ÐÓ°ÉÔ­´´

Advances in technology often depend on an ability to prepare novel materials
with specific properties. Over the past 150 years, the study of organic
chemistry has produced a vast repertoire of well-tested reactions and chemical
reagents, as well as a detailed understanding of how they work. By making
systematic use of this body of knowledge, chemists can construct complex
organic molecules which find use as drugs, insecticides or herbicides.

Techniques for designing useful inorganic compounds are much less advanced,
however. While organic chemistry concentrates on carbon-hydrogen compounds,
inorganic chemistry deals with the entire periodic table of elements. These
elements are now known to form a vast array of molecular structures many
of which were unknown just a few years ago. Undoubtedly, many more are waiting
to be discovered.

The principles that underlie molecular structure and chemical bonding
in inorganic materials are often far more complex than those related to
organic compounds, and are only partly understood. For example, the electronic
behaviour of the high-temperature superconducting ceramics is still largely
a mystery. Inorganic ‘designer chemistry’ is in its infancy, but the challenge
of making entirely new classes of compounds with unusual structures and
properties is an exciting prospect.

The rapidly expanding area of organometallic chemistry deals with compounds
containing bonds between metal atoms and carbon atoms. Chemical journals
are full of new organometallic compounds with unusual structures. Many such
materials are interesting theoretically, often because they involve unusual
bonding, but few of them have found commercial applications in the way that
new organic compounds have. This situation may soon change. Several research
groups are working with families of organometallic compounds that may lead
to new high-tech materials. One such class is the metal ‘sandwich’ compounds,
in which a metal atom is literally sandwiched between two, usually flat,
rings made up of carbon or other atoms. Our research group at the University
of Virginia is particularly interested in exotic ‘multidecker’ sandwiches
having two or more metals and three or more rings, and in which the middle
rings contain the element boron as well as carbon. We are excited about
these structures because of their potential for use as building blocks for
creating materials with useful electronic properties.

As often happens in chemistry, the first known sandwich complex was
prepared accidentally. In 1951 two research groups were trying independently
to make completely different compounds containing iron and the ring-shaped
cyclopentadienyl hydrocarbon group (C₅H₅). Unexpectedly, they
both obtained instead some strange orange crystals. X-ray crystallography
showed that the molecule had an iron atom symmetrically located between
two flat, pen-tagonal, cyclopentadiene rings (see Figure 1a). Nicknamed
ferrocene, this compound is more formally known as bis(cyclopentadienyl)iron
and its chemical formula is Fe(C₅H₅)₂. Chemists were astonished
by ferrocene’s structure. How was the iron atom bound to the two cyclopentadiene
rings? The important feature is that the cyclopentadienyl groups are ‘aromatic’,
a term used by organic chemists to signify that some of the electrons are
‘delocalised’ and so move freely around the ring. The delocalised electrons
interact with the atomic orbitals of the metal, forming very strong chemical
bonds between the iron atom and the carbon ring.

The discovery of ferrocene signalled the birth of modern organometallic
chemistry, revealing a vast potential for exotic chemical structures involving
metals bound in unusual ways to carbon and other elements. Since then, chemists
have prepared many similar compounds, generically called metallocenes, from
combinations of other metals and ring-shaped hydrocarbons. For example,
you can make a sandwich from chromium and benzene (C₆H₆), which
has a ring of six carbon atoms containing three double bonds. The compound
dibenzenechromium (see Figure lb), and others like it, were actually prepared
as early as 1919, but at the time no one realised that they had a sandwich
structure. Had their discoverer, F. Hein, conceived of such an unlikely
geometry, he would no doubt have rejected it, fearing ridicule from the
chemistry establishment of that time.

Compounds such as ferrocene are very stable. This has led chemists to
wonder whether larger stacks could be made: could another metal-cyclopentadiene
unit be added to a metallocene, for example, to form a triple-decker sandwich
having three rings and two metals (see Figures 1c, d, and e)? In 1972, Helmut
Werner and Albrecht Salzer at the University of Zurich made and identified
the first triple-decker. In their compound nickel was sandwiched between
cyclopentadiene molecules to form the dinickel ion Ni₂(C₅H₅)₃⁺.
It was prepared as a salt with a boron tetrafluoride anion (BF₄^–) which
is unstable and highly reactive towards air and moisture. Later in the same
year, our research group prepared the first electrically neutral triple-decker
compounds. These were different in that the middle pentagonal ring was a
carborane ring (C₂B₃H₅) rather than cyclopentadienyl, and the
metal atoms were cobalt. The overall formula was (C₅H₅)₂Co₂(C₂B₃H₅)
(see Figures 1d and e). These dicobalt compounds proved to be extraordinarily
stable, surviving temperatures as high as 400 °C and not reacting with
oxygen at room temperature. This stability is characteristic of a class
of compounds containing clusters of carbon and boron and metal atoms known
as metallacarboranes. A group of chemists at the University of California
at Riverside, led by M. Frederick Hawthorne, prepared the first such compounds
in 1965 in the form of large 12-atom icosahedral cages; Hawthorne subsequently
developed this class of structures into a huge area of research.

Boron the stabiliser

It was clear to us that the presence of boron in the central ring of
our triple-decker carborane complexes helps to stabilise these structures,
and this led us to prepare a whole series of stable triple-decker sandwich
compounds based on the same principle. Two groups of German chemists have
also made triple-decker compounds with rings containing boron. Walter Siebert
and his group at the University of Heidelberg have made boron-stabilised
triple-deckers in which the middle ring contains three carbon atoms and
two boron atoms, while Gerhard Herberich’s team at the University of Aachen
has made similar compounds incorporating borole (C₄B) rings. Recently, triple-deckers
have been made with benzene rings and rings containing phosphorus and/or
arsenic carborane atoms. Meanwhile, we succeeded in extending our family
of triple-deckers to include structures containing heavy metals such as
ruthenium and osmium, as well as top and bottom rings containing nitrogen
(C₄N) and phosphorus (C₄P).

Encouraged by the stability of these boron-containing sandwich compounds,
we have tried using them as building blocks to construct even larger structures.
Because some of the bonding electrons in the sandwich compounds are delocalised,
we think it should be possible to construct multidecker sandwich compounds
that are electrically conducting. Many research groups are looking at very
large conducting molecules (conducting polymers) for applications as diverse
as transistors, electrochromic materials (which change colour in response
to an electric current) and for use in rechargeable batteries. Most conducting
polymers being studied are based on conventional organic building-block
units such as pyrrole (C₄N) or thiophene (C₄S) rings, or even simple hydrocarbons,
as in polyacetylene ((CH)_n). Our carborane sandwiches could, in principle,
lead to an entirely new class of conducting polymers.

As it happens, boron is an excellent element for promoting electron
delocalisation in sandwich compounds because of its inherent ‘electron deficiency’.
This arises because boron’s valence electron shell, which takes part in
bonding, has only three electrons, one fewer than carbon’s. Boron also readily
shares its valence electrons with neighbouring atoms. Detailed studies of
triple-decker carborane sandwiches reveal that electron delocalisation is
extensive. Working in cooperation with William Geiger and his group at the
University of Vermont, we have shown that if there is an odd, ‘unpaired’
electron on one of the two metal atoms in the sandwich, it flips back and
forth between the metals so rapidly that, in effect, the electron resides
on both simultaneously. Strange as it may seem, this behaviour is allowed
under the rules of quantum mechanics which govern the world of electrons
in atoms. We would expect to find electron delocalisation in larger stacks
as well, and investigations of tetra-decker complexes indicate that this
is indeed the case.

The question is, if several such sandwiches are connected, will electrons
be able to travel easily between them? This depends on whether we can join
them with the right kind of linking units, or ‘molecular wires’. Fortunately
it is quite easy to add a wide variety of organic and inorganic groups of
atoms at specific locations in the carborane rings. Such versatility is
very important, because it allows us to make and examine many different
kinds of multi-sandwich compounds. This kind of ‘molecular engineering’
holds great promise for making novel materials tailored to have specific
properties.

In our laboratory we are exploring several ways in which metallacarborane
sandwich units can be linked using cyclopentadiene or benzene rings, as
shown in Figure 2. One way is to stack the sandwiches so that the metal
atoms are aligned, with hydrocarbon rings capping both ends of the structure
as in Figure 2a. So far there are few examples of tetra-decker or larger
molecules stacked in this way. The first tetra-decker sandwiches were prepared
by the Heidelberg group, who also produced a penta-decker, a hexa-decker,
and even polydecker compounds. All of these contain either diborolene (C₃B₂)
rings or thiadiborolene (C₂B₂S) rings; the central metal atoms
are usually cobalt, iron or nickel. In our research, we sought for years
to make carborane (C₂B₃) tetra-decker sandwiches, but were unsuccessful
until last year, when we unexpectedly found a way to do it while working
on a different problem.

To describe how this was accomplished, let me first explain how we make
our metallacarborane starting materials from the boron hydride pentaborane(9),
which has the formula B₅H₉. This compound was manufactured by the tonne
in the 1950s as part of the US government’s programme to develop boron hydrides
as rocket and jet fuels. After a few years, the programme was cancelled,
leaving a large stockpile of pentaborane available for research purposes.
As shown in Figure 3, the reaction of pentaborane with an alkyne (a hydrocarbon
containing a carbon-carbon triple bond) gives a small pyramidal carborane
containing four boron atoms and two carbon atoms with two bridging hydrogen
atoms located between adjacent boron atoms. These hydrogen bridges can be
removed to give a negatively charged ion which can react with a metal atom
attached to a hydrocarbon ring. In a process we informally call ‘decapitation’,
the boron atom at the apex of the pyramidal cage is removed and replaced
with two bridging hydrogen atoms to give a double-decker molecule (see Figure
3a). These molecules are the building blocks for the molecular engineering
that follows. For example, by ejecting the two bridging hydrogen atoms still
attached to the carboranedouble-decker sandwich, we can add a second metal-hydrocarbon
unit to create a triple-decker (see Figure 3a, bottom).

We then thought that it should be possible to form a tetra-decker stack
in a similar way by sandwiching two carborane double-deckers around a metal
ion by ejecting the remaining bridging hydrogen atoms. Although we tried
this reaction many times and failed, we eventually succeeded by first modifying
the electronic character of the open face on the double-decker complex.
We found by accident that if the hydrogen atom normally attached on the
middle boron atom is replaced by another atom (X) that reduces the electron
density on the carborane ring – a chlorine atom, for example – the reaction
gives the desired tetra-decker (see Figure 3b).

Conducting polydeckers

Figure 4a shows an example of a tetra-decker made in this way. As you
can see, the stack is not perfectly straight but bends slightly in the middle.
The analogous tetra-decker sandwich with three cobalt atoms has one less
electron; as a result, one electron must be unpaired, making the compound
paramagnetic. We can also make the cobalt-nickel tetra-decker paramagnetic
by adding or removing electrons. These paramagnetic sandwiches are particularly
interesting because their unpaired electrons are delocalised between the
metal atoms. We anticipate that our method of producing multi-decker sandwiches
with many metal atoms can eventually produce polymers in which electrons
can travel easily up and down the stack but not in other directions. Many
researchers are studying such materials, called one-dimensional conductors,
in the hope that they might be useful for novel electronic systems. The
Heidelberg chemists have made a nickel-diborolene polymer which is a semiconductor.
We think that we can make similar polydecker stacks from metallacarborane
units.

Multi-decker sandwiches are just one of many possible types of electron-delocalised
networks that could be constructed from small metallacarborane modules.
For example, linking triple-decker sandwiches through a bridge of hydrocarbon
rings that have delocalised electrons may allow electrons to migrate between
the two stacks. An example is shown in Figure 4b, in which a tetra-decker
and two double-decker sandwiches containing cobalt are tied together by
polycyclic hydrocarbon rings. We have succeeded in making very large molecular
structures in this way.

A different approach to assembling large metallacarborane systems involves
linking individual sandwich units via connecting groups attached to boron
or carbon atoms in the carborane rings. To do this we adapted some well-known
reactions to develop methods for replacing the hydrogen atoms of the carborane
rings with other groups of atoms, as in Figures 2c and d. This may allow
us to make a polymer consisting of linked carborane triple-decker units,
as in Figure 5.

What do these types of boron-based compound have to offer, and why does
the world need them? First, all of our metallacarborane materials are coloured,
crystalline solids that are extremely easy to work with. They are soluble
in organic solvents, which means that they can be used in a large variety
of chemical reactions. Unlike other metal-hydrocarbon sandwiches, they are
stable in air, which constitutes a major advantage over many organometallic
compounds in terms of molecular engineering. In metallacarborane sandwiches,
the carborane units stabilise a considerable variety of metal-organic ring
combinations. A direct consequence of this ‘boron stabilisation’ is that
it opens up a truly enormous range of possibilities for creating new molecular
structures. The molecules described here are but a very tiny tip of a huge
iceberg.

A second advantage is that we can introduce organic functional groups
systematically and thereby select and fine-tune their chemical and electronic
properties. For example, we can make a sandwich complex acidic by attaching
carboxylic acid (COOH) groups. Such molecular tailoring is commonplace in
organic chemistry, but has been rare in inorganic chemistry.

Finally, the special electronic nature of the carborane multidecker
sandwich systems opens the way to many possible future uses. Heavy metals
such as cobalt, iron and nickel have the interesting and useful property
that some of their valence electrons can easily be transferred to other
atoms or molecular fragments, including those bound to the metal. Metallacarborane
sandwiches may therefore have potential applications in the burgeoning field
of molecular electronics. For example, one dicobalt triple-decker readily
gains or loses electrons to give five differently charged molecules. Other
metallacarborane triple and tetra-decker sandwiches behave similarly. Such
‘electronic versatility’ might well be exploited in useful ways, such as
in polymeric sandwiches that could function as electrical conductors, insulators
or semiconductors in different electronic environments. What is more, the
different electronic states also generate different magnetic properties,
which could also be exploited. The future could bring entire families of
new materials designed to have specified combinations of these properties.

But this field is still at a very early stage of development. Both we
and other research groups have a lot more to learn about how to make multi-decker
molecules, and their chemical and physical properties, before they can usefully
be applied in technology. The evolution of this science should be fun to
watch, and may well lead to breakthroughs that we cannot foresee.

Russell Grimes is professor of chemistry at the University of Virginia,
Charlottesville.