Catalysis T&y, 18 ( 1993) 443-47 1 443

Elsevier Science Publishers B.V., Amsterdam

CAlTOD 230

Catalytic skeletal isomerization of linear butenes to

isobutene

A.C. Butler and C.P. Nicolaides*

CatalysisProgramme, Division of Energy Technology,CSIR, P 0 Box 395, Pretoria 0001
(Republicof South Africa)

The increased demand for isobutene, used for the production of the octane-enhancer methyl tert-
butyl ether, has generated tremendous interest in the catalytic conversion of the linear butenes to
isobutene. In this review we survey the progress made since the late 1970s in implementing the cata-
lytic skeletal isomerization reaction of these linear alkenes. Halogenated catalysts, especially those
based on alumina, and prepared using a variety of compounds of fluorine, chlorine or bromine, have
been shown to exhibit both high conversions and selectivities for the reaction, resulting in high yields
of isobutene, when water is added to the feed stream. Elution of the halogen from the catalyst leads to
the loss of catalytic activity and necessitates the continuous or discontinuous addition of the halogen
compound. As a consequence, environmental and other considerations are most likely to weigh against
the industrial usage of these catalysts. Another class of catalysts exhibiting high activities and selectiv-
ities, again in the presence of water, are the silicated altinas. No information is, however, available
on their long-term stability. Even alumina on its own displays high activity and selectivity, provided
water is co-fed with the hydrocarbon stream. More recent results obtained over other types of catalysts
such as zeolites and molecular sieves are also presented. Most promising are the results obtained with
the zeolite ferrierite which gives high yields of the branched isomer in the absence of any other addi-
tive or diluent. The catalyst also appears to be fairly stable showing no decrease in the yield of isobu-
tene after 14 days on-stream. The high yields of isobutene can be ascribed to the small channel diam-
eters which prevent the extensive dimerization or oligometization of the linear butenes or of the product
isobutene. Plans for the first large-scale demonstration plant to produce isobutene from n-butenes
using ferrierite as catalyst have already been announced in the United States.

1. INTRODUCTION

The last review on the catalytic skeletal isomerization of linear butenes to

isobutene appeared in 1977 [ 1 ] as a section in a review paper which dealt
with the activity and surface properties of fluorinated aluminas. Since that
time considerable interest has been shown in this reaction because isobutene
is the alkene precursor in the synthesis of the branched fuel ethers methyl tert-
butyl ether (MTBE) and ethyl tert-butyl ether (ETBE). MTBE in particular

l Corresponding author.

0920-5861/93/$06.00 0 1993 Elsevier Science Publishers B.V. All rights reserved.

444 A. C. Butler and C.P. Nicolaides/ Catal. Today 18 (1993) 443-4 71

is the octane-enhancer of choice to replace tetraethyllead in reformulated gas-

oline [ 2 1. Other current industrial uses of isobutene include those as the
monomer in the production of polyisobutene, as one of the monomers for
butyl rubber and polybutene manufacture, and for the production of a variety
of chemicals [ 3 1.
Currently for the production of MTBE, isobutene is predominantly ob-
tained from catalytic or steam cracking fractions [ 41. Since the supply of iso-
butene from these sources is insufficient to meet anticipated demand, alter-
native routes have been implemented or investigated. These include the
dehydrogenation of isobutane [ 5 1, the dehydration of tert-butanol [ 6 ] and
of isobutanol [ 71, which in turn can be obtained from isosynthesis [ 8-101,
the exhaustive metathetical ethenolysis of branched alkenes [ 111 and the cat-
alytic skeletal isomerization (SI) of linear butenes. In this review, we criti-
cally evaluate the progress achieved since the late 1970s in the catalytic con-
version of linear butenes to isobutene as reported in both the open and patent
literature. Our aim in preparing this non-exhaustive review of the literature
was to identify the types or classes of compounds or materials that have been
studied and used as catalysts for this reaction, to obtain an understanding of
their observed catalytic activities and selectivities and to gain an appreciation
of the reaction conditions required by each group of catalysts. The advantages
or disadvantages presented by each group are also discussed where
appropriate.
In an earlier survey [ 12 1, the kinetics and mechanism of the isomerization
reaction, the thermodynamic and physical properties of the four butenes and
the thermodynamics of the reaction system were reviewed. The reaction sys-
tem is a rather complex one since apart from the reversible skeletal isomeri-
zation reaction, the cis-trans and double-bond shift reactions of the linear
butenes are also occurring (see Fig. 1). In the reaction mechanism depicted
in Fig. 1 the rearrangement of the carbenium ion is shown to proceed via a
protonated cyclopropane ring which on opening forms the primary carben-
ium ion [ 13 1. The temperature dependence of the equilibrium concentra-
tions has also been reported [ 141 and the results show (see Fig. 2) that the
equilibrium concentration of isobutene in the mixture of butenes decreases
with increasing reaction temperature. In practice, in the catalytic conversion
of linear butenes to isobutene, the yield of isobutene is further lowered due to
the occurrence of side-reactions such as dimerization to Cs, transformation
of this product into C3 and Cs, hydrogen transfer and coking [ 151. As a con-
sequence, the SI reaction is conducted at low pressures which, as shown by
thermodynamic calculations [ 151, minimize the undesirable dimerization and
polymerization reactions.
The skeletal isomerization reaction has also been shown to be a more de-
manding reaction than both the cis-tram and double-bond shift isomeriza-
tion reactions, since it only proceeds over sites of higher acid strength [ 16,171.
A. C. Butler and C.P. Nicolaides / Catal. Today I8 (1993) 443-4 71 445

H,C - CH,- CH = CH,

II
tH+ -H+
H H
II +
H,C- CH= CH- CH,+HIH
f;ir C-;-h--H,
3 + H,C CH CH CH,
- = -

H2

II
CH3 C”3

H&-CH3 ~LH,c=c
+; -cH,

ii
Fig. 1. Reaction scheme for the isomerization reactions of the butenes.

The results, obtained using HR indicators, the dehydration of 4-methyl-2-pen-

tanol as the catalytic test reaction and a variety of acid catalysts, indicated
that there is a strong correlation between the occurrence of a given catalytic
reaction and the presence of sites of a given acid strength (see Table 1) . Sim-
ilar results were subsequently obtained by these authors using a series of sil-
ica-alumina catalysts with compositions ranging from 0 to 10096A.1203[ 17 1.
A confirmation of the above results was obtained in our laboratories by
carrying out some of the above-mentioned acid-catalyzed reactions, and some
other reactions, over the same type of catalyst as a function of temperature
[ 18 1. Acid-catalyzed reactions have been categorized as difficult/demanding
or facile from either (i) the temperature required to attain a particular reac-
tion rate over a specific catalyst [ 191, or (ii) the rate at which the reaction
proceeds over a catalyst at a fmed temperature [ 201. These previous evalua-
tions have shown that the cracking of alkanes (e.g. n-hexane) can be consid-
ered a difficult reaction requiring strongly acidic sites, whereas alcohol dehy-
dration is one of the most facile acid-catalyzed reactions [ 2 11. It has further
been demonstrated that the more difficult the reaction is, the higher must be
the acid strength of the sites [ 22,23 ] or the temperature [ 23 ] in order for the
catalytic reaction to proceed. We have therefore examined the conversion of
l-butene to &&runs-2-butene and to isobutene, l- and 2-butanol dehydra-
tion to the butenes, methanol conversion to dimethyl ether, as well as n-hex-
ane cracking as a function of the reaction temperature over the same catalyst,
viz. zeolite HZSM-5. The results in Fig. 3 show that the conversion of l-bu-
tene to ci.s/trans-Zbutene (the most facile reaction in this group of acid-cat-
alyzed reactions) and the alcohol dehydration reactions occur at lower tem-
peratures than the skeletal isomerization reaction or the n-hexane cracking
446 A.C. Butler and C.P. Nicolaides / Catal. Today 18 (1993) 443-471

0 I- BUTENE
0.5 0 ISOBUTENE
o TRB-2-BUTENE

BOO 1000
Temperature (K)

Fig. 2. Temperature dependence of the equilibrium concentrations for the isomerization reac-
tions of the butenes [ 141.

TABLE 1

Correlation between the nature of the catalytic reaction and the acid strength of the catalyst [ 17 ]

Acid strength Dehydration cis-trans Positional Positional Skeletal

Isomerization isomerization isomerization isomerization
(l)-(2) (beyond 2-
position)

4.75dH,>O.g2 +
0.g2bHR> -4.04 + + +
-4.04&HR> -6.63 + + + +
-6.63&H, + + + + +
A. C. Butler and C.P. Nicolaides / Catal. Today 18 (1993) 443-4 71 447

x l-W-- c#T~-2-EuTENE
0 2-wrbNoL-- BIJTENE
Cl METHANOL -0ME
0 I-WTANOL --BUTENE
0 I-BUTENE -I-NE
A M-HEXANE CRACKING

TempemturePC)

Fig. 3. Acid-catalyzed conversions as a function of temperature over zeolite HZSM-5 [ 18 1.

reaction (the latter being the most demanding reaction). From our results we
can rank the reactions in terms of their difficulty as follows: n-hexane crack-
ing > skeletal isomerization of linear butenes > alcohol dehydration > dou-
ble-bond shift. The SI reaction can therefore be considered to be of moderate
difficulty, requiring sites of relatively higher acid strength. This has been ex-
plained in terms of the skeletal isomerization reaction proceeding through the
thermodynamically less stable primary butyl carbenium ion intermediate
[ 12 1. Thus, whereas qalumina shows only low activity for the skeletal iso-
merization reaction, fluorinated @nninas are significantly more active [ 241.
This has been ascribed to the fact that fluorination increases the acid strength
of the aluminas [ 11. Chlorination of alumina also leads to the same results
[25].
An earlier review [ 121 surveyed the catalytic systems employed prior to
1974 in carrying out the SI reaction. These included metal halide catalysts,
alumina and aluminosilicate catalysts, phosphoric acid, phosphate and fluor-
inated alumina catalysts. From their own work as well [ 241, the authors con-
cluded that fluorinated *alumina, containing 1 wt.-% F, was the best catalyst,
exhibiting the highest activity (33.5% conversion), high selectivity (87.1%
for isobutene), a catalyst half-life of 6 1.5 hours and excellent regenerability.
Similar conclusions regarding the activity and selectivity of fluorinated cat-
alysts for this reaction were reached in that author’s 1977 review [ 1 ] in which
the preparation, catalytic activity, reactivation, surface properties and active
sites of these catalysts were described. In a report on the, upgrading of C4
cracking cuts with acid catalysts [ 3 1, the SI reaction of the n-butenes was also
briefly reviewed.
In another review by Ghosh and Kydd [26], the preparation, structure,
TABLE 2

Tabulated summaryofliterature (selectedresults)

Catalyst type Reaction Press. Space Feed and additives Conversion Xi Time-on Reference
temp. (atm) velocity (%) (%) stream
(“C) (Skh) selectivity (h)

1% F/AlrO, 450 1 ca. 0.5’ O.l2%C;, 16.24%n-C,, 40.9 89.4 - Juguin and
4.73%X,, 10.14% l-C,-, Miquel, 1981
67.67% 2-C;, 1.10% i-C,‘, (JJ=P)
H20/HC=2.2/1

0.1% F/@l,O, 450 1.1 ca. 0.8’ I-Butene, 10% H20, 51.0 63.8 1 Franz et al.,
40 ppm HF 1983 (Huels)

38.9 75.0 30

BiOCl/A&O, 480 1 ca. 3.9’ I-Butene, l-Ci/H2=1/1.5 49.4b 66.8b -

MUer et al.,
HWh03 450 1 ca. 3.8’ 0.5% i-C,-, 76% wCTS, 61.0b 42.ob 20 1982
23.5%Cg,Feed/H2=1/1.5 ( Erdiilchemie )

Ha/M203 450 1 ca. 3.8’ 0.5% &Ci, 16% u-Cis, 58.5b 42.1b 100
23.5%C4s,Feed/H2=1/1.5

CH3WGO3 475 1.05 9.4 16.3% n-C,, 4.0% i-C,, 33.7 96 45.3 Adams et al.,
44.1% l-C?, 17.5% tranr-2-Ci, 1983 (Polysar)
12.0% CiS-2-c,’ (in mol-%) ,
5.8% N2, 300 ppm CH3Cl

20 ppm Pd, 3% 450 1 ca. 0.5” O.l2%Cy, 16.24%~~C,, 40.9 90.3 - Juguin and
Si02/A1203 4.73% i-C,, 10.14% l-C,‘, Miquel, 1982
67.67% 2-C;, 1.10% i-CT, and 1984 (HP)
H20/HC=2.2/1

A12o3 482 1 ca. 1.8’ ZButene, 3.34% Hz0 30.9 85.1 3.0 Myers and
Strope, 1984
Al2o3 482 1 ca. 1.2. 2-Butene, 11.4% H20 33.2 83.5 6.25 (Phillips
Petroleum)
CCL,, 1% Pd/AlzOs 350 ca. 18 ca. 0.5 2-Butene, 2 000 ppm CC14, 31 64
Hz (4bar)
Eleazaret al.,
CCl,, 1% Pd/AlzOJ 350 ca. 13 0.5 2-Butene, Hz (ca. 1.1. bar), 34.1 85 1984
42.8% n-Butane (Engelhard)

CCl,, 1% Pd/AlzOs 350-375 IX. 26 0.5 2-Butene, H2 (ca. 1.1 bar), 51 74 682
47% n-Butane

Partially 525 0.5 7.6 1-Butene, 36.5 66.3 Sikkenga, 1985

Na+-exchanged H,/ 1-C; =2.0/l (StandardOil)
borosilicate
2.1% SiO,/A1,0, 475 2 n-Butene 49 72 1
44 76 6 Nilsen et al., 1985
6.1% Si02/Alz03 415 2 n-Butene 56 55 1 (Statoil and CIIR)
49 70 6
BPO, on 475 2 n-Butene 43 19 1
Si02/A120J 34 83 5.5

‘Silicalite’ 325 1.6 31.4 49% t&i, 49.6% n-G, 79.4 12.5 de Clippeleir et al., 1985
1.4% light HCs (Labofma)
HzO/feed=l.ll/l

0.5964 1.1% 500 2 1-Butene > 28.3 75 4 Stkker et al., 1986

Si/AlzOs (CIIR)

n-BuBr/AlzOs 447 ca. 2.6. 1-Butene,2 10 ppm n-BuBr 29.9 89.9 l-3 Sun and Gastinger, 1987
(Atlantic Richfield)

MeCl/BeO 430 ca. 0.6’ 1-Butene 29.2 76.2 Sun, 1988 (Atlantic
Richfield)
n-BuBrlMgO 430 ca.O.7’ 1-Butene/Z-Butenes= 1/ 1 38 68.0

6% WO,/AlzO, 360 ca. 1.5. lButene+ca. 0.025 atm. Hz0 42.6b 85.4b 1.25 Baker and Clark, 1987
(PlindersUniversity)

9.3%X),‘, 31.1% 300 4.2. I-Butene, Ci/Nz= l/5 1.5 Luy et al., 1988
zrO,/Aw, (INCAPE)
%
W
TABLE 2 (Continued) &

catalyst type Reaction Press. Space Feed and additives Conversion i-CT Time-on Reference
temp. (atm) velocity (%) (%) stream
(“0 W8’h) selectivity (h)

430 0.4 1-BUtC%W 25_lb*” 96.4b*c 1 La Ginestra, 1988

(Universiti ‘La
Sapienza,’ and CNR,
Rome)
20 ppm Pd, 3% 500 1 ca. 1.1’ 0.12% Propene, 16.24%n-Butane, 33.5 94.1 - Ju8uir.tand Martino, 1991
SiOz/A&O3 4.73% i-Butane, 10.14% l-Butene, (JFJ?
67.67% 2-Butene, 1.10%
i-Butene, H,G/HC= 1.331I
4.5%F/AlxOI 450 1 1.9 1-Butene, C; /H&k i/O.23 49.3 66.9 5 Sxabo et al., 1991
O.l2~N~jm~,~~~N~=ll9 (Universities of Le Havre
and Caeu, and Total)
Theta-l 339 1 1.9 1-Butene, 86.6 35.6 71 Bani et al.,
i-Cr/Nz= L/8.3 1987 (BP)
ZSM-22 550 1.6 21 I-Butene 39.8 94.5 - Rahmin et al.,
1992 (Mobil)
ZSM-23 500 1.2 160 I-Butene 38 90.6 - Haa et al.,
1992 (Mobil)
MgAFSG-31 550 1 93 1-Butene 44 77 - Gaffney and Jones,
1992 (Arco)
SAPW 1 343 1 4.7 2-Butene 50.1 53.3 - Gajda, 1992
(UGF)
Ferrierite 350 1.4 2 99.42% n-G, 50.9 80.9 336 Grandvalfet, 1992
0.44% i-CT (Shell)
*Approximatevalue calculated from specified GHSV or LHSV, based on an assumed catalyst density of 1.Og/ml. ‘r
Q&ilated from given data. 4
No indication given of amount of cracking or oligomerization products. z?
A. C. Butler and C.P. Nicolaides / Catal. Today 18 (1993) 443-471 451

acidity and catalytic properties of fluorinated catalysts were summarized.

Their review, however, did not cover the skeletal isomer&ion of linear al-
kenes. The kinetic modelling of the reaction has also been the subject of some
previous reports [ 27,28 1.
This review of the literature on the catalytic skeletal isomerization reaction
of the linear butenes is divided into three sections: the first and second sec-
tions dealing with h~ogenated and non-h~ogenated catalysts respectively
covering the period up to 1991. The more recent work on zeolites and molec-
ular sieves is covered in the third section. In Table 2, we have tabulated in
chronological order the patents and papers, listing the catalyst type, the reac-
tion conditions, the type of feed and additives, as well as selected catalytic
results from each paper or patent. For the results in Table 2 and ~ou~out
this review, the following definitions hold:
% conversion = % ~-CT+ % non-Cc in product stream
% selectivity = [% i-C,=/ (% i-C,= + % non-C,= ) ] X 100
96yield = % i-C,= in product stream = (% conv. x 96 sel. ) / 100
where ~-CT=isobutene, and non-CT =products formed other than the
butenes.
In our tabulation or discussion of the literature results and where only the
percentage conversion and percentage selectivity are given, the percentage
yield can be readily calculated from the equation given above.

2. HALOGENATED CATALYSTS

Most of the catalysts reviewed in this section were prepared by the addition
of a halogen compound to alumina, except for those catalysts described in one
of the patents, where the use of alkaline earth oxides is reported.
In a patent from IFP [29 1, the catalytic properties of ~uo~nated ahuninas,
for the skeletal isomerization reaction in the presence of water, were de-
scribed. The catalysts were prepared by impregnating activated alumina with
a solution of hydrofluoric acid. The solid was then dried at 100°C followed
by calcination at 530 “C. The olefin feed used was a C4cut whose composition
is listed in Table 3.
The catalytic reactions were conducted at 450°C and a pressure of 1 atm
using a catalyst with a fluorine content of 1% by mass. The influence of the
water, which is co-fed with the hydrocarbon stream, was examined by varying
the water/hydrocarbon (H,O/HC) molar ratio (see Table 4). In the absence
of water and at an LHSV= 2, high conversions of around 50% were obtained
together with relatively low isobutene selectivities (50.3%) giving a 25.5%
yield of isobutene. Under the above reaction conditions, rapid coking of the
catalyst was also observed. With HzO/HC= 2.2 and LHSV= 1, the conver-
sion dropped to 40.9% but the selectivity increased to 89.4% (see Tables 3
452 A.C. Butler and C.P. Nicolaides/ Catal. Today 18 (1993) 443-471

TABLE 3

Composition of feed and product stream’ obtained over fluorinated alumina [ 291

Compound Feed stream’ Product streamb

Methane 0.04
Ethane + ethylene - 0.17
Propane 0.01
Propene 0.12 2.17
n-Butane 16.24 17.41
Isobutane 4.73 4.64
1-Butene 10.14 12.39
2-Butene 67.67 33.58
Isobutene 1.10 29.59
Polymers
‘Obtained at 45O”C, 1 atm, LHSV= 1 and an H,O/HC=2.2.
bMass 46.

TABLE 4

Effect of water and percentage F on catalytic properties of fluorinated alumina* [ 29 ]

F 06) LHSV H,O/HC Conversion (%) Selectivity (% ) Yield (%) Coke (%)

1 2 1.35 30.6 93.0 28.5 0.5

1 1 2.2 40.9 89.4b 36.6 0.2
1 2 - 50.7 50.3 25.5 28.3
1 2 11 13.2 95.6 12.6 0.01
0.05 2 1.35 4.5 95.7 4.3 0.01
10 2 1.35 43.6 71.2 31.0 0.6
’ Obtained at 450°C and 1 atm.
b Detailed analysis is given in Table 3.

and 4). The yield of isobutene therefore increased from 25.5% to 36.6% (even
at the lower LHSV) upon the addition of water to the feed stream (which also
in effect reduces the partial pressure of the olefin to 0.3). At higher H,O/HC
levels, the conversion was found to decrease. Also in the presence of water,
coking of the catalyst was minimal. It was further shown that decreasing the
fluorine content to 0.05% resulted in a dramatic decrease in conversion
whereas higher fluorine loadings gave lower selectivities. The usefulness of
water in the feed stream and the levels of halogen required in fluorinated cat-
alysts for the SI reaction of linear butenes was thus demonstrated. Similar
results were obtained with this catalytic system in the skeletal isomerization
of 1-pentene (35.2% conversion and a selectivity to isopentenes of 91.4%).
However, since no information is given on the use of other additives or di-
luents, from the data given in this patent alone it is difficult to evaluate the
A. C. Butler and C.P. Nicolaides I Catal. Today 18 (1993) 443-4 71 453

role of water, whether it is acting only as a diluent, resulting in lower partial

pressures of the olefin, or actively influencing the catalytic sites.
This issue was addressed by Franz et al. [ 301 who examined in more detail
the beneficial effect of added water on the activity, selectivity and life-time of
fluorinated catalysts. Three separate experiments were conducted in which
pure 1-butene was used, or was diluted with hydrogen or water. In the last two
experiments the same level of dilution was used i.e. diluent/HC=0.4. The
results obtained are shown in Fig. 4 and it can be observed that the addition
of water (containing 40 ppm HF corresponding to the fluorine concentration
in the aqueous phase downstream of the reactor) to the 1-butene feed resulted
in a much slower deactivation of the catalyst while at the same time improv-

Time-on-stream (hours1

lb)

Tim-on-stream (hours)

il;-;‘i

5 15 25
Time-al-strean (hours)

Fig. 4. Catalytic properties of fluorinated alumina. (a) Plot of percentage conversion versus
time-on-stream; ( 1) with added water, (2) with hydrogen as diluent, and (3) with no added
water or diluent. (b) Plot of percentage yield versus time-on-stream. (c) Plot of percentage
selectivity versus time-on-stream; ( I)-( 3) as above [ 301.
454 A. C. Butler and C. P. Nicolaides/ Catal. Today 18 (I 993) 443-4 71

ing the yield of isobutene, as compared with the runs where either no diluent
or hydrogen was used. The use of water as additive therefore has a more pro-
nounced effect than just that of an inert diluent such as hydrogen. The addi-
tion of water was not found to deactivate or damage the catalyst irreversibly.
Fluorine is not immediately eluted from the catalyst and only a few ppm of
fluorine are detected in the aqueous phase downstream of the reactor. The
fluorine discharged from the catalyst is replenished by adding HF (or CF4)
to the fresh water stream. Regeneration of the catalyst by calcination at the
reaction temperature was shown to be possible, provided that the fluorine is
replenished, either continuously or discontinuously. Data on the activity and
selectivity of the catalyst after twelve regenerations and 1000 hours of total
operating time are reported [ 301, Furthermore, the reaction was shown to
proceed at 6 bar and with higher amounts of water (up to HIO/HC = 4 in the
feed stream). Recycling of the water containing the fluorine effluent was found
to be as effective as addition of the F-compounds to the feed water in main-
taining catalyst activity. Replacement of the l-butene feed with a Cq cut, or
an n-butene equilibrium mixture, or a cis/truns-2-butene mixture, or pure cis-
2-butene, yielded similar results. In the experiment with the C4cut, the results
show that the butanes behave as inerts. The isomerization of l-pentene to
isopentene was also carried out with results analogous to those obtained with
the butenes.
Mtiller et al. [ 3 1 ] reported on the use of aluminas specially prepared from
aluminium hydroxide and then impregnated with various halogen com-
pounds such as BiOCl, BF3, ZrOClz or HCl. The catalytic reactions were car-
ried out in the presence of hydrogen and were conducted in both fixed-bed
and fluidized-bed reactors, With the latter, portions of the catalyst can be
subjected to continuous or discontinuous removal to a regeneration unit and
then fed back to the reactor after regeneration. Selected catalytic results from
these systems are given in Table 2 where it can be observed that these cata-
lysts produce yields of between 25-32% at MHSVs of about 3.8. The space-
time yields of these catalysts are therefore higher than those of the catalytic
systems surveyed previously. With only A1203or with ZrOClJcommercial
alumina as catalysts, both lower conversions and selectivities were obtained.
The effect of particle size and pore radii on the activity of alumina catalysts
activated with methyl chloride (230-300 ppm in feed stream) was also ex-
amined [ 321. It was shown that increasing the particle size from 0.7x 10D3to
3 17.5~10~~ cm and the percentage of pores having radii between 100 and
1000 8, from 1’3to 32%, results in significantly lower yields of isobutene at
MHSVs > 10. Some of the results have been tabulated in Table 5.
Eleazar and his co-workers [ 331, in their attempts to design a catalyst for
the SI reaction that does not require frequent regeneration and high temper-
ature, and to strike a balance between the production of isobutene on the one
hand, and the production of alkanes on the other, investigated the use of a 1%
A.C. Butler and C.P. Nicolaides / Catal. Today 18 (1993) 443-471 455

TABLE 5

Effect of particle size and pore size distribution on catalytic activity of alumina catalysts activated
with methyl chloride [ 32 ]

Particle size (cmx 10S3) 0.7 80 317.5

Surface area ( m2/g) 177 189 259
Pore volume (cc/g) 0.48 0.86 0.91
Pore size ~~~bution:
1OOto 1ooOOA 1%) 13 23 32.5
Lessthan loo‘& (%) 87 77 67.5
Reactor temperature ( “C) 474-475 477-478 472-474
Inlet pressure (mmHg) 796 765-766 765
MHSV 10.3 15 15.6
Time-on-stream (h ) 3.3 3.3 17.5
Conversion (% ) 36.9 36.8 26.6
Selectivity (% ) 95 93 91
Yield (%) 35 34 24

TABLE 6

Effect of n-butane (%) feed on activity and selectivity of ~~~enat~ 1%Pd/q+Os catalyst f 33 3

Hydrogen pressure (psia ) 60 16

n-Butane in feed (%) 0.7 42.8
Total pressure (psig ) 156 180
Conversion (%) 30.5 34.1

~l~tivities (%)
c 3+ 1 1
n-Butane 16 4
Isobutane 2 2
Isobutene 78 85
Amylenes 1 2
c 2 6
Y;k;d (%) 24 29

Pd/tt_A1203 catalyst pretreated with halogen compounds. This two-compo-

nent catalyst, which is then treated with a third component (CCL), was used
at 350°C and an MHSV=O.5. The reactant also contained n-butane and hy-
drogen, the latter being used to suppress the formation of coke-forming hy-
drocarbons. The results obtained on increasing the amount of n-butane from
0.7 to 42.8%, i.e. to the dilution levels which will be encountered in industrial
applications, are given in Table 6. The effect of pressure was also studied and
it was observed that increasing the pressure from 13 to 26 atm increased the
conversion from 34 to 5 l%, at the expense of lower selectivities. A catalyst
cycle-time of 682 hours was indicated.
The preparation and use of brominated alumina catalysts have been de-
scribed by Sun and Gastinger [ 34 1. The bromine compounds employed for
456 A.C. Butler and C.P. Nicolaides/ Catal. Today 18 (1993) 443-471

the vapour-phase bromination of the aluminas included HBr, n-butyl bro-

mide, methyl bromide and bromohydrin. The bromination of the oxides was
carried out at 400°C and the catalytic evaluation at 425-450°C. From the
latter it was observed that in terms of the activity and selectivity of these SI
catalysts, the vapour-phase impregnation gave superior catalysts compared
with the liquid-phase method, using, for example, an aqueous solution of
NH4Br (see Table 7). Other results showed that the addition of the halogen
compounds can be carried out either in a continuous manner or in a pulse
mode. It was also demonstrated that the activity of the catalyst can be main-
tained even by the recycling of the Cs+ fraction which contains some of the
eluted halogens. From their comparative studies with chlorinated and fluori-
nated catalysts, the authors concluded that the brominated aluminas were
more active and selective for the SI reaction than either of the former two
types of catalyst. A comparison of the data given in this patent with those
obtained by Mtiller et al. and presented above, does not, however, show that
brominated catalysts do indeed offer higher yields or higher space-time yields
than, for example, chlorinated catalysts.
In another patent from the Atlantic Richfield Company [ 35 1, the use of
halogenated alkaline earth oxides such as magnesium oxide, calcium oxide
and beryllium oxide is reported. The skeletal isomerization catalysts were
prepared by treatment of the above oxides with either bromine or chlorine
compounds. The catalytic reactions were again conducted at temperatures
around 430-45O”C, but at much lower space velocities (GHSV=270 h-l)
than those employed for the alumina catalysts (see Table 7 ) . This could be
an indication that halogenated alkaline earth oxides are of lower activity than
the corresponding alumina catalysts. Further, a comparison of the results ob-
tained with the three types of alkaline earth catalysts indicates that those of
magnesium and beryllium (see Table 2) are more active than that of calcium.
Further information on the side-products arising from the parallel and con-
secutive reactions to the SI reaction, using a fluorinated alumina catalyst, was
recently reported by Szabo et al. [ 15 1. The reactions were studied as a func-

TABLE 7

Activities and selectivities obtained over brominated alumina catalysts [ 341

Br source HBr m&r ( aq.1 n-BuBr n-BuBr

Br quantity 50 ml 25ml 210 ppm
Addition method Pulse Impregnation Pulse Continuous
Temperature ( ’ C ) 450 425 449 448
GHSV 1250 1200 1050 1035
Conversion (%) 29 <2 31.9 29.9
Selectivity (%) 91.4 88.0 89.9
Yield (%) 21 28.1 26.9
A.C. Butler and C.P. Nicolaides / Catal. Today 18 (1993) 443-471 451

tion of temperature, 1-butene partial pressure, contact time and time-on-

stream. The composition of the products formed over a range of temperatures
is shown in Table 8. Formation of the dimer is favoured at low temperatures
but decreases above 150°C and when the hydrocarbon pressure is reduced. It
was also shown that the dimer formed is the source of the C3 and C5 species
observed, that the by-products are important at all temperatures and that the
yield of isobutene is highest (about 22%) between 400 and 450°C. At lower
temperatures, the skeletal isomerization is too slow whereas at higher temper-
atures, the thermodynamics are unfavourable and side-products become too
important even at moderate conversion levels. In this study, the effects of
butene dilution, with nitrogen and water, were again examined (see Table 8 ) .
In agreement with the previous work described above, it was observed that
the addition of water (H,O/HC=0.23) to the feed markedly decreases the
amount of by-products and enhances both the selectivity to isobutene (36%
at the maximum) and the catalyst life-time. In terms of the dramatic influ-
ence of the water, the authors speculated that in the presence of water, the
active sites are either blocked or converted into less active, but more seleo
tive, sites. Another conclusion reached is that as the activity drops with in-
creasing time-on-stream, the rates of the secondary reactions decrease more
markedly than that of the SI reaction, causing the concentration of isobutene
to go through a maximum.
From the above survey of the publications on halogenated catalysts for the
SI reaction, it can be concluded that this type of catalyst does exhibit suffl-
cient activity and selectivity (at relatively high temperatures) for the skeletal
isomerization reaction of both the linear butenes and pentenes. Further, the

TABLE 8

Conversion of l-butene over fluorinated alumina: composition (wt.-%) of the products (space time
7= 0.536 h; catalyst age 5h) as a function of reaction temperature [ 151

Temperature ( “C) 300 450 450 450

N2/C4 2 2 9 9( +IW)
c2 0.02 0.62 0.31 0.15
c3 2.5 13.6 6.9 4.9
Isobutane +butane 2.9 9.8 1.9 0.7
I-Butene 12.2 5.7 11.7 13.7
Isobutene 10.5 13.2 24.0 33.0
tranr-2-Butene 30.8 9.2 18.0 21.0
cis-2-Butene 20.8 9.6 13.8 15.9
Isopentane+pentane 0.15 0.85 1.3 0.24
Pentenes 8.9 22.0 12.1 8.1
CS 5.3 8.9 9.9 1.9
C7 +c* 5.9 6.0 0.1 0.3
mol C3/mol C, 0.47 0.99 0.86 0.98
mol CJmol Cs 0.01 0.21 0.094 0.24
458 A.C. Butler and C.P. Nicolaides/ Catal. Today 18 (1993) 443-471

deactivation of the catalyst due to the elution of the halogens appears to have
been minimized by either the continuous or intermittent addition of halogen
to the feed. However, only in two reports is an indication given of the long-
term stability of these catalysts. Also, the role of the halogens in corrosion
promotion, especially that of fluorine, and the deleterious environmental ef-
fects of many halogenated organic compounds are well known. For these rea-
sons, industrial processes which avoid the use of compounds of this type are
likely to be more favourably received by industry and environmentalists alike.
As a consequence, the following group of catalysts, described in Sections 3
and 4, represents a logical development away from the previous class, and it
can be safely assumed that future catalytic research will concentrate more on
non-halogenated catalysts.

3. NON-HALOGENATED CATALYSTS

As an alternative to, or an improvement on, the fluorinated alumina cata-

lysts for the skeletal isomerization of n-butenes, Juguin and Miquel [ 36,371
described the preparation and use of a catalyst consisting of a r-alumina of
low alkali and alkaline-earth metals content, which had been impregnated with
a solution containing ethyl silicate, aqueous palladium nitrate and ethanol.
The preparation of these silicated aluminas was earlier described by the group
at Snamprogetti [ 38 1. After hydrolysis, drying and calcining, the final cata-
lyst consisted of 3% silica, 20 ppm Pd and the balance alumina. The catalyst
consisted therefore of alumina to which a specific amount of silica is added.
The activity and selectivity of this silicated alumina catalyst were determined
in both the presence and absence of water. As in the case of their previous
work [ 29 1, the isomerization feed was a C4 cut of the following composition:
propene 0.196,butanes 2 1.O%,isobutene 1.1% and linear butenes 77.8%. Op-
erating temperature and pressure were 450 “C and 1 bar, respectively. The
results (Table 9) show that a significant increase in isobutene yield is ob-
tained (from 29.0 to 36.9%) for a simultaneous increase in water flow (HzO/
TABLE 9

Influence of water and SiOl (%) on activity and selectivity of Pd+ Si01/A1203 catalysts’ [ 36,371

Catalyst LHSV H,O/HC Coke (%) Conversion Selectivity Yield

additives after 10 h (%) (%) (%)
operation

3% SiOl + 20 ppm Pd 2 1.35 0.4 31.5 91.9 29.0

3% SiOz + 20 ppm Pd 1 2.2 0.3 40.9 90.3 36.9
3% Si02 +20 ppm Pd 2 - 25.5 51.7 46.1 24.1
ll%SiO,+20ppmPd 2 1.35 0.4 31.4 83.8 26.3
‘At 450°C and 1 atm pressure.
A.C. Butler and C.P. Nicolaides / Catal. Today 18 (1993) 443-471 459

HG2.2) and halving of the hydrocarbon flow. It can be noted that the yield
of isobutene obtained here with the silicated aluminas is almost exactly equal
to that obtained with the halogenated catalysts under the identical reaction
conditions (Table 4). The results also show that a significant increase in cok-
ing is achieved when the water co-feed is omitted. However, a 10% decrease
in isobutene yield is observed when the silica content is increased from 3% to
11%. The skeletal isomerization of 1-pentene was also examined over these
catalysts. As indicated by Juguin and Miquel, the purpose of the palladium in
the catalyst is to facilitate the regeneration of the catalyst, particularly for the
combustion of high-molecular-weight products and coke which are deposited
on the catalyst during the reaction. In the presence of palladium, lower regen-
eration temperatures are employed (starting temperatures of 200-250 ’ C ver-
sus 325-375”(Z) and also lower oxygen concentrations, resulting in a lower
combustion velocity.
Subsequently [ 391 it was shown that if the same type of catalyst (i.e.
Pd+ Si02/A1203) is prepared from spheres of alumina manufactured by co-
agulation in oil drops, rather than from cubic r-alumina, then higher conver-
sions and selectivities are obtained (Table 10). The experiments were again
conducted in the presence of water (H,O/HC = 1.33- 1.35 ). In all of these
studies on silicated aluminas [ 36,37,39’], however, no indication is given of
the catalyst cycle-times.
Silicated aluminas, prepared using different amounts of tetraetboxysilane,
were also investigated by Nilsen et al. [ 401. In this study, the catalytic exper-
iments were conducted in the absence of water and the resultant selectivities
were lower (Table 2) than those obtained by Juguin and his co-workers
[ 36,37,39]. It was also observed by these workers that none of the silicated
alumina catalysts showed higher selectivities than alumina itself. The cata-
lysts with a low proportion of silicon gave much higher conversions and yields
of isobutene than alumina. In the range of 0.75 to 6.1% SiOZ, the activity
increased with surface silicon amount, but the selectivity to isobutene de-
creased. Further, the silication of alumina was found to improve the deacti-
vation profile.

TABLE 10

Influence of type of alumina on activity and selectivity of Pd+ SiOJA1203 type catalysts in the pres-
ence of wate? [ 391

Alumina type Conversion (%) Selectivity (%) Yield (%)

Commercial cubic pAlzOJ 31.5 91.9 29.0

Alumina prepared by
coagulation in oil drops 33.5 94.1 31.5
‘At 5OO”C,1 atm, LHSV=2, H20/HC= 1.33-1.35.
460 A.C. Butler and C.P. Nicolaides/ Catal. Today 18 (1993) 443-471

The effect of lanthanum addition to silicated aluminas, via impregnation

with a lanthanum nitrate solution followed by calcination at 5 50 ’ C, was stud-
ied by Stiicker et al. [ 411. Once again it was shown that silicated alumina
catalysts gave high yields of isobutene, but the activity of the catalysts de-
creased with increasing La content (Table 11) . Using unmodified alumina as
catalyst, the yield of isobutene was approximately 1/ 10 of that of the silicated
material.
Another evaluation of silicated aluminas, which included characterization
by elemental analysis, X-ray photoelectron spectroscopy, infrared spectros-
copy and reaction kinetics using the SI reaction of l-butene as the catalytic
test reaction, has been reported by Nilsen et al. [ 421. The catalysts were pre-
pared by the condensation reaction of tetraethoxysilane with the surface of
the alumina. The silicon-containing groups that formed on the alumina sur-
face, at less than monolayer coverage, induced strong acidity. The’ presence
of both Bronsted and Lewis acid sites was detected by infrared spectroscopy.
The catalytic reaction was carried out in a flow reactor at atmospheric pres-
sure and at 450-525 ‘C. The reaction was found to be first order in the reac-
tant, and the activity increased with increasing surface acidity, i.e. with in-
creasing Si content of the surface and apparently with increasing Bronsted
acidity. A comparison of the activities of the commercial silica-alumina and
silicated-alumina (6 wt.-% SiOZ) catalysts prepared by the reaction of
Si(OC,Hs), with the A1203 surface indicates that activity exceeding that of
silica-alumina can be achieved when Si is restricted to a monolayer (see Fig.
5). It was inferred, on the basis of the similarities in catalytic activity and
infrared spectra, that the catalytic sites are nearly the same in the two classes
of materials.

TABLE 11

Skeletal isomexization of l-butene over La-modified silicated alumina at 475 ‘C [ 4 1 ]

Amount of Yield of Selectivity to

isobutene (% ) isobutene (%)
Si (%) La (%)

0 0 2.5 55
0 3.1 0.5 70
0 4.5 0.3 55
1.1 0 20 77
1.1 2.5 1.5 85
1.1 4.9 0.5 70
2.4 0 19 75
2.4 2.5 5.5 65
2.4 5.2 2.0 IO
*Measured at 6 h on-stream, WHSV = 2.
 461

HARSH&V “SILICATED A&’

io
Time -on-stream fmin)

Fig. 5. Activities and stabilities of commercial s~i~~~~a catalysts compared with those
prepared from [Si(OC,H,),] and Alz03 (samples 1 and 4; 2.9 and 0.96 Si groups per nm re-
spectively). Conversion determined at 5OO”C,1 atm, I-butene partial pressure of 0.57 atm and
spacevelocityof2.2~10-smoI/(gofcat* s) [42].

Recently, 3asini et al. f43 1 studied the gas-solid intem~ion phenomena

occurring during the catalytic reaction over silicated ~umiua by vibrations
spectroscopic experiments. This was done both with transmittance experi-
ments on self-supported wafers of solid samples and with diffuse reflectance
investigations inside the catalytic reaction environment in a plug-flow micro-
reactor. The spectra recorded during the interactions led the authors to pro-
pose that the catalytic isome~tion occurs through a tempemt~e-de~ndent
proton transfer equihbrium between an OH surface group and an H-bonded
internal olefin.
In a patent by Myers and Strope [ 44 1, it was demonstrated that even pure
alumina, without any other catalyst additive, was also an active and selective
catalyst for the SI reaction at 480°C provided water is added to the olefin
feed (at 3.34 mol-~) or to the r~eneration gas (N&ir= lo/ 1). The data
(Table 12) show that the presence of water in the n-butene feed results in an
increase in both the conversion and selectivity and that the presence of water
in the butene feed is more effective than the presence of water in the regen-
eration gas. In other experiments it was observed that water in the feed causes
a greater retention of catalyst activity than the presence of water only in the
462 A.C. Butler and C.P. Nicolaides/ Catal. Today 18 (1993) 443-471

TABLE 12

Influence of water on catalytic activity and selectivity of aluminaa for the I reactionb [ 44 ]

Hz0 content (mol-%) 3.34 3.34 0

Regeneration gas & dry wet wet
Time-on-stream (h) 1 1 1 0.5
Conversion (%) 14.0 35.5 40.3 25.9
Selectivity (%) 53.7 68.8 65.3 81.5
Yield (%) 7.5 24.4 26.3 21.1
‘Catapal alumina, surface area 2 19 m*/g, sodium content 0.003 mass%, pellets 1/ 16 in.
bAt ca. 48O”C, 1 atm and LHSV=2.0

regeneration gas. The only pretreatment carried out on the alumina was drying
at 350°C for four hours under nitrogen. This beneficial effect of the small
amounts of added water on the skeletal isomerization activity of alumina was
also presented in an earlier paper [ 45 1. It can thus be concluded that the
improvement in activity and selectivity for the SI reaction, upon addition of
water to the feed streams, applies to the cases of both halogenated and non-
halogenated catalysts and even to alumina itself. Further understanding of
this beneficial effect of the water could help in the designing of better catalysts
for this reaction.
Since the addition of SOS- to supports such as ZrOz, TiOz, FezOs and HfOz
produces an enhancement in their catalytic activities for acid-catalyzed reac-
tions, Luy et al. [ 461 studied the activity of a series of sulphated ZrOJAl203
samples. The catalysts were prepared by impregnating JJ-A~~O~ with repeated
applications of an aqueous solution of zirconyl chloride at room temperature,
followed by washing with NH,OH solution and impregnation with a dilute
solution of HzS04. The materials were then dried, calcined and kept under
helium until used. These catalysts, together with a sulphated alumina, a pure
alumina and a pure silica-alumina were tested in an isothermal plug-flow re-
actor at the relatively low temperature of 300 ’ C and at 1 atm, where the mass
of each catalyst was adjusted so as to obtain the same catalyst surface area in
the reactor (75 m* ). The catalysts were activated at 500°C for two hours in
air flow. Table 13 lists the initial conversions extrapolated to zero time and,
in parentheses, the conversion obtained at 90 minutes on-stream. Pure Al203
yielded equilibrium values for the three linear butene isomers, with no skele-
tal isomerization evident. With Si02-A1203, only traces of isobutene were
detected. The addition of Zr02 to A1203 produced a very sharp decrease in
the conversion to the cid truns-2-butene. However, when the sulphate was
added to Zr02/A1203, not only was the n-butenes equilibrium distribution
observed, but significant amounts of the skeletal isomer were also formed.
Conversion to isobutene showed an increase with increasing ZrOz content.
Further, sulphate ion addition to pure A1203 also yields isobutene, although
A. C. Butler and C. P. Nicolaides / Catal. Today I8 (1993) 443-4 71 463

TABLE 13

Specific smface areas of, and conversions’ obtained over, sulphated ZrOJAl,O, and other catalysts
at 300 “C 1461

Catalyst S, (BET) Conversion (%)

(m2g-‘)
cis-2-C,’ trans-2-C; i-C,-

4203 200 31.9 (31.9) 49.6 (49.6) 0 (0)

Si02-A1203 559 31.4 (31.4) 49.2 (49.2) Traces (tr)
1.6% zrOJAl2O3 190 12.4 (12.5) 6.3 (6.3) 0 (0)
6.5% ZrO2/Al2O3 177 0.8 (0.8) 1.1 (1.1) 0 (0)
15.4% ZtoJAl2O3 156 1.9 (1.9) 1.3 (1.3) 0 (0)
I .6% ZKXJAl2O3 + 7.1% SOi- 193 28.5 (30.4) 46.5 (48.1) 8.8 (4.3)
6.5% ZrOJAl2O3 +6.6% SOf- 184 30.3 (30.7) 47.8 (48.4) 9.0 (3.3)
14.3%ZlQ/Al203 +9.1%SOj- 172 22.6 (30.8) 48.9 (48.0) 10.0 (4.0)
31.1% ZlOJAl2O3 +9.3% SOf- 174 28.9 (31.1) 44.8 (46.1) 13.5 (5.8)
Al203 + 8.0% SOi- 184 30.2 (31.1) 46.6 (48.2) 7.8 (3.2)

“Initialconversion extrapolated to zero time and, in brackets, values obtained at 90 min on-stream.

the effect is not as significant as in the case of ZrOz/Al,03. For all these sys-
tems, however, the yield of isobutene obtained was far below the equilibrium
value.
The improvement of the selectivity of SI catalysts via the neutralization of
some of the acid sites has also received attention [ 47 1. The approach used
involves the exchange with alkali metal cations of the acid sites on the surface
of the support or binder, whereas the acid sites of the zeolite or molecular
sieve catalyst are not exchanged to a high degree. This selective exchange is
achieved through the employment of organic ligands which can complex the
metal cations but are too large to enter the pores of the molecular sieve. The
organic ligands, dissolved in toluene, can thus transport the complexed cation
to the unhindered support acid site, but because of their size, cannot transport
the cation to the acid sites situated in the pores of the sieve. As a consequence,
the latter type of acid sites are not neutralized to a large extent. For the cata-
lyst under investigation (20% borosilicate in alumina), it was shown that the
partially neutralized system exhibited a considerable increase in selectivity
(from 36.6 to 66.3%) with a concomitant decrease in activity (from 64.5 to
36.5% conversion) which gave, however, the same percentage yield of isobu-
tene. The data, therefore, did not demonstrate the beneficial effect of the neu-
tralization process other than to maintain the inverse relationship between
conversion and selectivity.
In another report [ 481 the use of a “silicalite type” catalyst for the skeletal
isomerization of n-butenes was described. It was shown that with this type of
catalyst, and under the reaction conditions employed (see Table 2 ) , high con-
464 A. C. Butler and C.P. Nicolaides/ Catal. Today 18 (1993) 443-4 71

version together with extremely low selectivities to isobutene ( < 16%) were
obtained, even in the presence of steam.
Another group of compounds that has been investigated as catalysts for the
SI reaction are the phosphates. In the paper by Nilsen et al. [ 401, results are
presented on the maximization of both the surface areas and catalytic activi-
ties of various boron phosphate preparations, silicated boron phosphate and
boron phosphate supported on both alumina and silicated alumina. The cat-
alytic evaluations were carried out at 475 ‘C and with an MHSV= 2. The re-
sults indicated that the silication of the boron phosphate appears to stabilize
the surface area and substantially improve the catalytic activity of the systems
(yield of isobutene increased from 24 to 36%). In contrast to the non-sili-
cated catalyst, the silicated equivalents maintained their high selectivities
throughout the run (up to 6.5 hours on-stream). The effects of the calcination
and silication temperatures, the type of silicating agent, steaming and the P/
B ratio were found to be small, as long as the silicon content was in the range
of 8-l 5%. The boron phosphate/alumina catalysts exhibited similar activi-
ties to, but better selectivities than, the silicated alumina catalysts. A signifi-
cant difference between these last two types of catalysts was in the type of by-
products produced. The silicated boron phosphate produced mainly Cs,
whereas the silicated alumina produced C 5+ , indicating a higher cracking ac-
tivity for the latter catalyst which in turn shows that the silicated aluminas
have sites of stronger acidity.
La Ginestra et al. [49] reported on their investigation into the catalytic
activity of several crystalline tin phosphates, prepared by different methods
such as refluxing or hydrothermal treatment. The results showed that the acidic
properties of these materials strongly depend on the values of the different
physical and chemical parameters characterizing their preparation or the
thermal pretreatment of the catalysts. Despite the similarity of the X-ray
powder patterns of the various tin phosphates, the chemical, thermoanalyti-
cal and catalytic data were different. The latter were obtained from the SI
reaction of 1-butene at 430 oC (after pretreatment in air at 450’ C ) . The re-
sults showed that the most active materials for the skeletal isomerization re-
action were the less crystalline materials. Their activity increased with in-
creasing P/Sri ratio above the stoichiometric P/Sri== 2. A typical result is given
in Table 2. No indication, however, is given of any side-products. The strong
acid sites of the tin phosphates, presumably responsible for the SI reaction,
were attributed to an excess of adsorbed phosphoric acid.
A study on the preparation, conditioning and use of a catalyst consisting of
tungsten oxide/alumina has also appeared [ 501. The catalyst was prepared
by impregnating an alumina sample with sodium tungstate, followed by drying,
washing with acid and drying at 160 oC. Catalytic activity was only achieved
after a partial reduction of the yellow W03 component of the catalyst to a
dark blue oxide. The catalytic isomerizations of 1-butene, 1pentene and l-
A.C. Butlerand C.P. ~ic~laides / Catai. Today 18 (1993) 443-471 465

hexene were carried out as a function of preconditioning treatment and type

of carrier gas. For l-butene, the use of Hz/H20 for precon~tioning at 380°C
and as carrier gas promoted the longest catalyst life-time. It appears that the
feed stream should contain some hydrogen, since running in atgon resulted
in rapid catalyst deactivation. Pretreatment in hydrogen only did not produce
any si~i~~t differences in the case of l-butene. It did, however, make a
difference in the reaction of 1-pentene. The different preconditioning and re-
generation procedures and carrier gas requirements for each of the two al-
kenes are discussed in detail by the authors. It was also noted that the reaction
conditions employed did not favour the metathesis reaction. The skeletal iso-
merization reaction of higher alkenes such as 1-octene and 1-dodecene and a
Fischer-Tropsch product was also examined.

4. ZEOLITES AND MOLECULAR SIEVES

Barri et al. [ 5 1] reported on the suitability of ZSM-5,ZSM-23 and Theta-

1 zeolites as catalysts for the SI reaction. Using a diluted stream (Nz/
HC= 8.3) it was found that Theta- 1 is more than three times more selective
for isobutene formation than ZSM-5 (at 377-379”C, MHSV= 7 and 1 atm,
see Table 14). The Theta-l zeolite produced considerably less of the undesir-
able iower and higher than C4 hydrocarbons. Xtwas also determined that the
Cs+ produced over Theta-l was mainly alkenes whereas that obtained over
ZSMS contained significant amounts of alkanes. The high yields of isobutene
observed with the Theta-l zeolite, relative to ZSM-5, can be ascribed to the
smaller channel diameters of the former (i.e. 5.5 x 4.4 A for Theta-l and
5.6~5.3and5.5x5.1 AforZSM-5).Thenarrowerchannelsthereforeappear
to be preventing the dimerization of the feed or of the product isobutene due
to steric constraints.
In the case of the undiluted feed, all these catalysts (ZSM-5,ZSM-23 and
Theta-l ) gave low selectivities to isobutene ( < 10 mol-%). Again using the
diluted feed, the effect of reaction temperature was studied as well as the re-
generation of the catalyst. The results showed (Table 15) that increasing yields
TABLE 14

Conversions and sekctivities*obtained over ZSM-5 and Theta-l zeolites [5 11

Catalyst Temperature Conversion Selectivities (%)

(“C) (%)
C,, CT CT C, ZC? i-C,- C,,

ZSM-5 377 96.6 6.2 4.8 20.9 14.5 10.9 11.2 31.5
Theta-l 379 90.0 0.9 0.8 10.6 3.8 33.7 35.0 15.2
‘In mol-%, at 1 atm, MHSV= 7.2 and N2/Cr = 8.3.
466 A.C. Butler and C.P. Nicolaides/ Catal. Today 18 (1993) 443-471

TABLE 15

Conversions and selectivities* obtained over Theta- 1 as a function of reaction temperature and regen-
eration [Sl]

Temperature Conversion Selectivity (%)

(“C) (%I
C 14 2-C; i-C,- C s+

234 85.1 0 92.1 7.1 0.8

276 84.5 0.8 83.6 13.9 1.7
326 87.9 3.8 54.3 36.0 5.9
376 89.2 6.6 41.8 43.7 7.9

Zeoliteregeneratedin air at S50°Cfor 16 h

376 87.5 5.3 41.9 40.6 12.2
426 87.7 7.9 41.7 42.1 8.3
475 86.1 8.6 42.8 42.0 6.6

“In mol-%, at 1 atm, MHSV=7.6-9.0 and NJC; =7.7.

of isobutene are obtained as the temperature is increased from 234 to 376 ‘C

and that regeneration of the zeolite, in air, does not have any deleterious ef-
fect on the activity of the catalyst. Other results given in the patent show that
the % of C5 + can be decreased by using higher space velocities (2 to 26 h- I )
which at the same time increase the 96 of isobutene. The performance of the
catalyst with time-on-stream, again using the diluted stream, was also evalu-
ated and it was shown that after 71 hours on-stream the 96 conversion de-
creased only slightly, from 89.5 to 86.696, whereas the selectivity to isobutene
increased from 29.0 to 35.6%. In other experiments it was observed that lower
calcination temperatures (e.g. 325 versus 500°C) result in higher yields of
isobutene (25.5 versus 4.6 mol-%) when using a pure l-butene feed. The use
of the lower calcination temperatures (which presumably give only partial
removal of the template) and lower oletin partial pressures also resulted in
higher selectivities to isobutene. From the results detailed in this study, it can
be concluded that at relatively low reaction temperatures (below 400 ’ C ) and
with diluted feed streams, zeolites such as Theta- 1 can produce high yields of
isobutene with fairly good stabilities with time-on-stream.
A substantial advantage arising from the use of zeolitic catalysts is that they
allow the reaction to proceed at lower reaction temperatures (probably due
to their stronger acidity relative to that of the halogenated and other amor-
phous catalysts described in the last two sections), thus affording higher equi-
librium concentrations of the branched isomer.
The catalytic properties of Theta-l were compared by Rahmin et al. [ 521
to those of its isotype ZSM-22 [ 53 ] at higher temperatures (400-550’ C) and
at considerably higher MHSVs (2 1 and 75 h- ’ ) and where the crystallite size
of the ZSM-22 used was below 0.5 microns and that of Theta-l was up to 2
A.C. Butler and C.P. Nicolaides / Catal. Today 18 (1993) 443-471 467

microns. From the results shown in Table 16 we can observe that the micro-
crystalline ZSM-22 catalyst has significantly higher activity than the large
crystal Theta- 1 at 400°C and MHSV= 75. Increasing the temperature to 550
’ C, however, results in approximately the same percentage yield of isobutene
by both catalysts.
The use of the narrower-pore zeolite ZSM-23 (4.5 x 5.6 A) was also exam-
ined [ 541 at high temperatures (SOO’C) and at even higher space velocities
( 160 h-l ). The yields of isobutene obtained ranged between 27% and 35%
(see Table 2 ) .
The application of molecular sieves as catalysts has also been the subject of
recent investigations. Gaffney and Jones [ 55 ] examined the catalytic prop-
erties of the sieve MgAPSO-31 at high temperatures and high space velocities.
Highest yield of isobutene (33.9%) was achieved at 55O”C, 1 atm and
MHSV= 93. In contrast, the study on SAPO-11, with pore diameters of
6.3 x 3.9 A, was carried out at low temperatures and low MHSVs [ 56 1. At
343°C and an MHSV=4.7, the yield of isobutene was 26.7% (Table 2).
The last catalytic system to be reviewed is that described in a recent Shell
patent [ 571 which is based on the zeolite ferrierite. It was found that with this
narrow pore zeolite (4.2 x 5.4 and 3.5 x 4.8 A) the conversion of the linear
butenes can be carried out at low temperatures (35O”C), a total olefin pres-
sure of 1.4 bar, MHSV = 2 and in the absence of any diluent, with high activ-
ity and selectivity and with excellent stability with time-on-stream. The re-
sults obtained show (Table 17) high yields of the branched alkene which
actually increase with time-on-stream.
As in the case of the Theta-l work, low temperatures were again employed
with ferrierite thus giving the higher equilibrium yields. The main difference

TABLE 16

Comparison of ZSM-22 and Theta-l* [ 52 ]

WHSV Temp. WHC Cow. Sel. Yield

(“0 0)) 06) (%I

Microcrystalline
ZSiU-22
75 400 3 44.3 61.1 27.1
75 550 3 40.6 88.1 35.8
21 400 10 40.4 83.3 33.6
21 550 10 39.8 94.5 37.6

Theta-l
75 400 3 3.6 87.7 3.2
75 550 3 34.9 94.5 33.0
21 400 10 26.6 90.5 24.1
21 550 10 35.2 94.5 33.3

‘At 1.6 aim.

468 A.C. Butler and C.P. Nicolaides/ Catal. Today 18 (1993) 443-471

TABLE 17

Conversion of n-butenes over ferrierite as a function of time-on-stream8 [ 571

Product Mass%

48h 336 h

C,-C~ paraffins 0 0
Ethene 0.04 0.01
Propane 0.07 0.02
Propene 1.7 0.8
Butane 1.3 0.8
n-Butene 40.1 49.1
Isobutene 37.0 41.2
C 5+ 19.9 8.0

aAt 35O”C, 1.4 atm andMHsV=2.

between the two zeolites is, however, that the ferrierite zeolite is able to pro-
duce the high yields of isobutene without any feed dilution. Another dissimi-
larity between the two zeolites is that whereas the calcination of Theta-l at
5OO”C, in order to remove the template, results in low yields of isobutene
when a pure butene feed is used due to extensive Cs+ formation, in the case
of ferrierite the same calcination temperature results in a zeolitic catalyst with
good SI properties. It can thus be concluded, from the studies by Barri [ 5 1 ]
and Grandvallet et al. [ 571, that the narrower-pore zeolites are the most
promising catalysts for industrial application. Further work is, however,
needed to ascertain whether the channel dimensions such as those found in
ferrierite are indeed the ideal ones for the skeletal isomerization of 1-butene,
or whether even narrower pore zeolites can offer higher yields of the branched
isomer.

5. CONCLUSIONS

Since the late 1970s tremendous progress has been achieved in the catalytic
conversion of linear butenes to isobutene. In Section 2 we surveyed the cata-
lytic properties of halogenated catalysts. Even though this type of catalyst, e.g.
halogenated alumina, has been shown to exhibit sufficient activity and selec-
tivity, the corrosion and environmental problems associated with these ma-
terials are most likely to prevent their industrial application. From the survey
on the non-halogenated catalysts and zeolites, presented in Sections 3 and 4,
it appears that the narrow-pore zeolites, e.g. ferrierite, is the group of catalysts
that are suitably active, selective and stable with time-on-stream in terms of
the SI reaction. Their catalytic properties can be ascribed to (i) the high acid
strength of the zeolitic active sites which allows the reaction to proceed at the
A.C. Butler and C.P. Nicolaides / Catal. Today 18 (1993) 443-471 469

relatively lower reaction temperatures thus affording the higher equilibrium

concentrations of the branched isomer, and (ii) the narrower channels which
appear to prevent the subsequent dimerization, etc of the feed or of the prod-
uct isobutene, due to steric constraints. As a consequence, these zeolitic cat-
alysts could therefore be the choice of type of catalyst for future industrial
implementation of the skeletal isomerization reaction. Indeed, ferrierite is
the type of catalyst used in Lyondell Petrochemical Company’s single-step
catalytic process for the conversion of linear butenes to isobutenes [ 58 ] which
was announced early in 1992 [ 59 1. Lyondell disclosed that a large-scale dem-
onstration plant for the process is planned for this year with the possible
building of a commercial unit at its Channelview location after two years.

REFERENCES

1 V.R. Choudhary, Ind. Eng. Chem., Prod. Res. Dev., 16 (1977) 12.
2 G.J. Hutchings, C.P. Nicolaides and MS. Scurrell, Catal. Today, 15 ( 1992) 23.
3 B. Juguin, B. Torck and G. Martino, in B. Imelik, C. Naccache, G. Coudurier, Y. Ben Taarit
and J.C. Vedrine (Editors), catalysis by Acids and Bases (Studies in Surface Science and
Catalysis, Vol. 20), Elsevier, Amsterdam, 1985, p. 253.
4 H.L. Brockwell, P.R. Sarathy and R. Trotta, Hydrocarbon Process., 70 (9) ( 1991) 133.
5 J.N. Armor, Appl. Catal., 78 (1991) 141.
6 J.F. Knifion and J.R. Sanderson, US Patent No. 5 059 725 (1991), assigned to Texaco
Chemical Company.
7 M.A. Makarova, C. Williams, V.N. Romannikov, ICI. Zamaraev and J.M. Thomas, J. Chem.
Sot., Faraday Trans., 86 (1990) 581.
8 K. Klier, R.G. Herman, R.D. Bartian, S. DeTavernier, M. Johansson, M. Kieke and O.C.
Feeley, Proc. Liquefaction Contractors’ Review Meeting, US Department of Energy, Pitts-
burgh Energy Technology Center, September 3-5, 199 1, p. 20.
9 W. Keim and W. Falter, Catal. I&t., 3 (1989) 59.
10 A. Sofianos, Catal. Today, 15 ( 1992) 149.
11 R.L. Banks, US Patent No. 3 696 163 ( 1972), assigned to Phillips Petroleum Company.
12 V.R. Choudhary, Chem. Ind. Dev., 8 ( 1974) 32.
13 D.M. Brouwer and H. Hogeveen, Prog. Phys. Org. Chem., 9 ( 1972) 179.
14 J.E. Kilpatrick, E.J. Prosen, KS. Pitzer and F.D. Rossini, J. Res. Natl. Bur. Standards, 36
(1946) 559.
15 J. Szabo, J. Perrotey, G. Szabo, J.-C. Duchet and D. Comet, J. Mol. Catal., 67 ( 1991) 79.
16 J.-P. Damon, J.-M. Bonnier and B. Delmon, J. Colloid Interface Sci., 55 ( 1976 ) 38 1.
17 J.-P. Damon, B. Delmon and J.-M. Bonnier, J. Chem. Sot., Faraday Trans., 1, 73 ( 1977)
372.
18 C.P. Nicolaides, C.J. Stotijn and E.R.A. Van der Veen, unpublished results.
19 P.A. Jacobs, Carboniogenic Activity of Zeolites, Elsevier, Amsterdam, 1977, p. 168.
20 W.O. Haag and N.Y. Chen in L.L. Hegedus (Editor), Catalyst Design - Progress and Per-
spectives, Wiley, New York, 1987, p. 163.
21 C.P. Nicolaides, M. Wapiennik, K.I.G. Weiss, H. Van den Akker, B. Van Zalk and P. Wie-
laard, Appl. Catal., 68 ( 199 1) 3 1.
470 A. C. Butler and C.P. Nicolaides / Catal. Today I8 (1993) 443-4 71

22 D. Barthomeuf, in B. Imelik, C. Naccache, G. Coudurier, Y. Ben Taarit and J.C. Vedrine

(Editors), Catalysis by Acids and Bases (Studies in Surface Science and Catalysis, Vol.
20), Elsevier, Amsterdam, 1985, p. 75.
23 M. Guisnet, ibid., p. 283.
24 V.R.ChoudharyandL.K.Doraiswamy, J.Catal.,23 (1971) 54.
25 A. Melchor, E. Garbowski, M.-V. Mathieu and M. Primet, J. Chem. Sot., Faraday Trans.,
82 (1986) 1893.
26 AK. Ghosh and R.A. Kydd, Catal. Rev.-S& Eng., 27 (1985) 539.
27 N.S. Raghavan and L.K. Doraiswamy, J. Catal., 48 ( 1977) 2 1.
28 V.R. Choudhary and L.K. Doraiswamy, Ind. Eng. Chem., Process Des. Dev., 14 ( 1975)
227.
29 B. Juguin and J. Miquel, FR Patent No. 2 484 400 ( 1981 ), assigned to Institut Francais du
Petrole.
30 G. Franz, F. Heinrich and H.-J. Ratajczak, US Patent No. 4 367 362 ( 1983), assigned to
Chemische Werke Huels Aktiengesellschaft.
31 H.J. Miiller, H. Honig and W. Horlitz, EP Patent No. 54 855 (1982), assigned to EC Er-
dolchemie GmbH.
32 J.R. Adams, A.B. Gelbein, R. Hansen, J. Peress and M.B. Sherwin, EP Patent No. 71 198
(1983) and 71 199 (1983), assigned to Polysar International S.A.
33 A.E. Eleazar, R.M. Heck, J.C. Dettling and Y.-K. Lui, US Patent No. 4 443 191 (1984),
assigned to Engelhard Corporation.
34 H. Sun and R.G. Gastinger, US Patent No. 4 654 463 (1987), assigned to Atlantic Rich-
field Company.
35 H. Sun, US Patent No. 4 778 943 ( 1988), assigned to Atlantic Richfield Company.
36 B. Juguin and J. Miquel, EP Patent No. 66 485 (1982), assigned to Institut Francais du
Petrole.
37 B. Juguin and J. Miquel, US Patent No. 4 434 3 15 ( 1984), assigned to Institut Francais du
Petrole.
38 F. Buonomo, V. Fattore and B. Notari, US Patent Nos. 4 013 589 (1977) and 4 013 590
(1977), assigned to Snamprogetti S.p.A.
39 B. Juguin and G. Martino, SA Patent No. 90/2870 ( 1991) and FR Patent No. 2 657 605
( 1991) , assigned to Institut Francais du Petrole.
40 B.P. Nilsen, M. Stocker and T. Riis, in B. Imelik, C. Naccache, G. Coudurier, Y. Ben Taarit
and J.C. Vedrine (Editors), Catalysis by Acids and Bases (Studies in Surface Science and
Catalysis, Vol. 20)) Elsevier, Amsterdam, 1985, p. 197.
41 M. Stocker, T. Riis and H. Hagen, Acta Chem. Scand., B40 ( 1986) 200.
42 B.P. Nilsen, J.H. Gnuferko and B.C. Gates, Ind. Eng. Chem. Fundam., 25 (1986) 337.
43 L. Basini, A. Aragno and A. RatTaelli,J. Phys. Chem., 95 ( 1991) 211.
44 J.W. Myers and D.J. Strope, US Patent No. 4 436 949 ( 1984), assigned to Phillips Petro-
leum Company.
45 J.W. Myers, Chem. Eng. Prog., Symp. Ser., 66 ( 1970) 94.
46 J.C. Luy, J.C. Yori, A.A. Castroand J.M. Parera, React. Kinet. Catal. Lett., 36 (1988) 275.
47 D.L. Sikkenga, US Patent No. 4 550 091 ( 1985), assigned to Standard Oil Company
(Indiana).
48 G.E.M.J.deClippeleir,S.P.Leeuw,R.M.CahenandG.L.G.Debras,DEPatentNo. 3512
057 (1985), assigned to Labofina S.A.
49 A. La Ginestra, P. Patrono, M.A. Massucci, P. Galli, C. Ferragina and C. Mancini, Proc.
9th International Congr. Catal., 1 ( 1988) 499.
50 B.G. Baker and N.J. Clark, in A. Crucq and A. Frennet (Editors), Catalysis and Automo-
tive Pollution Control (Studies in Surface Science and Catalysis, Vol. 30), Elsevier, Am-
sterdam, 1987, p. 483.
A. C. Butler and C.P. Nicolaides / Catal. Today 18 (1993) 443-471 471

51 S.A.I. Barri, D.W. Walker and R. Tahir, EP Patent No. 247 802 (1987), assigned to The
British Petroleum Company p.1.c.
52 I. Rahmin, A. HUSS,D.N. Lissy, D.J. Klocke and I.D. Johnson, US Patent No. 5 157 194
( 1992), assigned to Mobil Oil Corporation.
53 W.M. Meier and D.H. Olson, Zeolites, 12 ( 1992) 449.
54 W.O. Haag, M.N. Harandi and H. Owen, US Patent No. 5 132 467 (1992), assigned to
Mobil Oil Corporation.
55 A.M. Gaffney and A. Jones, US Patent No. 5 107 050 ( 1992), assigned to Arco Chemical
Technology, L.P.
56 G.J. Gajda, US Patent No. 5 132 484 ( 1992), assigned to UOP.
57 P. Grandvallet, KP. de Jong, H.H. Mooiweer, A.G.T.G. Kortbeek and B. Kraushaar-Czar-
netzki, EP Patent No. 501 577 (1992), assigned to Shell Intemationale Research Maat-
schappij B.V.
58 Chem. Eng. News (March 23,1992), p. 7.