¿ìè¶ÌÊÓƵ

For those who like their coffee strong, but don’t want to stay awake
all night suffering from the side effects, decaffeinated coffee has been
a familiar sight on supermarket shelves for decades. But when in the early
1960s the German chemist Kurt Zosel hit on a novel way of removing caffeine
from coffee beans, he could not have foreseen how the same technique would
become the basis of a method for destroying toxic waste, the inspiration
for ways of making industrially important chemicals without con-ventional,
toxic solvents, and a bone of contention for real ale aficionados.

Behind all these possibilities is the extraordinary behaviour of some
very ordinary fluids. If you place certain gases under enough pressure and
at the same time heat them beyond a certain temperature, then provided
they do not decompose first they become ‘supercritical’: they enter the
no-man’s-land between a liquid and gas, where they are neither one nor the
other. The same thing happens to liquids heated to this critical point under
high enough pressure. In the 19th century, scientists treated this phenomenon
as a curiosity. Only later did chemists discover its practical uses.

When Zosel passed a stream of supercritical carbon dioxide through wet
coffee beans he was already familiar with the way chemists used supercritical
fluids (SCFs) to purify mixtures of chemicals by dissolving out unwanted
components. Some inspired lateral thinking led him to the idea that an SCF
might also dissolve and remove caffeine from coffee. Until then the only
alternatives were organic solvents such as aromatic hydrocarbons or benzene
derivatives which could leave toxic residues.

More recently, the ability of supercritical fluids to dissolve things
that will not dissolve in everyday solvents has attracted the attention
of Britain’s brewing industry. In a bid for consistency that will make drinkers
of real ale turn pale, some brewers have begun to use a similar technique
to ‘clean up’ hops for the brewing industry. Originally a German idea, it
was taken up by a company called English Hops to remove some of the natural
flavours and colours from hops and so give a more consistent taste to beer.
The food industry uses SCFs in similar ways. Universal Flavours of Slough,
which extracts oily essences for use in everything from celery soup to aerosol
sprays, uses SCFs to dissolve out the chemicals responsible for flavour
from vegetable matter. SCFs are used similarly by analytical chemists to
separate tricky mixtures of chemicals. Now, thanks to recent developments
in chemists’ understanding of supercritical fluids, researchers are extending
their repertoire even further.

Water becomes supercritical at temperatures above 374 degree C and pressures
above 218 atmospheres. Though it still looks like a liquid, many of its
other properties have changed radically. It mixes with oily materials, but
will not dissolve ordinary table salt. It will still dissolve oxygen, though,
a property that means it is able to support a flameless variety of combustion.
This is being explored as a means of ‘clean’ incineration of organic toxic
waste (see Technology, 24 July 1993). Steve Rich at Sandia National Laboratories
in New Mexico is investigating the use of SCFs to destroy a stockpile of
artillery shells at McAlester Army Ammunition Unit in Oklahoma, while General
Atomics, a company in San Diego, is considering their potential for destroying
chemical weapons. Other chemists are exploiting the solvent properties of
SCFs to make new types of polymer and other molecules that could function
as industrial catalysts – without the need for the potentially harmful solvents
in which much conventional industrial chemistry is carried out.

An ideal solvent for a chemical reaction is an inert fluid that acts
as a medium for carrying molecules of reactants. The reaction occurs as
dissolved molecules collide. For a solvent to do its job its molecules
have to approach closely enough to solute molecules to interact with them.
This is why water is such a good solvent for some substances. Water molecules
are ‘polar’: there is a slight negative electric charge on the oxygen atom,
which is balanced by slight positive charges on the hydrogens. So a water
molecule will interact electrostatically with ions such as chloride, which
is why water will dissolve sodium chloride, common salt, a substance made
up exclusively of ions. The general rule is that like dissolves like: that
is, most organic chemicals, which are nonpolar or ‘oily’, are not soluble
in polar solvents like water, but dissolve in solvents such as hexane, chlorine-containing
hydrocarbons and ethers, which are themselves oily.

But such conventional solvents have disadvantages. Most organic solvents
are toxic, some are flammable or even explosive, and all of them cause
pollution if not properly disposed of. Some may decompose in the atmosphere
to produce, among other things, reactive forms of chlorine that damage the
ozone layer. They can be recovered at the end of a reaction and reused,
but there are limits to how many times this can be done. The infamous CFCs
are inert to most chemicals, and would be ideal for many applications but
for the damage they do to the ozone layer.

Many fluids can be made supercritical, but three in particular – water,
carbon dioxide and xenon – have the useful qualities for a solvent of being
inert and nontoxic. Some reactions that require high temperatures when carried
out in conventional industrial solvents will take place almost at room
temperature in supercritical carbon dioxide, which becomes supercritical
at 31 degree C. Also, SCFs flow more easily than other solvents and have
different electrostatic properties, which helps them to cluster round reactant
molecules more effectively. The result is that less energy is needed to
make these molecules collide, which is why reactions happen at lower temperatures.

SLIPPERY MOLECULES

At the University of North Carolina in Chapel Hill, a team of chemists
led by Joseph M. DeSimone is exploiting supercritical CO₂ to
bring together the components of new types of polymer. The spur for DeSimone
has been the technological possibilities of fluorine-containing polymers,
whose applications range from lubricating layers in computer disc drives
to sealants for aircraft fuel lines; the most familiar example is PTFE,
better known as Teflon. But fluoropolymers have one major drawback. The
fluorine atoms have a residual negative charge, which makes fluoropolymer
chains polar, so they are not soluble in conventional nonpolar organic
solvents such as hexane. A solvent is needed that dissolves both the starting
material and the product so that it can be processed further. Water cannot
be used because water and fluorine make an explosive, highly dangerous
combination. There are polar organic solvents that will do the trick, such
as the cyclic ether tetrahydrofuran (THF), but they have to be recovered
and disposed of safely. The alternative, used until recently, was CFCs.

Last year, DeSimone and his group succeeded in efficiently polymerising
various fluorine-containing monomers, such as 1,1-difluoroethylene, in supercritical
CO₂ to form fluoropolymers. What is more, by adding a compound
called 1-iodoperfluorobutane, they can control the size of the resulting
polymer molecules and their chemical structure because the iodine group
acts as a ‘signal’ to finish chain growth. This is important, because polymer
materials made up of a mixture of molecules of different lengths do not
always have the same strength and stability which leads to inconsistent
quality. Because they could control the length of the polymer chains, the
chemists could avoid the elaborate solvent-based processes that would otherwise
be needed to get rid of chains of the wrong length. Also, supercritical
carbon dioxide leaves no trace of itself in the polymer chains. This, says
DeSimone, is ‘further evidence of the inert nature of supercritical carbon
dioxide for such reactions’.

The same group of chemists has polymerised styrene and methyl methacrylate
in supercritical CO₂ to produce emulsions of the polymers polystyrene
and polymethyl methacrylate, best known as Perspex, but also used in lacquers,
paints and adhesives. Using an SCF enabled him to control the chain length
of these polymers which is otherwise difficult to do. Having tested the
potential of SCFs in this way he has now taken these studies a step further
to make new types of high-density polymer with shorter chains.

DeSimone believes that SCFs could help to improve standards in the aeronautics
industry, where stringent quality control applies. The solubility of fluoropolymers
in SCFs depends on the pressure of the solvent and also on the polymer’s
molecular mass – effectively the size of its molecules. So different-sized
polymer molecules could be separated by changing the pressure of the system
gradually, so that each size drops out of solution at a slightly different
time. The hope is that researchers might in this way produce more consistent
and purer materials for applications such as resilient engine seals and
fuel line seals.

Other chemists are exploring how SCFs could lead to new catalysts. Martyn
Poliakoff and his team at the University of Nottingham have discovered that
they can make organometallic compounds such as metal carbonyls, many of
which are too unstable to prepare by traditional methods, using SCFs as
a solvent. Metal carbonyls take part in many industrially important chemical
reactions as reactants and catalysts and are used for making basic industrial
chemicals like formic acid and formaldehyde.

CRITICAL PRESSURE

Chemists have begun to realise that carbonyls in which nitrogen or hydrogen
molecules have been substituted for a carbonyl group can catalyse more complex
reactions too. The problem is that at room temperature hydrogen and nitrogen
gases do not dissolve well in conventional reaction solvents such as THF
or toluene. But Poliakoff and his colleague Steve Howdle realised that hydrogen
does mix exceptionally well with supercritical CO₂ in a reactor
at 80 to 100 times atmospheric pressure: ‘At the operating pressure of our
reactor we find that the SCF is perhaps ten times better than a conventional
solvent,’ says Poliakoff. His team has made a tungsten carbonyl in which
a hydrogen molecule replaces a carbonyl group. It survives for only a second
at room temperature in a conventional solvent, but under supercritical conditions
it can survive for more than three minutes – long enough to be studied using
relatively simple analytical techniques. They hope this will give them an
insight into why the compound is so unstable at room temperature, and allow
them to design other, longer-lived compounds that might be useful catalysts
for industry. They have even surprised themselves, by making other organometallic
compounds such as a rhenium cyclopentadienyl in which there are three nitrogen
molecules attached to the metal atom, and small quantities of the manganese
version. Such organometallics have great potential as catalysts. ‘Neither
of these compounds could be made in conventional solvents,’ says Poliakoff.
‘And, researchers had previously believed that such species would be unstable
at room temperature.’

In an intriguing variation on a technique called matrix isolation, Howdle
also found that he could trap reaction products such as the fleetingly stable
dinitrogen complexes in an inert matrix of polyethylene. He did this by
simply adding solid polyethylene to a mixture of supercritical CO₂,
nitrogen gas and metal carbonyl. This idea of trapping high-energy, and
therefore unstable, compounds for further study in an inert material usually
involves the much more difficult process of freezing a compound to a few
degrees above absolute zero. Poliakoff suggests that the technique could
offer a cheap and efficient way of impregnating carrier materials with drug
molecules. Because SCFs are far less viscous than the organic solvents traditionally
used in drug impregnation processes, they would have the added advantage
of leaving minimal residue in the tablet at the end of the procedure.

Other chemists have also succeeded in altering the course of some important
reactions by replacing conventional, highly halogenated solvents with supercritical
CO₂. Chemical reactions in which some of the hydrogen atoms
in toluene or ethylbenzene are swapped for halogen atoms such as bromine
are used to make the precursors for a variety of compounds, with uses that
range from agrochemicals to pharmaceuticals. In the past, these halogenation
reactions have worked efficiently only in solvents such as CFCs and tetrachloro-methane,
which do not themselves have removable hydrogen atoms. But earlier this
year, James Tanko and Joseph Blackert of the Virginia Polytechnic Institute
and State University in Blacksburg, reported that they had used supercritical
CO₂ as a solvent for an important group of reactions of this
type, known as free-radical side-chain brominations, in which they were
able to exert almost as much control over which hydrogen atoms are substituted
as is possible with conventional solvents.

In March, a team led by Ryoji Noyori at the Research Development Corporation
of Japan in Toyota reported that as a solvent supercritical CO₂
offered a fast and efficient route to formic acid, which is an important
industrial feedstock in processes that lead to drugs, agrochemicals, plastics
and many other materials.

As well as finding applications for SCFs, chemists are beginning to
investigate how their unusual properties arise. Though supercritical water
retains many of the properties of an ordinary liquid, it is also gas-like,
in that it has expanded to three times the volume the liquid would occupy
under everyday conditions. This means that supercritical water is both more
compressible than ordinary liquid water and has weaker bonds between the
molecules, which is why it will dissolve oily molecules. The water molecules
simply pack themselves round each organic molecule, much as molecules of
an organic solvent would do.

ALTERED STATES

But there is still a long way to go before chemists understand precisely
how supercritical fluids work at the molecular level. For this they will
have to pin down exactly how the molecules pack together, and work out the
energy changes – the thermodynamics and kinetics – of the packing process.
They must look in detail at how polarity is reduced and in what way the
‘hydrogen bonding’ structure of liquid water – the network of weak bonds
that connects the hydrogen atoms of one water molecule and the oxygens of
others – changes in the supercritical state.

Keith Johnston at the University of Texas at Austin is among those focusing
their efforts on such fundamental studies. In the past few years, Johnston
and his colleagues have shown that it is possible to alter the outcome
of a reaction carried out in a supercritical solvent by slightly altering
the pressure, and therefore the density, of the solvent. As a result they
not only have a clearer insight into how SCFs affect the mechanism of reactions,
but have pointed to a useful practical technique for making organic molecules,
where precise control over the outcome is highly desirable.

For example, when they studied how molecules of isophorone, a fairly
simple organic compound used in resin manufacture, join or ‘dimerise’ under
ultraviolet light in supercritical trifluoromethane and supercritical CO₂
they found that the type of solvent and final reaction pressure strongly
influenced the three-dimensional structure of the dimer.

These and similar developments look set to give SCFs an increasingly
prominent place in laboratories and industrial chemicals plants. They have
some clear environmental benefits over conventional solvents and, as Poliakoff
points out, such ‘tunable’ solvents add ‘a new dimension to chemistry,
which has never been done before’

David Bradley is an editor at the Royal Society of Chemistry based in
Cambridge and a freelance science writer.

* * *

Vending-machine chemistry

Most chemicals are made in batches in which starting materials are added
to a solvent in bulky reaction vessels. To get the reaction going, the mixture
may have to be raised to high temperatures or put under pressure, and catalysts
or initiators may also be required. At the end of the reaction the products
must be separated from by-products, and solvent. This is efficient on an
industrial scale, but there is room for improvement.

Martyn Poliakoff of the University of Nottingham envisages a time when
chemistry will be as simple as operating a drinks vending machine: ‘The
chemist will simply press a button and the machine will add the appropriate
reagents to the supercritical CO₂ and pump the mixture into the
reactor,’ he says. His optimism stems from some pioneering work being carried
out in his laboratory by Steve Howdle, James Banister, Peter Lee and others
who have built a flow system for doing continuous chemistry in SCFs, helped
by sponsorship from BP, the petrochemicals company.

In this system SCF is squeezed into a reaction vessel containing the
starting material at 80 to 100 times atmospheric pressure, controllable
to within 1 atmosphere. The reactants dissolve in the SCF, and the flow
of the fluid carries them into a reaction vessel, which the researchers
can irradiate with ultraviolet light to convert reactants to product. The
chemists monitor progress using infrared spectroscopy, which reveals which
atoms are present by identifying characteristic absorption of light as peaks
on a graph. Metal carbonyls have particularly characteristic ‘fingerprints’.

Industrial flow reactors using traditional solvents already exist, but
in these the product has to be separated from the solvent and purified.
Reactions in SCFs do away with the problem. The solvent simply boils off
and is collected for reuse. In principle, the researchers believe their
reactor could easily be modified to run as a ‘closed cycle’ in which impurities
would be removed from the SCF at the end of the reaction by passing it through
pellets of porous, absorbent minerals known as molecular sieves; the SCF
could than be recycled.

The density and electrical and solvent properties of SCFs can be altered
by slightly changing their temperature or pressure to control the rate at
which the reaction runs, and the products that result. In conventional
industrial chemistry, by contrast, this sort of fine tuning means changing
solvents or adding expensive catalysts. There is no problem scaling up
a flow system to make it suitable for industry: it is simply run for longer,
to increase the amount of product. And the low viscosity of SCFs means that
all piping can be kept narrow, helping to keep costs down.