Dehydrogenation by Heterogeneous Catalysts

Daniel E. Resasco

School of Chemical Engineering and Materials Science

University of Oklahoma

Encyclopedia of Catalysis

January, 2000
1. INTRODUCTION

Catalytic dehydrogenation of alkanes is an endothermic reaction, which occurs

with an increase in the number of moles and can be represented by the expression

Alkane ! Olefin + Hydrogen

This reaction cannot be carried out thermally because it is highly unfavorable

compared to the cracking of the hydrocarbon, since the C-C bond strength (about 246

kJ/mol) is much lower than that of the C-H bond (about 363 kJ/mol). However, in the

presence of a suitable catalyst, dehydrogenation can be carried out with minimal C-C

bond rupture. The strong C-H bond is a closed-shell orbital that can be activated by

oxide or metal catalysts. Oxides can activate the C-H bond via hydrogen abstraction

because they can form O-H bonds, which can have strengths comparable to that of the C-

H bond. By contrast, metals cannot accomplish the hydrogen abstraction because the M-

H bonds are much weaker than the C-H bond. However, the sum of the M-H and M-C

bond strengths can exceed the C-H bond strength, making the process thermodynamically

possible. In this case, the reaction is thought to proceed via a three centered transition

state, which can be described as a metal atom inserting into the C-H bond. The C-H bond

bridges across the metal atom until it breaks, followed by the formation of the

corresponding M-H and M-C bonds.1 Therefore, dehydrogenation of alkanes can be

carried out on oxides as well as on metal catalysts. In fact, both types of dehydrogenation

catalysts are typically found in a number of important industrial applications.

Catalytic dehydrogenation is employed in the production of propylene and

isobutylene from propane and isobutane, in the production of C6 to C19 mono-olefins from

the corresponding normal alkanes, and of styrene from ethylbenzene. In this

1
contribution, the chemistry involved in these reactions will be discussed, with particular

emphasis on the effects of catalyst structure and composition on the catalytic properties.

2. THERMODYNAMIC EQUILIBRIUM LIMITATIONS

The dehydrogenation reactions are thermodynamically favored at high

temperatures and low pressures. The equilibrium conversion can be readily calculated by

the expression:

Kp = x (h+x) P (1)

(1 x ) (1+ i + h + x )

where Kp is the equilibrium constant, x the equilibrium conversion, i and h the

number of moles of inerts and hydrogen, per mole of alkane in the feed, respectively.

The equilibrium conversion for the dehydrogenation of propane as a function of

temperature is illustrated in Fig. 1 for typical hydrogen/alkane (h) and inert/alkane (i)

feed ratios. In some industrial operations, part of the hydrogen produced is recycled to

inhibit the formation of coke on the catalyst. However, at a given temperature, the

equilibrium conversion decreases with the hydrogen/alkane feed ratio. For example,

under typical dehydrogenation conditions (550C, 1 atm) the equilibrium conversion

decreases from about 44 % to 25 % when the H2/alkane ratio increases from 0 to 3.

Therefore, the hydrogen/alkane feed ratio is a compromise between coke suppression and

conversion. Contrarily, when inerts, such as steam or nitrogen, are added to the feed the

equilibrium is benefited. For example, if the inert/alkane ratio is increased from 0 to 3 at

550C and 1 atm, the equilibrium conversion increases from 44 % to 64 %.

2
 3. CATALYTIC DEHYDROGENATION OF ISOBUTANE AND

PROPANE

The catalytic dehydrogenation of isobutane and propane are important reactions

used commercially for the production of isobutylene and propylene, respectively. One of

the major applications of isobutylene is as a feedstock in the manufacture of methyl

tertiary butyl ether (MTBE). In the United States, MTBE has been used in relatively low

concentrations as an octane booster in gasoline for more than 25 years. However, since

1992, it has been used in high concentrations in a large number of cities, to meet

requirements of the Clean Air Act in the oxygenated and reformulated gasolines. In

those cases, up to about 15 % MTBE is added to gasoline to allow more complete

combustion and reduce emissions of carbon monoxide (CO) and volatile organic

compounds (VOC). In recent years, objections to potential health effects of MTBE

caused by water contamination have arisen, which may have an impact on the MTBE

demand.2 However, demand for polyisobutylene, and polybutenes in general, is in

continuous growth. Thus, the needs for C4 olefin production will probably remain high.

C4 olefins are mainly produced from FCC (about 50 %) and steam cracking (21 %).

However, in some industrial operations, it is necessary to have increased flexibility in the

olefin supply and so, production of a single specific alkene is sometimes required. In

those cases, the direct catalytic dehydrogenation is the ideal solution. Propylene is an

important basic chemical building block for plastics and resins. Its worldwide demand

has steadily grown for the last 15 years and it is projected that, in the coming years,

demand growth for propylene will be equal to or even higher than that for ethylene.3

Similar to isobutylene, propylene can be produced as a by-product from FCC and steam

3
cracking operations. However, in response to the growing demand and for the greater

flexibility in the olefin pool that a dedicated process has in comparison to cracking units,

several propane dehydrogenation plants are now being operated in the world.

The dehydrogenation of lower alkanes is typically carried out on two different

types of catalysts: a) Pt-based catalysts and b) chromia-based catalysts.4-6 The main

characteristics of these two types of catalysts will be discussed here, together with some

reference to other less common materials. Several commercial processes for

dehydrogenation of lower alkanes are currently available. The severely deactivating

conditions imposed by the dehydrogenation process have challenged the process

designers to develop efficient reactors for this difficult task. Several options have been

tested and several have found successful applications. Among the various commercial

processes available for dehydrogenation of propane and isobutane, one can find fixed bed

reactors, operated in isothermal or adiabatic form. Some include cyclic operations, others

include continuous catalyst regeneration with moving beds, or with fluidized beds.

3.1. Pt-based dehydrogenation catalysts:

A side reaction that frequently competes with dehydrogenation is hydrogenolysis.

Platinum is a primary component in many dehydrogenation catalysts due to its high

activity for activating C-H bonds, coupled with an inferior activity for the rupture of C-C

bonds, resulting in intrinsically high selectivities toward dehydrogenation. On a Pt

surface, only low-coordination number sites (steps, kinks) are able to catalyze the C-C

bond breaking, while essentially all Pt sites catalyze the rupture of the C-H bonds.

4
 Another undesired side reaction that competes with dehydrogenation is coke

formation. Since both, hydrogenolysis and coke formation, are more sensitive to the

support structure than dehydrogenation, any impurity or inactive species on the surface

may act as a site diluent and should increase the selectivity toward dehydrogenation.

This effect is well documented and has been observed in a number of bimetallic

systems.7, 8 In addition to the dilution of sites, other more subtle factors may play a role

in altering the activity of the Pt surface. For example, significant differences in the

magnitude of these effects have been detected in a series of Pt catalysts promoted by the

addition of different metals, which are themselves inactive, such as Sn, In, Pb, and Cu,

but can alter the properties of Pt.9 It was found that the promoting effect among these

metals was highest from the addition of Sn and lowest from Cu. In fact, Sn has been the

preferred promoter for most Pt dehydrogenation catalysts.10, 11

In section 3.1.2, we

discuss the effects of Sn in greater detail.

3.1.1. Reaction Mechanism and Kinetics on Pt Catalysts

The isobutane dehydrogenation on Pt-based catalysts has generally been

described by the following set of reaction steps:

(A) C4H10 + 2 * C4H9* + H*

(B) C4H9* + 2* C4H8** + H*

(C) C4H8** C4H8 + 2 *

(D) H2 + 2 * 2 H*

Cortright et al.12 have conducted deuterium tracing studies under reaction

conditions, which showed a much larger extent of deuterium in the isobutylene product

5
than in the isobutane. Also, it was seen that the small fraction of isobutane that contained

deuterium was fully deuterated, indicating that the dissociative adsorption of isobutane is

slow compared to the exchange of hydrogen/deuterium in the adsorbed isobutyl species.

These results support the hypothesis that step (A) is the slowest step and are in line with

earlier observations that show that as the hydrogen pressure increases, the

dehydrogenation rate decreases as a result of the competition of hydrogen and the alkane

for adsorption sites.13 Accordingly, the following rate expression should represent the

kinetics

rD = kA 2 [ PiC4H10 - ( P iC4H8 PH2 ) / Keq ] (2)

*

where rD is the net dehydrogenation rate, kA is the rate constant for the

dissociative adsorption of isobutane (step A), Keq is the overall equilibrium constant for

isobutane dehydrogenation, and is the fraction of sites free of adsorbed species. The
*

isobutane coverage under reaction conditions is practically negligible12 and the fraction

of free sites can be determined from a expression derived from the assumption that the

only species on the surface are hydrogen and isobutylene, which are in equilibrium with

the gas phase. Accordingly the expression for the fraction of free sites can be expressed

as:

=
*

KC 8PiC 4 H 8
1 + KDPH 2 1 + 2 KDPH 2 + KDPH 2 +
4 PiC 4 H 8 KC
(3)

where KC and KD are the equilibrium constants for isobutylene desorption (step C)

and hydrogen adsorption (step D) , respectively. Combining equations (2) and (3)

excellent fits have been obtained on a number of reaction data over a wide range of

6
temperatures, concentrations, and catalysts.14 All the fitting parameters thus obtained had

physical meaning and were used to compare thermodynamic functions on different

catalysts. This kinetic expression clearly explains the reaction orders obtained on most

previous studies, i.e., first order dependence with respect to the alkane and negative half

to zeroth order dependence in hydrogen.9, 13

3.1.2. Addition of Sn to Pt as a promoter for activity, selectivity, and catalyst life

The addition of Sn has important beneficial effects on the catalytic properties of

Pt for the dehydrogenation of lower alkanes. First of all, and as mentioned above, it

increases the selectivity towards dehydrogenation by inhibiting hydrogenolysis.

Similarly, the addition of Sn has a profound effect on the catalyst life. The Pt-Sn catalyst

retains a much higher activity than the pure Pt catalyst. The enhancement in selectivity

and stability, clearly illustrated in Fig. 2a and 2b, can be explained in geometric terms by

the dilution of Pt ensembles by Sn. As described above, this dilution greatly reduces the

activity towards reactions that require a large ensemble of Pt atoms to constitute the

active sites, such as hydrogenolysis and coking. Also, the increase in selectivity of the

pure Pt sample with time on stream shown in Fig. 2b can be explained by the same

geometric arguments. The carbon deposited during the reaction plays the role of an

inactive species that inhibits undesired reactions, including coking. This explanation

accounts for two observations on the pure Pt catalyst. Both, the selectivity and the

stability of the catalyst improve as a function of time on stream as the number of large

ensembles is reduced by the presence of carbon. On the pure Pt catalyst the initial

deactivation is very fast. Thus, it is difficult to measure the true initial activity of

unpromoted Pt and determine whether the initial dehydrogenation rate on Pt is higher or

7
lower than on the bimetallic catalyst. In fact, pure Pt may be initially more active, but

after a short time it deactivates while the bimetallic retains its activity. This phenomenon

is particularly pronounced in the absence of added H2, as in the experiments of Fig. 2.

The higher initial activity of pure Pt catalysts can only be observed when high

H2/hydrocarbon feed ratios are employed.15

3.1.3. Preparation of Pt-Sn Dehydrogenation Catalysts and its effects on performance

The beneficial role of adding Sn to Pt catalysts has been observed on different

supports and on catalysts prepared by various methods. In some preparations the addition

of Sn has been conducted by sequential impregnation, first of an aqueous solution

containing the Sn precursor (typically SnCl2) and then of another solution containing the

Pt precursor (typically H2PtCl6). In these preparations, the degree of Pt-Sn interaction is

relatively low because Sn tends to interact with the support becoming segregated from Pt.

Therefore, to maximize the metal-metal interaction, other preparations have been used.

For example, the use of an aqueous solution containing both Pt and Sn precursors and

excess HCl results in the formation of a bimetallic PtCl2(SnCl3)22- complex.16,17 In this

preparation, both metals are deposited on the surface in the same compound and, after the

thermal treatment, a high degree of alloyed metals can be obtained. Other methods

involve the Pt-catalyzed surface reduction of an organometallic tin compound, resulting

in the selective deposition of Sn over the Pt surface.18 For example, the catalyst can be

prepared from a solution of Sn(C4H9)4 in an organic solvent, such as n-hexane. This

solution is added onto a pre-reduced Pt/support sample, without exposure to air. The H

8
that remains adsorbed on Pt is responsible for the reduction of the Sn(C4H9)4 , causing the

selective deposition over the metal.19

The catalyst preparation may critically influence the Pt-Sn interaction and

consequently the catalytic behavior. As illustrated in Fig. 3 for a series of Pt-Sn catalysts

supported on silica, the promoting effect of Sn strongly depends on the method employed

in the preparation. Catalysts prepared by co-impregnation, particularly when the solvent

is an aqueous solution containing HCl, result in a relatively high extent of Pt-Sn

interaction. By contrast, sequential impregnation results in a large fraction of unalloyed

Pt.

Although alumina and silica have been the two supports most widely studied for

dehydrogenation reactions, other non-acidic supports have also shown promising

properties. For example, KL zeolite has been proposed to be an effective support for Pt-

Sn dehydrogenation catalysts. It has been reported20 that these catalysts maintain high

isobutane dehydrogenation activity and selectivity for extended reaction intervals.

Characterization of these materials indicates that a fraction of the Sn is alloyed with Pt

and the rest is in the form of Sn2+ ions, exchanged with K+ previously present in the

zeolite. It was further hypothesized that the K displaced from the zeolite framework can

interact with the Pt-Sn alloy particles, promoting the activity. The promoting effect of K

was demonstrated by deliberate addition of excess K to the catalysts, which greatly

enhanced the dehydrogenation activity and selectivity.

Table 1 makes a comparison of the isobutane dehydrogenation conversion on

several Pt and Pt-Sn catalysts.20 In the table we compare the reaction rates per gram of

Pt. It is generally accepted that the best form of comparing the activity of a series of

9
catalysts is in terms of turnover frequency (TOF), based on the number of sites as

measured by hydrogen chemisorption. However, when Pt-Sn alloys and unalloyed Pt

phases co-exist, the TOF may be misleading. The chemisorptive capacity of Pt is greatly

changed when it is alloyed with Sn. For example, it has been shown that the heat of CO

adsorption drops from Pt to PtSn alloy by as much as 20 kJ/mol.21 This energy difference

causes significant changes in the adsorption equilibrium constants at 298 K and,

consequently, in the observed adsorption capacity. In recent work,22 it was observed that

the CO adsorption capacity of Pt-Sn/SiO2 dropped to zero after 1 hour at 500C on

isobutane stream. However, the catalyst was almost as active for dehydrogenation as at

the beginning of the reaction. The reason for this apparent discrepancy is that the

unalloyed Pt is rapidly covered by coke, while the Pt-Sn alloys remain more or less free

of coke. Since alloyed Pt does not adsorb significant amounts of CO the CO/Pt measured

on the fresh catalyst was mainly due to the fraction of unalloyed Pt, which after a while

contributes little to the activity. The situation is very similar for the chemisorption of

hydrogen. Verbeek and Sachtler23 have shown that Pt-Sn alloys adsorb very little

hydrogen and have ascribed this decrease to a lowering of the heat of adsorption.

Microcalorimetry studies24 have shown that, even though the addition of Sn resulted in a

large decrease in the saturation adsorption coverages for H or CO, the heats of adsorption

at zero coverage on Pt:Sn samples were similar to those on pure Pt. However, it must be

noted that these SiO2-supported samples were prepared by sequential impregnation and,

as shown below, this technique leads to a large fraction of unalloyed Pt, which may be

responsible for the measured high heats of adsorption. For changes in the initial heats of

adsorption to be seen, the majority of Pt needs to be alloyed with Sn. In fact, when

10
higher Sn concentrations were used in those studies,24 the heats of adsorption at zero

coverage decreased significantly, in agreement with the idea mentioned above. Only at

very high H2/hydrocarbon ratios, at which the deactivation is less pronounced, the TOF

values for Pt and Pt-Sn catalysts are normally found to be similar. It is generally

observed that the initial deactivation increases with the amount of unalloyed Pt, then the

unalloyed Pt in the sample is not related to the catalytic activity. However, CO and H2

chemisorption primarily occurs on the unalloyed Pt. Therefore, for bimetallic Pt-Sn

catalysts, we prefer to report rates per total Pt.

A different methodology to estimate TOF has been proposed by Rajeshwer et al.25

These authors propose that hydrogen/oxygen titration on Pt-Sn alloys is an activated

process and, as a result, the hydrogen titration of preadsorbed oxygen conducted at room

temperature only reflects the exposed fraction of unalloyed Pt, but that conducted at

150C titrates both alloyed and unalloyed Pt. As a result, the difference between the two

uptakes reflects the exposed fraction of alloyed Pt. This is an interesting method,

although the possibility of hydrogen spillover onto the support,26 may lead to

overestimation of the density of alloyed Pt. To rule out this possibility, Rajeshwer et al.

conducted measurements on physical mixtures of monometallic Pt and Sn catalysts,

which showed the same uptake as the monometallic Pt catalyst. However, values of H/Pt

uptake ratios much greater than one have sometimes been observed, which is hard to

explain without invoking spillover onto the support or, as suggested by some authors,

hydrogen uptake by dissolution in the bulk of the Pt-Sn alloys. 23

11
3.1.4. Structure and Composition of Pt-Sn Catalysts

The assessment of Pt dispersion on monometallic catalysts and quantification of

its exposed sites is well established and can be readily accomplished by standard

techniques such as hydrogen chemisorption, electron microscopy, or EXAFS. As

described in the previous section, the same task is much less straightforward with

bimetallic Pt-Sn catalysts. Pt and Sn are able to form a number of ordered Pt-Sn alloys,

such as PtSn, PtSn2, PtSn4, Pt2Sn3, and Pt3Sn.27

However, the formation of alloys like these on high-surface-area supports

strongly depends on the nature of the support, thermal treatments, impregnation method,

anions present, etc. In addition, the chemisorptive properties of these alloys are less

understood than those of pure Pt. Another fundamental question related to these systems

is the oxidation state of Sn under reaction conditions. Upon reduction, tin may remain in

the SnII or SnIV forms or may get reduced to the metallic Sn0 state and form PtxSny alloys.

Pt-Sn alloys are more easily formed when the support interacts only weakly with

the Pt and Sn, such as in the case of silica-supported catalysts. One of the techniques

commonly employed to determine the oxidation state of supported metals and identify

metal-metal interactions is temperature programmed reduction (TPR). Fig. 4 shows the

TPR of a bimetallic Pt-Sn/SiO2 catalyst compared to a monometallic Pt/SiO2 catalyst

after calcination at 350C.22 The TPR for the pure Pt catalyst shows a H2 consumption

peak at about 125C. The profile for the bimetallic Pt-Sn catalyst has two H2

consumption peaks. The first one appears in the same position as that for pure Pt

(125C), while the second has a maximum at 200C. This peak is associated with the

reduction of a Pt-Sn alloy. EXAFS studies28 conducted on the same set of samples

12
confirm that the bimetallic Pt-Sn/SiO2 catalyst has both unalloyed Pt and alloyed Pt-Sn.

Fig. 5 shows that the radial distribution of the bimetallic sample clearly exhibits two

peaks. However, as shown in the simulated data of Fig. 6, neither the pure Pt or the PtSn

alloy alone exhibits those peaks. Only when both Pt and PtSn coexist is that the two

peaks appear. Therefore, the presence of the two peaks in the radial distribution of the

Pt-Sn/SiO2 catalyst demonstrates the presence of both unalloyed Pt and PtSn alloy. The

fitting of the experimental data indicated that, in that sample 35 % of the Pt was

unalloyed, and 65 % in the form of the PtSn alloy. Using other alloys, such as Pt3Sn and

PtSn2 resulted in poor fits. Similar conclusions were obtained from Mssbauer

spectroscopy and X-ray diffraction data.29 These results indicate that, on silica support,

PtSn alloy is formed upon reduction in hydrogen. By contrast, when alumina is used as a

support, the reduction of Sn to the metallic state is much more difficult. Studies of Pt-

Sn/Al2O3 catalysts with tin loadings in the range 0.3 5.0 wt % showed no evidence of

metallic tin, even after reduction at high temperatures. Only when the tin loading is

increased to very high values, a small fraction of metallic tin can be obtained. Fig. 7

shows XPS data of a Pt-Sn/SiO2 (5 wt % Pt, 5 wt % Sn) catalyst and a Pt-Sn/Al2O3 ( 5 wt

% Pt, 29 wt % Sn) catalyst.30 It is observed that, even with such a high Sn loading, the

fraction of metallic Sn is low compared to that obtained on silica.

3.1.5. Is the Sn promotion of Pt geometric or electronic in nature ?

The arguments in favor or against geometric or electronic effects have captured

the attention of Heterogeneous Catalysis researchers for decades. Today, it is widely

accepted that if a given metal of very low catalytic activity is distributed uniformly over

13
the surface of a more active metal, the activity of the active metal is greatly affected for

reactions requiring a large ensemble of atoms, but it is only moderately affected for

reactions that require a single site. This effect can be explained by simple geometric

dilution of the active sites by the presence of the inactive atoms. However, the question

that still generates controversy is whether the atoms of the active metal that remain

exposed to the gas phase are chemically (or electronically) altered by the presence of the

inactive metal. In many cases, the electronic modification of the active species is very

small compared to the more notable geometric effect. For example, large activity losses

are typically observed for the ethane hydrogenolysis on Ni when small amounts of Cu are

added to the catalyst. However, much lower effects are observed on the same catalysts

for the dehydrogenation of cyclohexane31. In that case, geometric effects are clearly

dominant and only small effects could be attributed to an electronic modification of Ni by

the presence of Cu.

In Pt-Sn alloys, both geometric and electronic effects seem to play significant

roles. An evidence in support of electronic modifications of Pt by the presence of Sn can

be obtained by X-ray Absorption Near-Edge Spectroscopy (XANES), which is usually

considered a good technique to probe the electronic state of an atom in its environment.

In the LIII XANES spectra, the so-called white line feature is a prominent pre-edge

maximum that shows significant variations with changes in the oxidation state of the

metal. Among the various contributions to absorption in this region, the most important

one is an intra-atomic p-to-d electronic transition, from occupied 2p states to unoccupied

5d states in the central Pt atom. An oxidized Pt sample has a higher density of

unoccupied d states and, as a result, the white line notably increases in comparison to that

14
of the metallic Pt. Accordingly, an alteration of the Pt LIII edge shape caused by addition

of Sn would indicate an electronic modification of Pt, caused by the presence of Sn. This

alteration has, in fact, been reported in several papers, and it is illustrated in Fig. 8. The

reduction in the of the white line intensity has been taken by some authors as an

indication of a decrease in the density of unoccupied d states in Pt, which means that an

electron transfer occurs from Sn to Pt. Other authors32 have postulated that the

intermetallic interaction may involve sp orbitals. The Pt-Sn bonding in Pt3Sn alloys

would involve the occupied Sn 5p and the unoccupied Pt 6sp. The resulting donation of p

electrons from Sn to Pt would only have an indirect effect on the d states of Pt.

Recent quantum chemical calculations support these results and show that Sn

donates electrons to both 6sp and 5d orbitals. These calculations used density functional

theory (DFT) and were conducted for two types of Pt and Pt-Sn clusters. In the first type,

19 Pt atoms were constrained in three layers of an FCC-like crystal configuration,

keeping a fixed interatomic distance of 2.77. To study the effect of Sn, three Pt atoms

were replaced in the corners of the top layer. These two clusters were called Pt19 and

Pt16Sn3, respectively. In the second type of cluster, the distribution of 10 Pt atoms was

optimized for energy minimization, letting the interatomic distances relax from the bulk

value of 2.77. Here, the effect of Sn was evaluated by replacing four Pt atoms by Sn

atoms, three at the corners of the top layer, and one at the bottom. Table 2 summarizes

the Mulliken electron population of the 6s, 6p, and 5d orbitals in three top Pt atoms for

the four different clusters. A clear increase in electron population is observed in both the

6sp and 5d orbitals when Sn is added in the cluster, indicating an electron donation from

Sn to Pt.

15
 Even when electronic interactions do occur, they may not necessarily have

significant effects on the catalytic properties. The analysis of the nature and energetics of

adsorption of certain adsorbates may help determine whether electronic effects have

important influences in catalysis. For example, in the Pt-Sn system, the adsorption of

ethylene has been a sensitive probe that has provided important information.33 It is

known that on pure platinum, ethylene can adsorb in different forms. At low

temperatures, -bonded ethylene and di- bonded ethylene species can be observed.

However, as the temperature is increased above room temperature, an ethylidyne species

appears. The evolution of these species as a function of temperature can be examined in

the following IR and microcalorimetry experiments. As shown in the IR spectra in Fig.

9, when the adsorption of ethylene is conducted on a Pt/SiO2 catalyst at 70C, two bands

are typically observed at 1506 and 1428 cm-1. These bands are due to -bonded ethylene

and di--bonded ethylene, respectively. Under these conditions, the measured initial heat

of adsorption is 125 kJ/mol. By contrast, when the temperature is increased to room

temperature (25C), the bands at 1506 and 1428 cm-1 almost completely disappear and a

new band at 1342 cm-1 dominates the spectrum. This band is due to ethylidyne. At this

temperature, the initial heat of adsorption increases to 157 kJ/mol, consistent with a

stronger interaction of the ethylidyne species with the surface. A different behavior is

observed on the Pt-Sn/SiO2 catalyst. For this catalyst, the same two bands at 1506 and

1428 cm-1 are observed at low temperatures, but they do not disappear when the

temperature is increased to 27C. At this temperature, the transformation of -bonded

ethylene and di--bonded ethylene species to ethylidyne is much less pronounced on the

bimetallic catalyst than on the pure Pt catalyst. Then, not only the bands corresponding

16
to -bonded ethylene and di--bonded ethylene species are more intense, but also the

band corresponding to ethylidyne is significantly weaker than on pure Pt. As mentioned

above, it is possible that a significant fraction of Pt remains as unalloyed Pt. Therefore,

the small ethylidyne band observed on the Pt-Sn catalyst may in fact be due to adsorption

sites on the small Pt fraction that is not alloyed with Sn. The microcalorimetric data on

the Pt-Sn catalysts are also consistent with the IR data. The observed heat of adsorption

at room temperature was only 135 kJ/mole, much lower than on pure Pt, which was 157

kJ/mol. The observed hindrance in the ability of Pt to form the ethylidyne surface species

when Sn is added could just be due to a geometric blockage by Sn of the 3-fold Pt sites

necessary to stabilize the ethylidyne species. In fact, that may certainly occur. However,

the DFT quantum chemical calculations have shown that, even when Sn is not occupying

the 3-fold Pt sites, it may exert an electronic effect that prevents the stabilization of

ethylidyne. The calculated changes in energy, per mole adsorbed, when -bonded

ethylene, di--bonded ethylene, or ethylidyne species form on Pt and Pt-Sn clusters are

summarized in Table 3. It is observed that the addition of Sn weakens the interaction of

ethylene with Pt for all three species. However, the effect is most pronounced for the

ethylidyne species. On the Pt-Sn clusters, very small, if any, exothermicity is predicted

by these calculations for the adsorption on the 3-fold Pt sites. Moreover, the Mulliken

population analysis suggests that the formation of ethylidyne species on Pt involves

electron donation from ethylidyne to the 6sp orbitals on Pt, accompanied by back-

donation from the 5d orbitals to the adsorbate. This analysis further shows that the

increased electronic population in Pt as a result of the addition of Sn, causes a more

repulsive interaction for the formation of ethylidyne than for the formation of -bonded

17
ethylene or di--bonded ethylene. Therefore, it seems that when Sn is added to Pt, both

geometric and electronic effects contribute to inhibit the stabilization of ethylidyne on the

surface.

3.1.6. Catalyst Deactivation and Regeneration

The major cause of catalyst deactivation in dehydrogenation processes is carbon

deposition. The high temperatures and low pressures necessary to achieve high

equilibrium conversions are conducive to the rapid formation of coke. The use of

promoters and the operating conditions, e.g., temperature and H2/alkane ratio, are

parameters that can be adjusted to modify the rate and extent of carbon deposition.

A typical kinetics expression that describes the rate of carbon formation has the

form:

rC = k1 PiC4H8 k-1 PH2 C (4)

(1 + Ka PiC4H8 + Kb PH1/2 + KcC )

where P iC4H8 and PH are the partial pressures of olefin and hydrogen, respectively,

C is the concentration of carbon on the surface, and k1, k-1, Ka, and Kb temperature-

dependent parameters that are a function of the catalyst. Here it is important to note that

the expression predicts a strong inhibition of coke formation by both a negative term in

the numerator, which reflects the elimination of carbonaceous deposits by hydrogenation

and a term in the denominator that result from the competition for adsorption sites. The

expression also predicts that the rate of coke formation is expected to be higher initially,

when the amount of carbon on the surface is low. Both trends are in fact observed

experimentally. What is important to discuss is that the concentration at which the

18
carbon on the surface begins to inhibit further deposition is a strong function of the

catalyst composition. For example, while on a Pt/Al2O3 catalyst, the rate of formation

starts slowing down at a carbon content of about 2 wt % C, on a bimetallic Pt-Sn/Al2O3

catalyst the coking rate remains unchanged up to about 7 wt % C.

Due to the substantial formation of coke, lower alkane dehydrogenation catalysts

require frequent regenerations. Different regeneration procedures have been tested, the

most common ones involving oxygen, but other alternative, or complementary methods

involve treatment in chlorine, steam or sulfur-containing streams.34 The most obvious

effect of the regeneration treatment with oxygen is the burning of the carbonaceous

deposits. It is generally accepted that two distinct coke burning processes occur as a

function of temperatures. Oxidation of carbonaceous deposits is a catalytic process that

is greatly accelerated by Pt, or other metals typically used in dehydrogenation reactions.

Therefore, the oxidation of coke in contact with (or in the close vicinity of ) the metal

particles can readily occur at relatively low temperatures. By contrast, the coke deposited

on the support, away from the metal particles, can only be oxidized at high temperatures

because it requires the spillover of activated oxygen from the metal onto the support.

The two oxidation regions can be clearly demonstrated by the temperature

programmed oxidation (TPO) technique. When the support is alumina two distinct

oxidation peaks are clearly observed, one at about 450C and another above 500C. A

clear demonstration that the one at the lower temperature is associated with the metal was

conducted on Pt and Pt-Sn catalysts.35 Both, the dehydrogenation activity and hydrogen

chemisorption capacity were completely restored after a partial oxidation of the coke

deposits at 450C. These results clearly demonstrate that the coke removed by high

19
temperature oxidation is located on the support, rather than on the metal. A second

important result obtained in the same investigation was the demonstration that the relative

amounts of each of the two types of coke is a strong function of the Sn/Pr ratio in the

catalyst.35 As illustrated in Fig. 10, an increase in the amount of Sn causes not only a

decrease in the amount of coke on the metal but also an increase in the amount of coke on

the support, indicating that the presence of Sn enhances the mobility of coke, or its

precursors, from the metal to the support.

In addition to the burning of coke, other secondary undesired effects may also

occur during the treatment in oxygen. For example, although at intermediate

temperatures in the presence of oxygen, sintering is relatively slow, at the high

temperatures required to completely eliminate the coke, rapid growth of Pt particles

occurs. The reason for this is that at the intermediate temperatures (e.g., around 400C)

Pt is in the oxide form, which interacts with the oxidic support, slowing down the Pt

mobility. However, above 500C, Pt oxide becomes unstable and metallic Pt is more

susceptible to sintering.

The oxidation treatment not only may cause metal sintering but it can also alter

the interaction between the metal and promoter due to a disruption between Pt and Sn

that tends to occur under oxidation conditions. For example, temperature programmed

reduction (TPR) studies of sequentially impregnated Pt-Sn/SiO2 catalysts, treated in air at

increasing temperatures and increasing periods of time, showed that the H2 consumption

peak associated with the reduction of Pt-Sn alloys, significantly decreased while those

ascribed to unalloyed Pt and unalloyed Sn increased. This trend can be explained in

terms of a segregation of the Pt and Sn caused by the high temperature oxidation. As

20
illustrated in the TPR of Fig. 11, the segregation was also observed in co-impregnated

catalysts, which started with a much higher extent of Pt-Sn interaction. The fresh catalyst

exhibited H2 consumption peaks at about 146 and 186C, which correspond to Pt-Sn

phases richer in Pt and Sn, respectively. After an oxidation treatment conducted at 500C

to eliminate the coke formed during reaction, the peaks clearly split, generating a peak at

125C, which corresponds exactly to the reduction of unalloyed Pt, and a broader peak

centered at about 240C, which can be ascribed to bimetallic phases rich in Sn. TEM

evidence also supports the proposed segregation phenomenon. TEM studies on model Pt-

Sn catalysts,36 have shown that oxidation at 550C for 1 hour resulted in severe

morphological changes in the intermetallic particles. The Sn in the alloys formed an

oxide and migrated to the surface of the particle. SnO2 formed a ring around the edge of

the particle in contact with the SiO2 support. The Pt did not form an oxide but became

metallic Pt at the core of the particle. Subsequent 18 hour reduction (650C) resulted in

complete recombination of the alloys.

An important point to discuss is whether the disruption of the alloys and

segregation phenomena have significant effects on catalytic properties. Experimental

results show that, in fact, the effects may be very large. As shown in Fig. 12, the

oxidative regeneration treatment results in an important drop in conversion, together with

a much higher propensity of the catalyst to form coke. As compared in the TPO profiles

of Fig. 13 for a Pt-Sn/SiO2 sample, after the first run and after an oxidation./reduction

cycle followed by reaction. A much greater amount of carbon was produced on the

catalyst that had been subjected to the oxidation/reduction regeneration treatment. Note

that on silica-supported catalysts only one TPO peak is observed, as opposed to the

21
alumina-supported catalysts described in Fig. 10, where one peak corresponds to coke on

the metal and one on the support. Therefore, all the coke formed in this case must be

associated with the metal. When the comparison is made between Pt-Sn bimetallic

catalysts and pure Pt catalysts, it is observed that the latter forms much more coke, and

the TPO peak appears at somewhat higher temperatures, due to a more refractory nature

of the carbon. In the case of the Pt-Sn sample, after the oxidation/reduction cycle, the

TPO peak shifts to the same position as that observed on pure Pt catalysts. This shift

indicates that some of the alloy has been destroyed, leaving behind patches of unalloyed

Pt, on which the formation of the graphitic carbon is more favorable. The regenerated Pt-

Sn catalyst still form less carbon than the pure Pt sample, but almost four times the

amount found on the fresh bimetallic sample. Therefore, the activity loss observed after

the oxidation/reduction cycle can be linked to the partial destruction of the Pt-Sn alloy,

which results in an increase in the deactivation due to coking. Under oxidizing conditions

at high temperatures, the Sn forms an oxide and, at the expense of the subsurface layer,

the exterior becomes rich in Sn. The Pt, which does not form an oxide, is metallic and

becomes the core of the particle. Reduction at high temperatures results in the

destruction of the oxide and possibly redistribution of the Sn, opening patches of pure Pt.

Although the thermodynamically most stable configuration is the complete

recombination of the alloy, short reduction times do not allow for this to occur.

Consequently, after the oxidation/reduction cycle, the bimetallic particles have become

heterogeneous, with Sn-rich regions interdispersed with Pt-rich regions. It is even

possible that pure Pt ensembles are present allowing for the formation of more refractory

carbon to form on the surface under subsequent exposure to reaction conditions. On

22
alumina-supported catalysts, the disruption of Pt-Sn alloys by the high temperature

oxidation is even more pronounced than on silica-supported catalysts.37

The deleterious effects of alloy disruption and Pt-Sn segregation, as well as

sintering of Pt may be overcome by the so-called oxychlorination treatment. In this

treatment chlorine-containing compounds are incorporated in the oxidation stream. This

is a process that has been employed for many years in the regeneration of naphtha

reforming catalysts, but has also been applied in the regeneration of dehydrogenation

catalysts. With monometallic Pt catalysts, the main effect of the oxychlorination

treatment is the redispersion of the Pt particles, which may have sintered during the coke

burning operation. During the oxychlorination treatment, volatile Pt oxychloride species

form and effectively re-disperse the Pt when these species are anchored by the alumina

support. When Sn is present on the catalyst, the effect of chlorine during the

oxidation/reduction cycles depends on the type of support and the Sn loading. For

example, on alumina, Sn may block the anchoring sites of the support, thus inhibiting the

redispersion process. But, at the same time, oxychlorination of bimetallic Pt-Sn catalysts

enhances the contact between Pt and Sn. Thus Cl may have the effect of minimizing the

Pt-Sn segregation that would normally occur in the oxidation treatment without

chlorine.38 The unwanted effect of the oxychlorination process is the generation of

acidity on the support, which must be neutralized to avoid the increase in coke formation

and isomerization products in the subsequent cycle.

Other regeneration methods include the use of steam, either during the

regeneration or continuously during the reaction. Steam can help to reduce the formation

of coke, via the reaction:

23
 C + 2 H2O CO2 + 2 H2

This reaction is accelerated by the presence of alkali ions, which are sometimes added to

the catalysts as promoters, not only to favor this reaction, but also to eliminate support

acidity.

3.1.7 Industrial units using Pt-based catalysts

Several commercial processes employ Pt-based catalysts. For example, Oleflex

from UOP employs promoted Pt/Al2O3 catalysts in a reaction system composed of 3 to 4

moving bed reactors. In this process, the catalyst is continuously regenerated in a separate

regeneration circuit. A typical Oleflex catalyst consists of spherical pellets of -alumina

(surface area about 100 m2/g) containing Pt (< 1 wt %) and promoted with Sn and alkali

metals (e.g. Li). In some patents,39 it is implied that Co and Zn can also been used as

promoters for these catalysts. The STAR process from Phillips Petroleum40 employs a

solid catalyst based on Pt, 0.2 to 0.6 wt %, supported on a Zn aluminate material, 10 to 50

m2/g. This support exhibits a unique stability at high temperatures in the presence of

steam, and this is essential in this case because the process involves the use of steam as a

diluent, at a high steam/alkane feed ratio and relatively high pressures, i.e. up to 3.5 atm..

In addition, due to its non-acidic nature, the support does not promote undesired side

reactions, such as isomerization or cracking. The catalyst contains Sn as a promoter to

reduce coke formation and increases selectivity. As mentioned above, when a diluent

such as steam is added, higher equilibrium conversions can be obtained. In this process,

the operating alkane conversion at 600C is close to 50 % and the overall selectivity is

reported to be 95 %.

24
3.2. Chromia-based dehydrogenation catalysts:

Several oxides are able to readily catalyze the activation of C-H bonds via hydrogen

abstraction. In this step, the oxygen species plays an important role in the formation of a

surface O-H bond after the rupture of the C-H bond. However, the crucial role is played

by the surface cations, which are thought to be responsible for the initial activation of the

C-H bond. Chromium oxide has been known as an effective dehydrogenation catalyst for

many years.41 The first chromia-based catalysts were developed at the onset of World

War II, when the sources of natural rubber were suddenly discontinued, and butadiene

had to be made by dehydrogenating n-butane or n-butene. In recent years, with the

increased demand for dehydrogenation processes, modified chromia-based catalysts are

being used in the dehydrogenation of isobutane and propane. In industrial applications,

alumina is the preferred support. Zirconia has been recently considered as a promising

support due to its high thermal and chemical stability.

3.2.1 Kinetics of Dehydrogenation on Chromia Catalysts

Several kinetics studies have been conducted for the dehydrogenation of lower

alkanes on chromia-alumina catalysts. A recent study conducted on a commercial

catalyst (Cr2O3 18 %, Al2O3 82%, ZrO2 0.25 %) has showed that the dehydrogenation of

isobutane on this type of catalysts can be described by a Langmuir-Hinshelwood type

equation:

PiC 4 H 8 PH 2
k ( PiC 4 H 10 )
rD = K
(1 + KEHPiC 4 H 8 PH 2 + KEPE + KHPH 2 ) (5)

25
Similar to expression (3), obtained for the Pt-based catalysts, this expression predicts a

first order with respect to the alkane concentration. However, as opposed to the situation

on the Pt catalysts, for which the inhibition by hydrogen was due to a competition for

adsorption sites with the alkane, on the chromia catalysts the best fits of the experimental

data were obtained when the hydrogen terms in the denominator were neglected. This

result is an indication that, on chromia catalysts, hydrogen does not strongly inhibit the

adsorption of alkane.

3.2.2. Active Species in Supported Chromia Catalysts

Chromium oxide supported on alumina has been extensively studied by a number

of characterization techniques (e.g. UV-Vis diffuse reflectance, Raman spectroscopy, X-

ray photoelectron spectroscopy, and chemical analysis). However, the exact nature of the

active site is still controversial. The nature of the chromium species existing on the

surface of a calcined catalyst are not only dependent on a number of parameters, such as

type of support, pre-calcination temperature, and chromium oxide loading, but they also

change during the reaction. On a calcined sample, two types of Cr6+ species have been

normally found. One of them is grafted Cr6+, anchored to the alumina support by

interaction with the surface OH groups. The second type of Cr6+ species on calcined

samples is not chemically bound to the support. The amount of each of these species can

be estimated by first extracting the unbound fraction with water, and then titrating (by

iodometry) the extracted and solid fractions. The remaining amount of chromium left on

the surface is Cr3+. This oxidation state can be either in the form of crystalline aggregates

of -Cr2O3, large enough to be detected by XRD, or as highly dispersed Cr3+ species,

26
undetected by XRD. A typical distribution of each of these species as a function of

chromium content on a chromia-alumina catalyst series is shown in Fig. 14.42 It can be

observed that, at low chromium loadings, the grafted Cr6+ is the dominant species, as the

amount of Cr increases, the fraction of soluble Cr6+ begins to increase, reaching a plateau

at about 2 wt % CrO3. Interestingly, when the Cr loading was kept constant on alumina

supports of varying surface area, the amount of grafted Cr6+ increased linearly with

surface area. In the present example, at Cr loadings higher than about 4 wt % CrO3 the

Cr3+ species begin to form. At the high-loading end, they are the dominant species on the

catalyst, including the presence of -Cr2O3, clearly observable by XRD.

In a reducing environment, as that encountered under reaction conditions, the Cr6+

species are reduced to Cr3+. Fig. 15a shows XPS spectra of a calcined 15.3 % CrO3 on

alumina catalyst. The peaks obtained in the binding energy region corresponding to Cr 2

p3/2 were fitted with contributions from Cr6+ and Cr3+ states, taking into consideration the

multiplet states that arise from the spin-orbit splitting. A significant fraction of Cr6+ has

been clearly observed in all calcined catalysts, being the dominant form on the catalysts

with lower Cr concentration. After a reduction treatment in H2, or after the

dehydrogenation reaction, the peak corresponding to Cr6+ decreased significantly. This

decrease is illustrated in Fig. 15b, which compares a K-promoted, 15.3 wt %

CrO3/alumina catalyst after calcination, reduction in H2, and reaction at 470C in

isobutane/He feed. An interesting difference between the two calcined samples in Fig.

15a and 15b is clearly evident. The K-promoted catalyst shows a higher Cr6+/Cr3+ ratio,

which was also confirmed by chemical analysis and was ascribed to the formation of a

potassium chromate phase at the expense of the Cr3+ species.

27
 The interesting point is that the Cr3+ species are supposed to be the active species,

but not all Cr3+species are equally active. In fact, it is believed that the chromium species

initially present on the calcined catalyst as dispersed Cr6+, are more active than -Cr2O3.

Therefore, the catalyst activation involves the Cr6+-to-Cr3+ transformation. However, in

many cases, the pre-reduction step is omitted because the reaction environment is

reducing and the reaction temperature is normally high enough to induce the

transformation. Although a pre-oxidized sample may exhibit a lower initial conversion,

it eventually reaches the same conversion as one that has been pre-reduced in hydrogen

or other reducing gas.

Industrial chromia-based dehydrogenation catalysts contain small amounts of

alkali promoters, typically 1 wt % K. The beneficial role of the alkali is related to its

ability to poison the acidic sites on the alumina and to inhibit the undesired cracking and

coke formation, associated with acidity. However, the alkali promoter plays a role in

determining the nature of the Cr species present on the catalyst surface. As mentioned

above, the presence of K may lead to the formation of a potassium chromate (Cr6+) phase.

Although this phase is destroyed under reaction conditions, depending on the K and Cr

concentrations, its formation may inhibit the generation of less active forms of Cr3+, e.g.

-Cr2O3. For Cr concentrations of the order of 15 wt % CrO3 and above, the addition of

K has a promoting effect on dehydrogenation activity, but only at low K concentrations.

When the K concentration exceeds about 1.2 wt % K, the formation of the less active -

Cr2O3 is enhanced. On the other hand, at low Cr concentrations, e.g. 10 wt % CrO3 and

lower, the addition of K does not result in an increase in activity.

28
3.2.3 Typical Preparation of a Chromia-based Dehydrogenation Catalyst

The -alumina support (surface area 80-100 m2/g) is impregnated with an aqueous

solution of CrO3 and K2CrO4 to derive a loading of chromium of about 5-7 wt % Cr and

of potassium of about 1 wt % K. After drying at room temperature, the catalyst is

calcined in flow of air at 600C for 6 h. During this calcination step, decomposition of

the chromium precursors occurs, together with an strengthening in the interaction

between the chromium and aluminum oxides. Before reaction, the catalyst is sometimes

activated in situ by reduction in hydrogen flow at 500C. Alternatively, the reduction can

be induced by the reaction mixture. After the deposition of chromium oxide, the surface

area typically remains almost unchanged with respect to the original alumina support,

which indicates that the chromium oxide does not agglomerate or plug the pores of

alumina.

Recent studies indicate that using zirconia (ZrO2) as a support of chromia

catalysts instead of alumina may have interesting advantages. Among the potential

advantages of chromia-zirconia catalysts Zirconia is a catalyst support that in recent

years has attracted the attention of researchers for applications in several reactions. In

most catalyst preparations, the support precursor is hydrous zirconium oxide, which is

commonly obtained by precipitation from ZrOCl2 solutions. Typically, the precipitation

is carried out by dropwise addition of ammonium hydroxide, or alternatively, by bubbling

in the solution a gaseous flow of ammonia in an inert carrier. The precipitate is finally

washed to eliminate traces of Cl-. Without the addition of any promoter, the zirconium

oxide rapidly loses its surface area upon heating to the high temperatures required by the

dehydrogenation reactions. This loss in surface area accompanies a crystallographic

29
phase transformation from the metastable tetragonal phase into the monoclinic phase.

However, the presence of foreign species, able to interact with the surface of zirconia,

significantly retards this transformation and helps retaining relatively high surface areas.

This resistance to surface area losses by addition of foreign species has been observed on

several systems, such as sulfate zirconia, tungstated zirconia, chromia-zirconia, as well as

ceria- and lanthanum-doped zirconias. In all these cases, the preservation of surface area

parallels the retention of tetragonal zirconia. Fig. 16 illustrates this trend for a series of

chromia-zirconia catalysts. As the Cr content increased, both the surface area and

fraction of tetragonal zirconia, retained in the sample after heating at 873 K, increased.

3.2.4. Industrial units using chromia-based catalysts

The three main industrial dehydrogenation processes that employ chromia-

alumina catalysts are the Catofin, the Linde-BASF, and the Snamprogetti-Yarsinetz FBD.

In the Catofin process, the reaction takes place at sub atmospheric pressure and in the

temperature range 525-675C in 3 to 8 fixed bed adiabatic reactors, depending on the

feedstock. The catalyst is of the chromia-alumina type, containing about 20 wt %

chromia. The process does not use hydrogen or steam as a diluent or to decrease coke

formation. So, the coking rate is very fast and a cyclic operation is required. Each cycle

includes the reaction period, discharge of the reactor, and regeneration/reheating in situ.

The regeneration/reheating involves the burning of the coke. This step not only

eliminates the coke from the catalyst, but also brings the temperature up, after cooling

down due to the endothermicity of the reaction. Typically, about 2 % of the feed goes to

coke and the reaction cycles are of the order of 8 min. The complete cycle, including

30
regeneration/reheating is finished in less than 20 min. Most of the activity is regained

after the regeneration/reheating step. However, a slow irreversible deactivation does take

place over the reaction-regeneration cycles. This deactivation is due to a decrease in

surface area and a phase transformation to the less active form of -chromia. To

compensate for the activity loss, the process temperature is increased, but this

temperature increase accelerates the phase transformation and consequently the rate of

deactivation. After 1.5 to 3 years the catalyst load has to be replaced.

The Snamprogetti-Yarsinetz FBD process employs fluidized-bed reactors, without

diluents and operating at atmospheric pressure. In this process, the chromia-alumina

catalyst is continuously transferred from the reactor to the regeneration zone and back to

the reactor. The catalyst contains 12-20 wt % Cr2O3 supported on a mixture of high

transition aluminas.

3.3. Other Lower-Alkane Dehydrogenation Catalysts

In addition to the Pt- and chromia-based catalysts other materials have reported to

be active and selective for the dehydrogenation of lower alkanes. For example, zinc

titanates have shown moderate activity for isobutane dehydrogenation.43 These materials

exhibit some economic and environmental advantages. Their cost is low in comparison

to the Pt-based catalysts and they do not exhibit the toxicity of the chromia-based

catalysts. A recent study44 conducted on zinc titanate films has shown that there is a clear

correlation between dehydrogenation activity and the catalyst structure. Depending on

the Zn/Ti ratio, several phases may be present on these catalysts. When the Zn/Ti ratio

was lower than 1, the phases present were hexagonal zinc metatitanate (ZnTiO3) and

31
titanium dioxide (TiO2). At higher Zn/Ti ratios, the main phases were cubic ZnTiO3 and

cubic Zn2TiO4. The maximum dehydrogenation activity occurred for a Zn/Ti ratio close

to 2, whose stoichiometry corresponds to the cubic phase Zn2TiO4.

Similarly, sulfided nickel has shown some promising characteristics in the

dehydrogenation of isobutane.45,46 Unlike Pt, unpromoted Ni is very unselective for

dehydrogenation reactions. However, even at low dosages S binds very strongly to the

surface of Ni and forms stable surface sulfides. When the sulfur concentration is high

enough, the typically high hydrogenolysis activity of Ni is eliminated and

dehydrogenation selectivity can be very high. One of the interesting characteristics of

these catalysts is that not only the dehydrogenation rate exhibits a zeroth-order

dependence with hydrogen, but also the rate of deactivation is much less affected by

hydrogen than most typical Pt-based catalysts. This difference is probably due to a much

weaker interaction of H2 with the sulfided nickel surface than with typical Pt catalysts.

3.4. Removal of hydrogen to shift equilibrium conversion

The dehydrogenation of lower alkane is in most cases limited by thermodynamic

equilibrium. Several approaches have been used to remove part of the hydrogen

produced during the reaction from the reacting mixture, thus shifting the equilibrium

conversions to higher values. The most promising approaches involve the use of

membrane reactors and the addition of hydrogen acceptor materials. For the selective

permeation of hydrogen, both inorganic membranes and non-porous Pd membranes have

been used in most investigations. Palladium membranes are particularly effective

because they only allow the hydrogen diffusion through the membrane due to the high

32
solubility of dissociated hydrogen in the Pd bulk. The addition of Ag improves the

mechanical properties of the membrane, reducing brittleness and modifies the hydrogen

solubility, with an optimal performance at about 23 % Ag. A typical membrane reactor

configuration47 employed in the dehydrogenation of lower alkanes is illustrated in Fig.

17. The dehydrogenation catalyst is placed inside the membrane tube. Part of the

hydrogen produced by the reaction diffuses out of the reaction zone and it is swept away

by a high flow of purge gas that passes by the shell side of the tube. Recent advances in

inorganic membranes are opening interesting possibilities to employ these materials in

dehydrogenation reactors. For example, combinations of mesoporous Al2O3 membranes

and microporous silicalite-alumina membranes prepared by hydrothermal synthesis of the

silicalite in the host material have been tested with promising results in dehydrogenation

reactions. Differences in the permeabilities of hydrogen and the hydrocarbons through

these materials are large enough to produce a high permselectivity and dramatic enhances

in dehydrogenation conversion.

The second interesting alternative is to conduct the dehydrogenation reaction in

the presence of hydrogen acceptor materials. For example, the conversion of isobutane to

isobutylene was found to double its equilibrium value in the presence of an intermetallic

compound, Zr2Fe, which is able to form a stable hydride, Zr2FeH2.6. An interesting

demonstration of the effect was conducted at 470C with a chromia-alumina catalyst.48

In the presence of Zr2Fe, the initial isobutylene yield was 45 %, which almost doubles the

25 % equilibrium conversion that would be obtained if hydrogen remained in the gas

phase. However, after the intermetallic compound was converted into the hydride form,

the isobutane conversion dropped to about 20 %.

33
4. OTHER DEHYDROGENATION REACTIONS

4.1. Dehydrogenation of C6-C19 normal alkanes:

Alkylsulfonate are widely used in industry for the manufacture of detergents and

emulsifiers. Dehydrogenation plays an important part in the production of

alkylsulfonates, since normal alkanes are an available feedstock, which can be

dehydrogenated to the corresponding mono-olefins. In a subsequent alkylation step, the

mono-olefins react with aromatics to produce alkylbenzenes, which are finally

sulfonated. The typical dehydrogenation of C6-C19 normal alkanes is carried out on Pt-

based catalysts at temperatures around 450-520C. Most of the concepts discussed above

for the dehydrogenation of propane and isobutane can be applied for these catalysts. The

typical catalyst is Pt supported on alumina, which is promoted with alkali (e.g., Li) to

neutralize the acid sites and to improve selectivity and catalyst life. As discussed above,

the main cause of catalyst deactivation is coke formation, and one way to extend the life

of the catalyst is to co-feed hydrogen at high hydrogen/hydrocarbon ratios. As in the case

of lower alkane dehydrogenation the use of promoters such as Sn or In greatly improve

the catalyst performance. As opposed to the case of lower alkane dehydrogenation,

where side reactions play a minor role, when using higher alkanes, the formation of side

products such as aromatics is not uncommon. Therefore, the reaction is operated at low

conversions (e.g., lower than 15 %). As a result, the post-reaction separation process

plays a very important role. After the dehydrogenation reactor, hydrogen and light

alkanes are first separated while the olefins and unreacted alkanes are sent to the

34
alkylation unit, where the olefins react with benzene to yield the alkylbenzene. After this

unit, the unreacted alkanes are recycled to the dehydrogenation reactor.

4.2. Dehydrogenation of Ethylbenzene to Styrene:

Styrene is a major commodity used in the manufacture of plastics, synthetic

rubber, resins, and insulators. The principal form of styrene production is the direct

dehydrogenation of ethylbenzene, which can be accomplished at around 600-630C on

iron oxide catalysts promoted with various oxides. This process is normally carried out

either isothermally or adiabatically in the presence of a large excess of steam, which has

several benefits in the reaction. First of all, as shown in section 2 (see Fig. 1), the

presence of an inert moves the equilibrium toward higher conversions. Second, the

addition of steam supplies the heat required by the endothermic reaction. Finally, the

presence of steam at high temperatures helps to inhibit the formation of coke on the

catalyst surface.

Among the most common promoters added to iron oxide, the alkali oxides have

the most noticeable effects on catalytic activity and selectivity. As illustrated in Table

4,49 the addition of K in the form of K2CO3 causes an increase in activity of about an

order of magnitude over the activity of unpromoted Fe2O3. This promoting effect reaches

a maximum at about 20 wt % K2CO3, at higher K concentrations the overall conversion

starts decreasing, while the selectivity to styrene slightly increases. The selectivity is

typically high. In addition to the main dehydrogenation reaction, some side reactions

occur. The most common side reactions are cracking to benzene and ethylene or

hydrogenolysis to toluene and methane, and coke formation.

35
 Although several species co-exist in the K-promoted catalysts under reaction

conditions, the most active phase is believed to be KFeO2. This phase forms from the

interaction of a highly dispersed F3O4, mixed with Fe2O3, and K2Fe22O34, which acts as a

K source.50 The active KFeO2 phase can be found on the edges of hexagonal platelets

consisting of a solid solution of F3O4/K2Fe22O34. However, this phase is unstable at

ambient conditions, so it is only detected by in-situ characterization techniques.51

Several forms of catalyst deactivation have been identified for this type of catalysts under

ethylbenzene dehydrogenation.52 The first one is the common coke deposition, typical of

all the dehydrogenation catalyst. The second type is more unique to this type of catalysts

and it is due to the blocking of sites by the CO2 produced from the reaction of steam with

the carbon deposits. These two forms of deactivation are considered reversible because

both, coke and adsorbed CO2, can be eliminated by the same steaming treatment.

However, in addition to this reversible deactivation, other irreversible phenomena occur

in these catalysts. The most important one is the loss of K from the active phase, KFeO2.

This reaction may occur by hydrolysis (with steam) or by direct reduction (with H2),

which results in the generation of KOH. In this form, K can migrate and becomes

segregated from the Fe-containing phases. Under industrial conditions, this irreversible

process occurs very slowly, and as a result the catalyst life can be of extended to up to

two years. To stabilize the active KFeO2 phase other promoters, such as Cr and Ce, have

been used, retarding the deactivation process.

36
References

1. D. B. Kang, A. B. Anderson, J. Amer. Chem. Soc., 107, 7858 (1985)

2. C. Andrews, Ground Water 36, 705 (1998)
3. J. Cosyns, J. Chodorge,D. Commereuc, B. Torck, Hydrocarb. Proc. 77, Mar.
1998
4. D. E. Resasco and G. L. Haller, in Catalysis, Royal Society of Chemistry, Vol. 11,
379
5. F. Buonomo, D. Sanfilippo, and F. Trifiro, in Handbook of Catalysis,
6. M. P. Atkins, G. R. Evans, Erdol, Erdgas, Kohle 111, 271 (1995)
7. P. Biloen, J. N. Helle, H. Verbeek, F. M. Dautzenberg, and W. M. H. Sachtler, J.
Catal. 63, 112 (1980)
8. Y. Soma-Noto and W. M. H. Sachtler, J. Catal. 32, 315 (1974)
9. L. C. Loc, N. A. Gaidai, and S. L. Kiperman, in Proc. IX International Congr.
Catal., Calgari, Canada, Vol. 3, p. 1261 (1988)
10. F. C. Wilhelm, U.S. Patent 3,909,451 (1974)
11. T. Imai and C. W. Hung, U.S. Patent 4,866,211 (1989)
12. R. D. Cortright, E. Bergene, P. Levin, M. Natal-Santiago, and J. A. Dumesic,
Stud. Surf. Sci. Catal. 101, 1185 (1996)
13. L. K. Lok, N. A. Gaidai, B. S. Gudkov, S. L. Kiperman, S. B. Kogan, Kinet.
Katal. 27, 1365 (1986)
14. R. D. Cortright, P. Levin, and J. A. Dumesic, Ind. Eng. Chem. Res. 37, 1717,
(1998)
15. R. D. Cortright, and J. A. Dumesic, J. Catal., 157, 576 (1995)
16. G. T. Baronetti, S. R. de Miguel, O. A. Scelza, A. A. Castro, Appl. Catal. 24, 109
(1986)
17. S. R. De Miguel, G. T. Baronetti, A. A. Castro, and O. A. Scelza, Appl. Catal., 45,
61 (1988).
18. J. Margifalvi, M. Hegedus, S. Gobolos, E. Kern-Talas, P. Szedlacsek, S. Szabo, F.
Nagy, in "Proceedings, 8th Inter, Cong. Catal., Berlin, 1984," Vol. IV, p. 903.
Dechema, Frankfurt-am-Main, 1984.
19. G. J. Siri, M. L. Casella, G. F. Santori, and O. A. Ferretti, Ind. Eng. Chem., 36,
4821 (1997)
20. R. D. Cortright, and J. A. Dumesic, Appl. Catal. A, 129, 101 (1995)
21. A. H. Haner, P. N. Ross, U. Bardi, and A. Atrei, J. Vac. Sci. Tech., 10, 2719
(1992)
22 S. M. Stagg, C. A. Querini, W. E. Alvarez, and D. E. Resasco, J. Catal. 168, 75
(1997)
23. H. Verbeek, and W. M. H. Sachtler, J. Catal., 42, 257 (1976)
24. R. D. Cortright, and J. A. Dumesic, J. Catal. 148, 771 (1994)
25. D. Rajeshwer, A. G. Basrur, D. T. Gokak, and K. R. Krishnamurthy, J. Catal. 150,
135 (1994)
26. O. A. Barias, A. Holmen, E. A. Blekkan, Catalysis Today, 24, 361 (1995)
27. T. B. Massalski, Ed. in Binary alloy phase diagrams; ASTM. Philadelphia,
1986

37
28. A. Borgna, S. M. Stagg, and D. E. Resasco, J. Phys. Chem. 102, 5077 (1998)
29. Y-X. Li, K. L. Klabunde, and B. H. Davis, J. Catal. 128, 1 (1991)
30. Y. Zhou and S. M. Davis, Catal. Lett. 15, 51 (1992)
31 J. H. Sinfelt, Adv. Catal. 23, 91 (1973)
32 . P. N. Ross, J. Vac. Sci. Tech., 10, 2546 (1992)
33. J. Shen, J. M. Hill, R. M. Watwe, B. E. Spiewak, and J. A. Dumesic, J. Phys.
Chem. B, 103, 3923 (1999)
34. F. Buonomo, D. Sanfilippo, and F. Trifir, in Handbook of Catalysis, p. 2140
(1998)
35 L. Lin, T. Zao, J. Zang, and Z. Xu, Appl. Catal. 67, 11 (1990)
36. T. P. Chojnacki,., and L. D. Schmidt, J Catal., 129, 473 (1991)
37. A. El Abed, S. El Qebbaj, M. Guerin, C. Kappenstein, M. Saouabe, and P.
Marecot, J. Chim. Phys. 92, 1307 (1995)
38. G. J. Arteaga, J. A. Anderson, C. H. Rochester, J. Catal. 184, 268 (1999)
39. G. J. Antos, U.S. Patent 4,216,346 (1980)
40. R. O. Dunn, G. F. Schuette, F. M. Brinkmeyer, W. Sund, Proc. De Witt
Petrochem. Review, 1992, 1
41. K. K. Kearby, in Catalysis (P. H. Emmet, editor) Reinhold, New York, Vol. 3
p. 453 (1955)
42. F. Cavani, M. Koutyrev, F. Trifiro, A. Bartolini, D. Ghisletti, R. Iezzi, A.
Santucci, and G. del Piero, J. Catal. 158, 236 (1996)
43. A. W. Aldag and B. Okla, U. S. Patent 4,524,144 (1985)
44. Z. X. Chen, A. Derkng, W. Koot, and M. P. van Dijk, J. Catal. 161, 730 (1996)
45 D. E. Resasco, B. K. Marcus, C. S. Huang and V. A. Durante, J. Catal. 146, 40
(1994)
46. D. E. Resasco, B. K. Marcus, C-S. Huang, and V. A. Durante, US Patent
5,468,710 (1995)
47. N. Itoh, AIChE Journal 33, 1577 (1987)
48. V. V. Lunin and O. V. Chetina, Neftekhimiya, 30, 202, 1990
49. T. Hirano, Appl. Catal. 26, 65 (1986)
50. M. Muhler, R. Schlgl, and G. Ertl, J. Catal. 138, 413 (1992)
51. M. Muhler, J. Schtze, M. Wesemann, T. Rayment, A. Dent, R. Schlgl, and G.
Ertl, J. Catal. 126, 339 (1990)
52. K. Kochloefl, , in Handbook of Catalysis, p. 2151 (1998)

38
Table 1: Isobutane conversion over a series of Pt and Pt-Sn catalysts. 400 C, 12.5 Torr isobutane, 75 Torr H2, and 760 Torr
total pressure (from ref. 20)

Catalyst WHSV conversion Pt wt % total rate selectivity

-1
(h ) (moles iso-C4/ g Pt h) to isubutylene
Pt/SiO2 6.5 0.190 1.2 1.77 0.20
PtSn/SiO2 3.8 0.042 1.2 0.23 0.98
1:2 Pt-Sn/ L zeolite 3.9 0.053 0.55 0.65 0.88
1:2.5 Pt-Sn/ L zeolite 3.9 0.064 0.44 0.98 0.93
1:3.5 Pt-Sn/ L zeolite 8.2 0.030 0.58 0.73 0.98
1:2 Pt-Sn/ L + K(OH) 3.9 0.069 0.58 0.80 0.99
Pt/K/SiO2 8 0.077 1.2 0.89 0.36

WHSV = (g isobutane/ h) / (g catal.)

Total rate = (moles of isobutane converted/ h ) / (g. Pt)
Selectivity = moles isobutylene/moles of isobutane reacted
Table 2: Mulliken Population of the 6s, 6p, and 5d orbitals on three Pt Atoms in Pt and PtSn clusters

cluster Pt-6s Pt-6p Pt-5d

19 atom clusters
(constrained)
Pt19 1.74 1.24 26.60
Pt16Sn3 2.02 1.81 27.22
10 atom clusters
(unconstrained)
Pt10 1.91 1.36 26.67
Pt6Sn4 2.40 1.65 27.51
Table 3: Calculated changes in electronic energies (kJ/mol) for adsorption of -bonded ethylene, di--bonded ethylene, or

ethylidyne species on Pt and Pt-Sn clusters

cluster -species di--species Ethylidyne + H

19 atom clusters
(constrained)
Pt19 - 71 - 116 - 94
Pt16Sn3 - 54 - 87 0
10 atom clusters
(unconstrained)
Pt10 - 103 - 149 - 109
Pt6Sn4 - 91 - 91 - 18
Table 4. Effect of K2CO3 addition on activity, selectivity, and surface area of Fe2O3 catalysts. From Ref. 49

Fe2O3 K2CO3 Dehydrogenation rate Selectivity to BET area

(wt %) (wt %) (mol/g min) x 105 Styrene (%) (m2/g)
100 0 3.2 95.4 15
80 20 33.8 96.6 7.2
75 25 29.6 96.9 6.7
72 28 26.6 96.9 6.3
69 31 24.7 97.0 4.7
EQUILIBRIUM CONVERSION (%) 80

60

i=3
40
h,i=0

h=1
20
h=3

0
600 650 700 750 800 850 900

TEMPERATURE (K)
HYDROGEN CONSUMPTION (a.u.) Pt-Sn

Pt x 2

0 50 100 150 200 250 300 350

TEMPERATURE (C)
 35

Pt-Sn
30
% CONVERSION

25

20 Pt

15
0 20 40 60 80 100 120

TIME (min)
 100

Pt-Sn
95
% SELECTIVITY

90

Pt
85

80

75
0 20 40 60 80 100 120

TIME (min)
ABSORPTION COEFFICIENT (a.u.)

-15 0 15 30 45
RELATIVE ENERGY [ E - Eo ] (eV)
 b
a c

27C
Pt-Sn
Absorbance

- 70C
Pt-Sn

27C
Pt

- 70C
Pt

1550 1500 1450 1400 1350 1300

Wavenumber (cm-1)
 Pt + PtSn
Fourier Transform (a.u.)

Pt cluster
PtSn

1 2 3 4
R/
 Fourier transform (a.u.)

0
50
100
150
200

1
2
R/
3
4
 240 C
HYDROGEN CONSUMPTION

125 C

REG500

186 C
Fresh
146 C

0 100 200 300 400 500 600

TEMPERATURE (C)