Chemistry Reference
In-Depth Information
detectable metal leakage. Polymethacrylate/phos-
phonamide catalyst is very selective with cyclo-
hexene and 65% H
2
O
2
—very little allylic product is
formed compared with the homogeneous analogue.
Titanium silicalite TS-1 continues to occupy a
central position in the research and development of
heterogeneous epoxidation catalysts, the aim being
either to improve its own performance further or
to reproduce its key features in other structures, as
noted earlier. Early work [122] established the scope
of the catalyst, which performs most strongly with
small substrates such as lower alkenes [10,55], allyl
chloride [180], allyl methacrylate [181] and, inter-
estingly, butadiene [182] (to monoepoxide)—all at
high conversion and selectivity. Hydrogen-peroxide-
based processes competitive with existing technology
can be envisaged in all these cases.
The purity and crystallite size are both critical to
epoxidation: the former because of selectivity and
the latter because of activity. Many patents and pub-
lications refer to the addition of mildly basic buffer-
ing agents to suppress any acid sites present at the
start or developing during use. Although deleterious
during TS-1 formation, the addition of alkali metal
cations in use can improve selectivity [129]. Amides
such as dimethylformamide recently have been
claimed [183], as has the addition of heavier cations
including zinc [184] and of amines or amine ox-
ides such as 2,6-lutidine, which clearly only reaches
external surfaces. The need for small crystallites but
tough catalyst bodies has led to many studies aimed
at supporting TS-1 on more robust (and cheaper)
solid substrates [185,186]. Many groups have con-
firmed that there is a maximum loading of Ti in
TS-1, which appears to correspond to the limit for
effective site isolation. However, the conventional
way of achieving this distribution (co-precipitation
of Ti and Si from esters) has been challenged by a
recent paper [187], claiming that post-substitution of
TiCl
4
into silanol nests in silicalite (high-silica ZSM-
5) formed from
n
-butylamine, tetraethylammonium
bromide or hexamethylenedianine templates gives
higher catalytic activity for propylene epoxidation
because of higher concentrations of isolated frame-
work Ti.
Jacobs' group [188] has attempted to improve
understanding of epoxidation by Ti zeolites through
quantitative sorption experiments including sub-
strate, solvent and product. For TS-1, methanol
allowed the biggest concentration of olefin inside the
pores, corresponding to the biggest initial rates of
reaction. Longer chain olefins epoxidise faster ini-
tially but the epoxide is very slow to desorb—acetone
is said to be a better solvent for these—and deacti-
vation therefore is fast in methanol. Larger pore
materials do not adsorb olefins as strongly, corre-
sponding to lower reactivity. Regarding solvents, an
interesting recent patent from Exxon [189] employs
methanol and CO
2
for carrying out propylene epox-
idation under supercritical conditions. Examples also
are given with 1-octene, and improved selectivities
on both olefin and H
2
O
2
are shown. It has been
claimed recently [190] that vanadium incorporation
into the TS-1 framework improves activity for propy-
lene epoxidation, even though VS materials alone
are worse than TS-1. For the Ti-b zeolite, a sharp
dependence of 1-hexene conversion on crystallite
size is shown, as in other results for TS-1, which
seems to confirm the product desorption limitation
for larger molecules. Addition of thallium fluoride
has been claimed to promote the epoxidation of
methallyl chloride by H
2
O
2
and TS-1 [191], e.g. at
40°C for 2 h a selectivity of 93.2% at a conversion of
52.3% is obtained.
As noted in Section 3, Ti-b is the best-characterised
large-pore analogue of TS-1 to date but the larger
cavities always accommodate solvent along with the
substrate. This increases the tendency for epoxide
ring-opening and means that the slightly basic ace-
tonitrile gives better results than the slightly acidic
methanol. The influence of zeolite hydrophobicity is
highlighted by studies on the epoxidation of unsat-
urated fatty esters [192]. Elimination of residual
aluminium (Brønsted acid) sites and of extra-
framework TiO
2
in Ti-b is also difficult, both con-
tributing to side reactions. Recent work by van Hooff
[193] has made Ti-b for epoxidation by post-synthe-
sis: dealuminisation is followed by TiCl
4
treatment.
For 1-octene, results in acetonitrile are comparable
with TS-1 (accepting that this is not the best solvent
for TS-1 reactions).
It is probably fair to say that other microporous
materials such as UTD-1 [194] and SSZ-33 have not
shown any advantages over Ti-b as hosts for frame-
work titanium. A recent addition by Corma's group
[195] is ITQ-6, prepared by delamination of fer-
rierite. This offers a very high
external
surface area
and Ti framework substitution is readily achieved,
the product being active and selective in 1-hexene
and norbornene epoxidation—comparable with Ti-
