Coal is more familiar to us as a fuel than as a source of chemicals.
Most feedstock chemicals – raw materials used in manufacturing a product
– come from oil or natural gas. These natural reserves contain hydrocarbons
that are basic molecules for the world’s chemicals industry. For example,
they are essential for making plastics, solvents and fertilisers. But world
supplies of petroleum and natural gas will not last indefinitely. World
reserves of oil are estimated to last for another 40 years; natural gas
for perhaps 60. Proven world reserves of coal, by contrast, are likely
to last for at least 200 years at present rates of consumption. Now chemists
are looking at ways of exploiting coal, not only as a fossil fuel, but also
as a long-term alternative source of basic materials for making important
industrial chemicals.
This is not the first time people have sought alternatives to petroleum
as a source of hydrocarbons. During the Second World War, when Germany’s
oil supplies were very limited, another source of hydrocarbons was needed
for use as fuel in internal combustion engines. Researchers there refined
a process named after the two German chemists, Franz Fischer and Hans Tropsch,
who invented it in the 1920s. The Fischer-Tropsch process converts a mixture
of hydrogen (H2) and carbon monoxide (CO) gases into hydrocarbons
(containing carbon and hydrogen) which can be used as fuel. A similar process
has been used more recently in South Africa: at the height of the oil embargo
in the 1970s, about 40 per cent of the country’s requirements for petrol-type
hydrocarbons were met using Fischer-Tropsch technology.
Advertisement
In both the German and the SASOL (South Africa Synthetic Oil) plants,
the mixture of H2 and CO, known as synthesis gas or syngas, was
generated from coal using a process called gasification . Industrial chemists
already use syngas, derived mainly from naphtha and natural gas, as a starting
mixture for a wide variety of important petrochemicals, ranging from ammonia
(for making fertilisers), through methanol and acetic acid (used as solvents,
and to make polyvinyl acetate) to butanal (important in making plastics).
So why can’t they simply adapt the method to produce other, larger hydrocarbon
molecules from coal?
The technology of coal gasification is well established (see ‘Coal comes
back as a gas’, New Scientist, 29 April 1989). But the fine details of the
Fischer-Tropsch reactions are still not entirely understood, even though
the reaction has been known for about 70 years. This means that chemists
cannot exert as much control as they need over the reactions. So researchers
in Britain, France, the US, the former Soviet Union, South Africa and other
countries, many of them working in industry, are trying to find out more
about Fischer-Tropsch chemistry.
To balance the chemical books, chemists say that the Fischer-Tropsch
reaction ‘removes water’ from the product of a mixture of hydrogen and carbon
monoxide, to yield hydrocarbons. Chemically, we can write the overall reaction
like this:
nC0 + (2n+1) H2 –> CnH2n+2 + nH20
If n=1, the reaction produces methane (CH4); if n=2, the
product is ethane (C2H6). Water is a very stable substance
and represents an energy minimum, so the process is in part driven by the
formation of water, with a large amount of energy being released as a by-product.
But there is no direct route available for the reactants, carbon monoxide
and hydrogen, to combine to give the products, hydrocarbons. Although the
overall reaction is ‘downhill’ in energy terms, there is a large energy
‘mountain’ to be climbed on the way from reactants to products, so a catalyst
is needed. Catalysts speed up chemical reactions without being changed themselves.
A catalyst is rather like a tunnel through the mountain – using it is not
nearly as much effort as climbing over the top. Without catalysts some reactions
happen so slowly as in effect not to happen at all.
Research done since the original German work, by British, American and
German chemists, has shown that many metals are good catalysts for Fischer-Tropsch
reactions. The best are ruthenium, iron and cobalt. These metals are found
in the middle of the group known as the transition elements. They can exist
in different oxidation states, and can cause reactant molecules to come
together to form various unstable intermediates, which then react further
on the catalytic surface. The metal is usually deposited on a support, an
inert metal oxide such as silica or alumina, to increase the surface area
of the catalyst and ensure that as many reactant molecules as possible are
exposed to it.
PURE PRODUCTS
But there is more to this chemistry than simply speeding up the reactions.
The most important thing is selectivity – to obtain useful amounts of products
which are pure. The key to better selectivity lies in the addition of various
‘promoters’, chemicals which enhance the catalysis. These are generally
oxides which are reduced to a more reactive state under the conditions of
the reaction (heat and an atmosphere of hydrogen). Indeed, the type and
amount of these additives can entirely change the course of the reaction,
and the catalytic ‘recipes’ are often closely guarded commercial secrets.
There are a few general rules, however. Nickel catalysts generally give
short hydrocarbons such as methane. Iron, cobalt or ruthenium, especially
when suitably promoted with complex mixtures – oxides, such as thorium oxide,
and potassium, for example – give longer hydrocarbons ranging from 5 to
as many as 1000 carbons. In the original German process, liquid hydrocarbons
with between 5 and 10 carbons were formed using a cobalt catalyst with a
thorium oxide promoter. The SASOL plant used an iron catalyst. Shell’s middle
distillate synthesis plant in Sarawak, Malaysia, which came on stream in
late 1992, uses Fischer-Tropsch technology to turn natural gas into a range
of products as part of a three-step process.
First, the natural gas is turned into syngas. Then the syngas is converted
into a long-chain hydrocarbon wax, using Fischer-Tropsch technology, with
a secret catalyst. This wax is then broken down (‘cracked’) into smaller
hydrocarbons: 15 per cent naphtha (containing between 6 and 10 carbons),
25 per cent kerosene (containing between 10 and 14 carbons) and 60 per cent
gasoil (hydrocarbons with more than 14 carbons). The wax is first broken
into short-chain linear hydrocarbons which are converted by an isomerisation
process into branched chain hydrocarbons. The various products are then
separated by distillation into ‘fractions’, just as crude oil is treated.
Fischer-Tropsch reactions normally produce quite complex mixtures, whereas
a single product is more easily isolated and so more valuable. For this
reason Fischer-Tropsch technology has largely been left waiting in the wings,
while chemists today are focusing mainly on how to produce one product of
good purity.
To do that chemists have to know how the reaction works. They agree
that the first stages of hydrocarbon synthesis involve the adsorption, on
the surface of the metal catalyst, of carbon monoxide and hydrogen. The
hydrogen molecule, H2, splits to give metal hydrides on the surface,
while the carbon monoxide breaks up giving a surface carbide, and water
(which is lost). Then the carbide combines on the surface with hydrides,
a process called hydrogenation, to give hydrocarbon fragments: methyne (CH),
then methylene (CH2), then methyl (CH3) groups on
the surface (see Figure 1).
There is much debate on what happens next. Somehow, chains grow by the
addition of more hydrocarbon fragments to these initial fragments, forming
the larger hydrocarbons which are then released. These include both ‘saturated’
hydrocarbons with single bonds between each carbon atom (alkanes, also called
paraffins) and more reactive ‘unsaturated’ ones with double bonds between
some of the carbons (alkenes, or olefins). It is generally agreed that alkenes
are formed first, and converted to alkanes in a separate hydrogenation step.
CHAIN GROWTH
But this still leaves us with the question of how the alkenes are formed.
Without even this basic knowledge, there is little hope of predicting –
and ultimately controlling – the outcome of the overall reaction. Early
studies by Fischer and Tropsch, and further work in 1980 by Robert Brady
and Roly Pettit, then working at the University of Texas at Austin, suggested
that methyl on the catalyst’s surface might trigger polymerisation (chain
growth) of methylene groups on the surface which ‘bridge’ two metal atoms
(see Figure 2). In this mechanism a surface methyl (CH3) reacts
with a methylene (CH2) to give an ethyl (C2H5), then
a propyl (C3H7), then a butyl (C4H9) and
so on. Finally, the group separates from the surface to form a complete
hydrocarbon molecule.
However, this mechanism raises more questions than it answers. To explain
how the primary products of the reaction are alkenes, the last step of this
process would have to be a reaction known as beta-hydride elimination. In
this reaction the second or ‘beta’ carbon of the alkyl group loses a hydrogen
atom, generating a double bond, and the group is released from the metal
surface as an alkene. Similar processes are known, but many chemists believe
such a step to be unlikely under the conditions present in the commercial
process, because the surface is already saturated with hydride so it would
be difficult to add any more.
Another problem is that model reactions based on the theory completely
fail to predict the observed number of product molecules with two carbons.
And this mechanism does not account for the formation of any branched hydrocarbons,
whereas these can comprise up to 5 per cent of the final product. The theory
also implies that methylenes that bridge two metal atoms should be highly
reactive. In fact, many such species have been isolated as stable metal
complexes, suggesting that this might not be the case.
In the early 1980s, our group at the University of Sheffield began to
explore some di-metal compounds. We began work on the chemistry of organo-metallic
rhodium (rhodium has the chemical symbol Rh), and made some interesting
compounds. Meanwhile, one of the newer Fischer-Tropsch catalysts, based
on rhodium, was being developed by groups in several countries: initially
by researchers working for Union Carbide in the US and later by Gabor Somorjai
and Alexis Bell at the University of California at Berkeley, V Ponec at
the University of Leiden in the Netherlands, Masaru Ichikawa in Japan and
Chen De-An in Xiamen in China. Metallic rhodium usually catalyses the production
of methane but, at low temperatures, when suitably promoted, it becomes
a good catalyst for medium length hydrocarbons containing between 2 and
7 carbons, and can even produce ethyl alcohol. Promotion is usually accomplished
by supporting the rhodium on a reducible metal oxide. Oxides of cerium (one
of the lanthanide group of elements) or of titanium and vanadium (both transition
elements) are commonly used as supports.
During the mid-1980s we looked at two processes in which methylenes
were converted into alkenes. First, we made a series of di-metal compounds,
such as Kiyoshi’s complex (Figure 3, top; named after the Japanese researcher
who discovered it in our laboratory). This contains two rhodium atoms joined
by two bridging methylenes (CH2), and also bound to methyls (CH3).
These compounds were very similar to the sort of intermediate fragments
thought to exist for a fleeting moment on metal surfaces during Fischer-Tropsch
reactions, and it looked as if they might act as a convenient model for
such intermediates. These model compounds were quite stable, but heating
them to 275 °C produced propene (CH2=CHCH3) and
methane, and by adding an oxidiser (such as iron sulphate) to decompose
the rhodium compound, the same products were obtained at room temperature.
Next, we traced which types of chemical group the carbon and hydrogen
atoms had come from, using isotopic ‘labels’ such as carbon-13 and deuterium.
We found that all the atoms in propene and methane had come from either
the bridging CH2 or the end CH3 groups. The C5(CH3)5
(pentamethylcyclopentadienyl) groups played no active part in the movement
of either hydrogens or carbons. We also found that all the propene came
from one of the methyls combining with two methylenes, and all the methane
came from the end methyls.
MODEL SYSTEM
Then we tried a similar molecule, in which vinyl groups (-CH=CH2)
replaced methyls. Its decomposition showed that the combination of a metal-vinyl
(Rh-CH=CH2) with a bridging methylene (Rh-CH2-Rh)
easily generates an allyl (-CH2-CH=CH2) in our model
system. In other words, vinyls combined with methylenes much more easily
than methyls did.
These reactions suggested that chain growth in the Fischer-Tropsch hydrocarbon
synthesis might happen differently (see Figure 4a). In this new mechanism,
the surface methylenes are formed in the same way as in the ‘classical’
mechanism. But we think the first step of the chain growth is by the combination
of a methyne (CH) and methylene (CH2) giving a surface vinyl
group (M-CH=CH2), where M is a metal atom. This then reacts with
a further surface methylene, giving a surface allyl (M-CH2-CH=CH2)
which then undergoes hydrogen migration (isomerisation) to generate a new
alkenyl (M-CH=CHCH3), which can then combine with another methylene
to give another link in the chain. The chain grows on in the same way until
the surface alkenyl reacts with a surface hydride. A variation on this also
explains how small amounts of branched-chain alkenes might be formed (see
Figure 4b). Overall then, this explains why alkenes rather than alkanes
are the primary products of Fischer-Tropsch reactions. It also explains
that while most of the hydrocarbons produced in Fischer-Tropsch reactions
are linear, some methyl branched hydrocarbons are formed too.
TENSE MOMENTS
We then set out to test our theory. This was quite difficult, as surface
catalysed reactions are notoriously hard to study. There is no way of detecting
what exactly is happening in, or on, the surface. We chose an indirect but
accurate method using carbon-13 as an isotopic label. A key step in the
theory was the triggering of chain growth by a vinyl group, so we decided
to ‘seed’ the reaction with tetravinyl silicon, which contained two carbon-13
atoms in each vinyl.
The experiments began tensely. The starting material for the synthesis
of the labelled tetravinyl silicon cost well over £1000 and an accident
would have been financially disastrous. But we succeeded in making a small
amount of labelled tetravinyl silicon, enough for a few careful Fischer-Tropsch
experiments. We fed minute amounts of the seed compound, together with syngas,
over our rhodium catalyst. By using a sensitive analytical technique which
combines gas chromatography and mass spectrometry (GC-MS), we could find
out where the two carbon-13 atoms from our vinyl ended up.
Much more work is still needed. But Futai Ma, who started it (now back
in Hangzhou in China), Peter Byers (a Tasmanian now in Birmingham) and Mike
Turner, who took over the project in late 1991, have independently obtained
similar results that are consistent with the predictions of our theory.
They found that two carbon-13 atoms are incorporated into the hydrocarbons
formed in the cerium-oxide-promoted rhodium catalyst, just as our theory
predicts would be the outcome of a chain reaction. Also, equal amounts of
carbon-13 end up in alkanes and alkenes, indicating a common origin for
both. By contrast, the incorporation of one carbon-13 into the hydrocarbons
happens rarely, so it is unlikely that the vinyls split.
It is only through a thorough understanding of the mechanism of the
Fischer-Tropsch process that the intricacies of the reaction can be understood.
Now we aim to build on our results and obtain more details about other reactions
on such catalysts. We hope to rationalise the mechanisms of a hitherto poorly
understood, but commercially important, process. The next step is to learn
how to control chain growth itself. It should then be possible to predict
all aspects of the Fischer-Tropsch process and selectively make the products
we want – from petroleum, from natural gas, or in the long term perhaps,
from coal.
Peter Maitlis is professor of inorganic chemistry and Jon Rourke is
a postdoctoral research assistant at the University of Sheffield.
* * *
Syngas chemistry
For Fischer-Tropsch chemistry we need a source of syngas. This can be
one of several fossil fuels: natural gas, oil, coal, lignites or oil shales.
The conversion of coal (largely carbon) to syngas is often known as coal
gasification. In this process the coal is treated with a mixture of steam
(H20) and air (which contains oxygen, 02). The overall reaction
is written chemically like this:
3C+H20+3/202 –> 2C0+H2+C02
This reaction produces energy in the form of heat (is exothermic) and
so is favourable.
When natural gas from the North Sea – largely methane, CH4
– or petroleum fractions such as naphtha are treated with steam (‘steam
reforming’), a syngas containing more hydrogen is obtained:
CH4 + H20 –> C0 + 3H2
The reaction uses up heat energy (is endothermic) and must be coupled
with a methane-burning reaction to provide it.
Different ratios of CO to H2 are needed for different purposes.
The ratios can be adjusted using a reaction called the water gas shift reaction:
C0 + H20 –> C02 + H2
This reaction is an equilibrium between carbon monoxide and water on
the one hand and carbon dioxide and hydrogen on the other. It can be driven
either way by adjusting reaction conditions such as temperature and pressure.
In this way syngas with any desired composition can be derived from whatever
carbon source is most convenient.
A British consortium comprising British Gas, National Power, PowerGen
and British Coal recently developed and tested a modern gasification plant
at Westfield in Fife, Scotland that can produce as much as 500 tonnes of
syngas per day. This plant, designed by British Gas in partnership with
the German firm Lurgi, successfully completed a test programme in 1990,
although no further programmes are planned. It converts bituminous coal
into syngas which contains about 28 per cent hydrogen, 56 per cent carbon
monoxide and 7 per cent methane.
One problem normally associated with soft or brown coals (such as lignite)
is that they are contaminated with a relatively large amount of sulphur.
If this coal were to be burnt it would produce unacceptably high levels
of sulphur dioxide, a major source of pollution and acid rain. A double
bonus of gasification is that the sulphur impurity in the coal is hydrogenated
into hydrogen sulphide which is then removed, converted into elemental
sulphur, and then into sulphuric acid, which can then be sold.