons with six or more
atoms, elegantly proven by 13C labelling by Gault et al. (see chapter 1)
With these 5C-intermediates, isomerization actually takes the form of a two- step reaction:
ring closure, followed by ring opening; the latter occurring at a different place on the ring
from that at which the closure took place. The 3C-intermediates suggested in the literature
are collected in figure 11 [5,39,55]. The 5C-intermediates are shown above in figure 10.

3C-Intermediates

/c2
C1

/c\
J'~C 3

II ~ "
1o)

of Isomerizotion

\ ,'cH3

C~C

/
(b)

(c)

/c

"C"

/
(d)

figure 11
The 3C-intermediates of isomerization (notice, not all bonds but only the structure is
indicated).
With the first three intermediates in figure 11, isomerization is again a 'ring closure - ring
opening' reaction but only three carbon atoms are involved. The last of the intermediates
shown (on the right side) has been suggested by Gault et al. [54d,56]; for isomerization it
should split into a carbene species and a rt-complexed alkene, which should then rotate,
and after the rotation the molecule should be reconstructed again. However, some doubts
arise about the last two steps [57].

604

chapter 13

13.1.4 Reactions on supported metal catalysts


Introduction of platinum on alumina catalysts in naphtha reforming revolutionized
the production of high quality gasoline. These catalysts dominated the scene for about
twenty years after their introduction [58] and, because of their enormous commercial
success, the amount of fundamental knowledge concerning their functioning is respectable.
Even in the early stages of research and use of these catalysts, it was suggested [58-60]
that the prevailing mechanism is as follows. An alkane is dehydrogenated on the metal
(this is called the metallic function of the catalyst) and alkenes migrate or are transferred
via the gas phase to the acidic centres on alumina (see chapters 6 and 7). There carbenium
ions are formed which induce skeletal rearrangements. Rearranged alkenes are then backhydrogenated on the metal. It appeared later that some isomerization and aromatization
also occurs on the metallic particles [60,61], but the prevailing mechanism with industrial
catalysts is still assumed to be the bifunctional catalysis as described above. The basic
facts concerning the acidic mechanism are the following. Alkenes are converted on
Br6nsted acid sites (OH groups) into secondary carbenium ions, for example:
CH2 = C H - C H 2 - C H 2 - C H 2 - C H 3 + H + =~
CH

3 -

+CH - CH 2 - CH 2 - CH 2 -

CH

3 ~

CH

3 -

(23)

CH2 - +CH - CH 2 - CH

- CH

The bond in the 13 position to the positive charge is activated and can be broken, whereby
an alkene and a shorter carbenium ion are formed; this constitutes acidic cracking.
Carbenium ions can also accept the arising fragment (alkene formed) and this leads to a
more stable tertiary carbenium ion. By hydride transfer from another molecule or by
splitting off of a proton, an isomerized or a shortened alkene is formed. The overall
scheme showing schematically reactions of n-hexane on a bifunctional catalyst is presented
in figure 12.

13.2

Model

reactions

of alkanes

on alloys

13.2.1 Nickel, cobalt and iron alloys with Group 11 elements.


Nickel-copper alloys, although of limited practical importance, are very useful for
fundamental research. Their electronic structures and surface compositions are quite well
known, and alloys well defined by XRD can be easily prepared without any support. The
latter aspect was important for hydrocarbon reactions with which one had always consider
the reactions on the support, i.e. the possibility of bifunctional catalysis. When, for
example, alloying of palladium with gold (0.6%wt Pd + 0.1; 0.6; 1.2%wt Au) causes in

Reactions of alkanes and reforming of naphtha

605

reactions of heptane (Hz/heptane ratio of 5.1; 4.6 at 454~ a shift from hydrogenolysis to
isomerization [61], one may ask whether this is because of alloying alone, all reactions
observed occurring on the metallic phase, or whether alloying suppressed hydrogenolysis
by palladium, which could still produce enough alkenes to sustain isomerization on the
support. Such problems were however not present when the study was made with unsupported nickel-copper alloys such as evaporated films [62,63].

n-HEXANE

CYCLOHEXANE

METHYLCYCLOPENTANE

1L

..,

n-HEXENE

ISOHEXANES

ISOHEXENES

1L

CYCLOHEXENE

CYCLOHEXADIENE

METHYLCYCLOPENTENE

ME T H Y L C Y C L O P E N T A D I E N E

BENZENE

figure 12

REACTIONS

ON

ACIDIC

CENTERS.

AFTER

MIGRATION

Reactions of n-hexane on a bifunctional catalyst (schematically, the stoichiometry is not


indicated). Horizontally: reactions on acidic centers. Vertically: reactions on the metal.

Cyclopentane-deuterium exchange has been studied [62] on nickel and on well


equilibrated nickel-copper alloys. It appeared that the exchange reaction was only modestly
influenced by alloying, but the side reaction of hydrogenolysis, yielding, amongst other
products deuterated methane, was dramatically suppressed. This selectivity effect produced
by alloying was seen as important and therefore a study with other molecules followed
soon. Roberti et al. [63] studied the exchange of methylcyclopentane and obtained results
which fully confirmed those obtained before [62].
Beelen et al. studied cyclopropane reactions [64]. Hydrogenation to propane can be
considered as simple addition of H 2 to an unsaturated bond [52], but production of ethane
and methane clearly requires a hydrogenolytic splitting of a C-C bond. Figure 13 shows
that the effect of alloying on the two types of reactions is very different. The general
character of these results was further confirmed by a study on hexane reactions, with
silica-supported nickel-copper alloys [65]. Results are summarized in figure 14.
In the same period of time, the group of Sinfelt in the EXXON laboratory worked
on exactly the same problems. Their first paper [66] on nickel-copper alloys even appeared

606

chapter 13

in the same issue of J.Catal. as the work described above [62]. The figure which summarized their findings on the selectivity effects of alloying was so convincing that research on
alloys was thereby greatly stimulated throughout the world. The figure in question (15)
compared the variations in the rates of cyclohexane dehydrogenation with those of ethane
hydrogenolysis.
o

H2/xe

Aso[

A.

H21q.d l.

ct

:b

~'o

6b

-1.0

-20

-0.I~

,12

.0.2

,4

8"OO/oCu

20

4o

60

80 % Cu

figure 13
Nickel-copper alloys (unsupported).
a) On the left. ct, the ratio of extents of hydrogen to xenon adsorption, established with
evaporated alloy films, prepared in uhv [62]. fl, the relative occupation of alloy surface,
by hydrogen adsorbed irreversibly at 293K; fl is taken unity for nickel [66]; for nickelcopper powder alloys.
b) On the right. The reactions of cyclopropane and hydrogen; activity in arbitrary units
A H - activity in hydrogenolysis to methane and A s - the total activity in all reactions,
including hydrogenation to propane. The latter is the highly prevailing reaction on alloys.
Support-free powder alloys are used.

While the experimental results obtained with different hydrocarbons and alloy
catalysts of different formulation gave a self-consistent and actually a very well reproducible picture, the explanations suggested by groups concerned were quite varied. One group
[66] believed first that their results could be explained just by considering the expected
variations in the d-character of bonds (see chapter 1 on this parameter) and the effect of
them on the chemisorption bond strength of hydrogen. However, when it later appeared
[67] that various metal pairs, including those which show hardly any miscibility and thus
perturb each other's electronic structure to the least possible extent, behave catalytically in
a very similar way, the original theoretical basis for the explanation was abandoned [68].

Reactions of alkanes and reforming of naphtha

607

On the other hand, another group suggested [62] that the selectivity effects are mainly due
to dilution of the active sites, which are nickel atoms or nickel ensembles, in the virtually
inactive matrix. However it was still believed at that time that there should be some room
for operation of electronic structure effects [65]. Later in the seventies a great volume of
information appeared (see chapters 1, 3 and 8) about the electronic structure of nickelcopper alloys, and it became clear that those features which were suspected [65] as being
a consequence of some electronic structure effect could be explained by dilution effects.

SIMIEL

x----'

%1 /<'~'1

17~
004 ,.4 k-x ~ x

/____ _

-I,
%

_.,.

'. s

-"

~I.
l

I'

- -

"

'

2'o

'

,b

'

A2

6B

80

100%Cu

figure 14
Reaction parameters o f n-hexane conversion by nickel and nickel-copper alloys. Rate per
g c a t a l y s t - r w, rate per cm 2 surface - r~; A 1 - log r w at 603K, A 2 = log r~ at 603K,
activation energy o f the overall reaction Eacp fission parameter M, selectivity parameter S;
Characterizing the extent o f non-destructive reactions (mainly isomerization); all as a
function o f alloy bulk composition (in atomic % copper). Support-free powder alloys used
as catalysts.

Results obtained with n-hexane have also been confirmed by those obtained in
reactions of smaller hydrocarbons such as ethane, propane and butane, studied on NiCu/SiO 2 alloy catalysts [69a]. A simple statistical treatment of ensemble size distribution
and its application to these reactions leads to a required ensemble size for hydrogenolysis
of 12-20 atoms, depending on the reaction (see also chapters 8 and 9). Silver is not
miscible with nickel but it also strongly suppresses hydrogenolysis of n-butane [69b].

608

chapter 13

10 6

Cyclohexane dehydrogenation
.
.
~
-=----=

i
Ethane hydrogenolys*s
10-

1-

L
0

20

40

60

80

100

Copper content (at. %)

figure 15
Activities of nickel-copper catalysts for the hydrogenolysis of ethane to methane and the
dehydrogenation of cyclohexane to benzene [66].

Reactions of cyclopropane have been studied on iron-copper and cobalt-copper


alloys, the results being comparable with those obtained with nickel-copper alloys [70a].
Unlike nickel-copper, the other pairs show only a limited miscibility but, nevertheless the
conclusions for all three systems are similar. (i) At low temperatures nickel can be more
active than alloys, but at higher temperatures the alloys are better since they are obviously
less poisoned by carbonaceous deposits. (ii) Addition of copper to iron, cobalt and nickel
always suppresses the selectivity to hydrogenolysis to methane and ethane. The selectivity
effect of copper on nickel has recently been again confirmed by Cole and Robertson [70b].
We have seen above (section 13.1.2) how kinetic equations can be derived and
related to the mechanism of dissociative adsorption of alkanes. Ollis and Taheri [71]
analyzed in this way the kinetics of ethane and pentane hydrogenolysis on nickel and
nickel-copper alloys, and concluded that on alloys the intermediates participating in the
rate-determining step, which is here assumed to be C-C bond splitting, always contain one

Reactions of alkanes and reforming of naphtha

609

hydrogen atom more than the analogous intermediates on pure nickel. It is as if the
presence of copper in the surface suppressed C-H dissociation and the lower hydrogenolysis activity were the consequence of it [52,53] (see section 13.1.3 on hydrogenolysis).
Langenbeck et al. [72] studied cyclohexane dehydrogenation to benzene on nickel
catalysts. Dehydrogenation is accompanied here by hydrogenolysis, but the latter reaction
could be strongly and selectively suppressed when oxides such as zinc oxide, titanium or
thorium oxides or zinc powder, were added to the nickel supported by silica, alumina or
other oxides. This was possibly the first observation of pronounced selectivity effects of
alloying on reforming reactions. A possible role of alloying was admitted, but finally an
explanation in terms of the theory of Schwab et al. [73] and Solymosi [74] was preferred.
They believed [73,74] that n-type semiconductors when used as supports inject their
electrons into the nickel particles and consequently the selectivity is modified. Having in
mind what has been said in chapters 5 and 8, the explanation in terms of alloying is more
likely ( a parallel SMSI effect, discussed in chapter 6, can also be expected with for
example titanium oxide). In any case, it is surprising how this important paper has been
ignored by all other authors in this field. Perhaps the reason is in the title ('Die Wirkung
von Metalloxyden mit Elektronentiberschussleitung auf die Katalytische Eigenschaften von
Nickel bei der Umwandlung von Cyclohexan'), which did not attract the attention of
people working on selectivity problems and on alloy catalysts. The same topic was studied
again [75] with the knowledge on catalysis by alloys gathered in the intervening time.
According to this study, nickel-copper alloys on alumina show in the reaction of cyclohexane an increase in selectivity to benzene, which remains constant over a broad
composition range. However with nickel-lead and nickel-antimony catalysts, the selectivity
to benzene, measured after two hours on stream, increases as a function of increasing
content of the metal added. However, even the maximum selectivity established in this
way is lower than the initial time-on-stream selectivity of pure nickel. Lowering of the
selectivity to benzene on bimetallic catalysts as compared with nickel is even more
pronounced at 573K. It was suggested [75] that selectivity is positively modified by the
second metal only when the self-poisoning by carbonaceous deposits plays an important
role. This is the case at high temperatures and low hydrogen pressures.
Suppression of hydrogenolysis leading to a higher aromatization selectivity in nhexane reactions at 773K has also been reported for nickel-tin on alumina catalysts [76],
probably due to a combined effect of alloying and of modification of the acidic sites on
alumina surfaces. When alumina-supported catalysts are used at high temperature, the
potential bi-functionality must not be forgotten.
Unsupported nickel-copper powder alloys have been used in a study of neohexane
(2,2 dimethylbutane) reactions. Alloys with about 50% nickel showed a slightly higher
isomerization selectivity than pure nickel (compare with figure 7, for iridium), but this was
not the main effect observed. The most pronounced effect of alloying was on the extent of

610

chapter 13

consecutive reactions (multiple breaking of bonds) occurring during one sojourn of


neohexane on the surface. The primary reaction produces methane and neopentane via the
2C-orB-adsorption mode. At higher temperatures more methane and less neopentane is
observed, which indicates that isopentyl species can react at those temperatures without
leaving the surface, as there is no influence of contact time. The point where consecutive
reactions suppressed neopentane production is at about 500K for pure nickel but about
550K for a Ni65Cu35 alloy. Obviously, copper suppresses the consecutive reactions of the
adsorbed isopentyl species.
A very interesting combination of two active metals appeared to be nickel with
palladium. Moss et al. [77], who worked with well-defined alloy films, took first a
reaction which should not be very sensitive to alloying, namely hydrogenation of ethene.
Then they went over to hydrogenolysis of ethane [78] in which reaction palladium and
nickel differ very much in their activities, even more than nickel and copper (see paragraph 13.2). In the first case the variation of activity with composition was modest, but in
the second case it was more pronounced. The most interesting result was that the rate of
product formation correlated very well with the percentage of nickel atoms in mixed
nickel-palladium ensembles comprising six atoms. With surface composition determined
by AES and by performing the statistical calculation of ensemble distribution by a MonteCarlo technique, the result shown in figure 16 was arrived at. The cobalt-palladium system
is in many respects similar to nickel-palladium. Continuous solutions are formed of a very
active metal with another which is almost inactive in hydrocarbon reactions. Both metals
of these systems are present in the surface of alloys. Results for various reactions are
shown in figure 17 for palladium-cobalt/silica catalysts [79]. In another section below,
results obtained with platinum-cobalt catalysts are discussed and the reader will notice
similarity between the systems. Here we can consider platinum as being the far less active
component of the couple. Variations in the aereal (specific) activities are of several orders
of magnitude, showing a minimum at certain concentrations of cobalt. However, selectivities show the picture one would expect: cobalt is active mainly for hydrogenolysis, but
palladium also in isomerization and dehydrocyclization. When comparing these results
[79] with those for the nickel-palladium system [78] one could come to the conclusion that
mixed ensembles of cobalt and palladium, unlike those of nickel and palladium, are of a

very low activity. The 'activity-composition' function does not show any kind of behaviour (a plateau or a curve with a maximum) which could be ascribed to the operation of
mixed ensembles in the palladium-cobalt system.

Reactions of alkanes and reforming of naphtha

.E
E

611

L.
L_
O

z
O

m
u.l
.__i
rn

X
9

~
rr
O

100

ixl

11

z
u.l
cl

\
\A

~t

-r
(,3
LU

50

OC

if)
A\

~_.

20

40

SURFACE

J.

COMPOSITION,

60

1.,

80
atom

F--

~ ,~."~_ o
Pd

Pd

figure 16
Above: rate of methane formation at 573K
over nickel-palladium alloys replotted against
apparent surface composition (symbols); percentage total surface atoms participating in
mbced ensembles is shown.
On the left: mixed nickel-palladium ensemble
on (111) surface, n = 6 with central nickel
doublet (nickel atoms shaded).

612

chapter 13

~,,.lO -5

100

k '~

-6

Bo

"~

g
60

o0_6
L

4'.-

9;

E
L,.

2 1 0 -7

.~
0

50
otorn */, cobalt

"]100
100

o
c

40

20

0 e
0

50
otom % cobott

100

~"

figure 17
Catalytic performance of lOwt% Pd-Co/Si02. (A) Catalytic activity for the reactions of
hydrogen with neopentane at 515K (v) n-hexane at 553K (o) and MCP at 553K (+). (B)
Selectivity changes - neopentane isomerization (13) and in n-hexane reactions: hydrogenolysis (o), isomerization (o), MCP formation (+), and fragmentation factor ~ (A), defined in
the same way as 9 in section 13.1.3.

13.2.2 Palladium alloys


We have already discussed some alloys of palladium in combination with nickel or
cobalt, where palladium played the role of the component of lower activity. Now we shall
turn our attention to alloys of palladium with a metal which, for the reaction in question,
shows a much lower activity than palladium.
Visser et al. [80] used unsupported palladium-gold alloys for skeletal rearrangements of n-hexane and found that alloying suppresses hydrogenolysis, thereby increasing
selectivity mainly to dehydrocyclization. Karpinski [81] studied butane reactions on alloys
and we can draw the same conclusion from both papers [80,81]: alloying palladium with
an inactive metal selectively suppresses the destructive reaction of hydrogenolysis. These
results are shown in the figures 18 and 19. Isomerization of n-butane and exchange
reactions were studied on palladium-gold films [81]. It is worth noting that isomerization
of n-butane is claimed to occur from adsorbed species which have lost only one hydrogen
upon chemisorption. We know already that such intermediates have been suggested for
isomerization (see chapter 10 and section 13.2). The large decrease observed [81] in rate
of methane exchange is ascribed to changes in the electronic structure of palladium. This
is of course in principle possible in a reaction involving hydrogen isotopes, and it would
not be very surprising, but one also must not forget that surface segregation of gold could
be very pronounced, and responsible for the strong suppression of exchange activity.
Clarke and Taylor [82a] have studied cyclopentane exchange reactions on palladium-gold

Reactions of alkanes and reforming of naphtha

613

alloys, with simultaneous deuterolysis of the cyclopentane ring as a side reaction; they saw
the same strong suppression of the ring fission as seen with nickel-copper alloy films [62].

,....> s

50-

MCP

30-

I 0 --~

J
I
I0

I
20

I
30

I
40

I
50

1
60

At%

Au

figure 18
Selectivity S in non-destructive reactions and the MCP content among products, under
comparable conditions. Pd, conversion ~ = 1.02%, 656K: 39% Au, ~ = 0.7%, 633K." 48%
Au, ~ = 0.6%, 646K.

20
0

l,(

x
q-

IU

20

,u

~
x

/s

ii '

,-

=3

~o

e-

10

t-

x
O

2
z
i

X
i

2o

30

40
9"o~ d

.... J

~so

lo

2o

3o
%

z~o
9okt

figure 19
Left: catalytic activity of Pd-Au(lll) alloys in hydrogenolysis of n-butane at 590K (o) and
of propane at 624K (x).
Right: catalytic activity of Pd-Au(111) alloys in isomerization of n-butane at 590K.

614

chapter 13

Alumina-supported palladium-gold alloys have also been studied by O Cinneide


and Gault [82b]. The results on selectivity in n-hexane reactions indicate the same
tendency as in the case of cyclopentane [82a]. By using carbon-labelled molecules, it was
possible to study the contribution to isomerization by the 5C-mechanism (chapter 1 and
section 13.1); this was high (more than 80%) with pure palladium and remained high in
the alloys.
Henriques et al. [83] studied the reactions of n-heptane using palladium-tin
containing HY-zeolites as catalysts; they observed a suppression of isomerization and of
coke deposition by tin. It is likely that in these

bifunctional catalysts

the addition of tin

mainly influenced the behaviour of the zeolite. The possibility that the added metal
influences substantially the catalytic behaviour of the zeolite must be always considered
with other combinations than this. Palladium-tin films [82a] showed a suppression of those
mechanisms of cyclopentane-deuterium exchange which could be suspected to occur on
ensembles of several palladium atoms. Thus, the inert component performs the same role
as others (Au, Cu in Ni or Pd) do: it suppresses the availability of ensembles of several
palladium atoms, with predictable consequences.
13.2.3 Ruthenium alloys
Ruthenium and copper have different crystallographic structures (hcp and fcc,
respectively) and are almost immiscible; yet addition of copper to ruthenium sometimes
causes very dramatic changes in chemisorption and catalytic behaviour. The reason is that,
although copper does not dissolve, it adheres to the surface of ruthenium crystallites.
Adsorption of copper on ruthenium is expected to be strongest on defects in the ruthenium
structure and places such as those around the edges of ruthenium microcrystals and other
similar places. Ruthenium-copper alloys were introduced by the EXXON laboratory and
since then have enjoyed great popularity amongst catalyst chemists.
Sinfelt et al. [84] prepared these alloys in the form of macroscopic aggregates by
co-reduction or by sequential reduction in aqueous solutions using hydrazine, and
subsequent reduction and sintering in hydrogen. These aggregates showed a very different
behaviour in cyclohexane dehydrogenation and hydrogenolysis as can be seen in figure 20.
Ethane hydrogenolysis was also suppressed strongly by addition of copper. The amount of
copper which would be necessary to cover the whole surface of the small aggregates - in
the case studied - would be about 1.5wt%. It is therefore obvious from figure 20 that
copper forms small microcrystals on the surface of ruthenium, because at 4wt% copper
there is activity observed which is still higher than that of pure copper. This picture by
Sinfelt et al. [84], developed from indirect catalytic information, has been perfectly
confirmed by analysis using XPS [85], EXAFS [86], and by adsorption of hydrogen [84].
The general qualitative conclusions made from this work on aggregates were later fully

Reactions of alkanes and reforming of naphtha

615

confirmed by studies using single crystal planes [87].

10 4
Q
o
i
0

I...

Deilydrogenation

___n

E
s
(.9
(l)

. dro 'en

10 2
z3
u
o

.._...
4.-.'
t~

figure 20

1
0

1
4

I
8

1
12

Copper content (at. %)

The rates of dehydrogenation and hydrogenolysis of cyclohexane on ruthenium-copper


aggregates as a function of composition [84].

Ruthenium-copper alloys prepared in the form of very small particles on supports


[67,86] were subjected to EXAFS studies. These reconfirmed the picture suggested on
basis of catalytic results of copper accumulating on the surface of small ruthenium crystals
[86]. The catalytic measurements with these catalysts showed extensive agreement with the
results obtained on various other supported alloys (called in [84-86] bimetallic clusters).
This attractive ruthenium-copper system has kept chemists interested and new papers are
still appearing. In spite of the general agreement on some points, some apparent controversies appeared, too. Let us turn our attention to them.
A series of papers has appeared from the laboratory of D.W.Goodman on the
behaviour of copper/ruthenium(0001) catalysts in hydrogenolysis of ethane [88,89] and
figure 21 is representative of the results and the conclusions drawn [89]. Copper suppresses hydrogenolysis almost proportionally to its surface coverage. Such behaviour could be
expected from a system in which copper forms islands on ruthenium, but does not
diminish the size of the ensembles of ruthenium on the remaining uncovered ruthenium
surface. The effect of sulfur which is known to block several sites simultaneously, and
also spreads over the surface, thus diminishing the availability of large ensembles, is more
pronounced [89]. Sinfelt analyzed [91] these results and pointed to the fact that, while

616

chapter 13

single crystal planes of ruthenium with copper on top do not exhibit a strong suppression
of ethane hydrogenolysis, supported alloy particles do, probably because of their heterogeneity. Sinfelt speculated that in powders copper blocks the active defect sites which would
survive selfpoisoning in the steady state, while copper islands formation occurs on the
single crystal plane. This was a very good guess by Sinfelt, as later appeared. Smale and
King [92] have probably solved these problems satisfactorily. They used 1H-NMR to
determine the ruthenium surface exposed and evaluated the rates of ethane hydrogenolysis
per surface ruthenium atom. Then they plotted these rates as a function of copper content
and obtained the picture shown in figure 22, below.
1.0

0.8
o

o COPPER
LLI
,<
n~
LU

9 SULFUR

>__

I,<
..J
LU
rr

0.2

0.0
0.0

figure 21

0 1
ADATOM

t .
0.2

0.3

0.4

0 5

0.6

,SURFACE C O V E R A G E ( M O N O L A Y E R S )

Relative rate of ethane hydrogenolysis on a Ru(O001) catalyst as a function of copper


coverage (reaction temperature 550K). Sulfur poisoning results from another paper have
been replotted for comparison in the same figure [89].

Monte-Carlo calculations (see chapter 4) performed in King's laboratory [92] showed that
the first portion of copper added to a system of ruthenium crystallites preferentially covers
the defects, edges and comers (specific decoration), the sites which are here assumed to
have a higher activity in hydrogenolysis. When the amount of copper added is increased, it
starts to form islands and finally microcrystals on the flat planes of ruthenium particles.
Such island formation has no selectivity effect and almost no effect on the activity
expressed per ruthenium atom. With increasing temperature self-poisoning occurs and then
the presence of copper can be beneficial by suppressing it (see chapter 9 and [93]).
Therefore, addition of copper can at higher temperatures cause an increase in the rate per

617

Reactions of alkanes and reforming of naphtha

ruthenium atom of the steady state reaction rate (see figure 22). This effect of copper on
ruthenium is thus very similar to that observed with nickel-copper alloys. Self-poisoning
can be influenced to a very important extent by alloying.
100

10

9 493

9 528

a 508

o 548

figure 22
Rate of ethane hydrogenolysis at
different temperatures (+ 1 standard
deviation) per surface ruthenium as
measured by proton NMR. Dependence
on copper surface concentration [92].

r...- { {

rc
D
ET
eg

0.1

0.01
0.0

02

0.4

06

0.8

Cu Atomic Fraction

Peden and Goodman have studied dehydrogenation and hydrogenolysis of cyclohexane on copper/ruthenium(0001) and have observed an enhancement of cyclohexane
dehydrogenation when ruthenium was covered by copper [88,89,93]. They suggest that this
could be either (a) an electronic structure or ligand effect on the adsorption bond
strengths or (b) a mutual synergism in the activities of the metals, for example, with
hydrogen supplied to copper by spill-over from ruthenium causing copper to become
active. Others [88,89,91,93] also suggest that the strained copper layer could itself become
active. Finally, one also cannot exclude a role for mixed ensembles in these reactions.
The conclusions drawn by Smale and King [92]

and their model of specific

decoration were confirmed by Sprocket al. on n-butane hydrogenolysis on rutheniumcopper/SiO2 catalysts [94]. A slight increase in internal fission was observed when the
fraction of copper in the catalysts was increased. They point out that surface coverage of
ruthenium by hydrogen, OH, depends on the copper content, and OH in its turn is known to
influence the selectivities. The values of selectivities in different pathways are correlated
with the order in hydrogen pressure and both should depend on the composition of the
catalysts.
Bond and Xu Yide [95] compared the behaviour of propane, n-butane and
isobutane on two types of catalyst: ruthenium on silica and ruthenium-copper on silica.

618

chapter 13

Evaluation of the results was performed by using the comprehensive scheme developed by
Kempling and Anderson [96]. The RusoCus0/SiO2 catalyst showed in its reduced state a
much lower activity at 463K than the pure ruthenium on silica catalyst; however the
steady state activity did not show any selectivity effects of copper on ruthenium. A model
of specific decoration described above [92,94] is compatible with all these results. It was
also demonstrated clearly [95] how minute details of the preparation procedure can
influence catalytic behaviour.
At the level of a general statement, probably all investigators in the field of
catalysis by alloys would admit that a support can influence alloy formation, the crystal
texture, surface composition, etc., but there are few detailed studies of the problems.
However, important information in this respect is slowly appearing. It has been shown
[97a,b] by TPR that the formation and stability of ruthenium-copper catalysts as well as
their activity depend very much on the support. Ruthenium-gold catalysts do the same
[97c,d]. The test reaction here was hydrogenolysis of propane. It was concluded that
suppression of hydrogenolysis acts preferentially on the multiple hydrogenolytic fission to
methane [97b].
~oo[

figure 23
Hydrogenation of cyclopropane on Ru-Au
alloys. Influence of temperature on selectivity
S. PH = 0.20 atm; Pcp = 0.03 atm. Alloys of
different Ru(-R) content: rq, R 100; o, R 089;
A, R 064; o, R 010; Pure Ru catalysts: |
0.6%wt Ru/Si02; I"1, Ru sponge. The region of
higher temperatures shows a pronounced effect

s(%)j
;

of the support.

40-

50

t00
T (*C)

~50

The ruthenium-gold system is expected to be similar to ruthenium-copper.


Galvagno et al. used cyclopropane reactions to examine ruthenium-gold/SiO e catalysts, as
discussed before in comparison with nickel-copper. They reported [97d] that the selectivity
changes produced by gold are most pronounced at higher temperatures, where secondary
reactions occur freely. The selectivity S to propane formation is high for all catalysts at
temperatures lower than 373K, but at 423K the ruthenium-rich catalysts show lower
values, while the copper-rich catalysts do not (see figure 23). This confirms that also with
this reaction hydrogenolysis requires larger active centres.

Reactions of alkanes and reforming of naphtha

619

13.2.4 Platinum in combination with Group 11 elements


Platinum is an universal, versatile catalyst, used in many reactions, and therefore
the literature concerning this metal is very extensive: a large part of it deals with reactions
of saturated hydrocarbons. No wonder, since before the advent of the catalysts for control
of vehicle exhaust, the largest application of platinum-containing catalysts was in naphtha
reforming. Besides investigations on these industrial catalysts containing rhodium, iridium,
gold or lead and modifiers, many studies were devoted to model systems such as platinum
-Group 11 alloys.
With regard to platinum-gold alloys and hydrocarbon reactions, one can speak of a
system comprising one active and one inactive metal. Comparison with the other similar
systems discussed above was therefore a great challenge. Experiments [98] with platinumgold alloys prepared by hydrazine co-reduction mixed with inert silica and finally annealed
at temperatures ensuring that the surfaces at least were equilibrated [99], revealed a pattern
already known from studies on other systems such as nickel-copper: platinum very diluted
(x~ < 17%) in gold shows a higher selectivity although at much lower total activity, for
non-destructive reactions. This can be seen in figure 24.
1.0-

figure 24
Reactions of n-hexane and hydrogen on pure Pt/SiO 2 and on
Pt-Au/Si02 equilibrated dilute
alloys (bulk xpt < 17%). Content
of platinum in %, average (bulk)
composition shown. Selectivities
to hydrogenolytic cracking, total
of cyclization reactions and
isomerziation.

0.8-~

0.6-

0.4S c y ,io /
/
/
/

0.2-

o,"
Scr

___ .~,/
ol o
0

to"'

_n
,

10

12

i / [ .

12

,-~

80

100

%Pt

At the higher pressures of n-hexane and hydrogen as used in experiments with supported
alloys [99], it was mainly isomerization which was favoured in this way. At low pressures
in experiments with evaporated and annealed films, it was mainly dehydrocyclization
[100]. However, the results collected in the early stage of studies with alloys, already

620

chapter 13

signaled that the results depend on the hydrocarbon used [98-101] and the details of
preparation conditions. The latter aspect was explicitly addressed by a comprehensive
investigation [102]; in this paper several platinum-gold catalysts prepared as before [98]
by hydrazine reduction showed a higher selectivity to destructive reactions and not lower
as with earlier catalysts [98]. The only difference between the methods used [98,102] was
that in the latter study platinum and gold were reduced in an adsorbed state on a support,
and this probably caused a complete reversal of the catalytic effects of gold on platinum
[102]. Substantial progress in understanding this apparent dichotomy has been made,
through the use of single crystal alloys made in situ by evaporating gold onto platinum
single crystal planes. Reactions with these catalysts have been monitored [103] in UHVapparatus equipped with an isolation valve, allowing work with mixtures of 200 Torr of
hydrogen and 20 Torr of n-hexane. The results have been summarized [103] by a
schematic graph shown in figure 25:
Au - Pt (111)
H2 , 573K

/VV

H2/HC= 1 0 , P t o t = 2 2 0 T o r r
Isomerization

C-5Cyclization

Hydrogenolysis

~,r omatization

1.0
>,

~Epitaxial
systems

-.-Alloys
"5

aJ 0.5

/"
9

" "--..-.

"~

~ .
"~.

"\
9 --..

'

gold

"

coverage

figure 25
Schematic summary of various results obtained with Au-Pt(lll) alloy catalysts in nhexane reactions, at the indicated conditions as a function of fractional Au-coverage.
Epitaxial grown gold layer, results shown by full line. Alloys were obtained by equilibration of that layer [1031.

We see very clearly that suppression of hydrogenolysis is only observed with annealed
surfaces in which gold is dissolved. Gold on the platinum surface (in a non annealed
system) has different effects from those of gold which is in the surface. This throws new
light on earlier results [98-102] and on the apparent contradictions. The single crystal
results obviously indicate to us what the origin could be of the poor reproducibility
sometimes observed in results obtained with powders. This is also the message which can

Reactions of alkanes and reforming of naphtha

621

be drawn from a later paper [104], from which it appeared that gold has a very different
effect when evaporated on Pt(100) plane rather than on the (111) plane and when it is
annealed or not. While alloying with gold increased the total activity of the P t ( l l l )
surface up to a surface coverage by gold of 0.4, the selectivity was influenced as shown in
figure 25 above. However, with Pt(100) surface the total activity decreased in proportion
to the gold coverage. Suppression of selectivity to hydrogenolysis was observed again, but
in contrast to the situation with P t ( l l l ) there was no enhancement in isomerization
selectivity. It was suggested that perhaps an ensemble in a form of a mixed triplet PtzAu is
particularly well suited for isomerization. Whatever the explanation, these results provide
yet another reason why platinum-gold alloys are so sensitive to the preparation procedure.
The differences in activities in various reactions between platinum and platinumcopper alloys are much smaller than with the platinum-gold couple. There are two possible
explanations: a possible spill-over of hydrogen from platinum sites to copper sites, thereby
making copper sites active is the first possibility, and further, one must consider seriously
the possibility of the operation in catalysis of mixed ensembles.
It has been reported [105] that, with platinum-copper/silica alloys, decrease in
overall activity of n-hexane-hydrogen reaction occurs, but that selectivity to hydrogenolysis increases as the platinum concentration decreases [105]. This is shown in figure
26:
T = 290~

0,8-

u.. -" I~i


---

o _

0.4

--C~

~ ~

o%
o :t~
0

1~

0
A
--- w - 0

~)

l"3

2;

6'0

a'o

100

%Pt
figure 26
Reactions of n-hexane and hydrogen on Pt-Cu/SiO 2 alloy catalysts. The selectivity to
isomerization (squares), hydrogenolytic cracking (triangles) and total cyclization (circles)
all as a function of average platinum content in the alloys.

The cracking patterns looked like that of platinum or of nickel diluted in copper [65].
Cyclization selectivity was increased by addition of copper as did also the contribution of
the cyclic 5C-mechanism to isomerization [106]. This indicates that in dilute alloys the
active sites are platinum atoms in very small ensembles of platinum atoms and that these

622

chapter 13

behave differently from those in pure platinum surfaces. There was no speculation [105106] on ligand effects influencing the properties of platinum atoms, since the temperature
dependence of all groups of reactions did not indicate variations in the corresponding
apparent activation energies. Results with n-hexane [105] were confirmed by those
obtained with neo-hexane (2,2-dimethylbutane) [107] as can be seen in figure 27:
1.0
hydrogenolysis
---o-isomerisotion
0

0.8

0.6
_

0.40

0.2-

'

2'0

'

-~'0

'

6'o

'

8'0

% Pt

'

100

figure 27
Neohexane reactions with hydrogen on Pt-Cu/Si02 catalysts, selectivities for hydrogenolysis and isomerization, both as a function of alloy bulk composition. These results were
obtained in the temperature region 553-593K, but for each catalyst in a region in which
the selectivities were almost independent of the temperature.

As mentioned in section 13.1, the product pattern of neo-hexane reactions also offers
information on the adsorption modes. On the basis of this information it could be
concluded [107] that the proportions of cz7 and cz~/'modes stay constant for all alloys. The
czg-adsorption mode is almost negligible over the whole range of composition [107]. This
means that the same species are always formed, but they participate in isomerization on
pure platinum and in hydrogenolytic cracking on platinum diluted in copper. Several
possible reasons were considered. First, this result could be an effect of a carbonaceous
layer present on platinum surfaces but missing to some extent on platinum-copper
surfaces. Den Hartog et al.[ 108] addressed this problem using a pulse-micro reactor. Under
these conditions the cracking pattern was in detail somewhat different from that seen
earlier [106], but the increase of hydrogenolytic selectivity in dilute alloys was confirmed
even for virtually carbon-free surfaces. Selectivity in n-hexane reactions did not depend on
the number of pulses, and the fraction of surface atoms covered by carbon atoms retained
from the first pulse of n-hexane/H 2 mixture was estimated to be less than 5%. Thus the
idea of selectivity being mainly determined by deposited carbon could be discarded and

623

Reactions of alkanes and reforming of naphtha

the idea of mixed ensembles proposed in earlier papers [106,107] was re-examined [108].

C/
\

C
C

\C /
/
Pt

Ca)

\ /
/ C\

C
/ C

\
/
/ C\

C
/C

C
l

C
t

C
i

C
i

P~

Pt

Cu

Pt

(b)

figure 28
Possible c~7' adsorption modes
of di-adsorbed neohexane.

(c)

Upon comparison of the different adsorption modes shown in figure 28, one possible
explanation which emerges is that complexes (a) and (b) lead to isomerization while
complex (c) leads to hydrogenolysis. Later [109], it was also admitted that complex (a)
might lead to hydrogenolysis. Originally, this possibility was disregarded [107] because
dilute platinum-gold alloys did not show enhanced hydrogenolysis. However, one has to
realize that platinum and gold form alloys endothermically, and that this stimulates
formation of platinum three-dimensional clusters in the bulk or two-dimensional ensembles, in the surface, and thus also in dilute platinum-gold alloys there could be present
platinum ensembles suited for isomerization by complexes such as (b). On the other hand
the smallest platinum particles show [109,5a-c] enhanced activity in hydrogenolysis but
not an enhanced activity in multiple-bond formation, a correlation which otherwise holds
for other metals. This points to metallocyclobutane complexes, such as (a) above. For the
time being, both these possibilities for explaining the enhanced hydrogenolysis should be
kept in mind.
13.2.5 Alloys containing rhodium, iridium and Group 11 metals
Rhodium and copper show a very limited miscibility, so that even at low contents
of copper the surface of rhodium can to a very large extent be covered. A very interesting
comparison has been made of reactions between ethane, pentane and neohexane and
hydrogen occuring on rhodium and rhodium-copper/SiO2 catalysts [110]. The reaction of
ethane was most sensitive to surface alloying. Adding 10 at% copper to rhodium suppressed the rate of ethane hydrogenolysis by two to three orders of magnitude. Ethane cannot
undergo reactions other than hydrogenolysis and dehydrogenation via 2C-~13 complexes.
With pentane, the main reaction was again hydrogenolysis, although there was a significant increase in isomerization selectivity, from 1-2% for pure rhodium to 5-15% for
alloys. Suppression of hydrogenolysis by copper was however much smaller with pentane
than with ethane. This was ascribed [110] to the existence of an one-site, 3C-ay adsorption mode, inducing hydrogenolysis as suggested by Rooney earlier [111]. The most
interesting with regard to the possible mechanism are the results obtained with neo-hexane

624

chapter 13

(2,2-dimethylbutane). Here we observe considerable ~B-hydrogenolysis to neopentane even


on those catalysts which show an almost zero activity in ethane hydrogenolysis. The same
situation exists with nickel-copper alloys [64]. There are several possible explanations to
be considered. (i) With neohexane, but not with ethane, one site 2C~B hydrogenolysis is
possible, perhaps due to the presence of the secondary carbon atom. (ii) Neohexane forms
not only a 3Cc~? intermediate (a metallacyclobutane-like complex), but also a 4Cc~8
intermediate (a metallacyclopentane-like complex). Both cycles are assumed to split upon
simultaneous formation of metal-carbenic species.
Coq et al. [116] have examined rhodium-copper/SiO 2 catalysts in reactions of nhexane, 2-methylpentane, 2,2-dimethylbutane, methylcyclopentane and 2,2,3,3-tetramethylbutane (TMB) with hydrogen. The presence of copper suppressed all reactions, but the
extent of suppression of hydrogenolysis, which is the principal reaction, depends on the
structure of the molecule. Particle size and alloying effects were compared for all these
reactants. For all reactants except the last, the effect of added copper was much more
pronounced than the particle size effect; however, with TMB it was the other way round.
An explanation was suggested which assumed that, to break the middle bond in the TMB
molecule, the molecule must be bound to the metal surface by carbenic bonds which
require large ensembles of rhodium atoms to be available. Such large ensembles can be
found only on flat faces of large rhodium particles and that is responsible for the particle
size effect. The other molecules can react on the same planes too, but they have also a fast
alternative, which is to react on the edge-, comer- and defect- sites. These sites are
however first occupied by copper when this metal is added to rhodium, and as a consequence

a much more pronounced effect of copper is observed with this group of

molecules, and vice-versa the particle size effect is less important with them.
Iridium-gold films were studied with n-butane and n-hexane by Plunkett and Clarke
[112]. The presence of gold mainly suppressed terminal hydrogenolysis, so that dehydrocyclization had a better chance to occur, and finally, at the highest temperature used and at
the highest content of gold in the films, it was a pronounced effect. Suppression of
hydrogenolysis of n-heptane was also observed at low pressures of heptane on iridium

foils covered by gold [113]. Results obtained with films [112] were confirmed by
experiments with iridium-gold/SiO2 catalysts [114], and the usual selectivity effects were
observed here, too.
Den Hartog compared iridium/SiO 2 (5%wt loading) with another iridium/SiO 2
catalyst in which 20% of the iridium had been replaced by copper [115]. The two metals
are practically immiscible so that a close similarity is to be expected with the ruthenium/copper system.

Reactions of alkanes and reforming of naphtha

625

figure 29
Reactions of n-hexane and hydrogen on

100I "

-z,.0

S(%)

Ir-Cu/Si02 alloy catalysts. Selectivities


for hydrogenolysis (-.-.-), dehydrocyclization ( - - -) and isomerization (full

50-

-2.0

line) are shown as functions of the


average iridium content. Also shown is
the total conversion per gram catalyst
(+) in a logarithmic scale. Temperature

/u7 ' s

-0

oYolr

of the reaction is 600K for all catalysts.

-01

The results shown in figure 29 show that upon addition of copper the total activity is
much decreased (right side scale), but the selectivity in hydrogenolysis is slightly

increased (left side scale). It is also important to see that the character of the fission
changes from internal (see sections 13.1 and 13.2) to terminal. This change is shown in
figure 30.

16-

figure 30
Multifission parameter My (see section

Mf

13.1.3 - f o r definition) as a function of


temperature, for 3 catalysts;
= 100% Ir

12-

8-

lr89CUll

= IrsoCu2o [115].
l.-

~o

s6o

s~o

060 rcKI

Relative participation of the internal (typical for iridium) and terminal (typical for nickel)
fission varies and is reflected by the parameter of multiple fission Mf. It is to be preferred
to use the more exact treatment suggested [96] and applied [95], but it is not easy to
analyse fully the kinetics of reactions on many alloys. It is worthwhile noticing that
addition of copper to either iridium or nickel causes with regard to M e opposite effects on
the character of fission. It decreases Mf in the case of iridium but increases it with nickel.

626

chapter 13

Let us stress that the exact reason is not known but a possible role of mixed ensembles of
iridium-copper, but not of nickel-copper, cannot be excluded. Another explanation could
be that edges of iridium particles are most active in internal fission, but they are the first
to be blocked by copper [116]. With nickel, the shift in selectivity could be due to an
ensemble size effect. The reader should notice that these are all speculative explanations.
Foger and Anderson [117] compared low-loaded (0.8-1.5wt% of metal) iridium/Si02 and iridium-gold/SiO2 catalysts having from zero up to 86% gold, and they also
observed a very pronounced dependence of the suppression of hydrogenolysis by gold on
the molecular structure of the reactant. Activation energies did not vary by addition of
gold in the range of temperature 440-550K, but the frequency factor did change: to only a
small extent over a broad range of composition when neopentane was the reactant, but it
varied much more strongly when either n-butane or neohexane (reacting here exclusively
by the 2C-o~B adsorption mode), was used. It was concluded [117] that neopentane
hydrogenolysis takes place solely via a 3C-~y intermediate (since there is no other
possibility) while neohexane and n-butane react preferentially via a 2C-c~B fission. It
would appear that the first reaction needs larger iridium ensembles. When we combine the
results of these investigations [115,117] we can consider that iridium-copper mixed
ensembles, on which we speculated above, are indeed operating in hydrogenolytic fission
on flat parts of the particle surface.
Brunelle et al. [118] prepared various rhenium-iridium and copper-iridium catalysts
with alumina as a support, and compared the selectivities with those of osmium, platinum
and palladium. The most essential results of this study are shown in figure 31.
It is a somewhat complex figure, containing much information. However, if we
focus on the most essential points, we conclude the following. Adding copper to iridium
causes an increase of selectivity for the non-destructive reactions (F), this selectivity
increase being accompanied by activity decrease (E). The suppression is most pronounced

with hydrogenolysis (E), the reaction which we already know as a large ensemble reaction
(see chapter 9). With iridium-rhenium alloys we observe synergetic effects which as we
shall see below can be ascribed to the mixed-ensembles (B,A). The window C shows that
the character of splitting is different on rhenium and iridium, but varies smoothly over the
alloys.

627

Reactions of alkanes and reforming of naphtha

00'
0

0.5

1 ,,t % O,

~176
~.,&

o 1~

or./" ,.~ ~
I

"

.! i

0.6 ...........
04

'

3.2
o
II

_
Ne

13

\t
N

05

0
Jr"
k-Os

Ir

O.S

I~

figure 31
Left: variations in properties of Re-Ir/AleO 2 catalysts with atomic fraction of iridium.
Comparison with Os-alumina catalysts.
A. Hydrogen chemisorption in cm~/g catalyst (o)
B. Turn-over measured at 303K for benzene hydrogenation (a) and 458K for n-pentane
hydrogenolysis (o) in mole/h.site.
C. Depth of n-pentane hydrogenolysis at 458K expressed by equation(1)(D).
Right: variations in properties of Ir-Cu/Al203 catalysts with the atomic fraction of copper.
Comparison with Pt-Pd/Si02 catalysts.
D. Hydrogen chemisorption in cm3/g catalyst (o). Carbon monoxide chemisorption
variation with wt% copper (*).
E. Turn-over measured at 373K for benzene hydrogenation (,) and 523K for pentane
hydrogenolysis (.), isomerization (V) and cyclization (n).
F. Isomerization (v) and cyclization (u) selectivities at 523K.

628

chapter 13

13.2.6 Platinum-rhenium model catalysts


We shall see below that in their working state in industrial reactors these catalysts
are always used with an alumina support (often 7-alumina) and modified by sulfur.
However, to understand the processes on these catalysts better, many have studied them in
a sulfur-free state and also supported by inert supports such as silica, or not supported at
all.
It was from the beginning very important to establish the particular role of sulfur
and of the acidic alumina support which is needed for bifunctional catalysis. The most
logical step to take is to study first well-defined support-free single crystal planes. These
are experiments which, ironically enough, have been performed last and only very recently
[119,120]. However, the results of these experiments form the best point at which to start
a general discussion on earlier results.
With ethane [120], cyclohexane and n-hexane [119] in the feed, platinum-rhenium
single crystal catalysts show a synergism: alloys are more active than the individual
components. Rhenium is very active in hydrogenolysis. Sulfur enhances the dehydrogenation activity of pure platinum as well as of platinum covered by rhenium. Amongst the
skeletal reactions occurring on platinum, C6- and Cs-dehydrocyclizations and isomerization
are suppressed most by rhenium, while hydrogenolysis is promoted. The presence of sulfur
promotes on all catalysts (pure platinum as well as platinum with rhenium), the C5- and C6
dehydrocyclizations, but isomerization remains low. The latter fact is very important for
the later discussion of results obtained with commercial platinum-rhenium-sulfur/alumina
catalysts working as bifunctional catalysts (see below). The main results obtained [119]
are summarized in figure 32.
1.0
0
[~.

O.Z.

.-

"--

0.5

,o Hydrogenolysis

o Cyclization

o Cyclization

9 Hydrogenolysis

9 Isomerization
121
9 --

~
r-

02

o Isomerization

o Dehydrocyclization

9 Dehydrocyclizalion

o
E

0.0

lr

0.0

~-

-,-

~- r

0.0

1.0
Re Coverage (ML)

2.0

.....

0.0

- .....
3?"

1.0
Re

~ ....

2.0

Coverage

v-

3.0
(M L )

figure 32
Reaction conditions: T = 573K, P,_hexa,e= 20 Torr and PH2= 200 Torr.
Left: rate of n-hexane conversion on P t ( l l l ) as a function of Re coverage.
Right: rate of n-hexane conversion on sulfided P t ( l l l ) as a function of Re coverage. The
catalysts were sulfided by depositing the saturation amount of sulfur.

Reactions of alkanes and reforming of naphtha

629

Let us now return to several other points of fundamental information that are
important for the understanding of industrial catalysts. Under the conditions used (UHV
apparatus, equipped with an isolation valve, reaction conditions: 100 Torr H 2, 10 Torr
ethane, T=573K), the rate of hydrogenolysis is very low on pure platinum(l 1 l) surface
[119,120]: (5x10 -3 tool methane site-is -1) and is considerably higher on rhenium(0001)
(0.55 mol methane site -1 s-1 on an annealed surface and 1.8 mol methane site -~ s-1 on a
surface roughened by argon bombardement). When two or more monolayers of rhenium
had been laid down on platinum(111), the activity was again 0.5 mol methane site -1 s -1. At
coverages of rhenium on platinum between 0.6 and 1.0 monolayer, a synergistic effect was
seen and the activities reached values of 2-3 in the above units. Since, at the temperature
of reaction, the penetration of rhenium into platinum is negligible, we can ascribe the
higher hydrogenolytic rates to the presence of rhenium

on

the surface of platinum. When,

vice versa, platinum is deposited on rhenium, again the rate of hydrogenolysis is higher
than that of either of the metals in the pure state. The most active platinum-on-rhenium
catalyst was comparable with a clean platinum(l 1 l) plane, when this plane had been
sputtered and there was

no

subsequent anneal. In another experiment the platinum-on-

rhenium surface was annealed at 923K, at which temperature platinum penetrates into
rhenium, which in its turn appears on the surface. Such alloy surfaces showed the highest
hydrogenolytic activity. The fact that platinum-on-rhenium catalysts show a lower activity
than the platinum-rhenium alloy should be kept in mind for further discussion, because
there are

theories which ascribe to platinum on rhenium a higher activity than that of

platinum itself. Also the fact that rhenium on platinum has an even higher hydrogenolytic
activity is worth remembering. It shows that rhenium manifests itself by its reaction
fingerprint in the product distribution, with hydrogenolysis prevailing.
Three monolayers of rhenium on platinum at 573K in a sulfur-free state also
showed a low but easily measurable activity for benzene formation [119]. This can be
compared with the results of a study of reactions with hydrogen and n-hexane, methylcyclopentane,

3-methylpentane,

n-pentane

and isobutane

on

evaporated

rhenium

and

rhenium-gold [121]. The films were inhomogeneous mixtures of various rhenium-gold


phases but the message was clear: adding of gold to rhenium suppresses hydrogenolysis,
so that it is possible to work at the higher temperatures (600-700K) which favour
dehydrogenation and benzene formation. It is not excluded that this dehydrocyclization
runs partially on a carbonaceous layer extending over the alloy surface, by a stepwise
dehydrogenation. The so-called 1-6 dehydrocyclization was indeed a prevailing pathway to
benzene [122]. Results are presented in table 7 [121].

630

a
=

"7

:~
m=

. . . .

gg
~5o

, ~ , ~

" ~ = m-

L~

,<5

r--

c-I

~5o,~

m.

tr)

oc5

~,

--

..=

fo

e-

e~

e-

e-

..g

.='~

chapter 13

Reactions of alkanes and reforming of naphtha

631

Results very similar to those with rhenium-gold films are also obtained with
tungsten-gold and tantalum-gold films; admixture of gold and the presence of carbonaceous deposits convert the hydrogenolytic catalysts into aromatization catalysts [121].
13.2.7 Platinum-rhenium on alumina (sulfur-free catalysts)
The presence of a support, so beneficial in industrial use, introduces many
complications which must all be considered in understanding the function of the industrial
catalysts (see chapter 7). Alumina can cause separation of metal precursors during
preparation of catalysts by the impregnation technique and it was at first not selfevident
that the components, platinum and rhenium, operate by being alloyed or at least in close
proximity to each other. Some workers [122-124] deny that and think that the two
elements act separately. It was in the beginning not evident at all that all, or a considerable part, of rhenium in the catalysts really becomes reduced under conditions of industrial
operation. At first glance, the literature offers a controversial picture, with conclusions
such as 'almost all rhenium is quantitatively reduced' up to 'under operating conditions
rhenium oxide is substantially unreduced and it is present as Re(IV)', which is thought by
many people to be beneficial for the performance of the catalyst [124-134]. However, a
closer inspection of the results allows us to make apparently consistent and reliable
conclusions. Such a consistent picture is built up from several ingredients which will now
be discussed in a logical but not a chronological sequence.
When a small amount of an oxide is mounted on alumina, which is a spinel with
many defects, both stoichiometric and non-stoichiometric, the first cations added occupy
the defect sites (see chapter 7). Only when these defect sites are filled and a complete
monolayer has been formed do crystallites of the active supported oxide start to appear.
This is illustrated in figure 33 by the MoO3-A1203 system, the results for which are being
taken here as representative of extended systematic studies performed at Beijing University
[135]. One can expect that rhenium oxides behave in a similar way to that of molybdenum
trioxide, the onset of whose crystal formation is clearly seen.
c,')

9 o 0,40121

>,

figure 33
Molybdenum

trioxide

in

crystalline

form, as a function of its content in the


MoO/Al20 ~ powders. Similar behaviour

0.20

can also be expected for the formation

"

O-O---C

0.20

0.40

0.60

Total Mo03,g/g ~ - A I 2 0 3

of a rhenium oxide layer on alumina.

632

chapter 13

Industrial catalysts usually contain only about 0.6%wt of total metal, while model
catalysts prepared for laboratory research contain up to 2%wt of metals. Rhenium, or other
cations on the defect sites, are difficult to reduce, and thus when rhenium loadings are
low, a number of rhenium cations can coincide with the expected (because the exact
number is not known) content of defect sites on the surface. Knowing this, one is not
surprised that in the literature quoted [125-134] those who studied catalysts with a very
low loading did not observe full or even appreciable reduction of rhenium, while those
who studied catalysts with a high metal loadings have seen that most of the rhenium was
reduced. These conclusions are mainly based on TPR and TPD measurements, but there
were also attempts to determine the content of ionic rhenium in a more direct chemical
way. Its extraction by acetylacetone is reliable, but since alumina becomes dissolved it is
difficult to avoid some reduced rhenium from alloy particles appearing in the gel-like
solution used subsequently for analysis by X-ray fluorescence. By this method up to 25%
of rhenium present was found to be ionic in reduced catalysts. However, this high figure
is, because of these difficulties, probably an overestimate [136]. When the extraction is
performed by another method, which appeared to be successful for ionic platinum, i.e. by
diethylamine/acetic acid (4%) mixture, the content of ionic rhenium was estimated as less
then 5% [137]. The concentration of Re(IV) has also been determined by calibrated ESR
signals [138]. Rhenium dioxide is the most stable oxide, so it can be expected that most of
the unreduced rhenium would be in the 4+ state. By using ESR, it was concluded that in
platinum-rhenium catalysts the fraction of unreduced rhenium was less than about 10%;
for pure rhenium or alumina catalysts less than about 20%. Rhenium could also have been
present in another valency, so that only a rough estimate could be made from all these
measurements. However, if we rely on the techniques used, the conclusion is that the
platinum-rhenium/alumina catalysts contain of 10_+5% ionic unreduced rhenium. This
amount is of such a small size that it allows us to speculate that platinum-rhenium alloys
are indeed formed, and operate also under standard industrial conditions, but on the other
hand it is large enough that it would be premature to deny the possible beneficial effect of
rhenium(IV) on reactions occuring on the support.
The proven presence of metallic rhenium still does not prove that alloys are formed
and operate in naphtha reforming; however a strong indirect evidence exists for alloy
formation. First, it has been shown that the presence of water, even in small amounts,
increases substantially the migration of rhenium oxide species on alumina [131,139-141].
When such mobile species collide upon reduction in progress with a platinum particle, and
platinum metal particles are certainly formed earlier than rhenium particles, the oxide will
be reduced by hydrogen delivered by spill-over from the neighbouring platinum. Once
rhenium metal appears, reduction can continue by hydrogen which is also dissociated on
rhenium. A very strong indication for the easy migration of rhenium has been presented
[141]. Using cyclopentane hydrogenolysis as a fingerprint reaction for alloy formation it

Reactions of alkanes and reforming of naphtha

633

was found that, when a physical mixture of Pt-Na-Y and Re-Na-Y zeolites was coreduced, the product patterns of exchange reactions clearly indicated that extensive alloy
formation indeed took place. This means that rhenium species can migrate from one
crystal to another and join platinum in other zeolite cages. It has also been established that
the presence of carbon in the system makes the reduction of rhenium easier [134]. This is
important in relation to the following. When reforming catalysts are deactivated they must
be regularly regenerated by burning off the carbon deposits. The platinum-rhenium alloy is
then re-converted into segregated oxides, but a repeat reduction in the presence of small
amounts of water obviously can restore the alloy. It is the water which is always present
in the system after the oxidation and re-reduction steps which very much helps alloy
formation.
Margitfalvi et al. [142] have studied the influence of an oxidation-reduction cycle
on the product distribution in the n-hexane-hydrogen reaction. They prepared the catalysts
shown in table 8 by depositing rhenium onto platinum covered by hydrogen (see chapter
7). These catalysts (no.l-4) showed only a small suppression of the total conversion and of
hydrogenolysis, the latter change being achieved by carbonaceous deposits. When the
system was treated by an oxidation-reduction cycle, that is, when the formation of alloys
was stimulated, the catalysts started to show the usual selectivities of Pt-Re/A1203
catalysts. It can be seen in the table that a part of the Cl-units produced by hydrogenolysis
appears in toluene, a CT-hydrocarbon formed from n-hexane; the presence of rhenium on
the platinum promotes this step.
The effects of oxidation-reduction cycles are complex. Botman et al. [137] saw that
upon calcination the rhenium oxide does not leave the support, as some authors earlier
suspected, but it penetrates into the interior of alumina. When the catalyst is being rereduced, then, depending on the duration of reduction, rhenium returns more or less
completely to the surface of alumina, where rhenium ions are reduced. This is probably
the background for the newest technology in platinum-rhenium catalyst preparation, which
makes use of a more crystalline 0~-A1203 and makes the catalysts with more rhenium than
platinum (typically 0.3wt% platinum; 0.6%wt rhenium), so that enough rhenium is all
times at the disposal on the outer surface of alumina.
Botman [137] also established that the presence of rhenium in the catalysts
increases the fraction of platinum which is in the ionic state after reduction, in particular,
when chlorine is also present in the system. Industrial catalysts are prepared with the use
of hexachloroplatinic acid and when it is not removed by extra steps, chlorine is always
partially retained by the carrier. Its presence is beneficial for the bifunctional operation of
the catalysts since chlorine creates more of strong acid sites on the alumina. This however
influences the reducibility of the platinum and the rhenium, irrespective of the fact that,
upon reduction of the catalysts with hydrogen, a part of the chlorine is removed as
hydrogen chloride.

table 8
Comparison of the catalytic properties of Pt/A1203 and Re-Pt/AI203 catalysts in hexane dehydrocyclization
reaction temperature: 753K, H 2' n-hexane = 5:1)
selectivity data

temperature of pretreatment (K)


catalyst

H2

O2

conversion

C~-C5

isohexane

benzene

methylcyclo-

toluene

pentane
I

Pt/A1203

773

0.39

0.190

0.60

0.061

0.069

0.062

Re-Pt/A1203

773

0.23

0.092

0.525

0.120

0.096

0.120

Pt/A1203

848

0.30

0.190

0.560

0.089

0.093

0.035

Re-Pt/A1203

848

0.25

0.160

0.473

0.100

0.140

0.094

Pt/A1203

673

773

0.61

0.077

0.690

0.084

0.027

0.098

Re-Pt/A1203

673

773

0.75

0.070

0.725

0.110

0.026

0.042

Re-Pt/AI203

773

673

0.72

0.61

0.740

0.092

0.024

0.056

Pt/AI203

773

773

0.78

0.110

0.700

0.096

0.024

0.044

Re-Pt/A1203

773

773

0.80

0.051

0.740

0.110

0.023

0.044

Re-Pt/A1203

773

848

0.73

Ii 0.060

0.690

0.150

0.029

0.036

* Calcination omitted; catalyst reduced after contact with moisture in air.


r

e.-.

Reactions of alkanes and reforming of naphtha

635

All this is very well illustrated by a detailed study performed by Munuera et al. [143,144]
which also revealed that by oxidation-reduction cycles one can produce catalysts with
different amounts of rhenium in the surface and that these catalysts differ in their cracking
patterns. Rhenium has a typical multiple'cracking character which appears when more
rhenium is in the surface. Such a state can be created by oxidation and mild reduction.
Platinum shows internal cracking and this appears only after a more thorough reduction.
Bond and Gelsthorpe studied hydrogenolysis of propane, n-butane and isobutane
and compared Pt/A1203 with Pt-Re/AI203 catalysts (EUROPT-3 and EUROPT-4). They
established that adding rhenium enhances the propensity to multiple hydrogenolysis [145a].
A comparison of results obtained with these catalysts by various laboratories has been
compiled [145b], this shows again how important to the performance of the catalysts are
the fine details of the preparation method.
A simple molecule such as n-butane probes mainly the hydrogenolytic properties of
these catalysts. The activity per unit metal surface area shows a very pronounced maximum when about 80% of metal content is rhenium [ 146]. We shall see more of this type
of activity pattern below. A detailed kinetic study has been performed on the reactions of
ethane and n-butane on Pt-Re-A1203; the rate varied according to the equation proposed by
Cimino et al. [29] in section 13.1.
Sharp maxima in the activity vs. catalyst composition curves have been confirmed
by several other groups [148-150] and have also been found for molecules other than
those already mentioned.

lO

figure 34
Specific activity (mole hlm-2), per unit
alloy surface area of Pt-Re/Al203 alloy
catalysts, in cyclopentane hydrogenolysis. Activity as a function of at% rhenium in the alloys on different carriers:

-2
10

Pt/~-Al203 catalyst (stars);


Pt/Al203 Degussa (full circles);

1 0 3_
.]

Pcyclopent = 0.1 bar, Pm = 0.9 bar,


T = 513K [1461.

-4
10

i
20

t.lO

610

810
%Re

10 0

636

chapter 13

Figure 34 shows the results for cyclopentane hydrogenolysis [146] and figure 35 shows
the selectivity pattern for heptane hydrogenolysis and isomerization [150] and similar
results are known for n-butane [151 ]. Explanations for this behaviour differ. Many authors
of the original and reviewing papers were inclined to see in the maxima, and in the
assumed underlying synergism of platinum and rhenium, a clear indication for ligand
effects of alloying. However, two other explanations seem to be more attractive. (i) Mixed

ensembles can be formed which allow a higher rate of catalytic reaction than ensembles of
any of the components taken alone [ 107,148,151 ]. Notice that this behaviour has been seen
also with support-free single crystal planes, so it can be ascribed to the metallic function
of the catalysts. (ii) There are different rate-determining steps on platinum and rhenium,
being C-C bond rupture on the former and removal of methane or other fragments on the
latter metal. This explanation seems to be well supported by some results for platinumrhenium catalysts [149], but similar maxima in activity vs composition curves are also
shown by the platinum-iridium and by iridium-osmium combinations [151]. In particular,
in the case of iridium-containing catalysts it does not seem likely that this active metal
would retain hydrocarbon fragments strongly, as does rhenium. Therefore, the first
explanation seems to be the most universal one.

@
(a)

figure 35
Conversion

OJ.-

of

n-heptane

on

(Pt+Re)/Al203 catalysts. Selectivities in isomerization (a) and


0.2

hydrogenolysis (b).A = 773K, [3


~ 1 7 6

O-

= 973K, 9 = 1073K. Selectivities


are plotted as functions of the

2'0

'

Z.'O

'

6'0

'

8'0

'

100

(at%) [150].

(b}

0.~

0.2
o

2'0

'

~'0

'

6'0

average rhenium content

'

8'0

'

~60

Reactions of alkanes and reforming of naphtha

637

Using neohexane (2,2-dimethylbutane) as a test molecule allows us to analyze the


prevailing surface complexes (see section 13.1). The overall activity pattern obtained with
this molecule again shows a maximum at about 75mo1% rhenium. This is shown in figure
36 [152], where the activity is expressed per mg of total metal, but the surface areas of the
metallic components do not vary much. The catalysts with 20% and 80% rhenium have an
average particle size of 1.3 nm and 1.0-1.7 nm respectively, but the activity per unit
weight differs more. The different catalyst compositions carry a fingerprint of either
platinum or rhenium in the product idstribution. On platinum-rich alloys, t~T-complexes are
observed and isomerization of neohexane, as well as its hydrogenolysis, takes place. On
rhenium-rich surfaces t~6-complexes and, of course, hydrogenolysis prevail: this is seen in
figures 37 and 38.
1,5

E
iS

I
I

>,.
..,=,
~

>

<

0,5

[]
ii 9
W
,
O

7-

-T-

40

20

Series

Series 2
Ketlen

60

80

100

% Re / R e + P t

figure 36
Activity for the reaction of neohexane with hydrogen (expressed as conversion per mg
catalyst) as a function of alloy composition at 548K. Series 1 catalysts are prepared from
dissolved commercially available salts, series 2 catalysts are prepared by dissolution of
metals in aqua regia and impregnation on A1203 and the 'Ketjen' are the commercial
catalysts. The two commercial catalysts are CK 455 (0.47wt% Pt, 0.475 wt% Re) and CK
433 (0.296 wt% Pt and 0.31 wt% Re).

The selectivity to neopentane is used as a diagnostic tool to establish the contribution of


the txl3-mechanism. However, in particular on pure rhenium, neopentane reacts further,
while still in the adsorbed state, to yield butane and propane. This leads to the apparent
decrease of the contribution of the t~6-mechanism in figure 38.

638

chapter 13

4o

3o

~o

c~'Y. s e r i e s

o~'r. s e r a e s

0,3'. K e t i e n

10

o~

0
0

20

40

60

80

.,5,,
loo

a'y".

seT,es

e3".

ser,es

~h",

Ketjen

% Re / R e + P t

figure 37

Neohexane - hydrogen reaction: selectivity in the 3C ~"~ and ot~[' mechanisms (isomerization and hydrogenolysis together) as a function of catalyst composition (catalysts as in
figure 36).
80

6O
\

.,...,
>
.,..,

\
\

40
Q.)
V)
2O

/.

C;' % c ~ . series 1
9

% r~h'. series 2

.1~ % n i l , c o m m e r c i a l

2O

>
(o

~5

20

40

60

80

'::,

% P + B. series 1

% P + B. ser,es 2

~.

% P + B. c o m m e r c i a l

100

% Re / R e + P t

figure 38
Neohexane - hydrogen reaction: the selectivity S (2C~fl) versus catalyst composition.
Below: the selectivity of propane and butane formation (both characteristic of multiple
splitting) versus catalyst composition (catalysts as in figure 36).

Reactions of alkanes and reforming of naphtha

639

Multiple fission of adsorbed intermediates formed from neohexane or neopentane


points likely to adsorption through three primary carbon atoms, viz. a 4C-complex, with a
tripodic structure.
With sulfur-free catalysts, the role of acidic sites is somewhat marginal: the nonacidic silica-supported platinum catalysts show approximately the same picture as the
alumina-supported catalysts. We have therefore neglected the role of acidity (bifunctionality) in the whole discussion above on the platinum-rhenium/alumina catalysts in reactions
at low temperatures. However, as we shall see below and also in figure 39, the situation
concerning hexane is very different with bifunctional catalysts modified by sulfur or sulfur
(see section 13.4).
100

200--Y

50

100

0
o

50

"I, Re

1oo

50

"/.Re

I~0

figure 39
Reaction of n-hexane with hydrogen: yields Y, defined as a product of conversion and the
corresponding selectivity, are proportional to the rates.
Left: Y as a function of rhenium content in Pt-Re/Cl-Al203, chlorine containing catalysts.
Hydrogenolysis (squares), isomerization (circles), cyclization (triangles).
Right: the same, for chlorine-free Pt-Re/Al203 catalysts.

13.3

Fundamental studies on reforming catalysts.

Even before the advent of platinum-rhenium/alumina catalysts, it was common


industrial practice to modify platinum/alumina catalysts before and during catalytic
reforming of naphtha by small amounts of sulfur, deposited in situ from molecules such
as, for example, thiophene. With platinum-rhenium/alumina the amount used is smaller
and it is added in a controlled way, before the reforming process has been started. Sulfur
treatment was found to be beneficial even with pure platinum catalysts, leading to a lower
hydrogenolysis and a higher yields of aromatic, but it is absolutely essential with
platinum-rhenium/alumina [153-165]. Without sulfur these catalysts show a too high

640

chapter 13

hydrogenolytic activity, until this starts to decay slowly due to the carbonaceous deposits
formed at high temperatures [163-165]. Under reforming conditions alumina is active
through its acid sites, the high number of which is due to the presence of chlorine (see
above). Obviously we have to investigate what are the roles of the individual components
in a catalyst containing platinum, rhenium, sulfur, chlorine and alumina, all together.
To have an idea which aspects of the results presented below should be inspected
most closely, let us start by listing the various suggestions formulated in the literature to
explain the superiority in reforming of the Pt-Re-S, C1/A1203 catalysts.
1) Rhenium ions can be built in the structure of alumina and once there are quite stable to
reduction. It is known that metal cations in alumina can serve as an anchor for platinum or
platinum-rhenium particles, making them more stable against sintering. Thus this could be
a positive role for rhenium in the catalyst [166].
2) Rhenium ions modify the surface of alumina in such a way that coke is formed either
in smaller amounts or it is more easily removable by hydrogen, or during regeneration by
air and hydrogen [ 124,159-165, 169-171 ].
3) Rhenium modifies the availability and the reactivity of hydrogen on the surface, by
influencing such phenomena as ' spill-over' of hydrogen [ 165,172].
4) Sulfur forms during sulfurization of the catalyst very strong Re-S and less strong Pt-S
bonds. Therefore,
and Pt~

the working surface of alloy particles would be formed by Re-S, Pt-S

The presence of sulfur partitions the surface into very small platinum

ensembles or even single platinum atoms, which are sufficiently active in dehydrogenation
of the feed for the bifunctional mechanism to operate freely, but which cannot be covered
by continuous carbon layers. The metallic function is mostly limited to aromatization via
dehydrogenation. Some isomerization accompanies it [155,156].
5) Partitioning of the platinum surface into small ensembles is indeed known to suppress
the formation of graphitic layers [173]. This effect too, has been suggested to operate on
platinum-rhenium/alumina catalysts and has been assumed to contribute to the superior
properties of these catalysts [156,174]. It has been shown [174] that most of the carbon is
deposited on the support and chlorine plays a role in anchoring the coke precursors.
However, the smaller amount of carbonaceous species which is on the metallic surface is
at the end of the day much more detrimental to the activity of the catalysts. When
rhenium is present, the residues have a higher H/C ratio [174]. Formation of graphitic
coke has been proven to be a structure-sensitive reaction [175,176] so that there is a quite
firm basis of evidence supporting this effect of rhenium.
6) It is still very popular to speculate that rhenium and sulfur both exert a ligand effect on
the adsorption of hydrogen atoms or hydrocarbon fragments on platinum [92,150]. For
example, it is said that rhenium donates its electrons to platinum, but it is partially
counteracted by electron shifts from platinum or rhenium towards sulfur [164]. One can
find remarks in the open literature and patents in which it is clearly assumed that shifts of

Reactions of alkanes and reforming of naphtha

641

electrons are mediated by the support. Such shifts have to explain the effect of other
additives to the Pt-Re/AI203 catalysts. However with regard to these ideas some scepticism
is not misplaced. Concluding this inventory, we would like to direct the attention of our
reader to some review papers, where a similar attempt has been made to list the possible
modes by which rhenium acts [177,178].
All the first five points have a sound basis in experimental results; one can trust
them all to play some role in reforming reactions. However, the picture is not yet
complete, because all these theories did not sufficiently acknowledge the bifunctionality of
Pt-Re-S,C1/AI203 catalysts. To complete the picture we shall present some results of a
study [179] on the specific role of bifunctionality. Figure 40 compares the variation in
yields, which are proportional to rates, in the n-hexane reactions on sulfurized Pt-Re/AI203
and Pt-Re,C1/AI203 catalysts as the rhenium content is varied [115,179].
,

, =

~oo I

S*II

I0~\

50-~

!/~

2.5-

5'O

"hRe

100

///'cr'''" ~

"-~9
.-~ .......~.

50

"h Re

100

figure 40
Left: yields (proportional to rates)for the various groups of reactions for sulfided
chlorine-free catalysts versus the rhenium content of Pt-Re/A1203. Indicated with arrows;
the physical mixture Pt/S/AI203 + Re/S/AI203. Yields in hydrogenolysis and cyclization with
the physical mixture catalyst have been omitted for reasons of clarity. They are similar to
those found for 50%Pt-50%Re/A1203.
Right: group selectivities for the sulfided chlorine-free catalyst as a function of the
rhenium content of Pt-Re/S/AI203. Indicated with arrows are the results obtained for the
physical mixture of Pt/S/AI203 + Re/S/AI203. (Symbols as in figure 39)

All catalysts contained l%wt of metals and were prepared by chemical impregnation of
alumina. Platinum nitrate, made from H2Pt(OH)6, and H3ReO5 were the metal precursors in
the preparation of chlorine-free catalysts. Figure 39 demonstrates that there is an effect of

642

chapter 13

chlorine on the yields but this does not influence the conclusions we are going to make.
The question to be answered now is whether in the chlorine-free catalysts the acid centres
of the alumina have some role. The answer can be obtained by selective poisoning
experiments using pyridine. Pyridine is irreversibly bound, under the conditions applied
[179], to the acid centres on the alumina, but it can be removed from the metallic part
almost completely. Therefore, by using pyridine one can make visible which part of the
reforming reactions is related to the acid centres, manifesting itself by irreversible
disappearance of activity, and which part can take place on metal sites, without any
assistance of the support; this is the slowly regenerating part of the reactions.
When pyridine is administered to the sulfur and chlorine-free Pt-Re/A1203, yields of
all products of the n-hexane/H 2 reactions fall, but two groups of reactions recover to a
great extent when admission of pyridine is stopped: dehydrocyclization and hydrogenolysis
[179,180]. In contrast to these reactions, isomerization reactions are hit most by pyridine,
and their suppression is irreversible. This is reflected by activity patterns as shown in
figure 41.
1.0

-3.0

to--O-o- I
20
Y
0.5
O

o
t(h )

2'o

rrt. I::~'R1DINE

-10

t (h )

}.

-0

o
t(h)

figure 41
Yields for a presulfided chlorine-free 25%Pt-75%Re/A1203 catalyst for the various groups
of reactions. Treac t .= 623K. First part: after reduction (in hours), 2nd part: an increasing
amount of pyridine added to the gas flow (ml), 3rd part: after the addition of pyridine has
been stopped (t in hours), 4th part." after a reduction for 5 hours at 673K (t in hours).
Symbols as in figure 39.

When the chlorine-free Pt-Re/AlzO 3 catalysts were sulfurized by thiophene (3 hrs. on


stream with 8% thiophene in H2 stream at 1 atm at 570K, followed by

15hrs. on stream

with pure hydrogen at 670K to remove weakly bound sulfur and hydrogenate carbon
species), the selectivity pattern changed from that shown on the left side of figure 42 to
that on the right side of the same figure.

Reactions of alkanes and reforming of naphtha

643

100
S'I,

loo 1

S('/,) /

/
/
/

50 ~

50 L

ar

, 9 " ".~],

-r

5~3

"/,Re

100

.~.1~

% Re

figure 42
Left: group selectivities as functions of the rhenium content for the chlorine-free catalysts
Pt-Re/Al203. Right: Group selectivities for the presulfided chlorine-free catalysts as
functions of the rhenium content in Pt-Re/S/AI203. Indicated with arrows are the results
obtained for the physical mixture of Pt/S/AI203 + Re/S/AI203.
Symbols: squares: hydrogenolysis; circles: isomerization; triangles: cyclization.

Sulfur brings about a very dramatic change in the selectivity pattern. While in the sulfurfree catalysts the selectivity to isomerization decreases monotonically when going from
pure platinum to rhenium, it increases in the series of sulfurized catalysts. We observe in
figure 42 that rhenium-rich sulfurized catalysts show an important suppression of
hydrogenolysis and dehydrocyclization, the two reactions which we identified (fig.41) as
taking place on the metallic surface. The reader will notice that without sulfur these are
very active catalysts for hydrogenolysis. The picture of yields in figure 40 leads us to
conclude that sulfur suppresses all reactions, both metallic and bifunctional, but suppression is less pronounced for isomerization. At 525K isomerization is largely an acidcatalyzed reaction on catalysts for which alumina is used as a support. It is possible that
suppression of the skeletal reactions on the metallic surface is accompanied by a slight
enhancement of the dehydrogenation reactions on metallic sites, from which effect the
selectivity to isomerization on acidic sites can also profit.
The same approach as used earlier in studies on platinum-rhenium catalysts has
been applied to platinum-iridium catalysts [180-181] and the same conclusions concerning
the role of sulfur and bifunctionality have been arrived at.
The overall picture of bifunctionality and of sulfur effects applies also with other
hydrocarbons; however, it is interesting to see some differences in points of detail. For
example, much of the aromatization of methylcyclopentane to benzene occurs on the
acidic centres of alumina [182]. This is true for platinum, platinum-iridium, platinumcobalt and platinum-copper catalysts [182]. With neohexane (2,2 dimethylbutane), sulfur

644

chapter 13

causes a switch from a strongly prevailing metal function to bifunctional behaviour [115].
With sulfurized catalysts, 2,3-dimethylbutane, almost absent when metal activity prevails,
becomes the most prominent product. This switch is most dramatic with chlorine-containing Pt-Re-S/A1203 catalysts.
Let us return to the basic information on the platinum-iridium catalyst. This very
important naphtha reforming catalyst is one of the few industrial catalysts where one can
speak about a rational design, in contrast to the more usual 'trial and arror' approach.
Sinfelt et al. [46,183] at the EXXON laboratory, New Jersey, systematically studied the
hydrogenolytic properties of metals and observed that iridium is very active, its deactivation slow and regeneration by burning off carbonaceous residues easy. On the other hand, it
was known that carbonaceous deposits are responsible for a great part of the deactivation
of the Pt/A1203 catalysts. Therefore, by putting some iridium into platinum it was intended
to decrease deactivation. Since iridium would be diluted in these catalysts by the much
less active platinum, the hydrogenolytic activity of iridium was expected to be suppressed
and should not harm the high selectivity to aromatization. The expectation was perfectly
fullfilled and the catalysts appeared to be very useful. The activity of the new platinumiridium catalysts on a volume basis was about three times as high as that of standard
platinum/alumina or platinum-rhenium/alumina catalysts, and moreover there was a very
high yield of aromatic, ensuring production of a high octane-number gasoline. Note that
these catalysts too are always used in a sulfurized state. Results of the very comprehensive
research performed are described in Sinfelt' s monograph [ 183] and key papers [ 185-188].
Sinfelt et al. studied platinum-iridium catalysts with a maximum of 1%wt platinum
and 1%wt iridium, in a high metal dispersion, i.e. with metal clusters in which practically
all atoms were exposed to the gas phase. This study also revealed the limits of the present
science in characterization of such catalysts. Some of the mono- and bimetallic catalysts
showed H/M (M = Ir or Pt) and CO/M ratios higher than one. With CO/M it is most
probably caused by the formation of stoichiometric or substoichiometric carbonyls, but
with hydrogen the reason is less obvious [184-186]. EXAFS studies showed that bimetallic
catalysts contain mixed clusters and a very thorough analysis of EXAFS spectra revealed
that the metals mix, but the solution is not homogeneous. Catalysts with 10wt% iridium
and 10%wt platinum could be also studied by XRD and they appeared to be solution
alloys [183,185,186]. It has been also seen that a pronounced gas-induced segregation of
components is caused by oxygen or other corrosive gases. In particular, with catalysts
consisting of large bimetallic particles, the restoration of the state before segregation can
be a lengthy process. M6ssbauer spectroscopy has also been used, with 57Fe as a probe.
This technique showed too that clusters consist of both platinum and iridium [187]. Using
hydrogen (0.97 atm) and ethane (0.03 atm) mixtures, it was shown that catalysts containing 0.3wt% iridium and 2.7wt% platinum gave a hydrogenolysis rate that was about
seven times lower than that for pure iridium catalysts, although the total amount of metals

Reactions of alkanes and reforming of naphtha

645

was ten times larger. This indicates that platinum suppresses the activity of iridium.
Those who pioneered research on platinum-iridium catalysts paid a great deal of
attention to whether or not the two metals were completely mixed; they even expressed
surprise when that happened to be the case. The reason for their doubts was that the older
metallurgical literature reported phase segregation and immiscibility at certain platinumiridium ratios.. However, one must not trust the older literature in all details; at the time
the work was done, vacuum technique was much below the present standards and
moreover a sealed quartz tube was used for long annealing procedures at very high
temperatures, since both platinum and iridium are high-melting metals. However, it is
known that a) quartz is permeable for oxygen at temperatures of 1300K and above; b)
traces of oxygen in sealed tubes stimulate segregation of iridium in the form of an oxide.
On the other hand the preparation of platinum-iridium bimetallic catalysts usually comprises some steps which stimulate alloy formation, such as mixing of metal precursors and
reduction by wet hydrogen, so that alloys can be formed and are indeed observed.
Conditions suitable for alloy formation are already created in process of drying the
precursors [188]. As a study by Raman spectroscopy showed, the presence of chlorine in
the system also stimulates mixing upon reduction [189]. This knowledge can now be used
to an advantage in choosing proper precursors. Regeneration of catalysts used in reforming
leads, due to the oxidation step, to the separation of components, and introduction of
chlorine-containing gases into the process of regeneration can help to restore good mixing
of the metals on the surface of alumina.
Growth of knowledge concerning platinum-iridium alloys is in many respects very
similar to the case of platinum-rhenium alloys [153]. Ramaswamy et al. [190] reported
that adding iridium to platinum only makes the catalysts worse from the point of view of
reforming. The sulfur-free alloy catalysts are too active in hydrogenolysis and only when
modified by sulfur do the alloys show properties superior to those of pure platinum on
alumina catalysts: this statement has been confirmed [151,191-193]. The similarity
between platinum-rhenium and other catalysts will be discussed below. Figure 43 shows
the activity of (Pt+Ir)/~-A1203 catalysts for hydrogenolysis of cyclopentane as a function
of iridium content; the reader will have already noticed how general is this picture. We
shall see it again for the iridium-osmium system, and what is found for cyclopentane is
also found for butane hydrogenolysis. An explanation based on assumed electronic
structure changes was favoured [151], but the possibility that the maxima in the activity
curves just mean an optimal composition and number of mixed ensembles was not
explicitly excluded: this seems to be at the moment the most attractive explanation.

646

chapter 13

Ar
15

figure 43
Activity of (Pt+Ir)/o~-Al203 (lO3.mol h-lm -2 metal) as a function of % iridium. Hydrogenolysis
of cyclopentane. T = 463K, Pc = 0.1 attn., PH2
= 0.9 atm [1511.

"o

2~

so

7~s

~6o% I r

Let us turn to the problem of selectivities and the influence thereon of the
sulfurization of alumina-supported catalysts. This can be seen in figure 44 [194a].

100 ~
Pt-Re/AI203

100

S%

S%

Pt-Re/AI203 .S
t

50-~

50

figure 44
Selectivities in the reacti-

100

Pt-Ir/AI203

so R~% 1do

1 0 0 /-rP t _ i r / A i z 0 3

5%

S%~

50

50

00
50 Ir% 100
100S%Pt-Co/ , . . ~ ~ 50

5'0 Re% 100

50 Ir% 100
S%10

50 ~ ~ 1 AI203

'I,"

50 Co% 100

,'~
50 Co% 100

ons of n-hexane with


hydrogen at 620K, standard reaction

conditions

[194], as functions of
catalyst
composition
(at%). On the left: indicated catalysts, sulfur
free. On the right: the
same
catalysts,
after
standard
sulfurization
(see the text). Symbols as
in figure 39. Notice the
crucial difference induced by sulfur.

Reactions of alkanes and reforming of naphtha

647

Conclusions derived from figure 44 are straightforward. Addition to platinum of a metal


active in hydrogenolysis always leads to an increase in selectivity to hydrogenolysis (left
side of figure 45). In this respect rhenium, iridium and cobalt behave exactly the same.
When sulfur is added to the catalyst the same effect again occurs with all systems:
isomerization selectivity increases and cyclization selectivity decreases. With rhenium-rich
and cobalt-rich sulfurized catalysts, the selectivity to hydrogenolysis is lower than that of a
sulfurized platinum/alumina catalyst. Only with a sulfurized Pt-Ir/A1203 catalyst is there a
slight increase in the selectivity to hydrogenolysis. To understand these similarities and
differences, one has to examine the thermochemistry of sulfide formation [155]. Thermochemical data (see table 9) show clearly [194b] that rhenium, iridium or cobalt placed in a
platinum surface capture sulfur preferentially and hold it more firmly than does platinum
alone. With iridium, as with platinum, deactivation by sulfur is less complete and therefore
the contribution to hydrogenolysis by iridium in iridium-platinum sulfurized catalysts is
not completely removed by sulfur. One must not forget that the activity of pure iridium in
hydrogenolytic cracking is several orders of magnitude higher than that of platinum, so
that even a small fraction of iridium atoms not covered by sulfur can induce a fast
hydrogenolysis.
table 9
Enthalpy of formation - AHf of some sulfides

-AHf, 298K in kJmo1-1


Pt S
Pt S 2

82
110

Re S 2
Re S 3
Re 2 S 7

178
208
450

Ir2 S 3
Ir S 2

105
133

Co S

358
119
153

Co 3 S4

Co $2

Important additional information on the working of these catalysts can be obtained


when, instead of only selectivities, rates also are compared [194]. This comparison has
been performed with a series of home-made catalysts and with two commercial catalysts
(EUROPT-3 and a Ketjen platinum-rhenium catalyst). If the measured activity at 620K of

648

chapter 13

each sulfur-free catalyst is put equal to 100, a picture is obtained with the same but now
sulfided catalysts, as shown in figure 45. In all cases, the rate of hydrogenolysis is
suppressed very strongly. The rate of isomerization is suppressed in the most pronounced
manner by sulfur with the catalyst which already shows a high selectivity to isomerization
before sulfurization. On the other hand, with catalysts which show the lowest selectivity to
isomerization before sulfurization (Ir, Pt-Re), the rate of isomerization, and thus not only
the selectivity, is higher on sulfided catalysts than on those that are sulfur-free. These
facts, and the knowledge [152] that with neohexane only 2,3-dimethylbutane is produced
by sulfided catalysts, have of course, led to the suspicion mentioned above that, in spite of
the relatively low temperature in these model reactions (620K), acid-catalyzed reactions

must play an important role with sulfided catalysts.

1.

120

[ s-

100

EE t

80-

80-

60

60-

~,0

t.0

20-

20
r"
Pt

Ir Pt-Ir
1WT%

.=_
Ru

[ PYRIDINE-FREE

100
%

bifunct.
mech.

==L
Pt

Pt-Re

0.3 IWT%
Euro Pt 3

Pt

Pt-Ir
IWT%

Ir

Ru

Pt

Pt-Re

0.3 IWT %
Euro Pt 3

figure 45
Left: relative rates (100 = given catalyst in the sulfur free state) after sulfurization
Right: relative rates (100 = given catalyst after sulfurization) after poisoning by pyridine:
the irreversible loss of particular activity indicates the extent of the bifunctional mechanism. In both parts: the left bar: hydrogenolysis, the right bar: isomerization.

To establish definitely the role of acid centres in the overall conversion of


hydrocarbons on various catalysts, poisoning by pyridine has been used [179-182]. When
the steady state of the catalysts was achieved after about 20 hours (let us call it state 'A'),
pyridine was administrated in pulses, until a new steady state was achieved (state 'B').
The reaction was then continued without pyridine for about 4 hours (state 'C' achieved), a
period which was followed by reduction in pure H 2 at 620K for about 5 hours (state 'D').
When the reaction was restarted thereafter with standard reaction mixtures and at a

Reactions of alkanes and reforming of naphtha

649

standard temperature, a new steady state was finally achieved (designed as state 'E'). The
results obtained in this way with n-hexane as reactant are evaluated in the following
manner.
An idea already mentioned above is followed, based on the paper by Robschlager
et al. [195], that pyridine poisons both metallic as well as acidic sites of a bifunctional
catalyst, but at suitably chosen temperatures the former is reversible and the latter
irreversible. By using this criterion the following picture of the importance of reactions
catalyzed by acid centres can be established. With sulfur-free catalysts, the activity in the
final state 'E' is always 90% or more of the steady state at 'A'. The poisoning can thus be
considered as essentially reversible and the activity of any of the sulfur-free catalysts is
almost purely due to the metallic sites. With sulfur-containing catalysts, the situation is
quite different. In some cases hydrogenolysis is still poisoned reversibly, see figure 45, left
bars, but with the two commercial catalysts the poisoning of hydrogenolysis is to a large
extent irreversible. Isomerization (right side of the bars) is in all cases more influenced by
irreversible pyridine poisoning than hydrogenolysis and with the two commercial catalysts
(0.3%Pt/A1203 the so-called EUROPT-3 and the platinum-rhenium catalyst EUROPT-4),
which are specially acidified during their production, the suppression of isomerization by
pyridine is most dramatic. Comparing of sulfur-free with the sulfided catalysts thus reveals
a clear shift in selectivities, from reactions catalyzed by metallic sites to reactions
catalyzed by acidic sites. Speculations on the existence of the shift demonstrated by figure
45 can already be found in some earlier literature [157]. The shift in selectivities caused
by sulfur deserves some more comment.
Reading the literature, one finds no indications that irreversibly bound sulfur (the
form which played a role in the above experiments) would increase the number or acid
strength of acid sites on the alumina. Our preliminary experiments with Pt-Re/A1203
confirmed this conclusion [196-197]. However, sulfur certainly decreases the number of
sites active in 'metallic' reactions, ((de-)hydrogenation, most of hydrogenolysis and
cyclization, and a part of isomerization). Dehydrogenation is necessary for acidic isomerization and yet the activity in isomerization increases in some cases after sulfurization
(figure 45). It means that these catalysts have possibly a lower total dehydrogenation
activity, but nevertheless after sulfurization produce more of those intermediates which
induce isomerization, obviously on account of the intermediates of other elementary
reactions. The speculation is that this shift in selectivity for the formation of various
intermediates should be due to something like that shown by the very schematic picture in
figure 46. Intermediates multiply bound to metallic sites are replaced on sulfided catalysts
by unsaturated intermediates which are more loosely bound, migrate more easily and can
form carbenium ions on acidic sites; these in their turn induce isomerization and other
reactions catalyzed by acid centres.
Recently, hydrogenolysis of ethane has been compared with hydrogenolysis of n-

650

chapter 13

hexane on Ir/SiO2 and Ir + S / S i O 2 catalysts. It appeared that the activity of iridium in nhexane reactions is suppressed by sulfur, but it is not eliminated completely. However, the
activity in ethane hydrogenolysis is suppressed totally by the same standard sulfurization.
It has been concluded from this observation [197] that the multiply-bound surface intermediates derived from the o~B binding, which is the only possibility for ethane, require a
larger ensemble of metal surface atoms in order to be formed than the complexes of the
metallocyclobutane structure, or n-complexed species.

\/

\/

c
c/
ii

~c /
i
-x--

~c
-

II
*

c/

/ c

I\

--c

\c /

I\

\ / \ /

--c--c

II
.)(_

c
c
/ \ / \

~I

/c
c____ c ~

C~c/

t c/c

-~ c

L -)(- l
C~C

-)(figure 46
A shift from multiply bound to weakly bound intermediates represents a transition from
'externally' to 'internally' unsaturated molecules.

The role of acidic centres in maintaining the isomerization activity was proven in
another way by experiments in which platinum-rhenium and platinum-copper catalysts
were prepared on inert silica. After a standard sulfurization procedure these catalysts
having no acid centres completely lost their activity, while catalysts prepared on alumina as we have seen above- did not.

13.4

Various combinations containing two transition metals

13.4.1 Combinations containing platinum


A very extended collection of results accompanied by a thorough analysis has been
published by Oliver et al. [198], who studied platinum-rhodium alloys on inert silica as

Reactions of alkanes and reforming of naphtha

651

support. By using propane, butane, 2-methylpropane (isobutane), 2-methylbutane and 2,2dimethylpropane (neopentane) they were able to establish the following reactivity sequence
of bonds between primary, secondary and tertiary carbon atoms on the series of alloy
catalysts:
(24)

CII -Cii > C I - CII > Cii - Cii I > C I - Cii I > C I - C i v

Since rhodium is much more active for hydrogenolysis than platinum (2x102 - 103 times,
according to the molecule used) an attempt was made first to correlate the activity with
the presence of certain rhodium ensembles. If their necessary size is m, the probability of
finding such ensembles is -- xmp.h, where x is the atomic fraction. However, this approach
appeared not to be successful. Then they considered m i x e d ensembles. For example, when
m is taken as 3 and the activity of the various mixed ensembles is assumed to vary as
indicated in table 10, the number of various ensembles can be calculated by using the
binominal statistics. Results are shown in table 10:
table 10
Characteristics of triatomic rhodium-platinum ensembles
Type of ensemble
Rel.no ensembles

Rh 3
X3Rh

Rh2Pt
3X2Rh(1-XRh)

RhPt2
3X~(1-XRh)2

Activity of ens.

aRh

a~'3~ aVSpt

aV3Rh

a~'Pt

Pt3
(1-XRh) 3

apt

The total activity of all ensembles taken together, is then given by a sum A (see below)
where R is a constant, m is the required ensemble size (number of atoms) and i shows
how many of these atoms are Pt-(i-atoms) and how many Rh-(m-i atoms)-atoms. Table 10
shows how the triatomic (m=3) ensembles are described, but in a full analogy description
can be extended to any value of m. Results obtained by using equation 25 are compared
with experimental results in figure 47.
A

i -m
E

m -i

m-i

R,,m.X ~

_ i]3
(1-xm) i aRh3 apt

(25)

i=0

In an earlier paper from the same laboratory, it was pointed out that platinum is
much better in forming 3CffT complexes (iso-units) than rhodium, which excells by its
propensity to form 2Cc~7 complxes, the ideal intermediates of hydrogenolysis [199].
Therefore, an exact quantitative analysis might be very complicated. Moreover, pure
platinum and pure rhodium tend to split molecules in different places; for example, with

652

chapter 13

n-alkanes platinum prefers internal splitting (see section 13.1) while rhodium is active in
multiple or terminal splitting.

-2

0
,

platinum (%)
40

80
v

-2

-3

m=6
.,.a

-3

-4

-4

E -4

.-"

"--4

"..4

-5

-5

m--8 ~

2
9

-6

()
figure 47

'

40

'

8'0

'

platinum (%)

Comparisons of the activities of the bimetallic catalysts with values calculated for
ensembles of m atom (indicated) assuming that the activity of an ensemble varies
geometrically with its composition: lines are calculated, points are experimental results:
(a), 2-methylpropane at 463K; (b), 2,2-dimethylpropane at 463K; (c), butane at 398K.

On the platinum-rhodium system information is also available concerning evaporated films and reactions of unbranched C4,C5 and C6-alkanes [200]. A faster than linear
decrease in hydrogenolysis rate was observed when platinum was added to rhodium (see
above). Only on platinum-rich films there was some isomerization, but traces of benzene
were found even on pure rhodium.
A closely related combination

is platinum-ruthenium. The difference is that the

latter system does not form a continuous series of homogeneous alloys, but the formation
of several phases can be accompanied by surface segregation of elements and this all can
lead to a pronounced heterogeneity of alloy particles. In spite of these potential difficulties, monotonically varying and mutually selfconsistent selectivity patterns have been
observed with Pt-Ru/SiO 2 and Pt-Ru/A1203 catalysts in reactions of n-heptane. The results
on the selectivity [201] for silica-supported alloys are shown in figure 48, the total
tumover number decreasing with increasing ruthenium content. This could indicate that at

653

Reactions of alkanes and reforming of naphtha

653K the catalyst worked under conditions of severe selfpoisoning, since ruthenium alone
is much more active in hydrogenolysis of ethane than platinum. Other workers [202] have
studied their catalysts most probably under conditions of a more limited selfpoisoning, as
can be expected for reactions of cyclopentane/H 2 at 573K, they observed what others have
observed with platinum-rhenium and platinum-iridium also: the total turnover number has
a maximum in the middle of the composition range.
S

1
X

0.75

SH
05 (

SC

SA

o 2~
N

Ru/(Pt + Ru)
25

50

15

IO0

figure 48
Conversion of n-heptane on (Pt-Ru)/Si02 catalyst. Selectivities: A aromatization (toluene);
C cyclization (ethyl- and dimethylcyclopentane); 1.isomerization; H hydrogenolysis (T =
653K).

Diaz et al. [203] have studied platinum-ruthenium alloys in order to establish how
the different mechanisms and intermediates (3C or 5C, see chapter 1) contribute to the
overall isomerization of hexanes and how these contributions can be influenced by mixing
these metals. Also the methylcyclopentane ring splitting is found to be influenced by alloy
formation. The most interesting point is that small amounts of ruthenium (strongly diluted
in platinum, thus not forming large ruthenium ensembles) strongly activate platinum
without inducing pronounced hydrogenolysis, so typical for pure ruthenium.
When discussing above the results concerning platinum-iridium, we have already
presented some information on the platinum-cobalt alloys, which in some respects recall
the platinum-iridium combination. Platinum-cobalt on alumina catalysts reduced at 973K,
were also studied by Zyade et al. [204]; some of their results obtained with 2-methylpentane at 573K are shown in table 11. We can see the frequently observed synergism in total
activities and suppression of isomerization on platinum by cobalt, and an enhancement of
terminal splitting and suppression of formation of 3C~7 complexes (iso-units) when cobalt

654

chapter 13

is added. This information also confirms the picture which we have already constructed on
the basis of results obtained with Pt-Re, Pt-Rh, Pt-Ru and Pt-Ir catalysts.
Let us now turn our attention to another closely related alloy, the platinum-nickel
system. As we have seen in chapter 4, there is extensive high quality information available
on this system and this is important for good understanding of the catalytic results. Most
of this information on surface composition as well as on catalysis by these alloys has been
produced by various French laboratories.
Platinum-nickel alloys have been studied with alumina [205] and active carbon
[206,207] as supports. An extended discussion of catalytic effects of alloying platinum
with nickel can be found [207]. For example, it was suggested that the frequently observed
synergetic activity maximum found here for isomerization and hydrogenolysis of neopentane (2,2-dimethylpropane) by alloys with about 10-20% platinum in nickel should be
ascribed to the electronic structure effects [207]. Very fine details of differences in
catalytic behaviour which are induced by alloying can be established by monitoring the
conversion of labelled molecules, for example, of 2-methyl(213C)pentane. The presence of
a molecule of 3-methyl(313C)pentane in products witnesses (see chapter 1) the operation of
a 5C(cyclic-) mechanism; the so-called bond-shift mechanism operating through 3C-~7complexes is related to the occurence of 3-methylpentanes labelled at another position in
the molecule. When the 13C label appears at a place which cannot be reached by either of
these mechanisms, the reason for it is most likely a repetitive conversion during one
sojourn of a molecule on the surface.
Some information does exist conceming the comparison of the kinetics of
hydrogenolysis of ethane [207] measured on 1 . 1 % Pt/SiO 2 and on a ( 1 . 1 % Pt + 0.7 %
Fe)/SiO2 catalyst. By a detailed analysis based of the models currently used (see section
13.1.2), it was concluded that ethane is adsorbed in a more dehydrogenated state at a low

PH2/Pc2H6ratio on Pt-Fe/SiO2 than on pure platinum catalysts.


An interesting subject from the point of view of fundamental research is the
platinum-palladium system, with which we can see again how the individual alloy
components impose their fingerprint on the product distribution and the overall selectivity
pattern. Palladium prefers, as do nickel, cobalt and ruthenium, terminal splitting (demethylation), while on pure platinum at low temperatures 3C0~7-complexes and internal fission
are favoured. By using hexane isomers, it has been shown [209] that the contribution of
terminal cracking is decreased by platinum in a clearly non-linear way. This could be
explained by stating that, as with the other metals mentioned above, terminal splitting
requires large ensembles of palladium (or of nickel, cobalt or ruthenium in other alloys),
although it was believed [209] that the results actually reveal the operation of electronic
structure effects in alloying.

M
c-,,.

i1-

i'

ir'

o o
t~.~

,oo

r,j

o~

,r ....

+
[ ]"
v

ir

o"
I

(:~

c~

It

,r

C,~

~'~

t~

r,j

,,_,

L"'-

~.~

9 ,~

(:~

t~

c~

v'~

c~

,~

v'~

~)

c~

~,

i~

i
,

i
J

!i

II

II

II

~s

,-:~

I.~

Reactions of alkanes and reforming of naphtha

--

655

656

chapter 13

Platinum-palladium alloy catalysts have also been studied by another group of


workers [210,211]. With alloys in the form of evaporated films and supported on silica,
reactions of n-pentane, n-hexane and 2,2-dimethylpropane (neopentane) with hydrogen
were monitored. Several interesting but not yet fully explicable effects were observed. For
example, in n-pentane conversion the selectivity to n-pentane cyclization was higher for
alloys than for either of the pure metals. The ratio of the selectivity to 1,6-closure to that
of the 1,5-closure was also higher on alloys than on pure metals. In the reaction of
hydrogen with n-hexane, 3-methylpentane and methylcyclopentane, saturated C6-products
prevailed as expected at higher hydrogen pressures, while alkenes and mainly benzene
prevailed in hydrogen-lean mixtures. The overall activity exhibited a minimum at about
the equimolar composition, where isomerization selectivity also showed a maximum. The
selectivities appeared to be influenced by i) hydrogen pressure, ii) the extent of selfpoisoning by carbonaceous deposits iii) alloy composition, so that all three factors are interrelated. This makes the picture very complex.
With platinum-palladium alloy catalysts the isomerization of 13C-labelled molecule
has also been studied on A1203-supported catalysts [209], and the variation between the
external splitting on palladium-rich and-amongst others- internal splitting on platinum-rich
alloys was again reported.
Catalysts containing platinum and a sixth group metal (molybdenum, tungsten)
have been studied by several groups [212-215]; explanation of the effects found is not
easy. It is usually uncertain whether complete reduction of molybdenum or tungsten oxides
occurs under conditions used, and this leads to additional problems, since unreduced
oxides can create or demolish acid centres on silica or alumina supports, respectively.
Yermakov et al. [212] studied ethane hydrogenolysis and found that Pt-Mo/SiO2 catalysts
showed a high activity which they ascribed to electronic structure effects. The same type
of catalyst (Pt-Mo/SiO2) was studied by Leclercq et al. [213] using n-butane hydrogenolysis; they ascribed the observed behaviour either to ligand effects and/or formation of
mixed ensembles. Y-zeolites which are active catalysts in their own right have been used
as supports for platinum and molybdenum containing catalysts [214]. Hydrogenolysis of nbutane was also faster on mixed metal catalysts than with monometallic catalysts with this
support. Kuznetsov et al. [215] studied reactions of neopentane (2,2-dimethylpropane) and
found a synergetic enhancement of the hydrogenolytic activity with bimetallic Pt-Mo/SiO2
catalysts. Results obtained with tungsten containing platinum catalysts show the same
general picture [215a]: addition of tungsten suppresses isomerization and promotes the
hydrogenolysis. However, for platinum-chromium catalysts a suppression of hydrogenolysis have been reported [215b].
With respect to the various catalysts discussed above (Ni-Pt; Pt-W, Pt-Mo) the
following results could be of interest. When nickel was evaporated on W(100) or W(110)
planes, effects were observed on the hydrogenolysis of butane, which were ascribed to the

Reactions of alkanes and reforming of naphtha

657

strained structure of nickel on W(100) [216]. Such effects of strain can in principle also
exist in some dispersed catalysts, when the components differ in all respects so much as
for example nickel and tungsten. In the same laboratory [217] nickel evaporated onto
platinum was observed to give a non-linear increase the activity in ethane hydrogenolysis
with surface coverage by nickel. Such non-linearity is often considered as an indication of
cluster formation.
Garin and Maire who studied the reaction of C6-alkanes on a long list of platinumbased alloys, published a summarizing review on this subject [218]. They concluded that
there are two main surface complexes inducing hydrogenolysis, the 2Ct~7- and 3Co~7complexes. The sequence of bonds in the order of decreasing reactivity is thought to be
[2181:

2CtxT: C I - Cii ,
3Cr C I - CiIi,

(26)
(27)

C I - CI, C i i - Cii
C i i - CN, C i i - Cii I

This sequence of reactivities differs from that which we saw above for platinum-rhodium
catalysts, but the reason for the discrepancy is not known. A summary of results in
graphic form is in figure 49.

iso
mode
C2mode

Ni

Pt-Pd

2-

Pt -W

Pd
8
I~

I -

%, ~, / / P t-N i
~%1,,/
II

iC o
Pt -Co

0.3

Pt-lr
Ir

0.2-

Pt-Ru

0.10.06
I

0.1

012 0.3 0.z.

3 45

~0

figure 49
2-Methylpentane (2MP) hydrogenolysis reactions [218]. The
ratio on the horizontal axis used
to display the results shows
which bond splitting is easier:
terminal splitting leading to 2methylbutane (supposed to be
induced by C2, oV3-adsorption
mode) or splitting at the isounit, leading to n-pentane. The
value of the parameter called
'iso-mode', used on the vertical
axis, is derived from propane
plus pentane concentrations;
that of the parameter called 'C2unit' from isopentane plus isobutane concentrations.

658

chapter 13

13.4.2 Transition metal alloys without platinum


Iridium-cobalt catalysts have been studied [219] using the reactions of alkanes, for
example, of 2-methylpentane. By using isotopic labelling the contribution of 3Cc~Tcomplexes to isomerization increased, as did hydrogenolysis activity, when cobalt was
added to iridium. This is the same effect as seen with platinum-cobalt catalysts. An
explanation assuming changes in the electronic structure of the components, induced by
alloying was also favoured here.
Iridium-ruthenium catalysts have been studied with n-butane hydrogenolysis.
Ruthenium is a metal with a strong preference for terminal and multiple splitting (section
13.1.1), while iridium at low temperatures prefers internal splitting. Alloy catalysts show a
transition from one characteristic pattern to the other [220].
Iridium-osmium catalysts with alumina as support show the same activity-composition dependence as several other platinum alloys (see above). This can be seen in figure
50 [151], where the results for benzene hydrogenation are shown. In these cases, the
possibility of mixed ensembles operating should not be forgotten. In any case this idea
seems to be the most universally applicable explanation.

r3~l

figure 50
Activity
of
(Ir+Os)/ct-Al203
(mol.h-l.g 1 of metal) as a

20

function of at% osmium. Hydrogenation of benzene. T = 373K,


0

P8 = 0.05 atm., Pm = 0.95 attn.

10

[1511.

50

I '1~

100
%Os

13.4.3 Multimetallic cocktails


The patent literature contains several suggestions to use more than two transition
metals in industrial catalysts. Your authors are not aware of an industrial application of
these multimetallic catalysts, but it might be nevertheless useful to know which combinati-

Reactions of alkanes and reforming of naphtha

659

ons have already been suggested. Without trying to be complete, we mention here that, for
example there is a proposal to use the almost obvious combination Pt-Re-Ir/S/A1203. In
these catalysts there is 0.1-0.2 wt% of each transition metal and 0.05-0.15 wt% of sulfur
[221]. EXXON laboratory owns patents [222] which suggest a combination such as: 2%wt
Rh, 0.01-2.0 Ir, 0-2wt Re, 0-2wt% Sn, 0-3.5wt% of halides, 0.01-2wt% Pt, with A1203 as
support. The support has been moreover modified by adding silica and alkaline earth
metals. The UOP company has another idea [223], namely, 0.05-1wt% Pt, 0.1-2wt% Ge,
Pd (as oxides) 0.05-1wt% Rh, Ir, Ru, Co or Ni 0.5-1.5wt% C1, the rest being A1203. We
have seen in the foregoing section what impact the presence of particular components has,
so that the reader can make for himself a picture of the complicated function of these
multimetallic catalysts. Owing to low loadings and the presence of elements which
increase the dispersion of metals on alumina, the fraction of metal atoms exposed would
approach unity in most of such catalysts.

13.5

Platinum-tin and other related catalysts

13.5.1 Platinum-tin catalysts


These are very important reforming catalysts and their popularity seems to be ever
growing [224]. The whole available arsenal of physical methods (chapter 2) has been
applied here, but the definitive picture of the working catalyst, including the mechanism of
reforming with them, is not yet available. However, the volume of knowledge gathered up
to now is very respectable and the albeit slowly emerging picture is very interesting.
In some respects the situation with these alloys is very similar to that with
platinum-rhenium catalysts. We can find papers which try to explain the total behaviour of
these catalysts by just considering the platinum-tin alloy formation [224,225], while others
prefer to assume the presence of unreduced tin and speculate on its effects on anchoring
metal particles, acidity of the support or coking and other side reactions [226,227]. With
the available knowledge one can conclude that most likely all these effects are operating
under reforming conditions. However, before presenting the catalytic results we shall turn
to the question of the state of the catalyst.
A very powerful tool to investigate the state of platinum-tin catalysts is MOssbauer
Spectroscopy. This is applicable also under in situ conditions which makes it particularly
attractive. Moreover, it is rather a short-range-order method than a long-range-order-XRDmethod and this is also an important factor for catalysts with very small particle size.
Analysis is performed by standard procedures of curve-fitting [226,228,233] and it is not a
trivial procedure. However, the results are in many respects rewarding.
The picture emerging from the most recent literature [231-233] is that one or more

660

chapter 13

alloy phases containing Sn ~ are present. In addition tin is present in bi- and tetravalent
oxides, and in chlorides, when chlorine containing precursors are used. The fraction of
platinum present in the alloy phase increases with increasing tin concentration. The first
portions of tin added are probably firmly bound to defects of alumina, dissolved in the
structure, etc. and are not easily reduced (if at all). The presence of chlorine in the
precursor facilitates alloy formation [232-234], as we saw also with iridium- or rheniumcontaining catalysts. M6ssbauer spectroscopy also revealed that reduction of tin is easier
and can even approach completness when silica is used as a support [233].
Another bulk technique, the common X-ray diffraction (XRD), can be used only
with model catalysts containing a higher platinum and tin content than the commercial
reforming catalysts [235]. The dominant phase seen is the stable intermetallic compound
PtSn, but not all tin is present as an alloy. Layers of tin aluminate may also be formed.
EXAFS can be easily performed with both the platinum and tin edges (L and K,
respectively) [234]. Again, the Pt-Sn alloy phase and Sn-O bonds have been observed.
The use of XPS should have been most informative, but it was not. Some papers
claimed the presence of exclusively ionic tin [236-238] and only under the pressure of
evidence coming from other techniques and due to the development of the XPS technique
and data evolution, did opinions started to change. Some later papers report the presence
of zero-valent tin, most of it forming an alloy with platinum [239-243].
Having in mind the summary of the results, namely that catalysts contain platinum
and tin in various phases, most of platinum being alloyed, we can now review the catalytic
behaviour. In advance the following can be said. In principle the presence of tin compounds can modify acid-catalyzed reactions and the presence of tin ions on platinum can
influence aromatization/cracking selectivity ratio. Moreover, it is not known whether the
Pt-Sn alloy is active in adsorption of hydrogen and hydrocarbons at all. It is quite possible
that only clusters of platinum in platinum-tin alloy particles or small platinum particles
(unalloyed platinum) are active, but not the alloy in the form of an intermetallic compound.
Early patents [224] as well as the early open literature point out what is the
advantage of using platinum-tin catalysts instead of pure platinum catalysts: at higher
temperatures and under more severe conditions, i.e. a higher hydrocarbon: hydrogen ratio,
the hydrogenolytic splitting is on alloy catalysts lower than on Pt/A1203 and the aromatization higher [244-246]. There are indications that the harmful coke precursors created on
the metallic surface, these precursors being dehydrogenated fragments of reacting molecules, are in the case of platinum-tin catalysts more mobile and less firmly attached to the
metallic surface. Further, it is claimed that coke is mostly deposited on the support and
that the metallic function of the catalyst stays preserved for a longer period of time [247].
Notice that a similar picture has also been suggested above for Pt-Re-S/A1203 catalysts.
Suppression of coke formation under more severe conditions is reported by several other

Reactions of alkanes and reforming of naphtha

661

workers and can be considered as very well established. The effect is more pronounced
with, say, n-heptane than with lower hydrocarbons [248-251], again in agreement with the
idea that the presence of tin suppresses the detrimental side reactions.
While the facts concerning platinum-tin catalysts seem to be clear, explanations
have been suggested based on a broad variety of ideas. While Bacaud et al. [226] stress
the effect on acidity, Beltramini et al. [251] deny the existence of this effect. Those who
concentrate their attention on the metallic part of the catalysts can be subdivided into two
groups. The first group offers an explanation based on the ensemble size effect: smaller
platinum ensembles should show a much lower hydrogenolysis rate, while dehydrogenation together with dehydrocyclization should be influenced to a much lesser extent
[225,250,251]. The second group believes that the beneficial effect of tin on the platinum
comes either from electronic interactions of zero-valent tin on platinum, or from the
interaction of platinum particles with tin ions [227,244,249].
An important paper for all future attempts to explain the behaviour of platinum-tin
alloys has been published by Coq et al. [253a], who studied Pt-M/AI203 catalysts, with M
being Sn, Pb, Ge, A1 or Zn. They used a series of hydrocarbons - n-hexane, methylcyclopentane and 2,2,3,3-tetramethylbutane. In particular, the last mentioned molecule was
interesting: with pure platinum catalysts, small particle catalysts show a 3Cc~y-dimethylation, leading to 2,3,3-trimethylbutane, while large particles split the molecule into two
isobutanes (2-methylpropanes). When a small amount of metal-M-precursor was adsorbed
on the metal (an organometallic compound was decomposed by reaction with hydrogen
preadsorbed on the platinum), then with M/Pt lower than 0.2 in atomic contents, small
particles started to behave as large ones. It means that the first dose of added M-metal,
occupied special sites abundantly present on small particles, likely at edges and defects. Of
the metals mentioned, tin seemed to be the most efficient. Essentially the same results
were obtained with ruthenium and rhodium as the transition metal component [253b,c].
Several groups of workers have compared tin as an additive to platinum, with some
other elements used as additives: Pb, Ge, Mo, W. Re. Tin seems to be the best additive for
enhancement of aromatization [250,254,255].
The problem with all platinum-containing alloy catalysts is regeneration. An
oxidative treatment clearly separates the components, but since it is possible to mix the
components as oxides or chlorides, regeneration is in principle possible. The fact that
platinum-tin catalysts are used in practice shows that the proper way of regeneration has
been already found by industrial laboratories. Tin and platinum form in an ionized state
several very stable coordination complexes and thus the anchoring of one component on
the 'template' formed by ions of the other component can operate here to our benefit

[2581.
The beneficial effect which tin has on platinum catalysts can also be achieved with
germanium. Germanium similar, to tin, exists in several valencies: Ge(0), Ge(II) and

662

chapter 13

Ge(IV), also forming platinum-germanium intermetallics [259-263]. With germaniumcontaining catalysts an enhanced isomerization is sometimes reported, but otherwise the
effects on selectivities and the suggested explanation of effects are mostly analogous to
those for platinum-tin systems.
Platinum on alumina catalysts can be also modified by elements such as tellurium
or antimony [256]. In some refineries, compounds containing antimony are indeed added
to the feedstock and they have a beneficial effect on reforming and fluidized catalytic
cracking, but the traces of these components appearing then in gasoline are a problem for
car exhaust catalysts.
The propane to benzene reaction can be performed on zeolites loaded with various
sp metals, semi-metals or their oxides (Ga, Te, Bi). It has been reported that platinum-tin
intermetallics are very good catalysts for this aromatization in combination with ZSMzeolites [264].
13.5.2 Other catalysts containing tin and related elements
Results of a detailed study of the iridium-tin system have been reported [265].
Conclusions

are very similar to those derived from the platinum-germanium and plati-

num-tin systems: C-C bond fission is suppressed and aromatization of higher hydrocarbons
is enhanced. The first point is of particular interest since iridium clearly prefers the 2Cc~Bsplitting at all temperatures, while platinum only at a very high temperature, being at low
temperatures mostly active through the formation of 3Cot"f-complexes. Nevertheless a great
similarity is observed.
Cyclohexane can be aromatized or the ring can be opened and further hydrogenolyzed. Addition of tin to iridium has an effect on these reactions as illustrated by table 12
[265].

table 12
Cyclohexane conversion at 526K; rates of formation (a.u.)
Benzene

Methane+Ethane Prooane

Butane/Pentane/Hexane

Ir/SiO 2

8.7

1.7

0.5

0.65

0.7

4.2

Ir+Sn/SiO 2

1.3

Reactions of alkanes and reforming of naphtha

13.6

663

Reforming of naphtha
Catalytic reforming of naphtha [266-269] forms a part of the oil refining network

and it stands in this network next to other units such as distillation, dehydrosulfurization,
cracking, etc. A scheme of a reforming unit is shown in figure 50.
COMP

-'~ Gas

R1

R3

H-E
NAPHTH

PI
Gasoline

figure 51
A simplified scheme of a reforming unit in a three-reactor configuration (R1-R3). The
position of the compressor (COMP), separation subumit (SEP), pump (P), heat exchange
(H-E) and re-heaters (H) are indicated.

It has been mentioned above (13.1) that reforming is a complex of dehydrogenation


processes accompanied by hydrogenolysis, isomerization and dehydrocyclization to the C 5
and the C 6 substituted alkyl rings and aromatics. These reactions occur partially on the
acidic support and partially on the metal, therefore we speak of bifunctionality of the
catalysts. Octane numbers of some hydrocarbons are given in table 13. Obviously, isomerizations, leading to branched hydrocarbons and aromatization, are the most desired
reactions. Let us mention some pieces of information concerning chemical and technological aspects of naphtha reforming; more details can be found in the literature [266-270].
As far as the chemistry is concerned, many isomerizations can reach equilibrium
and the higher the temperature the more branched molecules are present at equilibrium.
This is seen in table 14.

664

chapter 13

table 13
Octane numbers of pure hydrocarbons
research octane number (clear)

Hydrocarbon
Paraffins:
n-butane

113

n-pentane

62

n-hexane

19

n-heptane

n-octane

- 19

2-methylhexane

41

2,2-dimethylpentane

89

2,2,3-trimethylbutane

113
Aromatics:

Naphthenes(alkanes)"
n-butane

113

methylcyclpentane

107

benzene

99

n-pentane

62

1,1-dimethylcyclopentane

96

toluene

124

n-hexane

19

cyclohexane

110

1,3-dimethylbenzene 145

n-heptane

methylcyclohexane

104

isopropylbenzene

n-octane

- 19

ethylcyclohexane

43

1,3,5-trimethylbenzene 171

table 14
Equilibrium distribution of iso-alkanes at 750K.
number of branches [269]
hydrocarbon

C6

25

C7

15

45

30

45

40

C8

15

45

25

132

Reactions of alkanes and reforming of naphtha

665

Further, thermodynamic calculations reveal that for temperatures above 600K and
hydrogen pressures between 1 and 3 MPa the content of cycloalkanes is below 5%.
However, at 800K, aromatics are almost the only components favoured, except with C6
hydrocarbons. The rate of all reactions is favourably enhanced by an increased temperature, too, so that reforming is performed at somewhat high temperatures. However, the upper
limit of temperature is mainly dictated (apart from thermodynamic limitations) by the
extent of side reactions. These are induced by a too deep dehydrogenation of the feed,
leading to coke formation. A compromise is to work at about 750-800K. Due to coking,
deactivation of the catalysts slowly progresses and the temperature of reforming must be
increased to keep the conversion high. After a certain period of such operation, the
catalyst has to be regenerated semicontinuously or swing-reactors have to be used (i.e. two
parallel reactor units of which one is always being regenerated). After several regenerations, the catalyst has to be replaced, but this is a very costly operation [270-272]. Alloy
catalysts such as Pt-Re-S/A1203 have the life span 5-10 times as long as the originally
used Pt-S/A1203 catalysts. In all respects the new catalysts appear to be superior [273].
A typical three reactor configuration is shown in figure 51. It indicates that reactors
of unequal size are used and the units comprise external heaters (H), heat exchanger (HE), pumps (P) and compressors, the latter being necessary in recycling of feedstocks. The
reactors have an operating temperature of about 770K, and the pressure of about 30 atm.
hydrogen, needed at the inlet of the process, is produced by the process itself. The process
also supplies hydrogen for other processes of oil refining.
The inventors of new, robust and stable catalysts based on platinum and rhenium
also caused development in the engineering of the process: improvements which have been
introduced have been described [270-273]. The units presently used also contain provisions
for re-activation by oxygen and re-dispersion of the metallic components, using oxygen
and chlorine compounds. During these steps the support is also cleaned and recharged with
chlorine [274].

References
1
2
3a
b
C

C.Kemball, Proc.Roy.Soc., A 207 (1951)541; A 217 (1953) 376


J.R.Anderson, C.Kemball, Proc.Roy.Soc., A 223 (1954) 361
C.Kemball, Adv.Catal. 11 (1959)223
C.Kemball, Bull.Soc.Chim.Belg., 67 (1958) 373
C.Kemball, Catal.Rev., 5 (1971) 93
G.C.Bond in "Catalysis by Metals", Academic Press, London, (1962)
G.C.Bond in "Heterogeneous Catalysis", Oxford Sci.Publ.,Clarendon Press, Oxford,
2nd ed. (1987)

666

5a

b
C

d
e

f
g
6

9
10
11
12
13
14

15
16
17
18
19
20
21a

chapter 13
E.H.van Broekhoven, V.Ponec, Progr.Surf.Sci., 19(2) (1985) 351
E.H.van Broekhoven, V.Ponec, J.Molec.Catal., 25 (1984) 109
V.Ponec in "The Chemical Physics of Solid and Heterogeneous Catalysis"
(editors: D.A.King, D.P.Woodruff) Elsevier, Vol.4 (1982) 365
A.E.Shilov, Pure.Appl.Chem., 50 (1978) 725
R.L.Burwell Jr., Annu.Rev.Phys.Chem., 15 (1964) 131
R.L.Burwell Jr., Catal.Rev., 7 (1972) 25
L.Guczi, K.Ujszaszi, React.Kin.Catal.Lett., 8 (1978) 489
T.L.Cottrell in "The Strength of the Chemical Bond", Butterworths, London, (1958)
H.A.Skinner, Adv.Organomet.Chem., 2 (1964) 49
"Handbook of Chemistry and Physics", C.R.C.Press, Cleveland
"Comprehensive Organometallic Chemistry" (editors: G.Wilkinson, F.G.A.Store,
E.W.Abel) Pergamon Press, N.Y., Vol.6 (1982) 42
C.T.Mortimer in "Reaction Heats and Bond Strengths", Pergamon Press, Oxford
(1962)
R.T.Anderson in "Chemical Bond and Bond Energy", Academic Press, N.Y. (1971)
S.Cerny, M.Smutek, F.Buzek, J.Catal., 47 (1977) 166
M.Smutek, S.Cerny, J.Catal., 47 (1977) 178
P.Tetenyi, Proc.6th Int.Congr.on Catal.London, 1976, Chem.Soc.London, Vol.1
(1977) p.456
R.Evans, M.Polanyi, J.Chem.Soc., (1947) 252
M.E.Winfield, Australian J.Sci.Rev., A4 (1951) 385
S.T.Ceyers, Science, 249 (1990) 113
M.A.McKervey, J.J.Rooney, N.G.Samman, J.Catal., 30 (1963) 330
M.Vogelzang, M.J.P.Botman, V.Ponec, Faraday Disc.Chem.Soc., 72 (1981) 44
Z.Paal, P.Tetenyi, React.Kin.Catal.Lett., 12 (1979) 131; 7 (1977) 39
Z.Paal, P.Tetenyi, Nature, 267 (1977) 234
Z.Paal, P.Tetenyi, Appl.Catal. 1 (1981) 9
M.W.Vogelzang, V.Ponec, J.Catal., 111 (1988) 77
O.E.Finnlayson, J.K.A.Clarke, J.J.Rooney, J.Chem.Soc.Faraday Trans.I, 80 (1984)
191
C.O.Donohoe, J.K.A.Clarke, J.J.Rooney, J.Chem.Soc.Faraday Trans.I, 76 (1980)
345
H.Zimmer, P.Tetenyi, Z.Paal, J.Chem.Soc.Faraday Trans.I, 78 (1982) 3573
H.Zimmer, M.Dobrovolszky, P.Tetenyi, Z.Paal, J.Phys.Chem., 90 (1986) 4758
G.Leclercq, L.Leclercq, R.Maurel, J.Catal., 50 (1979) 87
J.E.Germain, 'Catalytic Conversion of Hydrocarbons', Academic Press, N.Y.
(1969)
H.Pines in "The Chemistry of Catalytic Hydrocarbon Conversions", Academic

Reactions of alkanes and reforming of naphtha

b
C

d
22

23
24
25
26
27

28
29
30
31

32

33
34
35

36

667

Press, N.Y. (1981)


H.Pines, C.T.Goetschel, J.Org.Chem., 30 (1965) 3530, 3546
L.Nogueira, H.Pines, J.Catal., 70 (1981) 404
H.Pines, J.W.Demkinski, J.Org.Chem., 30 (1965) 3537
Z.Paal, P.Tetenyi, Acta Chim.Acad.Sci.Hung., 53 (1967) 175, 193; 58 (1968) 105
V.A.Kazanski, Kin.Katal., 8 (1967) 977
I.V.Kalechitz, Kin.Katal., 8 (1967) 1114
B.H.Davis, J.Catal., 46 (1977) 348
P.Tetenyi, L.Guczi, Z.Paal, Acta.Chim.Acad.Sci.Hung., 83 (1974) 37
G.R.Lester, J.Catal., 13 (1969) 187
S.Engels, Z.Chem., 20 (1980) 212; 261
Z.Paal, P.Tetenyi, Acta Chim.Acad.Sci.Hung., 53 (1967) 193; 72 (1972) 277
O.V.Bragin, V.G.Tovmasyan, I.V.Gostunskaya, A.L.Liberman,
Izv.Akad.Nauk.USSR,Ser.Khim., 1340 (1977)
E.G.Christoffel, K.H.Robschlager, Ing.Eng.Chem.Prod.Res.Dev., 17 (1978) 331
F.Janowski, F.Wolf, Chem.Techn., Leipzig, 29 (1977) 273
A.J.Silvestri, P.A.Naro, R.L.Smith, J.Catal., 14 (1969) 386
W.L.Callender, S.G.Brandenberger, W.K.Meerbott, Proc.5th Int.Congr.on Catal.
Miami Beach 1972, North Holland Publ., vol.2 (1973) p.1265
J.H.Sinfelt, J.C.Rohrur, J.Phys.Chem., 65 (1961) 978
L.Guczi, A.Frennet, V.Ponec, Acta Chim.Hung., 112 (1983) 127
A.Cimino, M.Boudart, H.S.Taylor, J.Phys.Chem., 58 (1954) 796
J.H.Sinfelt, W.F.Taylor, Trans Faraday Soc., 64 (1968) 3086
J.H.Sinfelt, J.Catal., 27 (1977) 468
M.Boudart, AICHE Journal, 18 (1972) 465
A.Frennet, L.Degols, G.Lienard, A.Crucq, J.Catal., 35 (1974) 18
L.Guczi, K.Matusek, A.Sarkany, P.Tetenyi, Bull.Soc.Chim.Belg., 88 (1979) 494
A.A.Andreev, B.S.Gudkov, S.L.Kiperman, Kinet.i Katal., 2 (1968) 35
H.Matsumoto, Y.Saito, Y.Yoneda, J.Catal., 22 (1971) 182
C.Kemball, Ann.N.Y.Acad.Sci., 213 (1973) 204
G.A.Martin, B.Imelik, Surf.Sci., 42 (1974) 154
A.Frennet, G.Lienard, A.Crucq, L.Degols, J.Catal., 53 (1978) 150
B.S.Gudkov, L.Guczi, P.Tetenyi, J.Catal., 74 (1982) 207
F.Garin, F.G.Gault, J.Am.Chem.Soc., 97 (1975) 4466
F.G.Gault, V.Amir-Ebrahimi, F.Garin, P.Garayre, F.Weisang, Bull.Soc.Chim.Belg.,
88 (1979) 475
R.B.Levy, M.Boudart, Science, 181 (1973) 547
S.T.Oyama, G.L.Haller in "Specialist Periodical Reports: Catalysis", Chem.Soc.
London, 5 (1982) 333

668

37
38

39
40
41a
b
42
43
44
45
46

47
48
49
50
51
52
53
54a
b
C

d
55
56

chapter 13

P.N.Ross, P.Stonehart, J.Catal., 48 (1977) 42


A.Frennet, G.Leclercq et al., Proc.10th Int.Congr.on Catal. Budapest, 1992,
Elsevier, vol.B (1993) p.927
Hak Soo Kom, Chae Ho Shin, G.Bugli, M.Bureau-Tardy, G.Djega-Mariadassou,
Appl.Catal. A 119 (1994) 223
J.R.Anderson, B.G.Baker, Proc.Roy.Soc., London, A271 (1963) 402
J.R.Anderson, W.R.Avery, J.Catal., 5 (1966) 446
V.A.Kazanski, Uspekhi Khimii, 17 (1948) 655
F.M.Dautzenberg, J.C.Platteeuw, J.Catal., 19 (1970) 41
N.E.Shephard, J.J.Rooney, J.Catal., 3 (1964) 129
J.K.A.Clarke, C.O'Donohoe, Proc.R.Ir.Acad. 89b (1989) 405
H.Matsumoto, Y.Naito, Y.Yoneda, J.Catal., 22 (1971) 182
V.Ponec in "Electronic Structure and Reactivity of Metal Surfaces", (editors:
E.G.Derouane, A.A.Lucas), Plenum Press, N.Y.(1976) p.539
J.L.Carter, J.A.Cusumano, J.H.Sinfelt, J.Catal., 20 (1971) 223
E.H.van Broekhoven, Ph.D.thesis, Leiden University, The Netherlands, 1985
E.H.van Broekhoven, V.Ponec, J.Molec.Catal., 25 (1984) 109
J.H.Sinfelt, Adv.Catal., 23 (1973) 91
J.H.Sinfelt, AICHE Journal, 19 (1973) 673
S.Palfi, A.Sarkany, P.Tetenyi, J.Chem.Soc.Faraday Trans. 1 (1981) 77
R.S.Dowe, M.C.Gray, D.A.Whan, C.Kemball, J.Chem.Soc.Chem.Commun., (1971)
883
A.Sarkany, P.Tetenyi, React.Kin.Katal.Lett., 9 (1978) 315
H.Matsumoto, Y.Saito, Y.Yoneda, J.Catal., 19 (1970) 101;
Z.Paal, P.Tetenyi in "Specialist Periodical Reports: Catalysis", Chem.Soc.London, 5
(1982) 80
J.P.Brunell, R.E.Montarnal, A.A.Sugier, Proc.6th Int.Congr.on Catal., London,
1976, Chem.Soc.London, vol.2 (1977)p. 844
G.Leclercq, L.Leclercq, R.Maurel, J.Catal., 50 (1979) 87
R.Merta, V.Ponec, Proc.4th Int.Congr.on Catal., Moscow, 1968, Akademiai Kiado
Budapest, vol.2 (1971) p.53
A.Sarkany, P.Tetenyi, React.Kin.Catal.Lett., 9 (1978) 315
Z.Paal, Adv.Catal., 29 (1980) 273
V.A.Kazanski in "Mechanism of Hydrocarbon Reactions" (editors: F.Marte,
D.Kallo), Akademiai Kiado, Budapest, (1975) 15
G.B.Young, G.M.Whiteside, J.Am.Chem.Soc., 100 (1978) 5808
F.G.Gault, Adv.Catal., 30 (1981) 1
J.M.Muller, F.G.Gault, J.Catal., 24 (1972) 361
O.Zahraa, F.Garin, G.Maire, Faraday Soc.Disc., 71 (1981) 45

Reactions of alkanes and reforming of naphtha


57
58
59
60
61
62
63
64a
b
65
66
67
68
69a
b
70a
b
71
72
73
74
75
76
77
78
79
80
81
82a
b
83
84
85

669

V.Ponec, Faraday Soc.Disc., 71 (1981) 88


V.Haensel, G.R.Donaldson, Ind.Eng.Chem., 43 (1951) 2102
F.G.Ciapetta, J.B.Hunter, Ind.Eng.Chem., 45 (1953) 147
G.A.Mills, H.Heinemann, T.H.Milliken, A.G.Oblad, Ind.Eng.Chem., 45 (1953) 154
N.D.Zelinski, B.A.Kazanskii, A.F.Plate, Ber.Deutsch Che.Ges., 66 (1933) 1415
B.A.Kazanski, T.F.Bulanova, Izv.Akad.Nauk.USSR Ser.Khim., (1947) 29; (1948)
406
V.Ponec, W.M.H.Sachtler, J.Catal., 24 (1972) 250
A.Roberti, V.Ponec, W.M.H.Sachtler, J.Catal., 18 (1973) 381
J.M. Beelen, V.Ponec, W.M.H.Sachtler, J.Catal., 28 (1973) 376
M.W.Vogelzang, M.J.P.Botman, V.Ponec, Faraday Disc.Chem.Soc., 72 (1981) 33
V.Ponec, W.M.H.Sachtler, Proc.5th Int.Congr. on Catal., Miami Beach, 1972, North
Holland Publ. vol.1 (1973) p.645
J.H.Sinfelt, J.L.Carter, D.J.C.Yates, J.Catal., 24 (1972) 283
J.H.Sinfelt, J.Catal., 29 (1973) 308
J.H.Sinfelt, Acc.Chem.Res., 10 (1977) 15
J.A.Dalmon, G.A.Martin, J.Catal., 66 (1980) 214
A.Sarkany, J.Catal., 97 (1986) 407
P.Balaz, I.Sotak, R.Domansky, Chem.Zvesti, 32 (1978) 75, 444
T.S.Cale, J.T.Robertson, J.Catal., 94 (1985) 289
D.F.Ollis, H.Taheri, AICHE Journal, 22 (1976) 1112
W.Langenbeck, D.Nehring, H.Dreyer, Zeit.Anorg.Allgem.Chemie, 304 (1960) 37
G.M.Schwab, J.Block, D.Schulze, Angew.Chemie, 71 (1959) 101
F.Solymosi, Catal.Rev., 1 (1968) 233
H.D.Lanh, Ng.Khoai, H.S.Thoang, J.Volter, J.Catal., 129 (1991) 58
V.K.Duplyakin, Yu.I.Yermakov, A.S.Belyi, V.S.Alfeev, B.N.Kuznetsov, Kin.
Kataliz., 19 (1978) 1605
R.L.Moss, D.Pope, H.R.Gibbens, J.Catal., 46 (1977) 204
R.L.Moss, D.Pope, H.R.Gibbens, J.Catal, 55 (1978) 100
W.Juszczyk, Z.Karpinski, Z.Paal, J.Pielaszek, Proc.10th Int. Congr. on Catal., Budapest, 1992, Elsevier, vol.C (1993) p.1843
C.Visser, J.G.P.Zuidwijk, V.Ponec, J.Catal., 35 (1974) 407
Z.Karpinski, J.Catal., 77 (1982) 118
J.K.A.Clarke, J.F.Taylor, J.Chem.Soc.Faraday Trans.I, 72 (1976) 917
A.D.O Cinneide, F.G.Gault, J.Catal., 37 (1975) 311
C.Henriques, P.Dufresne, C.Marcilly, T.Ramoce Ribeiro, Appl.Catal., 21 (1986)
169
J.H.Sinfelt, Y.L.Lamn, J.A.Cusumano, A.E.Barnett, J.Catal., 42 (1976) 227
C.R.Helms, J.H.Sinfelt, Surf.Sci., 72 (1978) 229

670

86
87

88
89
90
91
92
93
94
95
96
97a

d
98
99
100
101
102
103
104
105
106
107
108
109
110

chapter 13
J.H.Sinfelt, G.H.Via, F.W.Lytle, J.Chem.Phys., 72 (1980) 4832
K.Christman, G.Ertl, H.Shimizu, J.Catal., 61 (1980) 397
H.Shimizu, K.Christman, G.Ertl, J.Catal., 61 (1980) 412
J.C.Vickerman, K.Christman, G.Ertl, J.Catal., 71 (1981) 175
C.H.F.Peden, D.W.Goodman, J.Catal., 104 (1987) 347
C.H.F.Peden, D.W.Goodman, Ind.Eng.Chem.Fundamentals, 25 (1986) 58
G.Haller, D.E.Resasco, J.Wang, J.Catal., 84 (1983) 477
J.H.Sinfelt, Int.Revs.Phys.Chem., 7 (1988) 281
M.W.Smale, T.S.King, J.Catal., 119 (1989) 441; 120 (1990) 331
D.W.Goodman, C.H.F.Peden, J.Chem.Soc.Faraday Trans, 83 (1987) 1967
M.Sprock, X.Wu, T.S.King, J.Catal., 138 (1992) 617
G.C.Bond, Xu Yide, Proc.8th Int.Congr.on Catal., Berlin, 1984, Verlag Chemie,
vol.IV (1984) p.577
J.C.Kempling, R.B.Anderson, Ind.Eng.Chem.Process dev., 11 (1972) 146
S.Galvagno, C.Crisafulli, R.Maggiore, A.Giannetto, J.Schwank, J.Thermal Anal., 32
(1987) 471
S.Galvagno, C.Crisafulli, R.Maggiore, G.R.Tauszik, A.Giannetto, J.Thermal Anal.,
30 (1985) 611
C.Crisafulli, R.Maggiore, G.Schembari, S.Scire, S.Galvagno, J.Molec.Catal., 50
(1989) 67
I.W.Bassi, F.Garbassi, G.Vlaic, A.Marzi, G.R.Tauszik, G.Cocco, S.Galvagno,
G.Parravano, J.Catal., 64 (1980) 405
S.Galvagno, J.Schwank, G.Parravano, J.Catal., 61 (1980) 223
J.R.H.van Schaik, R.P.Dessing, V.Ponec, J.Catal., 38 (1975) 273
H.C.de Jongste, F.J.Kuijers, V.Ponec, "Preparation of Catalysts" (editors:
B.Delmon, P.A.Jacobs), Elsevier, (1976) 207
R.P.Dessing, V.Ponec, React.Kinet.Catal.Lett., 5 (1976) 251
A.F.Kane, J.K.A.Clarke, J.Chem.Soc.Faraday Trans.I, 76 (1980) 1640
J.K.A.Clarke, I.Manniger, T.Baird, J.Catal., 54 (1978) 230
J.W.A.Sachtler, G.A.Somorjai, J.Catal., 81 (1983) 77
R.C.Yates, G.A.Somorjai, J.Catal., 103 (1987) 208
H.C.de Jongste, F.J.Kuijers, V.Ponec, Proc.6th Int.Congr.on Catal.,London, 1976,
Chem.Soc.London, vol.2 (1976) p.915
H.C.de Jongste, V.Ponec, Proc.7th Int.Congr.on Catal.,Tokyo, 1980, Kodansha/E1sevier, vol.A (1981) p. 186
M.J.P.Botman, H.C.de Jongste, V.Ponec, J.Catal., 68 (1981) 9
A.J.den Hartog, P.J.M.Rek, V.Ponec, J.Chem.Soc.Chem.Commun. (1988) 1470
A.J.den Hartog, A.G.T.M.Bastein, V.Ponec, J.Mol.Catal., 52 (1989) 129
J.K.A.Clarke, K.M.G.Rooney, T.Baird, J.Catal., 111 (1988) 374

Reactions of alkanes and reforming of naphtha


111
112
113
114
115

671

J.J.Rooney, J.Molec.Catal., 31 (1985) 147


T.J.Plunkett, J.K.A.Clarke, J.Catal., 35 (1974) 330
D.I.Hagen, G.A.Somorjai, J.Catal., 41 (1976) 466
Z.Karpinski, J.K.A.Clarke, J.Chem.Soc.Faraday Trans.I, 71 (1975) 2310
A.J.den Hartog, Ph.D.thesis, Leiden University, The Netherlands, 1990
A.J.den Hartog, P.J.M.Rek, M.J.P.Botman, C.de Vreugd, V.Ponec, Langmuir 4
(1988) 1100
A.J.den Hartog, M.Holderbush, E.Rappel, V.Ponec, Proc.9th Int.Congr.on Catal.,Calgary, 1988, Canad.Inst.Chem., vol.2 (1988) p.1174
116 B.Coq, R.Duarte, F.Figueras, A.Rouco, J.Phys.Chem., 93 (1989) 4904
117
K.Foger, J.R.Anderson, J.Catal., 64 (1980) 448
118 J.F.Brunelle, R.E.Montarnal, A.A.Supier, Proc.6th Int.Congr.on Catal., London,
1976, Chem.Soc.London, vol.2, (1976) p.844
119 C.Kim, G.A.Somorjai, J.Catal., 134 (1992) 179
120 D.J.Godbey, F.Garin, G.A.Somorjai, J.Catal., 117 (1989) 144
121
J.K.A.Clarke, J.F.Taylor, J.Chem.Soc.Faraday Trans.I, 71 (1975) 2063
122 M.J.Kelley, R.J.Freed, D.G.Swartzfager, J.Catal., 78 (1982) 445
J.B.Peri, J.Catal., 52 (1978) 144
123
P.Malet, G.Munuera, A.Caballero, J.Catal., 115 (1989) 567
124 R.J.Bertolacini, R.J.Pellet in "Catalyst Deactivation" (editors: B.Delmon, G.F.Froment), Elsevier, Amsterdam, (1980) 73
C.Bolivar, H.Charcosset, R.Frety, M.Primet, L.Tournayan, C.Betizeau, G.Leclerq,
125
R.Maurel, J.Catal., 45 (1976) 163
126 M.F.L.Johnson, V.M.LeRoy, J.Catal., 35 (1974) 43L
127 M.F.L.Johnson, J.Catal., 39 (1975) 487
128 A.N.Webb, J.Catal., 39 (1975) 485
129 B.D.McNicol, J.Catal., 46 (1977) 438
130a N.Wagstaff, R.Prins, J.Catal., 59 (1979) 434
b
H.Charcosset, AICHE Journal, 23 (1983) 187
131
B.H.Isaacs, E.E.Petersen, J.Catal. 43 (1982) 43; 85 (1984) 8
132 C.Bolivar, H.Charcosset, R.Frety, M.Primet, L.Tournayan, J.Catal., 39 (1975) 249
133 H.Charcosset, R.Frety, G.Leclercq, E.Mendes, M.Primet, J.Catal., 56 (1979) 468
134 P.S.Kirin, B.R.Strohmeier, B.C.Gates, J.Catal., 98 (1986) 308
135 Xie Youchang, Gui Linlin, Lin Yingjun, Zhao Biying, Yang Naifang, Zhang Yafen,
Guo Qinlin, Duan, Liahyun, Huang Huizhong, Cai Xiachai, Tang Xouchi, Proc.8th
Int.Congr.on Catal.,Berlin, 1984, Verlag Chemie, Weinheim, vol.V (1984) p.147
136 J.Y.Zhang, M.J.P.Botman, V.Ponec, React.Kinet.Catal.Lett,., 26 (1984) 189
137 M.J.P.Botman, Ph.D.thesis, Leiden University, The Netherlands, 1987
138 M.S.Nacheff, L.S.Kraus, M.Ichikawa, B.M.Hoffman, J.B.Butt, W.M.H.Sachtler,

672

chapter 13

J.Catal., 106 (1987) 263


M.A.Pacheco, E.E.Petersen, J.Catal., 96 (1985) 499
R.L.Mieville, J.Catal., 87 (1984) 437
S.M.Augustine, W.M.H.Sachtler, J.Catal., 116 (1989) 184
S.M.Augustine, M.S.Nacheff, C.M.Tsang, J.B.Butt, W.M.H.Sachtler, Proc.9th
Int.Congr.on Catal.Calgary, 1988, Canad.Chem.Inst.Ottawa, vol.3 (1988) p.l190
S.M.Ziemecki, G.A.Jones, J.M.Michel, J.Catal., 90 (1986) 207
C
142 J.Margitfalvi, M.Hegedus, S.Szabo, F.Nagy, React.Kinet.Catal.Lett., 18 (1981) 175
143 G.Munuera, P.Malet, A.Cabellero, Proc.10th Int.Congr.on Catal.,Budapest, 1992,
Akademiai Kiado, Elsevier, vol.A (1992) p.781
144 G.Munuera, private communication
145a G.C.Bond, M.R.Gelsthorpe, J.Chem.Soc.Faraday Trans, I85 (1989) 3763
b
G.C.Bond, J.Molec.Catal., 81 (1993) 99
146 C.Betizeau, G.Leclercq, R.Maurel, C.Bolivar, H.Charcosset, R.Frety. L.Tournayan,
J.Catal., 45 (1976) 179
147 G.C.Bond, R.H.Cunningham, E.L.Short, Proc.10th Int.Congr.on Catal.,Budapest,
1992, Akademiai Kiado, Elsevier, Amsterdam, vol. A (1992) p.849
148
S.M.Augustine, W.M.H.Sachtler, J.Catal., 106 (1987) 417
149 I.H.B.Haining, C.Kemball, D.A.Whan, J.Chem.Res.(M) (1977) 2056; (S) (1978)
364
150 L.Tournayan, R.Bacaud, H.Charcosset, G.Leclercq, J.Chem.Res.(S)(1978) 290; (M)
(1978) 3582
151
G.Leclercq, H.Charcosset, R.Maurel, C.Betizeau, C.Bolivar, R.Frety, D.Januay,
H.Mendez, L.Tournayan, Bull.Soc.Chim.Belg., 88(7-8) (1979) 577
152 M.J.P.Botman, K.de Vreugd, H.W.Zandbergen, R.de Block, V.Ponec, J.Catal., 116
(1989) 467
153 P.G.Menon, J.Prasad, Proc.6th Int.Congr.on Catal.,London, 1976, Chem.Soc.London, vol.2 (1977) p. 1061
154 R.W.Coughlin, K.Kawakami, A.Hasan, J.Catal., 88 (1984) 150, 163
155 W.M.H.Sachtler, J.Molec.Catal., 25 (1984) 1
P.Biloen, J.N.Heller, H.Verbeek, F.M.Dautzenberg, W.M.H.Sachtler, J.Catal., 63
(1980) 112
V.K.Shum, J.B.Butt, W.M.H.Sachtler, J.Catal., 96 (1985) 371; 99 (1986) 126
156 F.H.Ribeiro, A.L.Bonivardi, C.Kim, G.A.Somorjai, J.Catal., 150 (1994) 186
157 P.A.van Trimpont, G.B.Marin, G.F.Froment, Appl.Catal., 24 (1986) 53
158 P.G.Menon, G.B.Marin, G.F.Froment, Ind.Eng.Chem.Prod.Res.Dev., 21 (1982) 52
159 C.A.Querini, N.S.Figoli, J.M.Parera, Appl.Catal., 53 (1989) 53
160 J.M.Parera, C.A.Querini, J.N.Beltramini, N.S.Figoli, Appl.Catal., 32 (1987) 117
161
C.A.Querini, N.S.Figoli, J.M.Parera, Appl.Catal., 32 (1987) 133; 52 (1989) 249

139
140
141a
b

Reactions of alkanes and reforming of naphtha


162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192

673

J.Barbier, E.Churin, P.Marecot, J.Catal., 126 (1990) 228


J.M.Grau, E.L.Jablonski, C.L.Pieck, R.J.Verderone, J.M.Parera, J.Chem.Tech.
Biotechnol., 18 (1987) 105
J.M.Parera, J.N.Beltramini, C.A.Querini, E.E.Martinelli, J,Catal., 99 (1986) 39
E.M.Traffeno, J.M.Parera, Appl.Catal., 28 (1986) 193
Yu.I.Yermakov, Proc.7th Int. Congr. on Catal.,Tokyo, 1980, Kodansha/Elsevier,
vol.A (1981) p.57
Yu.I.Yermakov, B.N.Kuznetsov, J.Molec.Catal., 9 (1980) 13
A.G.Graham, S.E.Wanke, J.Catal., 68 (1981) 1
S.Engels, M.Wilde, Tran Kim Thanh., Z.Chemie, 17 (1977) 10
L.H.Ludlum, R.P.Eischens, preprints of the Am.Chem.Soc.Div.Petrol.Chem.meeting
(1976) 375
R.Burch, A.J.Mitchell, Appl.Catal., 6 (1983) 121
J.Margitfalvi, S.Gobolos, E.Kwaysser, H.Hegedus, F.Nagy, L.Koltai, React.Kinet.Catal.Lett., 24 (1984) 315
A.D.van Langeveld, F.C.M.J.M.van Delft, V.Ponec, Surf.Sci., 134 (1983) 665
S.M.Augustine, G.N.Allameddin, W.M.H.Sachtler, J.Catal., ! 15 (1989) 217
D.W.Blakeley, G.A.Somorjai, J.Catal., 42 (1976) 181
S.M.Davis, F.Zaera, G.A.Somorjai, J.Catal., 77 (1982) 439
J.H.Lunsford, L.W.Zingery, M.P.Rosynek, J.Catal., 38 (1975) 179
W.M.H.Sachtler, J.Molec.Catal., 25 (1984) 1
L.Guczi, J.Molec.Catal., 25 (1984) 13
P.G.Menon, G.F.Froment, J.Molec.Catal., 25 (1984) 59
P.J.M.Rek, A.J.den Hartog, V.Ponec, Appl.Catal., 46 (1989) 213
M.J.Dees, A.J.den Hartog, V.Ponec, Appl.Catal., 72 (1991) 343
M.J.Dees, V.Ponec, J.Catal., 115 (1989) 347
M.J.Dees, M.H.B.Bol, V.Ponec, Appl.Catal., 64 (1990) 279
J.H.Sinfelt in "Bimetallic Catalysts", Wiley & Sons, N.Y. (1983)
B.J.Kip, F.B.M.Duivenvoorden, D.C.Koningsberger, R.Prins, J.Am.Chem.Soc., 108
(1986) 5633
J.H.Sinfelt, G.H.Via, J.Catal., 56 (1979) 1
J.H.Sinfelt, G.H.Via, F.W.Lytle, J.Chem.Phys., 76 (1982) 2779
R.L.Garten, J.H.Sinfelt, J.Catal., 62 (1980) 127
S.C.Chan, S.C.Fung, J.H.Sinfelt, J.Catal., 113 (1988) 164
Y-J.Huang, S.C.Fung, J.Catal., 131 (1991) 378
A.V.Ramaswamy, P.Ratnasamy, S.Sivasanher, A.J.Leonard, Proc.6th Int.Congr.on
Catal.,London, 1976, Chem.Soc.London, vol.2 (1977) 855
R.W.Rice, Kang Lu, J.Catal., 77 (1982) 104
J.C.Rasser, W.H.Beindorff, J.J.F.Scholten, J.Catal., 59 (1979) 211

674

chapter 13

193
194
b

Y-J Huang, S.C.Fung, W.E.Gates, G.B.McVicker, J.Catal., 118 (1989) 192


V.Ponec, Catal.Today, 10 (1991) 251
K.C.Mills in "Thermodynamic Data for Inorganic Sulphides, Selenides and Tellurides", Butterworths, London (1974)
K.H.W.Robschlager, C.A.Emeis, R.A.van Santen, J.Catal., 86 (1984) 1
L.Spruit, M.Sc.thesis, Leiden University, The Netherlands, 1991
M.J.Dees, Leiden University, The Netherlands, unpublished
J.A.Oliver, C.Kemball, Proc.Roy.Soc.London, A 429 (1990) 17
T.ChenWong, L.C.Chang, G.L.Haller, J.A.Oliver, N.R.Scaife, C.Kemball, J.Catal.,
84 (1984) 389
Z.Karpinski, J.K.A.Clarke, Trans Faraday Soc., 71 (1975) 893
G.Blanchard, H.Charcosset, F.Garin, L.Tournayan, Nouv.J.Chimie, 5 (1981) 85
R.Gomez, G.Corro, G.Diaz, A.Maubert, F.Figueras, Nouv.J.Chimie, 4 (1980) 677
G.Diaz, F.Garin, G.Maire, J.Catal., 82 (1983) 13
G.Diaz, F.Garin, G.Maire, S.Alerasool, R.D.Gonzalez, Appl.Catal.A, 124 (1995) 33
G.Zyade, F.Garin, L.Hilaire, M.F.Ravet, G.Maire, Bull.Soc.Chim.France, 3 (1985)
341
E.J.Nowak, R.M.Koros, J.Catal., 7 (1967) 50
J.M.Dominguez, A.Vazguez, A.J.Renouprez, M.J.Yacaman, J.Catal., 75 (1982) 101
A.J.Renouprez, B.Moraweck, B.Imelik, V.Perrichon, J.M.Dominguez-Esquivlo,
J.Jablonski, Proc.7th Int.Congr.on Catal.,Tokyo, 1980, Kodansha/Elsevier, vol.A
(1981) p.173
S.Aeiyach, F.Garin, L.Hilaire, P.Legare, G.Maire, J.Molec.Catal., 25 (1984) 183
F.Garin, G.Maire, J.Molec.Catal., 52 (1989) 147
F.Garin, P.Girard, A.Chaqroune, F.Weisang, G.Maire, Proc.8th Int.Congr.on
Catal.Berlin, 1984, Verlag Chemie, Weinheim, 1984, vol.III (1984) p.405
Z.Karpinski, T.Koscielski, J.Catal., 56 (1979) 430; 63 (1980) 313
T.Koscielski, Z.Karpinski, Z.Paal, J.Catal., 77 (1982) 539
Ibid, Z.Phys.Chem.NF, 111 (1978) 125
Yu.I.Yermakov, B.N.Kuznetsov, Yu.A.Ryndin, J.Catal., 42 (1976) 73
G.Leclercq, T.Romero, S.Pietrzyk, S.Grimblot, L.Leclercq, J.Molec.Catal., 25
(1984) 67
T.M.Tri, J.Massardier, P.Gallezot, B.Imelik, J.Catal., 85 (1984) 244
B.N.Kuznetsov, Yu.I.Yermakov, M.Boudart, J.P.Collman, J.Molec.Catal., 4 (1978)
49
R.W.Joyner, E.S.Shpiro, Catal.Lett., 9 (1991) 239
D.W.Goodman, J.M.White, J.Phys.Chem., 91 (1987) 6669
J.A.Rodriguez, D.W.Goodman, J.Phys.Chem., 94 (1990) 5342
F.Garin, G.Maire, Acc.Chem.Res., 22 (1989) 100

195
196
197
198
199
200
201
202
203
204
205
206
207

208
209
210
211
212
213
214
215

216
217
218

Reactions of alkanes and reforming of naphtha


219
220
221
222
223
224

225
226

227
228
229
230
231
232
233

234
235
236
237
238
239
240
241
242
243

675

P.Esteban Puges, F.Garin, F.Weisang, P.Bernhardt, P.Girard, G.Maire, L.Guczi,


Z.Schay, J.Catal., 114 (1988) 153
H.Hamada, Appl.Catal., 27 (1986) 265
US Patent no. 5,066.632
US Patents nos. 4,966 682; 4,966 878; 4,968 408
US Patents nos. 4,791 087; 4,737 483
F.M.Dautzenberg, German Offenlegunsschrift, 2,121 763 (1971)
F.M.Dautzenberg, H.W.Kouwenhoven, German Offenlegungsschrift, 2,153 891
(1972)
F.M.Dautzenberg, J.N.Helle, P.Biloen, W.M.H.Sachtler, J.Catal., 63 (1980) 119
J.Volter, G.Lietz, M.Uhlemann, M.Hermann, J.Catal., 68 (1981) 42
Yang Weishen, Lin Litru, Fan Yineng, Zhang Jingling, Catal.Lett. 12 (1992) 267
R.Bacaud, P.Bussi6re, F.Figueras, J.M.Mathieu in "Preparation of Catalysts"
(editors: B.Delmon, P.A.Jacobs, G.Poncelet), Elsevier, (1976) 509
H.Lieske, J.Volter, J.Catal., 90 (1984) 96
R.Burch, L.C.Garla, J.Catal., 71 (1981) 360
R.Burch, J.Catal., 71 (1981) 318
R.Bacaud, P.Bussi6re, F.Figueras, J.P.Mathieu, C.R.Acad.Sci.Paris, ser.C, 281
(1975) 159
V.H.Berndt, H.Mehner, J.Volter, W.Meise, Z.Anorg.Allg.Chem., 429 (1977) 47
K.J.Klabunde,Y-X.Li, K.F.Purcell, Hyperfine Interact., 41 (1988) 649
Y-X.Li, Y-F.Zhang, K.J.Klabunde, Langmuir, 4 (1988) 385
Y-X.Li, K.J.Klabunde, B.H.Davis, J.Catal., 128 (1991) 1
V.I.Kuznetsov, A.S.Belyi, E.N.Yarchenko, M.D.Smolikov, M.T.Protasova,
E.V.Zatoloking, V.K.Duplyakin, J.Catal., 99 (1986) 159
N.A.Pakhonov, R.A.Buyanov, E.M.Moroz, E.N.Y.Yurchenko, A.P.Chernychev,
N.A.Zaitseva, G.R.Kotelnikov, React.Kinet.Catal.Lett., 14 (1980) 329
G.Meitzner, G.H.Via, F.W.Lytle, S.C.Fung, J.Phys.Chem., 92 (1988) 2925
R.Srinavasan, R.J.DeAngelis, B.H.Davis, J.Catal., 106, (1987) 449
B.A.Sexton, A.E.Highes, K.Foger, J.Catal., 88 (1989) 466
S.R.Adkins, B.H.Davis, J.Catal., 89 (1984) 371
J.M.Stencel, J.Goodman, B.H.Davis, Proc.9th Int.Congr.on Catal.,Calgary, 1988,
Canadian Chem.Inst., Ottawa, vol.3 (1988)p.1291
Y-X.Li, J.M.Stencel, B.H.Davis, React.Kinet.Catal.Lett., 37 (1988) 273
Y-X.Li, J.M.Stencel, Appl.Catal., 64 (1990) 71
R.Srinivasan, B.H.Davis, Platinum Metal Rev., 36 (1992) 151
H.A.Laitenen, J.R.Waggoner, C.Y.Chan, P.Kirzenztejn, D.A.Asbury, G.B.Hoflund,
J.Electrochem.Soc., 133 (1986) 1586
S.D.Gardner, G.B.Hoflund, D.R.Schryer, B.T.Upchurch, J.Phys.Chem., 95 (1991)

676

244
245
246
247
248
249
250
251
252
253a
b
C

254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269

chapter 13
835
B.H.Davis, G.A.Westfall, J.Watkins, J.Pezzanite Jr., J.Catal., 42 (1976) 247
B.N.Shelesnev, Y.W.Fornishev, M.E.Levinter, Neftekhimiya, 14 (1974) 205
Yu.N.Usov, L.G.Subranova, N.I.Kuskinova, Neftekhimiya, 17 (1977) 69
Lin Liwu, Zhang Tao, Zang Jingling, Xu Zhusheng, Appl.Catal., 67 (1990) 11
J.N.Beltramini, D.L.Trimm, React.Kinet.Catal.Lett., 37 (1988) 293
G.Leclercq, D.Haunay, R.Maurel, J.Chem.Res.(M) (1979) 713
J.Volter, G.Lietz, M.Uhlemann, M.Hermann, J.Catal., 68 (1981) 42
J.Beltramini, D.L.Trimm, Appl.Catal., 31 (1987) 113
B.Coq, F.Figueras,J.Catal., 85 (1984) 197
B.Coq, A.Chaqroune, F.Figueras, B.Nciri, Appl.Catal., A82 (1992) 231
B.Coq, A.Bittar, R.Dutartre, F.Figueras, J.Catal., 128 (1991) 275
B.Coq, A.Goursot, T.Tazi, F.Figueras, D.R.Salahab, J.Am.Chem.Soc., 113 (1991)
1485
Yu.A.Ryndin, J.Lorent, B.N.Kuznetsov, V.I.Kovalchuk, A.Pentenero, Yu.I.Ermakov, Kinet.Katal., 22 (1981) 513
B.N.Kuznetsov, Yu.I.Ermakov, Yu.A.Ryndin, J.Lotent, V.K.Duplyakin, A.Pentenero, Kinet.Katal., 22 (1981) 937
Chi H.Cheng, G.L.Prince, J.Catal., 118 (1989) 390
W.Uger, G.Lietz, H.Lieske, J.Volter, Appl.Surf.Sci., 45 (1990) 29
L.L.Murell, R.L.Garten, Appl.Surf.Sci., 19 (1984) 218
T.P.McCallister, K.R.O'Neal, German Offenlegungsschrift, 2,104 429 (1971)
J.Goldwasser, B.Arenas, C.Bolivar, C.Castro, A.Rodriguez, A.Fleitas, J.Giron,
J.Catal., 100 (1986) 75
S.Ruben de Miguel, O.A.Scelza, A.A.Castro, Appl.Catal., 44 (1988) 23
R.Bouwman, P.Biloen, J.Catal., 48 (1977) 209
C.Bolivar, H.Charcosset, M.Primet, B.Arenas, R.Torellas, Proc.VIIIth Ibero-American Congr.Catal.,Huelva, Spain (1982) 162
S.Scire, R.Maggiore, S.Galvagno, C.Crissafulli, L.Solarino, React.Kinet.Catal.Lett.,
40 (1989) 349
R.Frety, B.Benichouba, P.Bussi6re, D.Santos-Cunha, Y.L.Lam, J.Molec.Catal., 25
(1984) 173
R.E.Kirk, D.F.Othmer, Encyclopedia of Chem.Technol., Interscience Publ., Vol.10
(1953) 133 and later editions.
B.C.Gates, J.T.Katzer, G.C.A.Schuit in "Chemistry of Catalytic Processes",
McGraw Hill, N.Y. (1979)
M.Dean Edgar in "Applied Industrial Catalysis" (editor: B.E.Leach), Academic
Press, Vol.1 (chapter 5) (1983) 124
H.S.van der Baan in "Chemistry and Chemical Engineering of Catalytic Processes",

Reactions of alkanes and reforming of naphtha

270

271
272
273

274

677

(editors: R.Prins, G.C.A.Schuit), NATO Adv.Study Int.; Sijthoff & Noordhoff,


Alphen a/d Rijn, NL, (1980) 381
R.Prins, ibid, p.389
H.E.Kluksdahl, US Patent (1968) 3415 737
M.J.Sterba, V.Haensel, Ind.Eng.Chem.Prod.Res.Dev., 15 (1976) 2
J.H.Sinfelt, Adv.Chem.Engn., 5 (1964) 37
T.R.Hughes, R.L.Jacobson, K.R.Gibson, L.G.Schornack, J.R.McCabe, Hydrocarbon
Processing, 55(5) (1976) 75
A.R.Greenwood, K.D.Vesely, US Patent 3,647 680; 3,652 231
J.G.Ciapetta, D.N.Wallace, Catal.Rev., 5 (1971) 67
S.Engels, R.GOhring, W.Hager, D.Sager, G.Schr6der, R.Walter, M.Wilde, Chem.Techn., 26 (1984) 101
J.Biswas, G.M.Bickle, P.G.Gray, D.D.Do, J.Barbier, Cat.Rev.Sci.Eng., 30 (1988)
161
S.C.Fung, Chem.Tech. 1 (1994)40
D.R.Hogin, R.C.Morbeck, H.R.Sanders, US Patent (1959) 2916 440

Vous aimerez peut-être aussi