Anda di halaman 1dari 134

A dvances in Physics, 1999, Vol. 48, No.

1, 1± 134

Fullerenes under high pressures

B. Sundqvist
Department of Experimental Physics, UmeaÊ University, S-90187 UmeaÊ , Sweden

[Revision received 2 December 1997; accepted 11 December 1997]

Abstract
This paper reviews the properties and phases of fullerenes and their derivatives
and compounds under high pressures. For obvious reasons most of the paper deals
with C60 but the materials reviewed also include C70, simple derivatives of C60 ,
carbon nanotubes, and intercalation compounds of C60 with both acceptors and
donors, mainly alkali metals. After a brief overview of high-pressure techniques
and the structures and properties of C60 at atmospheric pressure, the structural
phase diagram of C60 from atmospheric pressure to above 40GPa (400kbar) is
reviewed. The evolution with pressure of the orientational and translational
structure of `normal’ molecular C60 in the range up to 1± 5 GPa (depending on
temperature) is discussed in some detail, as is the appearance of a large number of
polymeric phases at higher pressures and temperatures, some of them known to
have extreme mechanical properties. At very high static (or shock) pressures or
temperatures, C60 transforms into ordered or disordered forms of diamond or
graphite. The phase diagramis reasonably well investigated up to near 10GPa, but
at higher pressures there are still large gaps in our knowledge. Available experi-
mental data for the physical properties of both monomeric and polymeric C60
under high pressures are reviewed as far as possible. The compression behaviour of
C60 has been well investigated and is discussed in detail because of its basic
importance, but optical, electrical and lattice properties have also been studied for
several of the many structural phases of C60 . Whenever possible, experimental data
are compared with the results of theoretical calculations. The phase diagram and
properties of C70 are much less known because of the larger complexity caused by
the anisotropy of the molecule, and very little is known about most compounds of
C60 . However, noble-gas intercalation in C60 has been reasonably well investi-
gated. Finally, the high-pressure properties of superconducting alkali-metal-
intercalated C60 are brie¯ y reviewed.

Contents page
1. Introduction 2
2. Structures and phases of fullerenes at zero pressure 4
3. Some comments on high-pressure techniques 7
4. Structures and properties of molecular (monomeric) C60 9
4.1. Structural phase diagram 9
4.1.1. High-temperature behaviour and the fcc phase 10
4.1.2. Fcc-to-sc transition boundary 11
4.1.3. The orientational structure in the sc phase 14
4.1.4. The glassy crystal transition and the properties of the
glassy phase 19
4.1.5. Low-pressure phase diagram of C60 21
4.1.6. Does liquid C60 exist? 23
4.2. Physical properties of monomeric C60 under pressure 24
4.2.1. Compressibility and other elastic properties 24
0001± 8732/99 $12´00 Ñ 1999 Taylor & Francis Ltd
2 B. Sundqvist

4.2.2. Lattice vibrations 37


4.2.3. Thermal properties 44
4.2.4. Electronic band structure 46
4.2.4.1. Optical studies 46
4.2.4.2. Positron annihilation 48
4.2.4.3. Other studies 49
4.2.5. Electrical resistivity 50
4.2.6. Nuclear magnetic resonance 53
5. Structures and properties of polymeric C60 54
5.1. Photopolymerization 54
5.2. Early very-high-pressure studies 54
5.3. Low-pressure polymerized material ( p < 2 GPa) 56
5.4. Intermediate-pressure range (2± 8 GPa) 64
5.5. High-pressure range ( p > 8 GPa) 73
5.6. Transformations into other forms of carbon 79
5.6.1. High static pressures and/or temperatures 79
5.6.2. Catalytic conversion to diamond 81
5.6.3. Shock wave experiments 81
5.6.4. Other studies 82
5.7. Pressure± temperature phase diagram of C60 82
6. Structures and properties of C70 83
6.1. Phase diagram 83
6.2. Physical properties of C70 89
6.2.1. Thermophysical properties 89
6.2.2. Lattice vibrations 91
6.2.3. Electronic properties 92
7. Other fullerenes and fullerene compounds 93
7.1. Nanotubes 93
7.2. C60O, C61 H2, C60Hx and ( C59N) 2 94
7.3. Other C60 complexes and compounds 97
7.4. Intercalation of gases into C60 98
7.5. Endohedral fullerenes 100
8. Alkali-metal- and alkaline-earth-metal-doped C60 101
8.1. Normal-state properties 102
8.1.1. Bulk modulus 102
8.1.2. Electrical resistivity 104
8.1.3. Other studies 107
8.2. Superconducting-state properties 108
9. Comments, speculations and conclusions 113
Acknowledgments 115
References 115

1. Introduction
It is hardly original or controversial to state that the identi® cation of the C60
molecule by Kroto et al. [1]and the subsequent discovery by KraÈtschmer et al. [2]of
a simple and inexpensive method to produce large amounts of this and related
materials must be counted among the most important scienti® c discoveries during
the last two decades of this century. Over the last 6 years, research in this ® eld has
developed at a breathtaking rate, exceeded only by that on ceramic high-transition-
temperature superconductors, and the discoverers were recently honoured by being
awarded the 1996 Nobel Prize in chemistry. Not surprisingly, a large number of
books, reviews and conference proceedings have already been published on this
subject. A number of these will be referred to below and it should be stated here that
Fullerenes under high pressures 3

the selection to some extent re¯ ects the author’s own preferences, but in particular
the fact that these particular publications happened to be readily available at the
time of writing. For a particularly complete and up-to-date general review of the ® eld
the reader is referred to recent excellent book by Dresselhaus et al. [3], to which
reference will often be made below, but several other very general reviews [4± 7] or
collections of reviews [8± 10] have appeared over the last few years. However,
although the whole fullerene ® eld, various sub® elds, and particular properties of
various fullerenes have been reviewed in di€ erent publications, I am not aware of
any previous general review of the properties of C60 under high pressures. Only a
small number of short, rather specialized reviews [11± 15]and some sections on high-
pressure e€ ects in general reviews [3, 4, 7, 16] have so far been published. This is
somewhat surprising, since high-pressure studies have given very important informa-
tion on both pure and doped fullerenes and contributed signi® cantly to the rapid
developments in this ® eld, and many recent high-pressure studies hint at possibilities
for commercial exploitation of the structural phases existing at high pressures. In this
paper, I have tried to collect as much as possible of the information available on the
high-pressure properties and phases of both pure and doped solid fullerenes, as well
as compounds and derivatives. Fullerenes are here de® ned in a broad sense to
include both quasispherical molecular species such as C60, C70 and C76 and larger
complexes such as `buckytubes’ and `buckyonions’. Most high-pressure work so far
has of course been carried out on C60 because of the easy availability of this material
in pure form, but some information is also available on other fullerenes. `High
pressure’ is also very loosely de® ned, but as a general rule I have arbitrarily selected
100MPa (= 1 kbar) as the lower limit for inclusion, although in some cases reference
is made to work at lower pressures. The paper will also concentrate on experimental
results and discuss theoretical work only brie¯ y, partly because I am myself an
experimentalist, partly because only a small number of theoretical papers have
explicitly discussed the pressure or volume dependence and partly because I feel that
at the present level of understanding the most important task is to collect and
evaluate available experimental data to build a reference platform on which to build
future work, experimental or theoretical. The ® rst version of this review attempted to
give the reader a view of the state of our knowledge in the ® eld up to the autumn of
1996. During the refereeing process the ® eld has developed rapidly in some areas and
the Editor has been kind enough to permit me to update the paper. In this ® nal
version I have therefore extended the text, in particular in section 5, to take into
account also the most important developments between the initial submission of the
paper and the summer of 1997. I make no claim that the paper is complete and I
apologize for any oversights or errors that have succeeded in creeping in, but I hope
that I have included references to most of the important work that has been carried
out under high pressures.
The structure of the paper is as follows. A very brief and incomplete overview is
® rst given of the ambient-pressure structural phases of undoped fullerenes to set the
stage for subsequent sections. Since excellent reviews already exist in this area,
reference will be made either to original work or to available reviews rather than
presenting too many details. Before starting with the actual review, some brief
comments on high-pressure studies and techniques are also given in section 3. In
sections 4 and 5 the high-pressure phases, structures and properties of C60 will then
be discussed in some detail since these have now been extensively investigated by
many groups. The discussion will be divided into two main parts, with section 4
4 B. Sundqvist

discussing ® rst the rather well known structures of the phases based on molecular (or
monomeric) C60 and then various physical properties of these phases under pressure,
and section 5 discussing the structure and properties of the less well studied
polymeric forms of C60 . It should be noted that most physical properties of these
phases have only been measured on metastable material at `zero pressure’, here
de® ned as any pressure near or below atmospheric. The short sections 6 and 7 then
discuss available information on the phase diagrams and physical properties of C70
and other fullerenes, derivatives and compounds respectively, before we turn to
doped fullerenes and their properties under high pressures in section 8. Finally, the
paper is rounded o€ by a short section 9 containing some general comments and
conclusions.

2. Structures and phases of the fullerenes at zero pressure


This review is focused on solid fullerenes only and, since little information is
available for materials other than C60 and C70 under high pressures, most of the text
deals with these. In this section a very brief introduction will be given to the
structures of the zero-pressure phases as a help to understand the modi® cations
brought about by the application of a high pressure. Information on the structural
and dynamic properties of C60 at zero pressure can be found in a number of reviews
[3, 6, 16± 21]which will also be referred to where necessary below.
The most convenient method to produce fullerenes such as C60 in large amounts
is to evaporate graphite electrodes to soot under a low-pressure He gas [2], usually
using an electric arc discharge [22]. The soluble fullerenes are then washed out of the
soot using toluene which on evaporation leaves solid fullerene deposits behind. The
production, separation and puri® cation of fullerenes and related materials have
recently been discussed in some detail by Dresselhaus et al. [3]. Fullerenes can also be
produced in ¯ ames [23], although normally most molecules burn before leaving the
¯ ame, and claims have been made for the detection of natural fullerenes in carbon-
rich minerals (Shungite) [24], in fulgurites (rocks produced by melting minerals
during lightning discharges) [25], in sediments containing traces of intense burning
from the Cretaceous± Tertiary boundary layer [26], and even in trace amounts at
minuscule impact craters on spacecrafts [27].
Whatever the source, solid fullerenes form black or brownish powders or
crystals. As discussed recently by Dresselhaus et al. [3] and many others, the most
stable crystal structure at high temperatures is the same for all quasispherical
fullerenes (C60 , C70, C76 , etc. ). Because the molecules rotate almost completely freely
they are e€ ectively spherical and the molecules form a close-packed face-centred
cubic (fcc) structure. The room-temperature ( RT) structure of C60 has been
investigated in very close detail by X-ray and neutron di€ raction. These studies
show [3, 21, 28± 31] that the molecular rotation is not, in fact, completely free, but
that there is a strong intermolecular orientational correlation even in this phase.
Close to the orientational transition these correlated clusters may reach 40 A Ê in size
[30, 31]and thus consist of several dozen molecules. For our purposes, however, the
description of the molecules as freely rotating is good enough. On cooling the
rotation slows down and, below an orientational transition temperature [3, 16± 19,
29, 32, 33] To = 260K, the rotational motion is replaced by a combination of rapid
jumping motion between a number of well de® ned orientational states and a
librational motion about the equilibrium orientation in each such state. The
Fullerenes under high pressures 5

realization by Heiney et al. [32]that in this low-temperature phase the molecules are
orientationally ordered was one of the ® rst of the many surprising and exciting
discoveries concerning solid C60 . The basic structure of the low-temperature phase is
identical with that of the fcc phase except that the four molecules in the cubic unit
cell have di€ erent orientations, changing the fcc symmetry into a simple cubic (sc)
structure. However, there are two possible orientations [3, 29, 33] obtained by
rotating the molecules by about 38ë and 98ë respectively around the [111]axis of the
crystal from the `standard’ orientation. (A recent study [34]showed that the actual
angles may in fact be closer to 42ë and 102ë , values optimizing nearest-neighbour
orientations rather than long-range orientational order. ) In the resulting ordered
structures, electron-rich double bonds on one molecule face the electron poor centres
of hexagons (H orientation) and pentagons (P orientation) respectively on its
neighbour, as shown in the simpli® ed sketch in ® gure 1 (A). Since the latter
description is easier to visualize for non-specialists (including the author), I shall
refer in the following to H- or P-oriented molecules. The di€ erence in energy is very
small and the fraction of molecules in the more stable P orientation is only about
60% near 260K (® gure 1 (B)), increasing to about 84% near Tg = 90 K [33]. Below
this the thermal energy becomes too small compared with the energy threshold
between the two states for further reorientation to be possible, and below a glass
transition (or glassy crystal transition) temperature Tg the remaining orientational
disorder becomes frozen in creating an orientational glass. Interestingly, the lattice
parameter of H-oriented sc C60 is slightly smaller than for the P-oriented state. This
can be observed in ® gure 2 as an anomalously low thermal expansion coe cient a in
the sc phase [29, 35], since for every increase in temperature a certain number of
molecules reorient from the majority P orientation to the minority H orientation,
giving a small volume decrease which adds to the normal volume increase due to the
increase in temperature. Most physical properties of C60 have already been well
studied in these three phases at zero pressure and data can be found in the reviews
given above.
In addition to the fcc, sc and `glassy’ states, a fourth state, photopolymerized C60,
insoluble in common solvents, can be obtained by submitting thin ® lms of fcc C60 to
high-intensity visible or ultraviolet (UV) light [36]. This treatment is believed to
break up the sp2 double bonds joining adjacent hexagons on the molecules and to re-
form the bonds to form four-membered carbon rings linking neighbouring molecules
by two single carbon bonds with a strong sp3 character. Because C60 is a strong
absorber of light, only very thin ® lms can be completely polymerized in this way and
this has seriously hampered the study of this interesting material, but it is believed
that the polymer chains formed in this way are probably quite disordered and that
chain length is highly variable with polymerization conditions.
The situation regarding the structure of C70 is much more complicated. Although
molecular dynamics calculations [37± 40] have shown that the fcc structure should
theoretically be the most stable high-temperature structure, in real crystals the other
close-packed structure, hexagonal close packed (hcp), very often appears. Because of
the small di€ erence in energy, most crystals contain enough stacking faults to be
considered as mixtures of the fcc and hcp phases. Molecular rotation in C70 can
also give a much more complicated phase diagram because of the anisotropy of
the molecule. `Free’ molecular rotation (with the same caveats as above) seems to be
the rule at high temperatures, where the fcc and hcp structures are both based on the
resulting quasispherical molecular shape. Below about 350K, rotation around the
6 B. Sundqvist

(A)

(B)
Figure 1. (A) Molecules may be oriented with the double bond on the black molecule facing
either a pentagon (top) or hexagon (bottom) on the grey neighbouring molecule. (B)
Temperature dependence of the fraction of C60 molecules in the more stable P-
oriented state at atmospheric pressure. Insets (a) and (b) show projections of
molecules in the P- and H-oriented states respectively. (Reprinted with permission
from David et al. [33].)

short molecular axes is believed to stop and the molecules rotate around the long axis
only. This rotation persists at least down to about 280K. Taking the fcc phase as an
example, most workers agree that this phase is stable above 350K. Below this, the
axes of the uniaxially rotating molecules can either remain in random directions or
line up along random [111]directions, in both cases giving a fcc structure, or line up
Fullerenes under high pressures 7

Figure 2. Cubic lattice parameter of C60 plotted against temperature, showing clearly the
glass transition temperature Tg near 90K and the orientational transition tem-
perature To near 260K. ( Reprinted with permission from David et al. [33].)

in parallel along one particular [111] direction giving a rhombohedral (rh) crystal
structure. When the uniaxial rotation ® nally stops below 280K, any of these
structures may remain, or molecules in the rh phase may order orientationally
relative to their nearest neighbours in much the same way as in C60 to form a
monoclinic (mc) phase. However, there are many other possibilities. For example, it
has been suggested that in the uniaxially rotating phase (280± 350K) the molecules
behave as rotating tops, carrying out a nutational movement where the long axis
rotates around its equilibrium orientation [40, 41], and two hcp phases have been
reported to occur at RT [42]. Recent studies also show that uniaxial rotation
continues down to about 150K [43± 46], but it is still unclear whether this
temperature corresponds to the arrest of `free’ uniaxial rotation or to a glass
transition similar to that observed near 90 K in C60. Starting with the less stable
hcp structure at high temperatures a di€ erent structural evolution not discussed here
is believed to occur on cooling. It is thus obvious that the structural properties of C70
are much less well known than those of C60 , mainly because most samples are
probably mixtures of initially fcc and hcp materials and transitions belonging to both
structural sequences are often observed [47]. In fact, it may well be thermodynami-
cally impossible to create a pure single-phase crystal of C70. As will be seen below,
this is also re¯ ected in the present status regarding the knowledge of the phase
diagram and the physical properties of C70 .

3. Some comments on high-pressure techniques


Before discussing the phase diagrams and properties of fullerenes under high
pressures a few remarks should be made about high-pressure experiments and
techniques. Many experimental techniques have been used in studies of fullerenes
and it has sometimes been suggested that the technique used has a large e€ ect on the
8 B. Sundqvist

results obtained. The most obvious problem that may occur is deformation of the
sample. If the pressure is transmitted by a ¯ uid medium (gas or liquid) with zero
shear strength, the deformation will be homogeneous and a cubic single crystal such
as C60 will simply su€ er a decrease in volume without any change in shape.
(However, relative changes in atomic positions within the molecule or unit cell
may still occur. ) In practice, low-pressure experiments are often carried out under
such hydrostatic conditions but, at su ciently high pressures (greater than 12 GPa),
only non-hydrostatic solid media are available, and in this case shear stress is
unavoidable. The question is then: how important is shear stress?
Let us take the compressibility of C60, discussed in detail in section 4.2.1, as an
example. Many methods have been devised for such studies and the most common
method today is to measure directly the lattice constant using X-ray or neutron
di€ raction methods which give information on the lattice structure in the same
experiment. Di€ raction methods can be used with or without a pressure-transmitting
medium, ¯ uid or solid. In very high-pressure studies the sample is usually
compressed without the use of an additional medium, even if a `soft’ quasihydro-
static material such as NaCl can be added in the cell to work as a combined pressure-
transmitting medium and in-situ pressure calibrant, but at lower pressures a ¯ uid
pressure medium is often used. The compression properties can also be obtained by
measuring the volume of a sample mechanically at known pressures. In the piston-
and-cylinder geometry the piston displacement is measured during compression and
the maximum pressure is limited to 3± 4 GPa. Usually no pressure-transmitting
medium is used since the presence of such media makes it necessary to carry out
additional corrections for their often not very well de® ned compression properties.
The compressibility of C60 has been investigated experimentally by a large number of
groups using all the above methods and a variety of pressure-transmitting media,
ranging from noble gases to relying simply on the very small shear strength of C60
itself [48] to produce quasihydrostatic conditions in the experiment without a
pressure-transmitting medium. As will be shown below, measurements on the sc
phase below 1 GPa give almost identical results independent of whether noble gases,
Freons, NaCl or the C60 sample itself are used as pressure-transmitting medium,
showing that in this pressure range this property does not depend on whether the
pressure is hydrostatic or not. However, surprisingly, at the lowest pressures, in the
fcc phase, the results from the same experiments di€ er signi® cantly, such that the
resulting bulk moduli di€ er by a factor of up to two. Turning to the higher-pressure
range, it was observed already in the ® rst very-high-pressure compression experi-
ments on C60 by Duclos et al. [49] that the compression properties were strongly
dependent on the pressure medium used, and it has been veri® ed in many
experiments that the crystal structures observed can be very di€ erent under
hydrostatic and non-hydrostatic conditions since high non-hydrostatic pressures
seem to promote the formation of the polymeric high-pressure phases discussed in
section 5. The evaluation of data measured under non-hydrostatic conditions thus
needs some care, since shear stress sometimes has a large e€ ect on the measured
properties and sometimes little or no e€ ect at all. The most important e€ ect of non-
hydrostatic pressure seems to be to promote the formation of the high-pressure
polymeric phases of C60 , and the di€ erences observed between data from experi-
ments under di€ erent conditions at very high pressures can usually be ascribed to
this e€ ect. The di€ erences observed between results obtained under di€ erent
Fullerenes under high pressures 9

conditions in the fcc phase have not yet been explained but possible causes are
discussed below in section 4.2.1.
The upper pressure limits for hydrostatic conditions depend on temperature. At
room temperature, most light hydrocarbons, alcohols and oils give hydrostatic
conditions only below 0.5± 2 GPa. Mixtures of light hydrocarbons or alcohols do
not crystallize easily and can be used up to higher pressures. Common examples are
the 50± 50 mixture of isopentane and n-pentane which can be used up to 5± 7 GPa,
depending on how we de® ne `hydrostatic’, and the 4 : 1 mixture of methanol and
ethanol which is hydrostatic up to greater than 10 GPa [50]. Although even He
solidi® es already near 11.5 GPa [51] the use of liquid or solid noble gases allows
pressures to remain approximately hydrostatic up to about 60 GPa [52]. However, as
will be shown in section 7.4 such media cannot be used in the case of fullerenes.
A number of good books on experimental high-pressure techniques have been
published recently [53]. To give a brief overview, at low pressures (less than 3 GPa),
either piston-and-cylinder or directly pumped cylinder devices are usually used.
Depending on the design and material such devices can be used for almost any type
of scienti® c study, from optical spectroscopy (using quartz, sapphire or other
windows) and electrical transport studies to mechanical compressibility measure-
ments and neutron scattering studies, over a temperature range from below 4 K to
2000K. If a higher pressure is desired, large-volume belt-type devices [53], often
made from tungsten carbide, can be used over similar ranges in temperature. These
can also be used for neutron and X-ray scattering studies and for resistivity
measurements. The upper pressure limit is usually 8± 10 GPa, but maximum
pressures approaching 20 GPa have been achieved [54]. Finally, for the highest
pressures, anvil devices are used. Most present-day studies are carried out in
diamond anvil cells (DACs) [52, 53] consisting of two natural diamonds, about 1
carat in size, between the surfaces of which the sample is squeezed. These devices
have a very convenient size and can usually be compressed by hand using a screw.
The sample is usually placed in a small hole (up to 0.5 mm in diameter) in a simple
metal gasket between the diamonds, and this hole also gives the possibility of using a
¯ uid pressure-transmitting medium. Because DACs in many cases are capable of
generating pressures well in excess of 100GPa (1Mbar), techniques have been
developed to use liquid noble gases, such as liquid He, as pressure media even at
RT and above. If a larger sample but a lower maximum pressure is desired, anvil
devices [53]based on sintered diamond, sapphire, tungsten carbide or even steel [55]
can be used. Although resistance or neutron scattering measurements are sometimes
carried out in anvil devices, most anvil-type devices are constructed from sapphire or
diamond and mainly designed for optical and/or X-ray di€ raction studies.

4. Structure and properties of molecular (monomeric) C60


4.1. Structural phase diagram
As described in section 2 above, solid C60 forms two crystal phases at zero
pressure, a fcc phase with quasifree molecular rotation above To = 260K and a
nominally orientationally ordered sc phase with a strongly temperature-dependent
orientational disorder below. Near Tg < 90 K, molecular reorientation becomes
su ciently slow that the orientational disorder appears `frozen’ in most experiments,
and the resulting low-temperature structure is usually described as an orientational
glass. The application of pressure changes the intermolecular distances and thus the
10 B. Sundqvist

intermolecular interactions, leading to large changes in the transition temperatures


and the orientational order. This section will discuss the e€ ects of pressure on the
translational and orientational structure, starting in the fcc phase and working down
towards low temperatures, and map the low-pressure part of the presssure±
temperature phase diagram. At high pressures, polymeric phases are formed but,
in this section, only monomeric C60 , that is materials built up from free molecules
without covalent intermolecular bonds, will be discussed. A small number of reviews
of the structural evolution with pressure in this range have been published, dealing
mainly with the fcc-to-sc transition [13] and the orientational structure in the sc
phase [14, 15].

4.1.1. High-temperature behaviour and the fcc phase


A simple phase diagram, drawn on logarithmic scales to bring out clearly the
low-pressure behaviour, has been given by Poirier et al. [56]and is shown in ® gure 3.
The diagram shows that solid fcc C60 sublimes on heating without forming a liquid
phase at normal pressures. (The possible existence of a liquid phase under pressure
will be discussed in section 4.1.6.) The vapour pressure of C60 has been measured by
several groups [56± 60] and may be extrapolated as in the ® gure to ® nd an e€ ective
upper phase boundary for the fcc phase. Vorob’ev and Eletskii [61, 62]have shown
that this curve can be redrawn in reduced coordinates to coincide with that for noble
gases. A molecular dynamics simulation of the sublimation of C60 at 2700K and

Figure 3. Phase diagram of C60 drawn on a logarithmic pressure scale to show the low-
pressure range: (Ð ), data from experiment; (---), theoretical curves or extrapolations
from experiment. The liquid phase indicated has not yet been observed (see text).
(Reprinted with permission from Poirier et al. [56].)
Fullerenes under high pressures 11

50MPa has also been published [63], but it is uncertain to what extent this is realistic
since the C60 molecules have been reported by Sundar et al. [64]to start to break up
into amorphous carbon near 700ë C (973K). However, Kolodney et al. [65]reported
that free molecules are thermally stable to much higher temperatures. It may be
speculated that material produced by the standard solvent extraction method has a
lower amorphization temperature than sublimed C60 because of the unavoidable
presence of remaining solvent molecules in the lattice.
Little is known about the detailed e€ ects of pressure on the dynamic (rotational)
properties in the fcc phase, but it is reasonable to assume that the increase in
intermolecular interaction with pressure will slow down the molecular rotation and
increase the rotational anisotropy observed even at zero pressure [21, 28± 31]. At high
pressures the rotational motion will probably approach the ratcheting motion
observed in the sc phase, where the molecules instead of rotating quasifreely will
jump between a number of discrete orientational states. The decrease in the
intermolecular distance with increasing pressure and the continuation of molecular
rotation increase the probability that double bonds on neighbouring molecules can
interact to form intermolecular covalent bonds, and as a result the application of
pressures greater than 1 GPa at high temperatures usually leads to the formation of
polymeric phases, to be discussed in some detail in section 5. At lower temperatures,
the application of pressure and/or cooling at any pressure ® rst gives a transition into
the well known sc phase.

4.1.2. Fcc-to-sc transition boundary


The orientational ordering or fcc-to-sc transition was initially suspected [32]to be
a continuous transition but thermal and structural studies soon showed it to be ® rst
order with a small hysteresis and connected with rather large volume changes
D V < 1% [29, 33, 66] as well as entropy changes D S [67]. The transition can thus
be detected fairly easily by a variety of methods, although both the transition
temperature and the size and shape of the transition anomalies are very sensitive to
deformation and impurities. Since the ® rst measurements of the compressibility of
C60 [49, 68] were carried out almost simultaneously with the discovery of the
orientationally ordered sc phase [32], it was immediately realized that the application
of pressure to this soft material should lead to signi® cant changes in the inter-
molecular interactions, which in turn might give signi® cant e€ ects on the transition
temperature To. To investigate this idea, Samara et al. [69], and almost simul-
taneously Kriza et al. [70], used di€ erential thermal analysis (DTA) to measure the
fcc-to-sc transition temperature as a function of pressure. The two groups used
virtually identical techniques and found similar results for the slope of the phase line,
dTo /dp = 104KGPa- 1 [69] and 117KGPa- 1 [70], the very large values re¯ ecting
the large e€ ects of pressure on the intermolecular potential. However, later
investigations have shown that these values are not, in fact, intrinsic to pure C60
for which the true slope of the phase line is even larger. Both groups applied pressure
using helium gas which rapidly di€ uses into the interstitial cavities of solid C60 ,
changing the physical properties of the material and also the phase diagram. Such
e€ ects are discussed in some detail in section 7.4.
The slope of the fcc-to-sc phase line has later been reinvestigated by a large
number of techniques, including DTA, nuclear magnetic resonance (NMR), electron
paramagnetic resonance (EPR), compression measurements and resistivity meas-
urements, using several di€ erent pressure-transmitting media. The results from a
12 B. Sundqvist

Table 1. Experimental data for the slope dTo /dp of the fcc-to-sc phase boundary. If only one
pressure value is given, the experiment was carried out at a single temperature or press-
ure and a literature value for the zero-pressure value taken as a second reference point.
The error limits given are estimates by the original authors

Pressure
range Pressure dTc /dp
(GPa) medium (K GPa- 1) Technique used Reference

0± 0.8 He 104 6 2 DTAa Samara et al. [69]


0± 0.5 He 117 DTA Kriza et al. [70]
0± 0.8 He 109 6 4 DTAa Samara et al. [71]
0.6 He 111 Compressibility Schirber et al. [72]
0± 0.5 He 119.5 Thermal expansion Grube [73]
0.6 Ne 118 Compressibility Schirber et al. [72]
0.6 Ar 141 Compressibility Schirber et al. [72]
0± 0.3 Ar 174.2 Thermal expansion Grube [73]
0± 0.5 Ar 165 Compressibility Pintschovius et al. [74]
0± 0.8 N2 164 6 2 DTAa Samara et al. [71, 75]
0± 1.4 Pentane 159 6 3 DTAa Samara et al. [71, 75]
0± 0.5 Isopentane 160 NMR Kerkoud and co-workers [76, 77]
0± 0.5 Oil 132 6 3 Resistance Matsuura et al. [78]
0± 0.5 Ð b 163 Compressibility Lundin and Sundqvist [79, 80]
0.4 Alcohol 90 Raman Meletov et al. [81± 83]
0± 0.2 Te¯ on 120 Thermal conductivity Andersson et al. [84]
0± 1.26 Talc± epoxy 99 Resistancea Ramasesha and Singh [85]
0.5 (Solid) 6
65 5 Positron annihilation Jean et al. [86]
0.4 ? 100 6 10 EPR Kempinski et al. [87]
a
Double transition observed (see text).
b
Pressure applied directly to the C60 sample itself.

number of such studies are collected in table 1, subdivided into three groups. The
® rst group contains studies carried out in the intercalating noble gases neon and
helium, the second the results of studies using non-intercalating hydrostatic media
and methods that give well de® ned transition pressures, and the third group contains
studies carried out in solid media or over small ranges in T (such as RT only). While
the results given in the third group show a very large scatter, as might be expected,
those given in the ® rst two groups are relatively well de® ned but di€ er between the
groups. This is well illustrated in ® gure 4, reprinted from Samara et al. [71] and
showing the di€ erence between phase lines obtained in helium and N2 gas
respectively. Inspection of table 1 shows that ® ve of the values in the second group
are in almost perfect agreement and thus that the `best’ value for the phase line slope
of pure C60 can be taken as 162 6 2K GPa- 1 , while in helium and neon a
signi® cantly smaller average slope dTo /dp = 113KGPa- 1 is found. The accepted
explanation for the lower slope under these conditions [71, 75]is that the presence of
intercalated gas atoms in the lattice interstitials increases the bulk modulus of the
lattice. Since the transition is observed to occur at the same intermolecular distance
in all noble gases [72], this means that at any given temperature the transition occurs
at a higher pressure in the intercalated materials.
The slope dTo /dP can be calculated from simple thermodynamic considerations.
Samara et al. [71] used the known volume change D V at the transition [33, 66]
Fullerenes under high pressures 13

Figure 4. Pressure-temperature phase diagram of C60, showing the fcc-to-sc phase boundary
as measured in helium and in N2. (Reprinted with permission from Samara et al.
[71]. ) (1GPa= 10kbar. )

together with the Clausius± Clapeyron relation dTo /dp = D V /D S to calculate a


value for D S at the transition which turned out to be in very good agreement with
experimental data [67]. The volume change at the transition is close to 1% at zero
pressure [33, 66]but there are reports that indicate that D V decreases with increasing
pressure. Lundin and Sundqvist [80] found that, for pure C60 , D V decreased to
about 0.7% near 0.5 GPa, while Grube [73] found a similar rate of decrease in an
experiment in helium gas to 0.4 GPa. A similar rapid decrease in D V under pressure
was also observed at the fcc-to-rh transition in C70 [88], where D V was even found to
approach zero near 1.5GPa. Since the slope is independent of pressure, this indicates
that D S must also decrease, and Lundin and Sundqvist [80] speculated that in C60
the transition might change character under pressure from ® rst order to continuous
in analogy with the orientational transition in NH4 Cl [89]. However, very recent
accurate neutron studies of C60 by Pintschovius et al. [74]showed no such decrease in
D V with increasing pressure.
Surprisingly, some studies of V against p in C60 fail to observe any transition
anomaly at all [90± 92]. For the C60 ± C70 mixture studied by Lundin et al. [90]this is
not unexpected, since in such mixtures the transition is often quite smeared in T (or
p), but both the `mechanical’ study of Bao et al. [91] and the neutron work by
Blaschko et al. [92] were carried out on pure C60 which should have a clearly
detectable volume anomaly.
The slope of the phase line has also been calculated theoretically using other
methods. Several groups have calculated this slope as a test of di€ erent inter-
molecular potentials, basically van der Waals-type potentials with various added
electrostatic interactions (to be brie¯ y discussed in section 4.2.1). Although Burgos
et al. [93] calculate a very low dTo /dp = 40 KGPa- 1 , better agreement with
experiments is found in other calculations with slightly di€ erent potentials; Lu et
al. [94]reported a value of 115KGPa- 1 and Cheng et al. [37]120KGPa- 1 . Lamoen
14 B. Sundqvist

and Michel [95] instead calculated the slope in a model which assumed that the
transition originates in a coupling between molecular orientations and acoustic
lattice displacements and found dTo /dp = 182KGPa- 1 , which like the two former
values is in reasonable agreement with experiments.
Finally, it should be pointed out that many researchers have observed anomalies
more complicated than would be expected for a simple ® rst-order transformation.
Already in their ® rst study Samara et al. [69]found that the DTA peaks developed
from being slightly anisotropic at zero pressure to having a de® nite shoulder,
perhaps indicating the presence of more than one peak, at the highest pressures
used. Although they could not rule out a direct impurity e€ ect, Samara et al.
suggested that the double-peak structure may result from partial inhibition of the
molecular rotation by interstitial impurities, such that the molecules can have two
rotational states, hindered and unhindered rotation, about 5 K apart in energy.
Rasolt [96]instead discussed the precursor e€ ects observed by Samara et al. in terms
of ¯ uctuations in short-range orientational order and deduced that the transition
was a ¯ uctuation-driven transition. More clearly de® ned double transitions were
observed in the resistance studies by Ramasesha and Singh [85], who found that the
di€ erence in transition temperature depended on pressure and approached zero at
zero pressure, and very recently a double transition with a splitting of 0.1± 0.3 K has
also been observed in highly pure C60 at zero pressure in a thermal study using
modulated di€ erential scanning calorimetry [97]. It is thus possible that an
intermediate structural state exists in a very narrow range close to To. (The anomaly
could in principle be connected with surface `melting’, but the surface layers have
been observed to be rotationally disordered already at temperatures signi® cantly
below To [98].)

4.1.3. The orientational structure in the sc phase


As discussed in section 2 there are two possible orientational states, the P
(pentagon) and H (hexagon) orientations, in the sc phase at zero pressure. Far
below 260K the relative fractions of P- and H-oriented molecules are given by the
usual thermal distribution
1
f (T) =
1 + exp (- ¢ /kT )
, (4.1)

where ¢ is the energy di€ erence between the two states. (Close to To ¯ uctuations or
other e€ ects tend to bring the actual measured f closer to 50%.) The P-oriented state
has a slightly lower energy and is the preferred orientation at zero pressure, but the
H state has a slightly smaller molecular volume and should thus be energetically
preferable at high pressures. Because ¢ is very small, pressure has a large e€ ect on
the details of the intermolecular interaction in C60 . Using the observed di€ erence in
molecular volume and simple thermodynamic arguments, David and Ibberson [99]
showed that the relative energies of the two states should cross already at quite low
pressures. They also veri® ed this experimentally by neutron di€ raction studies under
pressure between 150 and 200K. At 150K, their data showed that the ratio [P]/[H]of
P- to H-oriented molecules decreased from about 70/30 at zero pressure to 50/50 at
an `equilibrium pressure’ peq = 191MPa (where obviously ¢ = 0). The equilibrium
pressure is probably weakly temperature dependent, since the orientational potential
should primarily depend on the intermolecular distance and thus the volume rather
than the pressure. If we assume that the equilibrium occurs at a certain (temperature-
Fullerenes under high pressures 15

Figure 5. Low-temperature phase diagram of C60 as suggested by Sundqvist et al. [103].

independent) molecular volume, we can use data for the thermal expansion [29, 35]
and the compressibility (section 4.2.1) to ® nd that peq should decrease to near
165MPa at 100K and to increase to about 217MPa and 242MPa at 200K and
250K respectively. This is, of course, a ® rst approximation only, but later high-
pressure studies of the thermal expansion [73], the compressibility [74] and the
thermal conductivity [100], as well as high-pressure Raman studies [101]all indicate
that the equilibrium pressure near 90± 100K indeed is closer to 150MPa than to
200MPa.
Above peq , the stable orientational state should be the H orientation. At zero
pressure the energy di€ erence between the two states is too small for any completely
P-oriented state to be observed at any temperature but, since peq is quite low, it
should be possible to apply pressures high enough for ¢ to be much greater than kT,
and thus an almost completely H-oriented structure. This was ® rst suggested by
Sundqvist et al. [102, 103]to explain anomalies observed in their experimental data
for the compression properties [79]and the thermal conductivity [104]of C60 under
pressure. By making a simple linear extrapolation of the fraction f of P-oriented
molecules against pressure at 150K they deduced an approximate `phase line’ for the
formation of such an H-oriented phase and obtained the low-temperature phase
diagram shown in ® gure 5, as well as an approximate value for f ( T ) at any pressure
[14]. As discussed in detail elsewhere [14], the phase diagram obtained agreed with
their observations of anomalies in the measured bulk modulus under high pressures
(® gure 19 below) and the thermal conductivity. Soon after, Wolk et al. [101] made
Raman scattering studies as a function of pressure and temperature and observed the
appearance of new lines due to libron modes. From these they also deduced the
existence of a preferably H-oriented phase above peq < 150MPa, as shown in ® gure
6. Blaschko et al. [92]later carried out high-pressure neutron scattering studies which
showed much better agreement with an assumed predominantly H-oriented phase
[P]/[H]= 70/30 than with a predominantly P-oriented phase at 1.6 GPa and RT.
Although the existence of a predominantly H-oriented phase has thus been
veri® ed, the form of equation (4.1) in principle forbids a completely P- or H-oriented
16 B. Sundqvist

Figure 6. Low-temperature phase diagram of C60 as suggested by Wolk et al. [101]


(reprinted with permission from the authors).

Figure 7. The function f ( T ) from equation (4.1), showing the quasilinear low-T behaviour
(---) and the observed constant f below Tg for C60 (- ´ -). ( Reprinted from Lundin and
Sundqvist [80].)

phase to exist at any temperature above 0K, and no well de® ned `phase boundary’
should exist for an H-oriented phase. It is therefore surprising that anomalies in bulk
physical properties have indeed been observed [79, 80]near pressures corresponding
to extrapolated `phase transition’ pressures. Figure 7 shows the theoretical fraction
of P-oriented molecules as a universal function of kT /¢ [80]. Because of the form of
the equation there is always a quasilinear increase or decrease in f ( T ) over [P]/[H]
ratios from about 80/20 to about 98/2, after which there is a ® nal more gradual
exponential decrease in the minority orientation. Any property sensitive to the
orientational structure should thus experience a fairly sharp change in its tempera-
ture dependence near kT < 0.25¢. From equation (4.1) together with data for ¢
Fullerenes under high pressures 17

Figure 8. The function f ( T , p) calculated from equation (4.1) and data for D (p) given by
David and Ibberson [99] at pressures from 0 to 1 GPa in steps of 100MPa.

against p given by David and Ibberson [99] it is possible to calculate approximate


values for the fraction f of P-oriented molecules at any pressure or temperature. The
result is shown in ® gure 8, which veri® es that an almost perfectly (H) orientationally
ordered phase must exist above some critical pressure. (The corresponding ® gure in
[14]was calculated in a simpli® ed model assuming a linear dependence of f on p and
is thus less accurate.) Although in principle no perfectly ordered phase exists, in
practice any structure with, say, 1% or 0.1% misoriented molecules must be
considered very well ordered. Note also that this phase is the only truly orientation-
ally ordered phase existing in the phase diagram of C60 .
The pressure dependence of f at constant temperature is shown in ® gure 9, which
has been calculated for pure C60 at a temperature of 150K, that is with 70% P-
oriented molecules at zero pressure. Near 5± 10% there is a fairly sharp bend in the
curve, but again it is di cult to see how this rather smooth change in slope can give
rise to the sharp anomalies in B(p) reported by Lundin et al. [79, 80] (see above).
However, as discussed by Burgos et al. [93] the crystal potential for molecular
reorientation is more complicated than usually assumed. These workers calculated
the intermolecular energy not only for the rotation of one molecule surrounded by
neighbours all in the P orientation, but also for several other cases. As an example,
® gure 10 shows the intermolecular energy for one molecule as a function of the
rotation angle about the [111] crystal axis for these cases. The dotted curve is the
standard result assuming all 12 nearest neighbours are in the P-oriented state, while
the full curve assumes coherent collective rotation of all molecules and the chain
curve assumes all neighbours in the H orientation. In the last curve there is only one
minimum; that is, if all (or possibly even a majority?) of the neighbours are H
oriented, the molecule studied will also end up in the same state. Burgos et al.
concluded that sc C60 will probably contain a large number of well oriented (both P
and H) microdomains rather than consist of a mixture of randomly oriented
molecules. Since the energy of the H-oriented state also decreases with increasing
pressure, it is not unreasonable that a ® nal `lock-in’ to a completely H-oriented phase
might also occur when the fraction of H-oriented molecules exceeds some critical
value which we may guess is not very far from 11 /12 < 0.92 (i.e. on average all
18 B. Sundqvist

Figure 9. The function f calculated from equation (4.1) as a function of pressure p at a


nominal temperature of 150K.

Figure 10. Intermolecular energy as a function of rotation angle about the [111]crystal axis:
(Ð ) curve obtained assuming all molecules rotate coherently; ( ´´´´´´) curve assuming
neighbouring molecules are all in the 98ë (P) orientation; (- ´ -) curve assuming
neighbours are in the 38ë (H) orientation. (Reprinted with permission from Burgos et
al. [93].)

nearest neighbours except one are in the H orientation). One interesting and testable
consequence of this model is that, once formed, the H-oriented state should be very
stable, and a `completely’ H-oriented phase might remain stable down to peq or the
fcc transition pressure, whichever is higher, and at low T possibly even at zero
pressure. Very recent calculations by L. Pintschovius (1997, private communication),
assuming ¢ to be a linear function of f , also showed that the combination of such an
e€ ect and the di€ erent compressibilities of the two states (section 4.2.1) will result in
an abrupt ® nal transition. As hinted at above, the P± H reorientation under pressure
has been observed to have signi® cant e€ ects on thermophysical properties such as
Fullerenes under high pressures 19

the bulk modulus and the thermal conductivity. This will be discussed further in the
relevant parts of section 4.2 below.
Before closing this section we note that a thermodynamic model which describes
the relative fractions of frozen-in, rotating and ratcheting molecules in the lattice has
been developed by Saito et al. [105]but, since this model has not yet been applied to
the material under high pressures, it will not be discussed further here. Also, the
activation enthalpy D Ha for molecular rotation at constant pressure and other
thermodynamic quantities have been measured by NMR in the sc phase under
pressure and will be discussed further in section 4.2.6. Finally, it should be noted that
anomalies are observed in many physical properties, mechanical [106, 107],
structural [108], acoustic [109], dielectric [110] and others [3], in the sc phase near
150K at zero pressure. The cause of these anomalies is still not undisputed, but in
some cases it can be shown that they are connected with the glass transition which at
high frequencies falls in this temperature range. The only study showing anomalies in
this range under pressure is the thermal conductivity study by Andersson et al. [84].
The anomaly shifts to higher temperatures with increasing pressure, indicating either
a correlation with the sti€ ening of the lattice under pressure or with the increase in
glass transition temperature Tg because of the decrease in molecular reorientation
frequency.

4.1.4. The glassy crystal transition and the properties of the glassy phase
The glass transition near Tg = 90K has also been studied in a few experiments
under high pressures. In the glassy crystal phase (or orientational glass) below Tg ,
molecular motion has become slow enough that no reorientational motion can
usually be detected during the course of a normal measurement. This means, for
example, that, once a sample has been cooled through the transition, the fraction of
P-oriented molecules can be considered constant at the limiting value obtained on
approaching Tg from above.
The pressure dependence of Tg has been found from thermal conductivity studies
by Andersson et al. [84, 100, 104]. ¸ shows de® nite anomalies at Tg because of its
sensitivity to orientational disorder. To a ® rst approximation, the thermal resistivity
Wdis due to orientational disorder should be given by Wdis ~ f (1 - f ) and thus have
a maximum at f = 0.5 in analogy with the electrical resistivity of concentrated alloys
[111], and on cooling towards Tg there must always be a decrease in W dis . When Wdis
suddenly becomes constant at Tg, there will be a sharp anomaly in d¸ /dT. From
data such as those shown in ® gure 11 the slope dTg /dp of the glass transition line has
been deduced in the range up to 0.7GPa, with the result dTg /dp = +62 KGPa- 1 [84,
100]. (Similar slopes have later been reported by Grube [73]and by Pintschovius et
al. [74]. ) Above 0.7 GPa the glass transition anomalies become small because of the
very low fraction of P-oriented molecules (see above). In the range 0.15± 0.2 GPa the
anomalies are also very small, since the [P]/[H] ratio is close to 50/50 at all
temperatures. Again, the molecular reorientation in sc C60 has interesting e€ ects
on the thermodynamic properties. For most glasses, the slope dTg /dp can be
determined from the formula [112]
dTg D a
dp
= Tg V g
D cp
, (4.2)

where D a and D cp are the changes in the thermal expansivity and the speci® c heat
capacity respectively at the transition. For C60, direct use of this formula gives a
20 B. Sundqvist

Figure 11. Thermal conductivity of C60 against temperature at the pressures (in gigapascals)
indicated, showing typical glass transition anomalies (arrows). ( Reprinted from
Andersson et al. [100].)

large negative value dTg /dp < - 130KGPa- 1 [104] because the P-to-H molecular
reorientation above Tg leads to a negative D a at the transition (see ® gure 2 above).
Correcting for this e€ ect a positive slope in better agreement with theory can be
found [104].
If the pressure is changed while the sample is kept at a temperature below Tg , no
change in the orientational structure can occur. On heating such a sample towards
Tg, interesting relaxation anomalies are observed [100], as shown in ® gure 12. A
comparison with ® gure 11 shows that the character of the anomalies changes, and
both minima and maxima occur in ¸ when the molecules relax towards the normal
orientational structure at the new pressure. If the sample is cooled at low pressures, a
majority of the molecules will be in the P orientation. If the pressure is then increased
to a pressure where H orientation should normally dominate, the fraction of P-
oriented molecules will decrease from its initial high value, through 50% where ¸ will
have a minimum because of the form of Wdis, towards a low value. This will be
observed as an initial decrease in ¸, followed by a ® nal increase. Close to Tg , such a
molecular relaxation can also be observed by measuring ¸ as a function of time [100].
Since the H orientation is connected with a smaller molecular volume, the
reorientation of each extra P oriented molecule to the H state increases the volume
available for other molecules to reorient, and as a result the e€ ective glass transition
temperature becomes lower if the low-temperature phase has a surplus of P-oriented
molecules, and vice versa. Wolk et al. [101] tried to determine the glass transition
slope by cooling at zero pressure, then applying pressure at low temperature and
measuring the glass transition temperature under pressure during heating. As
pointed out by Sundqvist et al. [113], the resulting slope, dTg /dp = 35 KGPa- 1 ,
shown in ® gure 6 above, is signi® cantly lower than the value of 62 KGPa- 1 obtained
by Andersson et al. [84, 100] by measurements at constant pressure (® gure 5) but
agrees well with the slope obtained from the position of the minima in ¸ observed at
Fullerenes under high pressures 21

Figure 12. Thermal conductivity of C60 against temperature at the pressures (in gigapascals)
indicated, showing modi® ed glass transition anomalies. In all cases, the samples had
been cooled near 0.1GPa and pressurized to the ® nal pressure below 90K before the
measurements were carried out during heating. (Reprinted from Andersson et al.
[100].)

[P]/[H]= 50/50 in the experiments with a frozen-in orientational distribution [100].


An extrapolation of the `50/50 line’ to lower pressures shows that it crosses the
`normal’ glass transition line at 0.15 6 0.05GPa, indicating that this is the equi-
librium pressure peq near 100K.

4.1.5. L ow-pressure phase diagram of C6 0


The information given in the preceding sections can be collected and condensed
in the form of the pressure± temperature phase diagram for C60 shown in ® gure 13.
The various phases have already been discussed above. The slope of the fcc-to-sc
phase line is taken as 162KGPa- 1 as discussed in section 4.1.2 and that of the glass
transition line is taken from Andersson et al. [84, 100]. To describe the orientational
states in the sc phase, two `boundaries’ are given. The almost vertical broken line
near 0.2 GPa is a weakly temperature-dependent boundary line corresponding to
peq (¢ = 0) , calculated for a constant transition volume as discussed in section 4.1.3
above. For p < peq the P-oriented state is more stable while above this the H
orientation dominates. The shaded area at higher pressures corresponds to the
region of the transition into the H-oriented state discussed above. The transition
region boundaries shown have been calculated in the following way. The lower-
pressure side corresponds to a fractional P state occupancy f = 15%, calculated
using a constant peq to obtain a lower limit. The high-pressure side corresponds to
f = 5% but calculated using the temperature-dependent peq to ensure that we ® nd an
upper limit. These limits on f were chosen because inspection of the theoretical curve
for f (p) in ® gure 9 indicates that the macroscopic `transition’ should occur near
f = 10%, very close to the value f = 0. 92 for which I speculated (from the
theoretical potential given by Burgos et al. [93] and shown in ® gure 10) that a
`lock-in’ transition to an oriented phase might occur. The observed anomalies in the
22 B. Sundqvist

Figure 13. Phase diagram of monomeric C60 as discussed in the text. The full line [71]
denotes the fcc-to-sc transformation and the dotted line the glass transition [100]. The
almost vertical broken line delineates the equilibrium line between regions with P
(low p) and H (high p) orientations and the shaded area is the region where a
transition into an H-oriented phase should occur (see text). Symbols denote
anomalies observed in various properties by Lundin and Sundqvist [80] ( j ), Jeon et
al. [117](n ), Huang et al. [118](, ), Meletov et al. [115]( r ), Meletov et al. [82](h ),
Bao et al. [91] ( d ), Blank et al. [114] ( s ) and Jephcoat et al. [116] ( m ).

macroscopic properties should thus fall between these limits if the assumptions made
above are correct. Above the high-pressure side of the shaded area less than 5% of
the molecules should be P oriented even in a pessimistic calculation and, if an
orientational `lock-in’ transition occurs as suggested by the potential of Burgos et al.,
the crystal should to a very good approximation be `completely’ H oriented. In both
cases the dominant type of disorder is probably no longer orientational but probably
stacking faults, or vacancies and interstitials. If domains of well ordered material
form, as suggested by Burgos et al., spectroscopic anomalies may occur even at
pressures below this region.
The fcc-to-sc phase line is well known and has been well studied, and the same is
true to a smaller extent for the low-pressure part of the glass transition line.
However, although the existence of the P ® H orientational ordering transition is
proved beyond doubt [92, 101], the actual transition `line’ or area is not well
investigated. The symbols in ® gure 13 denote compression anomalies observed by
Lundin and Sundqvist [80] at low temperatures as well as spectroscopic and
structural anomalies observed at room temperature. Several groups have reported
[81± 83, 91, 114, 115]the existence of anomalies in spectroscopic and other properties
between 2 and 2.5 GPa at room temperature, and Jephcoat et al. [116]observed the
formation of a structurally ordered phase near 2.5 GPa. Surprisingly, this high-
pressure phase is reported to be completely P oriented, although Jephcoat et al.
stated that data at 1.3 GPa are in better agreement with an H-oriented state.
Fullerenes under high pressures 23

Whether the unexpected orientation reported is correct, a mistake or owing to the


use of He as pressure medium (section 7.4) is not clear. Anomalies have also been
reported in the Raman [117]and infrared (IR) [118]spectra of C60 in the range above
1 GPa. It has sometimes been assumed [81± 83] that the anomalies observed near
2 GPa correspond to the glass transition, but the slope of the glass transition line is
only 62KGPa- 1 at low pressures and a linear extrapolation thus gives a glass
transition pressure of 3.4 GPa at 293K. Since the normal curvature of glass
transition lines is in such a direction as to increase this pressure value, it is more
likely that the anomalies observed are connected with the formation of an (H)
orientationally ordered phase. However, this still remains to be proved by low-
temperature high-pressure experiments.

4.1.6. Does liquid C6 0 exist?


Most neutral materials exist in three states: gas, liquid and solid. As already
discussed in section 4.1.1 above, C60 sublimes near 700K without forming a liquid
phase. As shown in ® gure 14 [119], the standard graphite± diamond phase diagram of
C contains a liquid phase at very high temperatures above 5000K, but no such phase
has yet been observed for C60 . It might be argued that it should be possible to
observe such a state if sublimation was suppressed by the application of a high
pressure, and a number of theoretical studies have been carried out to try to predict
the properties [120]and range of stability of such a phase. Interestingly, there is still
no consensus even on the question of whether a liquid phase can exist at all. Most

Figure 14. Phase diagram of carbon. Solid lines are equilibrium phase boundaries and the
broken curve BFG the threshold for rapid graphite± diamond transformation.
Commercial diamond synthesis is carried out at A, C denotes rapid diamond-to-
graphite conversion, DE is the area where graphite is converted into hexagonal
diamond and, along the dotted line, HIJ graphite reversibly transforms into
diamond-like structures. ( Reprinted with permission from Bundy et al. [119].)
24 B. Sundqvist

investigations have used the same intermolecular potential [121], but even so the
results di€ er between di€ erent groups. Hagen et al. [122]found that no liquid phase
can exist, since in their calculation the liquid± vapour coexistence line always fell at
lower temperatures than the solid± ¯ uid coexistence curve, while almost simul-
taneously Cheng et al. [123] found a narrow range above 1800K where a liquid
phase could possibly exist under a low pressure of a few megapascals (tens of bars).
Molecular dynamics simulations by Abramo and Caccamo [124] also show an
anomaly near 2250K at 2.2 MPa which might be associated with melting. Although
theoretically the C60 molecule should be stable to about 4000K [125], experiments
have shown it to break down at very much lower temperatures (section 4.1.1) and it
is not clear whether liquid C60 at 1800± 2300K would be stable on a molecular scale
long enough to be observed even in a rapid transient heating experiment. Later
theoretical studies [126± 129]have been even less conclusive and the question is still
open. Rasco  n et al. [130]suggested that the reason for the di€ erent results obtained
by di€ erent groups might be the form of the intermolecular potential used [121],
which they showed is close to the critical potential for which no liquid phase can
occur. (Although Shchelkacheva [128]used a di€ erent potential due to Yakub [131]
and found results very similar to those obtained in previous studies, it should be
noted that the Yakub potential is only an approximation to that of Girifalco [121].)
Finally, Vorob’ev and Eletskii [61, 62] argued from thermodynamic scaling argu-
ments that the similarity between C60 and the noble gases shows that a liquid phase
should indeed exist. The question of whether a liquid phase does exist or not has thus
not yet been answered but, since the physical properties of such a phase should be
quite unusual [120], the search will probably continue until either such a phase is
found or its existence has been conclusively disproved.

4.2. Physical properties of C6 0 under high pressures


4.2.1. Compressibility and other elastic properties
Having discussed how changes in temperature and pressure modify the structure
of C60, I now turn to the interesting question of how the physical properties of C60
are modi® ed by high pressures. As long as the molecules remain stable it is obvious
from the enormous strength of the intramolecular bonds that the molecular proper-
ties will change little under an applied pressure, but it is equally obvious that the
weakness of the intermolecular bonds means that the lattice properties will be quite
sensitive to the applied pressure. The pressure in itself is not, of course, important;
what is important is the change in volume V (and thus intermolecular distance)
brought about by the applied pressure. In this section the important relation between
volume and pressure will therefore be discussed in some detail. If the pressure is
hydrostatic and the material is isotropic, the compression will also be isotropic, there
will be no change in sample shape and the resulting volume decrease can be described
completely by the isothermal bulk modulus B
dp
B= -V dV
, (4.3)
T

or, equivalently, the isothermal compressibility · = B- 1. As discussed in section 3,


this ideal situation rarely occurs in real experiments. Although C60 itself is isotropic
at low pressures because of its cubic structure, the pressure in real experiments is not
often perfectly hydrostatic but contains uniaxial or shear stress components. In
Fullerenes under high pressures 25

particular, as mentioned in section 3, all pressure media solidify below 12 GPa [51±
53], and soft solid noble gases such as helium cannot be used to extend the
quasihydrostatic pressure range [52] because they intercalate into the C60 lattice
under high pressures. High-pressure studies of C60 are thus often carried out using
liquid media such as pentane, Freons or oils, or by compressing pure C60 without a
medium, relying on its small shear strength [48] to produce quasihydrostatic
conditions. The discussion below will concentrate on data for the isothermal bulk
modulus B obtained by direct measurements of the volume change under pressures,
but available data for the adiabatic bulk modulus Bs obtained from acoustic or
optical studies will also be discussed brie¯ y. The relation between B and Bs is most
clearly written in terms of the associated compressibilities · and ·s through the
thermodynamic relation [132, 133]
9TV a 2 B2
· = ·s +
cV
, (4.4)

and Bs = ·-s 1 should thus be expected to be slightly larger than B.


The compressibility or bulk modulus of a material not only is of technical and
practical interest but is also important from a basic scienti® c point of view. The
compression behaviour is determined by the intermolecular potential, and measured
data for B are therefore very important in forming a benchmark against which
theoretical models for the intermolecular interactions can be tested. A large number
of calculations have therefore been carried out for the bulk modulus of C60 and will
be brie¯ y discussed below. Many experimental methods have been devised for
studies of B, and the most common method today is to measure simultaneously
the lattice constant and the structure using X-ray or neutron di€ raction methods.
Unfortunately, this is a relatively slow method which gives only a limited number of
experimental points. The most common method is to use X-ray di€ raction and to
pressurize the sample in a DAC but the small sample size (less than 0.001mm3)
implies very long exposure times and most studies are therefore carried out using
synchrotron sources. With the small aperture available and the low intensity of the
scattered radiation, only a small number of di€ raction lines can usually be seen and
it can be di cult to identify small changes in the lattice structure. Neutron scattering
studies require larger sample volumes and are limited to the signi® cantly lower
pressures produced by piston-and-cylinder or large anvil devices. In very-high-
pressure studies the sample is usually compressed without the use of an additional
medium, even if a quasihydrostatic material such as NaCl can be added in the cell to
function as a combined pressure-transmitting medium and in-situ pressure calibrant,
but at lower pressures a ¯ uid pressure medium is often used. An alternative method
is to measure the compression properties mechanically in the piston-and-cylinder
geometry, where the piston movement can be measured very accurately but the
maximum pressure is limited to 3± 4 GPa. With large samples the piston-and-cylinder
methods give a very high resolution and accuracy provided that the device is
carefully calibrated and that accurate corrections for the compression of pistons
and gaskets as well as for the radial expansion of the cylinder are applied [134]. In
most cases, no pressure transmitting medium is used since it is then necessary to
carry out additional corrections for the usually not very well de® ned compression of
the medium. This method gives a semicontinuous record of V against pressure and
can thus be used to detect phase transitions and to map the pressure± temperature
phase diagram, but the accuracy is sometimes limited because of hysteresis e€ ects
26 B. Sundqvist

arising from friction both in the gaskets and inside the sample itself, which su€ ers
plastic ¯ ow in the experiment.
After the suggestion by Ruo€ and Ruo€ [135, 136] that compressed C60 might
have a bulk modulus signi® cantly higher than the value of 441GPa for diamond
[137], the bulk modulus of C60 has been investigated experimentally by a large
number of groups using all the above methods and a variety of pressure-transmitting
media. Surprisingly, although the methods used should all have a similar accuracy,
the results from di€ erent groups often di€ er by signi® cant amounts.
The compressibility study by Fischer et al. [68]was the ® rst high-pressure study
of C60 . Using X-ray di€ raction in a DAC under hydrostatic pressure they obtained
one single experimental point at 1.2 GPa, with a pressure calibration based on an
empirical pressure against load relation. This method is rarely used with DACs
because of the large inherent inaccuracy. The phase diagram in section 4.1.5 shows
that 1.2GPa is well into the stability range of the fcc phase and the measured volume
change up to 1.2GPa thus included both the continuous volume changes within the
two phases involved and the volume drop (about 1%) at the fcc-to-sc transition. The
average linear compressibility up to 1.2 GPa was d(ln a) /dp = - 2.3 ´ 10- 12 cm2
dyn- 1 = - 2.3 ´ 10- 2 GPa- 1 , corresponding to a volume compressibility · =
6.74 ´ 10- 2 GPa- 1 or a bulk modulus B = 14. 8 GPa. These values con® rmed the
basic assumption that the intermolecular interaction in C60 must be very similar to
the interplanar interaction in graphite, since the c-axis compressibility of graphite is
d(ln c) /dp = - 2.8 ´ 10- 2 GPa- 1 [138]. Because of the very small in-plane compress-
ibility of graphite the bulk modulus is close to the inverse of this value, at
B = 33.8 GPa [138].
Very soon afterwards Duclos et al. [49]presented the results of a more complete
investigation of the compressive properties up to 20GPa, under both hydrostatic and
non-hydrostatic conditions as deduced from the deformation of the ruby chip used
for pressure calibration. As shown in ® gure 15, the relative volume V / V o is a smooth
function of p at all pressures investigated under hydrostatic conditions (experiment
1) but not under non-hydrostatic conditions (experiments 2 and 3). The data from
experiment 1 were ® tted to both the Vinet et al. [139] and the Murnaghan [140]
equations of state with identical results, yielding an initial zero pressure bulk
modulus B(0) = 18.1 GPa with a pressure derivative B = dB /dp = 5. 7. Because of
Â
the experimental procedure used, no data were obtained below about 0.5 GPa and
contrary to statements in the original paper the measured values should thus apply
to the sc phase only. Again, no account was taken of the volume change at the fcc-to-
sc transition which makes the low-pressure value B(0) rather dubious. At very high
pressures, Duclos et al. observed repeatable changes in the X-ray di€ raction
diagrams above 20 and 16.5 GPa in experiments 2 and 3 respectively. The
compression data also indicate a larger V / V o at very high pressures in these
experiments than in the hydrostatic experiment. Since a volume increase at a
structural transition is highly unlikely for stability reasons, such a change in
molecular volume should be connected with a transition at lower pressures. If the
high-pressure compression data are extrapolated to lower pressures, the extrapola-
tions cross the data from experiment 1 between 5 and 7GPa, in good agreement with
the observations of the formation of an orthorhombic polymeric phase to be
discussed in section 5. From the slopes of the curves it is also possible to deduce
very uncertain values B > 200GPa for the average bulk modulus between 5 and
20GPa in this high-pressure phase. These values agree to within an order of
Fullerenes under high pressures 27

Figure 15. RT compression behaviour of C60. Hydrostatic conditions are believed to extend
to 20GPa in experiment 1 and to 10GPa in experiment 2. Experiment 3 was carried
out without a pressure-transmitting medium. The curve has been ® tted to the data
using the Vinet et al. [139]equation of state. (Reprinted with permission from Duclos
et al. [49].)

magnitude with the predictions of Ruo€ and Ruo€ [135, 136]and others (see below)
for the compressibility of the C60 molecule itself. Although more accurate numerical
data have been obtained in later experiments, these two early studies are still widely
cited because they illustrate both the fact that C60 is basically `three-dimensional
(3D) graphite’ in its intermolecular properties and the fact that measurements of the
high-pressure properties are complicated by the formation of high-pressure phases.
Later experiments have concentrated on either increasing the accuracy of the
measurements or on measuring the compression properties over larger ranges in
temperature and in other structural phases, or all of these. To give some examples,
David and Ibberson [99] carried out a highly accurate neutron di€ raction study of
the compressibility up to an Ar pressure of 0.4GPa at 150± 200K, with the primary
objective of studying the molecular orientation as a function of pressure (see section
4.1.3). Lundin and Sundqvist [79]were the ® rst to measure the bulk modulus in the
fcc phase at and above RT and also carried out careful studies of the compression
properties of sc C60 up to 1 GPa over a wide range in T, from 150 to 430K [80]. The
pressure range was extended signi® cantly by Ludwig et al. [141], who measured V
against p in both sc and fcc C60 up to above 8 GPa at temperatures down to 70 K.
Very careful studies of the compressibility in both the sc and the fcc phases up to
0.6 GPa at RT were carried out using neutron scattering in a gas environment by
Schirber et al. [72]. This study also included the e€ ects of gas intercalation on the
structure, bulk modulus and phase diagram on C60 (see section 7.4). Finally,
Pintschovius et al. [74] have very recently carried out a neutron scattering study
up to a pressure of 0.5GPa over an extended temperature range and answered
several questions that remained unanswered in the ® rst version of this review. To
date, over 20 publications have been devoted to measurements of the bulk modulus
of C60, some by direct compression as those presented above, some by other
methods. Numerical data from most of these are presented in table 2, and digitalized
28 B. Sundqvist

Â
Table 2. Experimental data for the bulk modulus B and for B = dB /dp for C60 under various
conditions. The values given are usually the initial bulk modulus B(0) at atmospheric
(zero) pressure. RT denotes room temperature; X indicates X-ray diffraction in a DAC,
M mechanical measurements and N neutron scattering.

Pressure
T range Comments and
(K) Phase (GPa) B(0) B Â Reference techniques

RT fcc+ sc 0± 1.2 14.8a Ð Fischer et al. [68] X


RT fcc 0 6.4b Ð Coufal et al. [142] Surface acoustic
wave technique,
thin ® lm
RT fcc 0 8.4b Ð Kobelev et al. [143, 144] Ultrasound
RT fcc 0 11.3b Ð Fioretto et al. [145] Brillouin
scattering, thin
® lm
RT fcc 0± 0.38 8.4c 6.7c Lundin et al. [90] M
RT fcc 0.05± 0.2 8.5 32 Grube [73] M, in Ar
298 fcc 0± 0.13 6.8 29 Lundin and Sundqvist M
[79, 80]
317 fcc 0± 0.25 6.7 24 Lundin and Sundqvist M
[79, 80]
336 fcc 0± 0.3 6.8 21 Lundin and Sundqvist M
[79, 80]
a
RT fcc 0± 0.35 6.7 Ð Komori and Miyamoto X
[146]
a
RT fcc 0± 0.3 13.4 Ð Ludwig et al. [141] X
RT fcc 0± 0.25 12.5 15 Schirber et al. [72] N, in He
RT fcc 0± 0.25 12 12 Schirber et al. [72] N, in Ne
RT fcc 0± 0.2 11.8 5 Schirber et al. [72] N, in Ar
RT fcc 0± 0.2 9.6 20 Pintschovius et al. [74] N, in Ar
RT sc 0± 20 18.1 5.7 Duclos et al. [49] X; hydrostatic
pressure
RT sc 0.38± 1.1 11.6c 4.2c Lundin et al. [90] M
298 sc 0.5± 1.1 9.5 11.1 Lundin and Sundqvist M
[79, 80]
RT sc? 0± 2 15.4 6.9 Bao et al. [91] M
290 sc 0± 2.7 16.4 Ð Bashkin et al. [147] M
RT sc 0± 13.7 20.9 5.7 Wang et al. [148] X
RT sc 0± 10 14.4 10.6 Haines et al. [149] X
RT sc 0.3± 12 13.4d 2.53d Ludwig et al. [141] X
RT sc 0.35± 1.6 28a Ð Komori and Miyamoto X
[146]
RT sc 0± 1.8 18.5a Ð Blaschko et al. [92] N
RT sc 0.4± 0.6 16.5a Ð Schirber et al. [72] N, in He
RT sc 0.4± 0.6 15.3a Ð Schirber et al. [72] N, in Ne
RT sc 0.3± 0.55 13.2a Ð Schirber et al. [72] N, in Ar
RT sc 0.23± 0.5 8.3 14 Pintschovius et al [74] N, in Ar
262 sc 0± 0.5 9.5 13 Pintschovius et al. [74] N, in Ar
236 sc 0± 0.5 9.6 13.1 Lundin and Sundqvist M
[79, 80]
180 sc 0± 0.1 11.5 Ð Grube [73] M, in Ar
170 sc 0± 9 14.2d Ð Ludwig et al. [141] X
152 sc 0± 0.6 10.4 16.2 Lundin and Sundqvist M
[79, 80]
150 sc 0± 0.28 12.75a Ð David and Ibberson [99] N
continued
Fullerenes under high pressures 29

Table 2Ð (concluded )

Pressure
T range Comments and
(K) Phase (GPa) B(0) B Â Reference techniques
e
150 sc 0± 0.5 12 ± Pintschovius et al. [74] N, in Hef
110 sc 0± 0.1 12.5 Ð Grube [73] M, in Hef
e
110 sc 0± 0.5 13 Ð Pintschovius et al. [74] N, in Hef
70 sc 0± 9 14.7d Ð Ludwig et al. [141] X
60 sc 0± 0.1 13.75 Ð Grube [73] M, in He; in glass
phase
RT orhg 0± 3 10.5 9.1 Shimomura et al. [150] X
290 orhh 0± 2.7 32.9 Ð Bashkin et al. [147] M
RT orh?h 5± 20 > 200a Ð Duclos et al. [49] X; for polymer
phase? (see text)
RT Amorphous 0 540b Ð Blank et al. [151] Ultrasound
2h
a
Average value over pressure range indicated (linear approximation).
b
Adiabatic bulk modulus BS .
c
Mixture of C60 and C70; no sharp fcc-to-sc transition observed.
d
Non-standard modi® ed Birch equation of state.
e
B nonlinear in p; negative initial dB /dp.
f
He reported not to intercalate in C60 below 180 K [73].
g
Orthorhombic C60 crystal obtained by crystallization from solution.
h
Polymeric material; see section 5.5.

data from most of those carried out at RT have been plotted together to bring out
the di€ erences and similarities. Some of these curves are shown in ® gures 16± 18,
plotted on di€ erent scales to enable comparisons in di€ erent ranges of pressure. Of
the values shown in table 2, some are given in the original papers, while some have
been extracted with reduced accuracy from graphs. It is obvious from table 2 that the
scatter in the data is surprisingly large, considering the accuracy expected in such
experiments. Because of the large number of investigations all these data cannot be
discussed in detail, but a number of comments should be made on the results.
First, Lundin and Sundqvist [80] noted that an internally consistent set of data
now exists for the compression properties of the sc phase up to above 1 GPa. This is
clearly seen in ® gure 16, showing the similarity of the RT data of Lundin and
Sundqvist [79, 80], Ludwig et al. [141]and Schirber et al. [72], which all agree very
closely above 0.25 GPa. To these data sets we can add the very recent data of
Pintschovius et al. [74]; in the sc phase, the data from [72], [74]and [80]are, in fact,
almost indistinguishable at all points in their common pressure range. The low-
temperature data of Ludwig et al., Lundin and Sundqvist and Pintschovius et al. also
agree very well with those of David and Ibberson [99]. To give some numerical
examples, the relative volumes V / V o observed agree extremely well between the
groups. At 150± 152K and 280MPa, Lundin and Sundqvist obtained
V / V o = 0.9780, Pintschovius et al. V / V o = 0. 9778, and David and Ibberson
V / V o = 0.9776 while at 1 GPa Lundin and Sundqvist obtained V / V o = 0.9460 at
152K and Ludwig et al. V /V o = 0.9464 at 170K. The RT data for the bulk modulus
B given in table 2 appear to di€ er greatly, but this seems to result mainly from
di€ erences in the extrapolations of the data since the data for the volume
compression in ® gure 16 actually agree very well between di€ erent groups. A
30 B. Sundqvist

Figure 16. Measured volume change against pressure for C60 in the low-pressure range at
RT. The `reference set’ for the sc phase discussed in the text contains data from
Schirber et al. [72] (d ), Pintschovius et al. [74] (---), Lundin and Sundqvist [79, 80]
(Ð ) and Ludwig et al. ( j ). Other data are from Duclos et al. [49](n ), Blaschko et al.
[92] ( s ) and Bashkin et al. [147] ( h ).

Figure 17. Measured volume change against pressure for C60 up to 3 GPa at RT. Data are
from Duclos et al. [49] ( n ), Fischer et al. [68] (, ), Schirber et al. [72] (d ), Lundin
and Sundqvist [79, 80] (Ð ), Bao et al. (---), Blaschko et al. [92] (s ), Ludwig et al.
[141] ( j ) and Bashkin et al. [147] (h ).

calculation of the average bulk modulus B over the common pressure range 0.3±
0.6 GPa gives values greater than 13.2 GPa from Schirber et al., 14.5 GPa from
Ludwig et al., 14.7 GPa from Pintschovius et al. and 14.5GPa from Lundin and
Sundqvist, all in very good agreement. Below 260K, on the other hand, the sc phase
is stable at zero pressure and no extrapolations are involved in the calculation of
B(0) . Data for the zero-pressure bulk modulus B(0) from the latter three data sets,
plus the low-temperature data of Grube [73] and David and Ibberson [99], have
therefore been plotted against temperature in ® gure 19. All data points in the range
Fullerenes under high pressures 31

Figure 18. Measured volume change against pressure for C60 up to 28GPa at RT. Data are
from Duclos et al. [49] ( n ), Ludwig et al. [141] ( j ), Wang et al. [148] (---), Haines
and Le ger [149] (s ) and Nguyen et al. [152] (d ).

Figure 19. Low-temperature zero-pressure bulk modulus B(0) for sc C60. Data are from
Grube [73] (n ), Lundin and Sundqvist [79, 80] (, ), Pintschovius et al. [74] ( d ),
David and Ibberson [99](h ) and Ludwig et al. [141]( j ). The full line has been ® tted
to the data up to 262K only.

below 260K can be well described by a straight line


B(0) = (15.6 - 0.023T ) GPa, (4.5)
with T in kelvins. An extrapolation of this line to RT gives a value for B(0) of
8.9 GPa which agrees very well with the data of Lundin et al. and Pintschovius et al.
(also shown on the plot). This shows again that we have a very good agreement
between six recent sets of data which thus seem to form a reliable reference set for sc
C60 in the low-pressure range. In particular, it should be noted both in ® gures 16 and
19 and in table 2 that all recent accurate values for B are signi® cantly (25± 55%)
lower than the value of 18.1 GPa given by Duclos et al. [49], which unfortunately is
32 B. Sundqvist

Figure 20. Measured bulk modulus B as a function of pressure at 150 and 236K. The
broken line has been corrected to show B against p for a hypothetical state with
constant f . (Reprinted from Lundin and Sundqvist [80].)

still widely used as a reference standard in calculations (see below). The temperature
dependence of the bulk modulus also seems to be unusually strong in C60 .
The P± H reorientation with pressure and temperature has a large e€ ect on the
measured compression data in the sc phase. First, the two states have recently been
shown to have di€ erent compressibilities [74], with the P-oriented state being about
10% `harder’. Second, every increase in pressure is accompanied by a reorientation
of a number of molecules from the P- to the H-oriented state. Because the H
orientation has a lower molecular volume, this gives an extra volume decrease which
adds to the volume change that would have been observed at a constant [P]/[H]ratio.
The sc phase thus has an anomalously low B in the pressure and temperature range
where this reorientation occurs, and theoretical calculations assuming a constant [P]/
[H] ratio should be expected to give higher values for B(0) than those observed in
experiments. Figure 20 shows the measured B as a function of pressure at 152 and
236K [80]. (Because of the high experimental resolution in this experiment, B could
be calculated directly from consecutive measured pairs of V and p.) At both
temperatures, anomalies are observed in the pressure range where the reorientational
transformation is expected to be approaching completion, as discussed in section
4.1.3. The reorientation in fact mainly a€ ects the slope B = dB /dp, as shown by the
Â
broken line which is the result of a simple approximate calculation of how B would
depend on the pressure if the [P]/[H] ratio (see section 4.1.3) had stayed constant.
Â
Even more pronounced anomalies, including changes in the sign of B , have recently
been observed below 150K by Pintschovius et al. [74]. Below the glass transition no
further molecular reorientation occurs, and both Pinschovius et al. and Grube [73]
showed that this results in a strong increase in B on cooling through Tg , as would be
expected.
At pressures well above 1 GPa the di€ erence between the results of di€ erent
studies become larger but data from all groups tend to fall on or in between the
extreme curves found by Duclos et al. [49] for hydrostatic and non-hydrostatic
conditions respectively (® gures 17 and 18; see also ® gure 15). Ludwig et al. [141]
found a very strong curvature in V against pressure and noted that no reasonable
standard equation of state is able to describe their data over the full range of
Fullerenes under high pressures 33

pressures studied. Considering the C60 molecules as incompressible spheres they


suggested a modi® ed Birch± Murnaghan equation of state in which the constant
volume of the spheres is subtracted from the crystal lattice. This modi® ed equation is
then able to describe the compression of the `compressible fraction’ of the lattice very
well at all temperatures, but the structure of the equation makes it di cult to
compare their numerical data with data from other sources. (The actual physical
e€ ect behind the observed rapid increase in B with increasing pressure is probably a
continuous polymerization with increasing pressure, as will be discussed further in
section 5.) The individual C60 molecules indeed seem to be quite incompressible. This
has been shown both theoretically, as will be discussed brie¯ y below, and experi-
mentally. Duclos et al. [49]showed from the disappearance of the (200) di€ raction
peak that the molecular form factor [32] must be almost unchanged near 20 GPa,
indicating a decrease in molecular radius of the order of 1% and a `molecular’ B of at
least 670GPa. Although Singh [153] tried to explain the di€ erences between
hydrostatic and non-hydrostatic studies from elasticity theory, the e€ ects discussed
above and in sections 4.1.3 and 5 suggested that the large di€ erences observed simply
re¯ ect the fact that the translational, the orientational and even the molecular
structure of C60 may di€ er between experiments depending on the rate of compres-
sion and the pressure medium used. These factors lead to di€ erent structures,
depending on the degree of polymerization, the degree of lattice disorder created
by inhomogeneous pressure and possibly pressure quenching of the orientational
state, all factors which will a€ ect the measured compression properties.
Returning to the low-pressure range, data for fcc C60 are quite scarce but the four
groups in the reference set de® ned above have all published data also for this phase.
Surprisingly, although the data for the sc phase agreed extremely well between these
groups, large di€ erences are found between the results for the fcc phase. The most
accurate results should be those obtained in truly hydrostatic media, and the data for
V / V 0 from Schirber et al. [72] and Pintschovius et al. [74] obtained under Ar gas
pressures are indeed in excellent agreement and intermediate between those of
Ludwig et al. [141]and those of Lundin and Sundqvist [79, 80], obtained under less
hydrostatic conditions (® gure 16). However, because of di€ erences in the experi-
mental resolution the numerical data for B given di€ er. Schirber et al. found a bulk
modulus B(0) near 11.8 GPa while the more closely spaced data of Pintschovius et al.
gave B(0) = 9. 6 GPa. The latter value is in good agreement with the single-crystal
data of Grube [73], who found B(0) = 8.5 GPa using a capacitive technique. The best
estimate for the true bulk modulus of the fcc phase at present is thus probably the
value of 9.6 GPa given by Pintschovius et al. (Note that this value is higher than the
value of 8.9 GPa recommended for the sc phase at the same temperature, in spite of
the larger intermolecular distance in the fcc phase.) As expected, all these values are
intermediate between the value of 13.4GPa found by Ludwig et al. from a linear ® t
to their X-ray data between 0 and 0.3 GPa and the value of 6.8 GPa found by Lundin
and Sundqvist using a high-resolution mechanical method. Results obtained by
other groups show a similarly large scatter. Optical (Brillouin scattering) zero-
pressure studies [145] indicated that B(0) = 11.3 GPa, almost identical with values
found by zero-pressure studies by ultrasonic methods [142, 144]. Theoretical
calculations [154] tend to support slightly higher values (see below). The large
di€ erences between di€ erent studies is remarkable, especially since the same
experiments in several cases gave very good agreement between the properties in
the sc phase (see above and in ® gure 16). However, there is a possible correlation
34 B. Sundqvist

between the structural state of most of the samples used and the measured bulk
moduli. The experiments giving high values (9± 13 GPa) for B(0) have all used well
ordered materials, either annealed powder or crystals or epitaxial ® lms, while in the
studies giving lower bulk moduli (6± 7 GPa) the materials had usually been submitted
to strong shear stress in the experiment [79, 80, 90, 147], ground [146]or uniaxially
pressed [144] before the experiment or consisted of disordered ® lms [142]. Since a
disordered sample should have a lower density than a close-packed crystal of the
same material, it is not unreasonable to connect low values for B with structural
disorder, but it is not clear why the correlation between disorder and low B should be
stronger for the fcc phase than for the sc phase. In the mechanical measurements [79,
80, 90, 147]it might also be possible that strong shear stress locks molecular rotation
until well below the nominal transition pressure on decreasing pressure, giving an
asymmetry in the transition behaviour and thus an arti® cially low volume close to
and below the transition pressure or even induces a small degree of polymerization in
the sample (section 5). Very recent compression studies on C60 under hydrostatic
conditions in a liquid medium using the same apparatus as Lundin and Sundqvist
(A. Soldatov 1997, private communication) showed better agreement with the data
of Pintschovius et al., indicating that non-hydrostatic stress and/or disorder are
indeed responsible for the di€ erences observed.
Somewhat surprisingly, some RT experiments failed to detect the fcc-to-sc
transformation (with a 1% change in volume) near 0.3 GPa but sometimes showed
unexplained anomalies at other pressures. In some cases the absence of signi® cant
transformation anomalies is clearly due to high levels of impurities, as in the early
study on a C60 ± C70 mixture by Lundin et al. [90], but in other cases no explanation
can be found. The mechanical compression study by Bao et al. [91]showed no trace
of anomalies below 1GPa but a large volume anomaly with a signi® cant hysteresis
near 2.2 GPa, while a neutron scattering study by Blaschko et al. [92] shows the
lattice compression to be almost linear in pressure from 0.45 to 1.6 GPa with a slope
that indicates a very high B in the fcc phase and a very small transition anomaly.
For completeness, I note that measurements of other mechanical properties, such
as Young’s modulus E and other elastic moduli, as well as the hardness of C60 have
also been carried out, but only at zero pressure. The results often show a wide
scatter. For E, values of 15.9 GPa [109] and 20 GPa [155] were obtained for single
crystals and 9.8 GPa for compacted powder [143], while thin-® lm studies indicated
values of 9.99 GPa [145] and 15± 18 GPa [156] for supported ® lms and signi® cantly
higher values for free-standing ® lms [157]. Di€ erent studies also show very di€ erent
temperature dependences of E. For other moduli the situation is similar.
As already stated above, a large number of calculations of the compression
properties of C60 have also been carried out, mainly to test models for the
intermolecular potential. Unfortunately, most workers ® xed the parameters of their
models by comparing their data with the very high B(0) = 18.1 GPa and low
B = 5. 7 given by Duclos et al. Early studies in this ® eld were often aimed at testing
Â
the prediction by Ruo€ and Ruo€ [135] that C60 at high pressures might be less
compressible than diamond. Usually, a simple van der Waals-type intermolecular
potential with Lennard-Jones (6± 12) or exponential (Buckingham or Born± Meyer)
atom± atom interactions were used with parameter values taken from studies of
graphite or hydrocarbons. These models gave reasonable values for the bulk
modulus [158± 162] but, if ® tted to give B = 18 GPa, they gave very large values
for B (10± 15) and thus signi® cantly too high values for B above 2 GPa, while
Â
Fullerenes under high pressures 35

molecular dynamics calculations gave very low values for To and an incorrect
structure for the low-temperature phase. In order to reproduce correctly the
transition temperature and the low-temperature structure it was necessary to add
an electrostatic interaction between the molecules. This can be done in many ways,
but usually negative charges are put in the centre of the (electron-rich) double CÐÐ C
bonds. Sprik and co-workers [37, 163] added the positive charges to the C atoms
while Lu et al. [94] added these to the centre of the single CÐ C bonds on the
pentagons and Burgos et al. [164] to the centres of the pentagons. Yildirim and
Harris [165] used the potentials of both Sprik and co-workers and Lu et al. and
found bulk moduli of 8.7± 9.6 GPa, in excellent agreement with recent experiments,
while Li et al. [166] use the Lu et al. potential and found B(0) = 19.3 GPa with
B = 18.1. (However, Pintschovius and Chaplot [167]noted that this value of B does
Â
not agree with the calculated dispersion curves and might be a factor of two too
high.) La Rocca [168] took into account the electron distribution on the molecule
and found B = 12GPa with B = 13, also in good agreement with experiments, while
Â
the very recent calculation by Prilutski and Shapovalov [169]used a Lu et al. type of
potential for ® nd B = 13.5 GPa. To show that theoretical calculations can be quite
successful, ® gure 21 shows the volume against pressure for C60 at RT as calculated
by Burgos et al. [164]. This calculation reproduced quite successfully the cell
parameters against both temperature and pressure, the transition temperature To,
the change in volume at To, the phonon frequencies, the densities of state, the speci® c
heat capacity and the sound velocities. A comparison with ® gures 16± 18 shows that
the calculated volume against pressure is very close to the experimental data. They
reported that B < 13GPa, but it is not clear to what this value refers. Inspection of
the ® gure shows that the initial B(0) in the fcc phase is probably below 10GPa, and
from data for elastic constants from a later calculation [93](using slightly modi® ed
parameters) a value near 9 GPa is found (see table 4 below). They also found a
change in the orientational structure and the [P]/[H] ratio (see section 4.1.3) with
pressure and predicted an equilibrium pressure Peq = 0. 6 GPa. Comparisons
between calculated an experimental data are, of course, complicated by the mol-

Figure 21. Calculated volume against pressure: (Ð ), calculated; (d ), [49], and (s ), [68].
(Reprinted with permission from Burgos et al. [164]. )
36 B. Sundqvist

Table 3. Theoretical temperature dependence of B at zero pressure for C60, taken from the
works of Burgos et al. [93], Yakub [131], Zubov et al. [133] and Girifalco [154]. For
comparison, experimental data from Pintschovius et al. [74], Lundin and Sundqvist [80]
and Ludwig et al. [141]are also shown.

B(0) (GPa) from the following references

T (K) Phase [93] [131] [133] [154] [74] [80] [141]

0 sc 16.4a Ð Ð Ð Ð Ð Ð
70 sc Ð Ð Ð Ð 15.1b Ð 14.7b
120 sc 15.4a Ð Ð Ð Ð Ð Ð
150 sc Ð Ð Ð Ð 12b 10.4b Ð
170 sc Ð Ð Ð Ð Ð Ð 14.2b
200 sc 14.5a Ð Ð Ð Ð Ð Ð
236 sc Ð Ð Ð Ð Ð 9.6b Ð
275 (sc) 13.7a Ð Ð Ð Ð Ð Ð
250 (fcc) Ð 14.3 Ð Ð Ð Ð Ð
261 fcc Ð Ð 14.05 Ð 9.5b Ð Ð
275 fcc 9.4a Ð Ð Ð Ð Ð Ð
a b b b
300 fcc 9.0 13.7 13.7 11.9 9.6 6.8 13.4
335 fcc Ð Ð Ð Ð Ð 6.8b Ð
350 fcc Ð 13.3 Ð Ð Ð Ð Ð
400 fcc Ð 12.8 12.9 Ð Ð Ð Ð
500 fcc Ð 11.9 Ð 11.7 Ð Ð Ð
600 fcc Ð Ð 11.3 Ð Ð Ð Ð
700 fcc Ð Ð Ð 11.5 Ð Ð Ð
800 fcc Ð Ð 9.75 Ð Ð Ð Ð
900 fcc Ð Ð Ð 11.3 Ð Ð Ð
1000 fcc Ð 9.6 8.27 Ð Ð Ð Ð
1100 fcc Ð Ð Ð 11.1 Ð Ð Ð
1200 fcc Ð Ð 6.82 Ð Ð Ð Ð
1300 fcc Ð Ð Ð 10.9 Ð Ð Ð
1400 fcc Ð Ð 5.37 Ð Ð Ð Ð
1500 fcc Ð Ð Ð 10.7 Ð Ð Ð
1600 fcc Ð Ð 3.86 Ð Ð Ð Ð
1800 fcc Ð Ð 2.11 Ð Ð Ð Ð
1900 fcc Ð Ð 0.74 Ð Ð Ð Ð
a
Calculated from theoretical elastic constants using equation (4.5).
b
Experimental value.

ecular reorientation under pressure in the sc phase (section 4.1.3) and by the fact that
calculations usually give the adiabatic value Bs (see equation (4.4)), features not
usually taken into account in the literature. The intermolecular potential of C60 has
recently attracted increasing interest and a number of recent papers discussed
various ways of improving the agreement with experiments [170].
The properties of fcc (i.e. rotationally disordered) C60 have also been investigated
theoretically. Since the intermolecular potential of this phase does not have to
reproduce the orientational structure the problem is here, in principle, simpler.
Girifalco [154]calculated V against pressure and B in the GruÈneisen approximation
and found B = 11.9 GPa at 300K, about 25% higher than the `best’ experimental
value of 9.6 GPa [74] discussed above. The calculated volume compression was in
very good agreement with the experimental data obtained by Ludwig et al. [141]
(® gure 16). Girifalco also calculated the temperature dependence of B up to 1500K.
Fullerenes under high pressures 37

Table 4. Theoretical elastic constants for C60 at RT.

c11 c12 c44 B


Reference (GPa) (GPa) (GPa) (GPa)

Burgos et al. [93]a 13.8 6.6 7.0 9.0


Zubov et al. [133]b 21.5 9.8 11.0 13.7
Yildirim and Harris [165]c 14.9 6.9 8.1 9.6
Yildirim and Harris [165]d 14.1 6.0 7.7 8.7
Li et al. [166] 29.9 14.1 15.9 19.3
Prilutski and Shapovalov [169]e 22.4 9.1 11.6 13.5
Venkatesh and Gopala Rao [173] 22 11 11 14.7
Yu et al. [174, 175] 26.4 13.8 11.7 18.0
a
Data given also at lower temperatures and in the sc phase.
b
Data given also at higher temperatures.
c
Potential due to Sprik et al. [63] (see text).
d
Potential due to Lu et al. [194] (see text).
e
At 4.2 K.

The bulk moduli B and/or Bs have also been calculated by a number of other groups
[124, 131, 133, 171± 173]. Available data for the temperature dependence of B have
been collected in table 3; to simplify comparisons, some zero-pressure experimental
data from table 2 have also been included. The experimental and theoretical
temperature dependences are rather similar and the agreement between di€ erent
works is reasonable except that Girifalco seems to ® nd an unusually small
temperature dependence. Several researchers have also calculated the elastic constant
c11, c12 and c44 , in both the sc and the fcc phases. RT data are presented in table 4
together with values for B calculated from the standard formula
c11 + 2c12
B= (4.6)
3
Finally, even the bulk modulus of the C60 molecule itself has been calculated. Ruo€
and Ruo€ [135, 136] used simple mechanical arguments to deduce a value of
843GPa, while Woo et al. [176] used a tight binding method to ® nd 717GPa and
Jishi et al. [177] used a phenomenological model [177, 178] ® tted to spectroscopic
data to ® nd 900GPa.

4.2.2. L attice vibrations


Like the compressibility, the lattice vibrations are determined by intermolecular
and interatomic interactions. So far, the pressure dependence of the phonon
spectrum of C60 has only been investigated by optical methods, that is Raman
and IR spectroscopy. These methods give two basic types of information. First,
spectroscopic studies show of course the pressure dependence of the vibrational
modes. Since the molecules themselves are almost incompressible (see above), neither
intramolecular bond lengths nor the corresponding vibrational frequencies should be
expected to change much with pressure. A high experimental resolution is therefore
needed to obtain accurate results. Second, these types of study also give information
on structural phase transitions under pressure and in some cases allow the
identi® cation of the basic type of structure (amorphous, graphite like or diamond
like).
38 B. Sundqvist

Very complete reviews of Raman and IR spectroscopy and lattice vibrations in


fullerenes have recently been given by Dresselhaus et al. [3, 179]. These reviews also
include brief notes on the pressure dependence of the active modes. Martin et al.
[180]have also given a very complete overview of the possible vibration frequencies
of solid monomeric C60, Raman-active, IR-active and normally silent modes. The
latter can often be observed in highly sensitive studies on real solids because the high
abundance of the isotope 12 C breaks the symmetry of the molecule. In their IR
absorption experiments Martin et al. [180]observed and identi® ed over 180 vibration
modes in the range 100± 4000cm- 1 and studied their dependence on temperature
from 77 to 300K and on pressure up to 2.5 GPa. Many low-frequency (inter-
molecular) modes are observed to sharpen appreciably below the fcc-to-sc transition
at 260, and below the glassy crystal transition many broad lines also split into
several well resolved peaks. In principle, these kinds of study can thus be used to
determine both phase and glass transition boundaries under pressure. Figure 22
shows the pressure dependence of 46 strong intermolecular and intramolecular
modes, measured by Martin et al. during a pressure decrease from 2.5 to 0.5 GPa
and thus in the sc phase. Some of the modes shown in this ® gure, in particular that at
611cm- 1 , were not observed in pristine C60 at zero pressure and were thus suggested
to be connected with the formation of a polymer phase under pressure. Since
polymerization has not been observed by other groups in this range of pressures
and temperatures, it is possible that this mode is instead connected with the expected
transition into an (H-)oriented structure at these pressures. The frequencies of most

Figure 22. The pressure dependence of a large number of lattice modes in C60 . (Reprinted
with permission from Martin et al. [180]. )
Fullerenes under high pressures 39

Figure 23. The pressure dependence of the IR-active modes of C60. ( Reprinted with per-
mission from Klug et al. [182].)

modes increase with increasing pressure, as would be expected for sti€ ening lattice,
but for many modes a decrease is observed instead. However, no detailed analysis of
the pressure dependence was carried out and as shown in the ® gure the pressure
range did not extend to su ciently low pressures to allow the detection of the fcc-to-
sc phase boundary.
Three other groups [118, 181± 183] have carried out IR spectroscopy studies
focused on the four strong IR-active modes under pressure. As an example, ® gure 23
shows the results of Klug et al. [182]which extend over the largest pressure range up
to 20GPa. The maximum pressures in the other experiments are 3.2 GPa [118, 181]
and 6 GPa [183]respectively. All these experiments show that the frequencies of the
three highest modes increase under pressure while that of the lowest mode decreases.
The latter e€ ect is interpreted by Aoki et al. [181]as resulting from an increasingly
`covalent-bond-like’ intermolecular interaction between opposing faces of neigh-
bouring molecules. The results from the three groups are collected in table 5 in terms
of the initial pressure slopes dx /dp (sometimes deduced from the graphs given in the
original papers). Data from a theoretical calculation (see below) are also given as a
comparison. The data of Huang et al. [118] and Klug et al. [182] are in good
agreement while Aoki et al. [181]found a slightly larger pressure dependence for all
40 B. Sundqvist

Table 5. Experimental data for the initial pressure slopes of the frequencies of the IR-active
modes in C60. Data were taken from Aoki et al. [181], Huang et al. [118]and Klug et al.
[182]and compared with theoretical results from Jishi et al. [177].

dx /dp (cm- 1 /GPa- 1) from the following references

Mode x 0 (cm- 1) [181] [118] [182] [177]a

F1u ( 1) 526 - 1.55 - 0.87 - 0.5 0.0


F1u ( 2) 576 2.87 2.5 2.3 1.2
F1u ( 3) 1183 4.39 3.4 3.8 2.4
F1u ( 4) 1429 6.28 4.4 4.3 4.5
a
Theoretical calculation.

modes. The data shown are also in good general agreement with the results of
Martin et al. [180]. The pressure dependences are rather weak, with mode GruÈneisen
parameters g i = B d (ln x i ) dp having magnitudes well below 0.1 [181, 183], as
expected from the small molecular compressibility. Klug et al. [182] found that the
two intermediate modes broadened and disappeared near 9 GPa, at which pressure
the slope of the lowest mode also seems to change, while the highest and lowest
modes remain observable up to the maximum pressures studied. On returning to
zero pressure the spectrum returned to its initial form, except for a slight broadening
of the lines and a strong increase in intensity for a group of minor peaks between 550
and 850cm- 1 now known to indicate polymerization. Huang et al. [118] reported
that the 576cm- 1 line splits into two components above 1.4 GPa with the two
components being similar in strength at the highest pressures (2.6 GPa or higher). As
discussed above in section 4.1.3, this e€ ect is probably connected with an orienta-
tional ordering at high pressures similar to that observed at low temperatures but to
a H-oriented rather than a P-oriented state.
A large number of Raman studies have also been carried out on the monomer
phases of C60 under pressure. Theory predicts ten Raman-active modes in C60 but, as
mentioned above, a larger number of modes are often observed in good crystals. In
high-pressure studies, however, only a small number of high-intensity lines are often
observed because of experimental di culties, in particular the presence of the strong
Raman line of diamond near 1333cm- 1 which usually makes it impossible to detect
the C60 lines at 1100 and 1250cm- 1. Available Raman data can be conveniently
divided into two groups. The ® rst includes a number of early exploratory studies
[117, 184± 188] while the second more recent group mainly concentrates on well
de® ned questions such as the pressure dependence of libron modes [189] or the
structural and orientational phase diagram under pressure [81± 83, 101, 190]. (In this
section no attention will be given to Raman studies on polymeric very-high-pressure
phases, which are discussed instead in section 5.) The ® rst group of experiments were
often aimed at establishing the existence of a transformation into an amorphous
state near 20 GPa but, at lower pressures, information on the pressure dependence of
the mode frequencies was also obtained. However, the slopes dx /dp obtained often
di€ ered signi® cantly between di€ erent groups, even as to the sign of the slopes. To a
large extent this seems to be an e€ ect of limited experimental accuracy in both
frequency and pressure since the pressure dependences of the Raman frequencies are
very sensitive to the details of the lattice structure. Also, it should be noted that these
Fullerenes under high pressures 41

Figure 24. Pressure dependence of some Raman frequencies in C60. (Reprinted with per-
mission from Meletov et al. [82].)

workers were not aware of the risk for polymerization of the C60 molecules because
of the action of both pressure and the exciting laser beam. The most detailed
investigation so far of the Raman spectrum of C60 under pressure is the recent study
of Meletov et al. [81± 83, 190]at RT up to 16GPa. Meletov et al. found that in this
range their results indicated the existence of three well de® ned phases: one below
0.4 GPa, identi® ed as the fcc phase, another between 0.4 and 2.4GPa, and the third
above 2.4 GPa. The 2.4 GPa transition was tentatively identi® ed with the glassy
crystal transition which at zero pressure occurs at 90 K but, as discussed in section
4.1.5, it is more probable that it is connected with the orientational ordering under
pressure. Typical data are shown in ® gure 24, reprinted from [82]. For most modes
the 0.4 GPa fcc-to-sc transition involves a step decrease in frequency, as also
observed by others [191]. In the sc phase, above 0.4 GPa, ten new lines were observed
because of the reduction in symmetry when orientational order appears, and all lines
observed broadened with increasing pressure. A systematic redistribution of in-
tensities between some of the lines also occurs, notably between the high-frequency
Ag (2) and Hg (7) modes which interact strongly above 2.4 GPa to avoid band
crossing. All these `anomalies’ were observed to be reversible on decreasing the
pressure. In table 6, available data for the pressure dependence of the eight Raman
modes that can usually be observed in DACs have been collected. These data should
all be valid for the sc phase and with a few exceptions they are all in good agreement.
Most workers give average slopes between zero and the maximum pressure only but
the data of Meletov et al. are given above and below 2.4 GPa. Data for other lines
are given in tabular form by Meletov et al. [83] and Snoke et al. [186]. The
frequencies of most lines increase slowly with increasing pressure, but for the lines
42

Table 6. Pressure dependence of eight Raman-active modes of C60 in the sc phase. Experimental data were taken from Jeon et al. [117], Chandrabhas et
al. [184], Raptis et al. [185], Snoke et al. [186], Tolbert et al. [187], Yoo et al. [188]and Meletov et al. [83]. For comparison, theoretical slopes from
Jishi et al. [177]are also shown.

d(x ) dp cm- 1 GPa- 1) from the following references


[83]

Mode x (cm- 1) [117] [184] [185] [186] [187] [188] < 2.4 GPa > 2.4 GPa [177]a

Hg ( 1) 273 1.1 3.2 3.3 0.1


421 0.16 2.4 2.4 0.5 0.0
B. Sundqvist

Hg ( 2)
Ag ( 1) 496 0 0.75 0.94 4.2 0.5 1.0
Hg ( 3) 710 - 0.92 - 0.55 - 0.8 - 0.8 0.5
Hg ( 4) 774 - 0.71 - 0.50 - 2.7 0.1 1.2
Hg ( 7) 1428 4.12 2.4 4.4 6.4 9.8 3.9 4.5
Ag ( 2) 1469 4.1 3.5 3.11 1.7 5.2 6.4 5.5 5.5 5.0
Hg ( 8) 1575 2.73 3.7 4.9 4.7 4.8 4.8 4.0
a
Theoretical calculation.
Fullerenes under high pressures 43

at intermediate frequencies a softening can be seen. However, in all cases the


frequency shifts are small and, except for the lowest-frequency modes, absolute
values for the mode GruÈneisen parameters are again below about 0.1 [83]. The fcc
phase has been much less studied because of its small pressure range of existence.
Although most studies agree that the pressure slopes of most modes are small and
negative, no accurate data yet exist.
In addition to the frequency shifts with pressure, many researchers observed
broadening of the lines or other types of anomaly. Many of the anomalous results
observed in early experiments, such as the observed increase in disorder under
increasing pressure which were suggested to be related to the formation of an
orientational glass [184, 187] or the hysteresis and/or change in zero-pressure
frequency after the high-pressure experiment [117, 188]are probably connected with
polymerization owing to the combined e€ ects of a high radiation intensity and a high
pressure since the zero-pressure `pentagon pinch’ mode Raman frequencies after the
experiment are usually below 1460cm- 1 .
On the theoretical side, Jishi et al. [177, 178]constructed a force-constant model
for intramolecular vibrations in C60 based on bond stretching and angle bending and
obtained parameter values by ® tting to Raman data. The model was then used to
calculate [177] the bulk modulus of the molecule (section 4.2.1) and the pressure
dependence of the Raman and IR modes. The results are given in tables 5 and 6 and
agree reasonably well with the experimental data, considering that the value of B
from Duclos et al. [49] was used (see section 4.2.1). Yu and co-workers [174, 175,
192, 193]calculated the pressure dependence of the whole phonon spectrum using a
tight-binding approach for intramolecular modes and van der Waals-type inter-
molecular interactions. They predicted a strong broadening of the spectrum and
positive frequency shifts up to about 9 cm- 1 GPa- 1 for intramolecular modes, in
general agreement with the experimental results shown above, and also splittings of
many IR and Raman lines. The low-frequency intermolecular modes should also be
expected to increase rapidly with increasing pressure, and calculations show a
broadening by a factor of about two up to 5.6GPa. Prilutski and Shapovalov
[169] also calculated the pressure dependence of the acoustic intermolecular modes
and the sound velocities and found results similar to those of Yu and co-workers.
While the positive pressure shifts observed in most experiments are reasonably well
reproduced by calculations, there is still no obvious explanation for the negative
pressure shifts observed in the intermediate range between 650 and 800cm- 1 .
As already mentioned above, Horoyski et al. [189]have also carried out a Raman
study of the extremely-low-frequency (15± 20 cm- 1) Raman-active libron modes in
sc C60 at 85 K up to 1.5 GPa. In contrast with the intramolecular modes, the
intermolecular modes are of course very sensitive to pressure because of the large
lattice compressibility. This is also true for the libron modes which are observed to
increase very rapidly with increasing pressure at the rates of 3.7cm- 1 GPa- 1 and
5.2 cm- 1 GPa- 1 for the modes at 17.6cm- 1 and 21.6 cm- 1 respectively (at 85 K).
These values are quite large, corresponding to mode GruÈneisen parameters g > 3
which is an order of magnitude larger than for intramolecular modes, but in very
good agreement with calculations by Vashishta et al. [193], Yu et al. [174, 175, 192]
and Prilutski and Shapovalov [169]. Cheng and Klein [160] have predicted an even
larger pressure dependence for both librational and low-frequency translational
modes. Wolk et al. [101]have also carried out a study of pressure-induced changes in
the molecular orientation at low pressures through Raman studies of libron and
44 B. Sundqvist

intermediate-frequency (about 500cm- 1 ) modes. Since this study was mainly aimed
at mapping the orientational phase diagram of C60 at low temperatures it is discussed
in some detail in section 4.1.3 instead of here.
Elastic and inelastic neutron scattering studies on C60 have been reviewed by
Prassides et al. [18, 194] and more recently by Pintschovius [21]. No neutron
scattering studies have been carried out on the pressure dependence of the acoustical
vibration modes of the spectrum. However, a weighted average of the pressure
dependence of these modes can be obtained from the thermal GruÈneisen parameter
g = 3a B /dcp (= - d(ln H D) /d(ln V )) , where a is the linear thermal expansivity, d
the density and H D the Debye temperature. A small number of calculations and
experimental derivations of this quantity has been carried out and, in contrast with
the values given above for the GruÈneisen parameters of the optical modes, very large
g values are usually found for the acoustic modes because of the weak and strongly
anharmonic intermolecular interactions. Experimental values near 10 have been
found for fcc C60 by Sundqvist [195] and for sc C60 by White et al. [196], values
veri® ed by theoretical calculations on fcc C60 by Girifalco [154] (g = 9.15) and
Zubov et al. [171, 172] (g = 8. 1) . These calculations also showed a rather rapid
decrease in g with increasing pressure [154]and increasing temperature [171, 172]. A
strong pressure dependence of the sound velocity was also found by Prilutski and
Shapovalov [169].

4.2.3. Thermal properties


Since the thermal properties of a material are determined by the phonon
spectrum the pressure dependence of the former can in principle be calculated from
that of the latter. Andersson et al. [84]have carried out low-accuracy measurements
of the speci® c heat per unit volume under high pressures, using the data to help
mapping the structural phase diagram, but because of the low accuracy they give no
numerical data. Theoretical calculations also exist for the speci® c heat under high
pressures in the range 0± 20K, where strongly pressure-dependent librational modes
give a large pressure dependence [192]. NMR measurements under high pressures
(section 4.2.6) have also been used as a tool to ® nd thermodynamic quantities
connected with the orientational dynamics [76, 77]as discussed in section 4.1.3, and a
calculation of Gibbs free energy for C60 up to 20 GPa and 1000K has been carried
out [197]. In general, however, very little information is available on the thermal
properties of C60 under high pressures except for the thermal conductivity.
The thermal transport properties of a basically non-conducting material such as
C60 potentially give important information on the lattice phonon spectrum,
especially the intermolecular acoustic modes and thus the intermolecular potential.
Like all transport measurements, thermal conductivity studies are also able to
provide information on phase transitions and phase boundaries since transport
properties usually change signi® cantly at phase transitions while such studies also
give a continuous record of the magnitude of the property studied. The thermal
conductivity of C60 has been studied in a number of high-pressure experiments, and
the results have been of great importance for our present understanding of the
structures and phases of C60 under high pressures. In this section, only basic
information on the pressure dependence of the thermal conductivity will be
discussed; the use of thermal conductivity measurements for phase diagram studies
were brie¯ y discussed above in section 4.1.
Fullerenes under high pressures 45

Figure 25. Thermal conductivity of C60 at the pressures (in gigapascals) indicated. (Re-
printed from Andersson et al. [84].)

Andersson et al. [84, 104]have measured the thermal conductivity ¸ in situ under
pressures up to greater than 1 GPa and over a large temperature range (50± 300K)
using a highly accurate hot-wire probe method [198]. A drawback with this method
is that single crystals cannot be studied and the measurements must be carried out on
compressed powder. The measured data, such as those shown in ® gure 25, are thus
signi® cantly a€ ected by lattice defects and grain boundaries. While data for single
crystals show the ¸ ~ T - 1 dependence typical for high-quality insulating crystals
[199], ® gure 25 shows a very `¯ at’ behaviour against temperature which is more
typical for disordered solids. In fcc C60 , the quasifree rotation of the molecules
provides a very strong phonon scattering mechanism and the phonon mean free path
K is almost constant and equal to the intermolecular distance. According to the
simple Debye formula
¸ = cp vsK , (4.7)
where vs is the speed of sound; the thermal conductivity is thus almost constant in
this phase, except for a slow increase with increasing temperature owing to cp in both
single crystals [199]and polycrystals (® gure 25). The pressure dependence of ¸ in this
phase, expressed in terms of the Bridgman parameter g = - d(ln ¸) /d(ln V ) , was
g = 5.5, a value typical for such orientationally disordered phases. In the sc phase,
on the other hand, there is a signi® cant temperature dependence and the sc-to-fcc
transition can thus easily be detected in the data at any pressure. In the sc phase a
stronger pressure dependence g = 9.5 was observed as would be expected for a
normal crystalline but disordered phase. Although the general features of the data
can thus be understood, no detailed analysis of the temperature and pressure
dependence could be carried out by Andersson et al. [84]. In particular, the break
in d¸ /dT near 150± 170K could not be modelled by any reasonable model, and these
workers showed that an anomaly exists also in the single-crystal data of Yu et al.
[199]at the same temperature. Models tested included the standard relaxation time
model including both phonon, impurity, defect, boundary and orientational disorder
scattering with a Debye model for the lattice, as well as an Einstein oscillator model
often used for disordered materials. It was suggested that either some other
46 B. Sundqvist

scattering mechanism, such as libron scattering, must exist above 170K, or that
some extra parallel channel for heat transport (libron or optical mode heat
transport?) must exist between about 100 and 170K. More theoretical work is
clearly needed to understand both the temperature and the pressure dependences
of ¸.

4.2.4. Electronic band structure


The electronic band structure of C60 has been studied directly by optical and
other methods and indirectly by resistivity measurements under pressure. The
resistivity data will be discussed in section 4.2.5, while this section will brie¯ y discuss
studies giving direct information on the band structure, starting with optical studies
of the band structure changes with pressure.

4.2.4.1. Optical studies. A number of studies have been carried out on the
pressure dependence of the electronic bandgap in C60 by measurements of either
the optical absorption edge [83, 115, 190, 200± 206] or the photoluminescence [83,
190, 207, 208]. Several of these studies have also been carried out to investigate the
properties of the various very-high-pressure phases of C60 , and these will be
discussed further in the relevant sections.
Only a small number of studies have been carried out on the photoluminescence
of C60 under hydrostatic pressure, all using argon ion lasers to excite the sample.
Sood et al. [207, 208] calculated the shift of the bandgap from the position of the
luminescence peak and found that in the low-pressure range up to 0.6 GPa the
bandgap energy E decreased with increasing pressure at the rate dE /dp = - 0.14 eV
GPa- 1 . No data could be collected at higher pressures because the intensity of the
luminescence was observed to decrease very rapidly with increasing pressure such
that no signal was observed at 3.2 GPa. Because of limitations in the detector
bandwidth, Meletov et al. [83, 190]used instead an extrapolation of the high-energy
side of the peak and found dE /dp = - 0.074eVGPa- 1 .
Most optical bandgap studies have been carried out by measuring the optical
absorption edge near 1.7 eV. Although some early studies [186, 200]were carried out
on mixtures of C60 and C70 , most studies have been carried out on highly pure C60 .
Moshary et al. [201], Snoke et al. [186, 202]and Hess et al. [203, 204]all measured
absorption spectra of C60 ® lms, Moshary et al. up to 35GPa in (hydrostatic) liquid
xenon, Hess et al. up to 5.3 GPa in (hydrostatic) unspeci® ed alcohol, and Snoke et al.
up to 40 GPa in quasihydrostatic CsCl. Meletov et al. [115, 190, 205] made similar
studies on small crystals between 1.1 and 3.1 eV up to 19GPa, mainly under
hydrostatic conditions. Their pressure medium was the standard 4 : 1 mixture of
methanol and ethanol known to be hydrostatic up to 12 GPa [50]. All groups agree
that the absorption edge of C60 has a two-step structure at atmospheric pressure with
a ® rst onset just above 1.7 eV and an extrapolated second threshold near 2.1 eV, and
that the absorption thresholds shift rapidly towards lower energies with increasing
pressure as the compression of the lattice broadens the electron energy bands of the
material, as shown in ® gure 26 (taken from Meletov et al. [115]). However, again
rather di€ erent results were obtained for the pressure dependence of the bandgap.
The two absorption thresholds shift at di€ erent rates under pressure such that the
1.7 eV feature disappears at high pressures as discussed below, and most studies have
been concentrated on the second high-energy threshold only. For this, Moshary et al.
[201](® gure 27) found that E was linear in pressure in the whole pressure range up to
Fullerenes under high pressures 47

Figure 26. Optical absorption spectra of a C60 crystal at RT obtained at pressures of 0 GPa
(curve 1), 0.9GPa (curve 2), 1.4GPa (curve 3) and 2.4GPa (curve 4). (Reprinted with
permission from Meletov et al. [115].)

Figure 27. Optical absorption in C60 at RT and the pressures indicated. CF denotes
`collapsed fullerite’. ( Reprinted with permission from Moshary et al. [201]. )

17GPa with a slope d(ln E) /dp = - 0. 03 GPa- 1 (dE /dp = - 0.055eV GPa- 1 ) , much
smaller than that found in the luminescence studies and indicating by extrapolation
an insulator-to-metal transition near 33GPa. Snoke et al. [186] found a larger
average slope dE /dp = - 0. 1eV GPa- 1 up to 10 GPa, while Hess et al. [203, 204]
found a linear behaviour to 5 GPa with an intermediate dE /dp = - 0.07 eV GPa- 1 .
In a later more detailed study, Snoke et al. [202]reanalysed their previous data and
carried out further studies that showed the gap energy to be a nonlinear function of
pressure with an initial zero-pressure slope near dE /dp = - 0.14eV GPa- 1 , smoothly
decreasing to about - 0.02 eV GPa- 1 above 10 GPa. These values are almost identical
with those found by Meletov et al. [115]at the same pressures ( - 0.15eV GPa- 1 and
- 0.019eVGPa- 1 respectively), and the initial slope is also in excellent agreement
48 B. Sundqvist

with the data of Sood et al. [207, 208]. A large majority of the optical studies carried
out so far thus agree on a nonlinear decrease in the bandgap energy with increasing
pressure with an initial zero-pressure slope dE /dp = - 0.14 6 0.01 eV GPa- 1 , corre-
sponding to a decrease by 0.1 eV for a 1% decrease in lattice constant. A rapid
decrease in the slope dE /dp with increasing pressure in this range is expected because
of the rapid increase in bulk modulus with p (section 4.2.1) and the strong
nonlinearity rules out a semiconductor-to-metal transition below at least 50 GPa.
Direct theoretical calculations [162, 209]of the bandgap against pressure show that
the absorption threshold energy [162] decreases strongly with an initial slope near
- 0.12 eVGPa- 1 and a rather large curvature, in excellent agreement with the
majority of the experiments.
Meletov et al. [115]also carried out a careful study of the behaviour in the low-
pressure range, where the initial low-energy threshold near 1.7 eV referred to above
can be observed. This feature was found to shift quite slowly with pressure at a rate
dE /dp = - 0.055eV GPa- 1 . Near 2 GPa the two thresholds merge, as shown in
® gure 26, and between 2 and 10GPa a single well de® ned threshold was observed. A
later study [83, 190] showed the same general features but gave a slightly revised
value dE /dp = - 0.071eVGPa- 1 . The low-energy structure and its evolution with
pressure were discussed in some detail in [115], where it was concluded that the
behaviour observed was not connected with the rotational fcc-to-sc transition, nor
with exciton transitions or impurity absorption. Since the disappearance of the low-
energy step coincides with the ® nal stage of the P± H reorientation transition (section
4.1.3), it might be speculated that this structure could be due to the orientational
structure and disorder in sc C60. It is well known that the molecular orientation has a
large e€ ect on the band structure [210, 211]and it is thus possible that the H-oriented
sc phase has a signi® cantly larger bandgap than either fcc or P-oriented sc C60 .
However, it is probably even more likely that the orientational disorder in the sc
phase could either give rise to (localized?) disorder states in the gap which disappear
when the orientational order improves with increasing pressure, or that the disorder
breaks the symmetry and allows otherwise forbidden transitions over the smaller
bandgap near the X point [209].
Above 10± 25 GPa all experiments show signs of a transition into very-high-
pressure polymerized phases as discussed in section 5.5. In the quasihydrostatic
experiment of Snoke et al. [186] the transition occurs at 13 GPa while in the
hydrostatic xenon medium used by Moshary et al. [201]the original phase is stable
to 25 GPa (® gure 27). Meletov et al. [115] observed only a broadening of the
spectrum above 10 GPa, indicating that the sample is never fully transformed up
to their maximum pressure of 19GPa.
Finally, it should be mentioned that a hydrostatic low-temperature (30 K)
experiment up to 1.5GPa using liquid xenon as the pressure medium has also
recently been carried out [206]and yielded pressure coe cients between - 0.045 and
- 0.080eVGPa- 1 for the energies of various exciton transitions near the gap edge.
Although the possible error in these values is large because of the small pressure
range and the large experimental di culties, they are in quite good agreement with
the results of a recent ab initio calculation [212].

4.2.4.2. Positron annihilation. In principle the electronic band structure can be


investigated in detail by positron annihilation, but the high-pressure studies carried
out so far in this area have been of a general exploratory type, exploiting the fact
Fullerenes under high pressures 49

that annihilation preferably occurs in regions of low electron concentration. Jean et


al. [86] studied positron annihilation in C60 powder containing less than 3% C70
under pressures up to 1.5GPa at RT. For the majority of the positrons the lifetime
¿ decreased rapidly with increasing pressure, indicating that they annihilated in the
lattice interstitials, but about 2% of the positrons showed a signi® cantly shorter
and pressure-independent lifetime and might thus annihilate in the cavities inside
the (almost incompressible) molecules. As noted by Sundar et al. [213], the data of
Jean et al. show a clear anomaly near 0.5GPa and a careful analysis of the
temperature dependence at zero pressure also shows a corresponding anomaly near
260K. We may reasonably assume that these anomalies correspond to the fcc-to-sc
transition and we can therefore derive approximate pressure coe cients
d(ln ¿) /dp = - 0. 11 GPa- 1 and - 0.04GPa- 1 in the fcc and sc phases respectively.
(These values di€ er from those given in the original paper because of the di€ erent
assumptions made in the analyses.) The values for d(ln ¿) /dp are similar in
magnitude to the volume compressibilities (section 4.2.1), but it should be noted
that the compressibility of the interstitial sites is larger than this because of the
small molecular compressibility. It is also possible to ® nd an approximate slope of
the phase line as dTo /dp = 65 6 5 KGPa- 1 , an unusually low value re¯ ecting
either the presence of a signi® cant amount of C70 or simply the use of non-
hydrostatic conditions. Schaefer et al. [214± 216] made a single high-pressure
experiment at 3.2 GPa and deduced a signi® cantly smaller average d(ln ¿) /dp =
- 0.019GPa- 1 between 0 and 3.2 GPa, possibly re¯ ecting the high fraction (15%)
of C70 in their sample but came to the same conclusions regarding the annihilation
sites. The di€ erence between the d(ln ¿) /dp values of the experiments may possibly
be due to the use of non-hydrostatic pressure conditions in the experiments, and it
would be interesting to see the experiment repeated under hydrostatic conditions in
order to ® nd out whether the pressure coe cients are truly di€ erent in the fcc and
sc phases.

4.2.4.3. Other studies. A small number of other types of study of the electronic
structure have also been carried out on C60 under high pressures. Hess et al. [203,
204] measured the exciton dynamics in the picosecond range at three pressures up
to 6.2 GPa. Although the zero-pressure results could be ® tted to a stretched-
exponential formula the high-pressure relaxation behaviour was found to be in
better agreement with a power law. The experiment showed that there was no
strong e€ ect of pressure on exciton decay in C60 and these researchers concluded
that the decay is probably not signi® cantly in¯ uenced by hopping or tunnelling
between localized states. Experimental studies have also been carried out using
EPR. Although the C60 molecule should not in principle give any signal a weak
EPR or electron spin resonance (ESR) signal is usually observed, possibly owing to
small amounts of impurities or e€ ects of surface photopolymerization (see section
5). Kempinski et al. [87]show that the g factor has a minimum near the rotational
transition, decreasing rapidly with increasing pressure in the fcc phase and
increasing rather more slowly in the sc phase. Since the molecular motion slows
down from quasifree rotation to a slower ratcheting on passing from the fcc to the
sc phase there is also a rather large increase in linewidth at the transition. Finally,
in addition to their calculations of the optical absorption e€ ects referred to above,
Xu et al. [162] also calculated the dielectric functions ²1 and ²2 and showed that
the static dielectric function ²0 should increase rapidly with increasing pressure as
50 B. Sundqvist

d²0 /dp = 0. 47 GPa- 1, or d(ln ²0 ) /dp = 0. 11 GPa- 1. They also calculated the
electron-energy-loss function as a function of pressure.

4.2.5. Electrical resistivity


Because of the weak interaction between the molecules, pure C60 is a semi-
conductor [3, 20] with an optically measured bandgap of about 1.7 eV. Since the
compressibility is rather high (section 4.2.1), the application of a high pressure
should give a rapid increase in the overlap of the molecular orbitals which might
eventually lead to a closing of the gap and thus a semiconductor-to-metal
transformation. This was ® rst investigated by Nu  nÄ ez-Regueiro et al. [217] who
carried out measurements of the electrical resistivity over a large range in
temperatures and up to 20 GPa. Although the measurements did show the expected
decrease in the bandgap with increasing pressure they also showed that no
metallization occurred over the range investigated. Instead, a transition into an
insulating state was observed near 15± 20 GPa (see section 5). Only a handful of
investigations were carried out later on the conduction properties of C60 [78, 85, 218±
220] and, although there is a general consensus that the bandgap decreases with
increasing pressure, there is no detailed agreement between the results from di€ erent
studies. Some of these di€ erences may be explained by external factors, such as the
fact that both the magnitude of the resistivity and the e€ ective bandgap are very
sensitive to the presence of intercalated gases, especially oxygen [221± 225] but also
argon [226]and organic solvents [225], but there still remain a number of questions
to be answered.
Since the band structure of C60 shows a relatively large bandgap, the electrical
resistivity q might be expected to follow the standard formula for semiconducting
materials:

q = q 0 exp - kEBaT . (4.8)

with an activation energy of the order of the bandgap. (Note that a factor of two
sometimes appears in the denominator and some care is thus needed in evaluating
the literature data.) However, this is not observed in practice. First, the value of Ea
deduced from resistivity measurements is usually between 0.15 and 0.4 eV, almost an
order of magnitude smaller than the optical gap. The low values are often explained
by assuming that carriers are derived from localized disorder or impurity states in the
gap. Also, if the logarithm of q is plotted against the inverse of the absolute
temperature over an extended temperature range, strong nonlinearities are often
found, usually attributed to a strong temperature dependence of Ea caused by
structural and/or orientational disorder [217], to volume expansion e€ ects on the
band structure [222], or to the presence of several conduction mechanisms acting in
parallel [224]. In spite of these anomalies the semiconductor model is still generally
accepted. However, Nu  nÄ ez-Regueiro et al. [219] suggested that the resistivity was
better described by Mott’s 3D variable-range hopping model [227]with
1 /4
T0
q = q 0 exp 2 . (4.9)
T

Plotting log q against T - 1 /4 was found to give a signi® cantly better linearity at all
pressures than the standard equation (4.6) for C60, C70 and C60 I4 and, assuming
Fullerenes under high pressures 51

Figure 28. Electrical resistivity of C60 against pressure at several temperatures. ( Reprinted
with permission from Matsuura et al. [78].)

carriers to be localized on individual molecules, they used their data to deduce a


pressure dependence of the density of states of the materials investigated. Although
this model thus seems reasonably successful, the analysis below will be made in terms
of the standard model in order to compare data from di€ erent sources.
Most high-pressure resistivity studies have been carried out in the sc phase near
and below RT, but fcc and sc C60 have very similar properties with a slightly higher
value for Ea in the fcc phase. The only study on single-crystal C60 is that of Matsuura
et al. [78], carried out on large crystals (several cubic millimetres) in both the fcc and
the sc phases between 200 and 350K and under hydrostatic (oil) pressures up to
1 GPa. The data shown in ® gure 28 show a clear drop in q on going from the initial
fcc phase to the sc phase at high p or low T, with zero-pressure values for Ea of
0.32eV and 0.27 eV in the fcc and sc phases respectively. In the sc phase, little or no
variation with pressure was found over the range investigated, but in the fcc phase Ea
decreased very rapidly with increasing pressure. All other studies have been carried
out using anvil devices, mainly in solid environments, and in most experiments
measurements have not been carried out at low pressures because of the high
resistances involved. Ramasesha and Singh [85] studied powder compacts in the
range 0.6± 1.3 GPa and between 290 and 425K and deduced energy gaps between 0.4
and 0.75 eV (de® ned as in equation (4.6)) with little systematic pressure dependence.
Bao et al. [220], Saito et al. [218]and Nu  nÄ ez-Regueiro et al. [217, 219]all carried out
measurements to very high pressures of the order of 25 GPa. As already mentioned
above, there is no detailed agreement between the results. As shown in ® gure 29,
Saito et al. found an increase in the resistivity between 0.2 and 8 GPa, in contrast
with the low-p results shown in ® gure 28, but between 8 and 20 GPa q decreased
exponentially with increasing q (i.e. log q is a linear function of p). The latter
behaviour is probably typical for C60 under very high pressures since a similar but
52 B. Sundqvist

Figure 29. Electrical resistivity of C60 against pressure. (Reprinted with permission from
Saito et al. [218]. )

more nonlinear behaviour was found also by Nu  nÄ ez-Regueiro et al. and Bao et al.
Finally, above some pressure in the range from 15 GPa [217, 219]to 23 GPa [220], q
starts to increase with increasing p and time, signalling the onset of a partial or
complete transformation into an amorphous insulating phase with a time constant of
the order of hours [219, 220]. The resistance against pressure curves often show
minor kinks, etc., but no clear anomalies have been observed that can be directly
connected with the formation of a polymerized phase at lower pressures near RT (see
section 5), except possibly the anomalous resistance increase found by Saito et al.
below 8 GPa. Referring brie¯ y to section 5 it should be noted that at 8 GPa the
material should already have transformed partially or completely into the ortho-
rhombic polymeric phase.
The RT results of Bao et al. [220] and Saito et al. [218]in the `log-linear’ range
between 8 and 16± 20 GPa show a reasonably similar pressure dependence. The
average decrease in 10 log q is approximately 0.375GPa- 1 in the experiment of Bao et
al. and 0.35GPa- 1 in that of Saito et al., in quite good agreement with the data for sc
C60 from Matsuura et al. (® gure 27). A reasonable preliminary conclusion is
therefore that d(10log q ) /dp = 0.36 6 0. 02 GPa- 1 (or d(ln q ) /dp = 0.83 GPa- 1 ) for
C60 at RT below about 15 GPa and that the observed deviations from this behaviour
are due to the use of non-hydrostatic pressure media. Nu  nÄ ez-Regueiro et al. [217]
found that Ea scales with volume, a behaviour which also agrees within the
experimental uncertainty with that found by Matsuura et al. As pointed out by
the latter, optical studies show that the bandgap has a much stronger pressure
dependence (section 4.2.4.1) which indicates that the energies of the disorder or
impurity states in the gap do not scale with pressure in the same way as the band
structure in general. Again, it should be noted that above 8 GPa the sample should
be at least partially in the orthorhombic polymerized state and it is thus uncertain to
what extent these results are valid for monomeric solid C60 .
Fullerenes under high pressures 53

In contrast with the other groups, Saito et al. [218]deduced the parameter Ea at
3, 6 and 10 GPa from measurements of the temperature dependence of q above RT
and up to 700K. As will be shown in section 5, under these conditions the sample
should de® nitely polymerize into the orthorhombic polymer phase at all pressures
studied, and at the highest temperature a small amount of the denser rhombohedral
phase may also be formed. In fact, the experimental data for q against T do show
very large nonlinearities, but surprisingly these seem to be reversible albeit with some
hysteresis. If the nonlinearities correspond to polymerization into the orthorhombic
phase, the phase transition temperature changes with q such that the transitions
occur at lower T at higher pressures, which is reasonable since in other studies
polymerization into the orthorhombic phase has been observed to occur above
100ë C at 3 GPa [228] and at RT above 6 GPa [229]. However, the observed
reversibility does not agree with the metastability usually observed (section 5). The
values obtained for Ea should thus probably be valid for polymeric orthorhombic
C60 , which thus has a large and almost pressure-independent Ea (0.8 eV at 3 and
6 GPa, and 0.6 eV at 10 GPa). Finally, Ma and Zou [230] carried out a resistance
measurement against temperature at 5GPa to signi® cantly higher temperatures and
found anomalies that can probably be connected with transformations into other
polymeric and disordered phases. This will be discussed further in section 5.4.

4.2.6. Nuclear magnetic resonance


A very complete review of NMR studies on both pure and alkali-metal-doped
C60 has recently been published by Pennington and Stenger [231]. For pure C60, only
one NMR line near 144ppm is observed, and in the fcc phase this line is extremely
sharp because of the rapid molecular rotation. On cooling into the sc phase the line
gradually becomes broader as the molecular motion freezes out and the 60 carbon
atoms sample slightly di€ erent environments [232]. Only one NMR study has yet
been carried out on C60 under high pressures up to 0.5 GPa [76, 77] in the
temperature range between 190 and 340K. In this range a sharp line of width 2±
6 ppm was observed at all temperatures. The temperature of the fcc-to-sc transition
can be determined very easily, since the spin± lattice relaxation time T1 is quite large
and increases with temperature in the fcc phase while it is small near To in the sc
phase and increases very rapidly with decreasing temperature. The data obtained
were used to ® nd an accurate value for the slope of the phase line (see table1). The
results were also used to deduce both the activation enthalpy D Ha for molecular
rotation at constant pressure and the corresponding activation volume D V a at
constant temperature. The former corresponds to the energy barrier to be overcome
when the molecule rotates from one to another orientational position while the latter
corresponds to the additional volume necessary for molecular reorientation. D Ha
increases rapidly with increasing pressure with initial slopes of approximately
0.7 GPa- 1 and 0.22 GPa- 1 in the sc and fcc phases respectively, as might be expected
since the compression of the lattice under pressure makes molecular reorientations
more di cult. D V a is large and independent of temperature in the sc phase but small
in the fcc phase, re¯ ecting the rapid quasi-isotropic rotation. In the latter phase it
increases slowly with increasing temperature.
NMR has also been used to study the e€ ect of oxygen intercalation in the lattice
of pure C60 (section 7.4) as well as various alkali-metal-doped phases (section 8).
54 B. Sundqvist

5. Structures and properties of polymeric C60


5.1. Photopolymerization
As mentioned in section 2, Rao et al. [36]showed in 1993 that thin ® lms of C60
could be polymerized by the action of intense visible or UV light, implying that any
C60 specimen submitted to daylight must contain some fraction of a polymeric phase.
Because C60 absorbs light very e ciently, only ® lms thinner than 10 m m can be
completely polymerized and the limited sample volumes available have made a full
characterization of photopolymerized C60 rather di cult. The polymerized material
is insoluble in common solvents but reverts to normal monomeric C60 on heating to
well above 400K. Photopolymerization is observed only in the fcc phase (above
[ ]
260K) and Rao et al. suggested a 2 + 2 photoaddition mechanism in which double
bonds on neighbouring molecules break up and re-form between the molecules,
forming a covalently bound four-membered intermolecular carbon ring. This
mechanism works only if the molecules rotate freely such that there is a small
probability that two double bonds come into su ciently close contact for this to
occur. The resulting decrease in molecular symmetry leads to the appearance of a
large number of previously forbidden lines in the IR and Raman spectra of C60 .
Recent Raman [233] and atomic force microscopy (AFM) studies [234] also
suggested that two di€ erent photopolymerized states may exist. Photopolymeriza-
tion at temperatures above 350K seems to produce mainly dimers, while material
polymerized at 320K and below contains larger clusters. These suggestions have not
yet been veri® ed because, although dimers and polymers should give di€ erent optical
and Raman spectra, no detailed information on these di€ erences is yet available.
However, a recent report that pure (dimeric) C120 has been synthesized [235]
promises that the spectrum of dimeric C60 will soon be available for comparison
purposes. The properties of photopolymerized C60 and the dimer± polymer question
will be discussed further in section 5.3 below in connection with low-pressure
polymerization of C60 into an orthorhombic polymeric phase.

5.2. Early very-high-pressure studies


After the suggestion by Ruo€ and Ruo€ [135, 136] that compressed C60 might
have extreme mechanical properties, several exploratory very-high-pressure studies
were carried out. Already in the ® rst of these, Duclos et al. [49]observed a transition
into a disordered phase above 16 GPa and also found that the compression
properties were very di€ erent under di€ erent experimental conditions, indicating
that di€ erent phases were created under hydrostatic and non-hydrostatic conditions
(® gure 15). Nu nÄ ez-Regueiro et al. [217], looking for a transition into a metallic state,
veri® ed the existence of a transition between 15 and 20GPa but found the high-
pressure phase to be insulating. In later studies [236, 237] they found evidence that
the material produced above 20GPa consisted of polycrystalline diamond. Tolbert et
al. [187]saw no such e€ ects up to 18 GPa but concluded that an orientational glass
was formed under high pressures, and Yoo and Nellis [188]found a reversible partial
transition into a semitransparent phase in the range 15± 25 GPa, a transition which
became irreversible above 27 GPa. Material recovered from these pressures was
diamond like and described as forming large disordered agglomerates of C60
molecules. Several attempts to characterize these phases by spectroscopic and
structural methods were carried out, but the results were rather disappointing in
that the materials appeared almost featureless. X-ray di€ raction [49, 149, 152, 238,
239], both in situ and on recovered metastable high-pressure-treated material,
Fullerenes under high pressures 55

showed the appearance of a di€ use ring characteristic of an amorphous phase at 10±
25GPa, or on decreasing pressure below 6 GPa. Optical absorption studies showed
an abrupt transition to either a transparent [201]or a black [185, 202]phase in the
same range, while Raman or IR studies usually showed a gradual broadening and
disappearance of spectral lines with increasing pressure in the range below this
(section 4.2.2). Also, no lines characteristic of crystalline diamond, C60 , or graphite
were usually observed [201], but sometimes a broad band typical of amorphous
carbon [185, 186, 188, 240]. Early work in the very-high-pressure ® eld has been
brie¯ y reviewed by Nu  nÄ ez-Regueiro et al. [11, 12].
After polymerization of C60 had been observed at atmospheric pressure, it was
realized that similar mechanisms might be active under pressure. While carrying out
spectroscopic studies on C60 at RT, Yamawaki et al. [229]found that, above 6 GPa,
new lines and bands appeared. They suggested that these resulted from a lowering of
the molecular symmetry due to polymerization under pressure and reported that the
new lines disappeared after heating the sample to above 400K in vacuo. It is not yet
clear whether the e€ ects observed were induced by the high pressure alone or by the
e€ ect of the excitation laser. Somewhat later, Iwasa et al. [241]reported that heating
C60 under a pressure of 5GPa resulted in signi® cant changes in its spectroscopic
properties and made the material insoluble in common solvents. They suggested that
at least two di€ erent polymeric phases could be formed by heating under a high
pressure. At low temperatures a strongly disordered phase, believed to be basically
fcc but with some distortion of the unit cell (`soft fcc’ ), was formed, while at higher
temperatures a harder, rhombohedral (rh) structure was observed. The materials
were characterized by Raman and IR spectroscopy, X-ray di€ raction and NMR,
and in all cases the tests showed new features which agreed with the suggested
changes in structure from monomeric to polymeric C60 . Both materials also reverted
to normal C60 on heating; for the rh material this occurred near 500K.
Several groups have since explored the phase diagram of C60 more systematically,
usually by heating samples to a series of di€ erent maximum temperatures at a
number of pressures, then quenching back to room temperature and investigating
the resulting materials, and a number of interesting phases have been discovered.
These will be discussed in the following sections. To obtain some degree of logical
order, these sections will be subdivided according to the pressure range studied, and
within each section the discussion will if possible start with RT studies and progress
towards higher temperatures. The ranges chosen partly re¯ ect the di€ erent tech-
niques used and partly also delineate di€ erent phase regions, although a signi® cant
degree of overlap exists. To name but a few samples of the work carried out, in
alphabetical order, Bashkin et al. [147, 242]investigated the low-pressure limits for
formation of the `soft fcc’ polymerized phase, while Blank et al. [114, 242]reported
that material formed at very high pressures and temperatures was harder than and
easily scratched natural diamond, and they later carried out a very complete survey
of the structures and phases of C60 at very high pressures and temperatures and
investigated the properties of the various phases found [244]. Kozlov and co-workers
carried out extensive studies on `fcc’, rh [228]and amorphous [245]materials formed
at high temperatures in the intermediate pressure range 2± 3GPa, and Nu  nÄ ez-
Regueiro et al. made systematic structural studies, suggesting a ® rst high-pressure
phase diagram [12]and comparing measured di€ raction patterns with the results of
numerical simulations [246]. The results from these and many other studies have
56 B. Sundqvist

recently been used to synthesize a preliminary high-pressure phase diagram for


polymerized C60 [247].

5.3. Low-pressure polymerized material ( p < 2GPa)


Starting at the low-pressure end, it should ® rst be commented that intentional
photopolymerization has only been carried out at zero pressure, although unin-
tended e€ ects can be observed whenever visible light is used for excitation in Raman
and other studies. However, there is every reason to assume that photopolymeriza-
tion can also be carried out under pressure, at least in the fcc phase.
As already mentioned above, Bashkin et al. [147, 242] carried out compression
studies on C60 above RT and observed a large abrupt change in volume when the
material was compressed to above 1 GPa at temperatures exceeding 500K. Further
studies showed that the volume change was connected with a change in structure into
an insoluble and thus probably polymeric phase and also enabled Bashkin et al. [147]
to deduce the `phase map’ shown in ® gure 30. The new polymer phase is created
whenever C60 is compressed and heated into the range above about 1 GPa and
reverts to normal monomeric C60 on heating to above the upper transition
boundary. However, since both the formation and the breakdown of the polymer
are thermally activated processes, the actual upper limit depends on time and heating
rate and, in order to obtain well reacted material, reaction times of several hours are
often needed [248]. The broken `equilibrium’ phase line has a slope dT /dp =
387KGPa- 1 and extrapolates to near 295K at zero pressure. It should also be
noted that this phase was probably ® rst observed by Samara et al. [71]in their DTA
studies of the fcc-to-sc phase line, but they concluded that the anomalies observed
were due to the formation of a C60 ± pentane clathrate compound.
Because the pressure range up to 2 GPa is easily accessible with standard piston-
and-cylinder devices and samples of masses up to several grams can be treated

Figure 30. Phase map for low-pressure polymerization of C60. The material polymerizes on
increasing pressure through the right-hand curve and depolymerizes if heated through
the upper (left-hand) curve. (Reprinted with permission from Bashkin et al. [147].)
Fullerenes under high pressures 57

Figure 31. NMR spectrum for pressure-polymerized C60 . ssb denotes spinning side band;
the main peak near 145ppm is the sp2 peak typical for normal C60 while the peak due
to sp3-type intermolecular bonds is indicated by a vertical line near 77ppm.
(Reprinted from Persson et al. [248].)

without di culty, the physical properties of the low-pressure-polymerized phase


have recently been investigated by several groups, and a number of review papers
[247, 249± 253] have been published. First, the molecules are de® nitely linked by
covalent bonds in this phase. This was proposed both by Rao et al. [36] for
photopolymerized C60 and by Iwasa et al. [241] for pressure-polymerized material
on the basis of spectroscopic data, but de® nite proof was given by the NMR results
of Persson et al. [248, 254]shown in ® gure 31. In addition to the usual sp2 peak near
145ppm, which seemed to be split into several unresolved subpeaks, these data also
showed a small peak near 78 ppm indicating the presence of a small fraction of
(intermolecular) covalent sp3-type bonds. Almost identical results with peaks at 146
and 73.5 ppm have also been obtained by Rachdi et al. [255, 256]for similar material
produced at higher pressures. Surprisingly, the exact lattice structure of this low-
pressure phase has been quite di cult to deduce and, as discussed above, the phase
has long been referred to as fcc (pC60) , as designated by Iwasa et al., or as `soft fcc’.
The very small sp3 peak in ® gure 31 shows that the number of intermolecular bonds
per molecule is small, which agrees with early suggestions that linear molecular
chains are formed. Bashkin et al. [147] showed di€ raction patterns both for newly
reacted material and for partially depolymerized (ground) material, while Persson et
al. [248] showed the evolution of the di€ raction diagram with time during
polymerization and Moravsky et al. [249]the same type of data during depolymer-
ization. However, in neither case could the structure of the polymerized phase be
deduced. Both groups suggested that the structure might be orthorhombic but other
related structures could not be excluded. Later structural studies have indicated that
the low-pressure polymerized material is often very disordered, and di€ raction
studies can usually be interpreted in terms of mixtures of orthorhombic (orh), rh
and tetragonal (tg) phases [251, 257, 258]. For a phase assumed to consist of linear
chains of polymerized molecules a possible explanation may be that random lengths
and directions of the chains create disorder that makes identi® cation impossible.
Random disorder is not unexpected, since polymerization is almost always carried
out by heating a sample in the fcc phase, with almost freely rotating molecules and
thus no preferred bonding directions. (We return to this question in the next section.)
X-ray di€ raction studies indicate that the nearest-neighbour distance decreases to
about 9.2 AÊ , but the macroscopic increase in density varies between about 1% [248]
and 15% [147], indicating a variable degree of polymerization or polymerization-
induced amorphization. In principle, mass spectroscopy might be able to de® ne an
average cluster size or chain length for di€ erent types of polymer but so far this has
58 B. Sundqvist

not been possible since the methods used for vaporization may themselves either
create or break down polymer chains and clusters [251].
Although almost all experiments show the pressure-treated material to be very
disordered, it was recently reported [250] that large single crystals could be
polymerized into well ordered states if carefully reacted under hydrostatic conditions
in silicone oil at 1.1 GPa and 550± 585K. Recent structural studies both on crystals
[259, 260] and on powder samples [261, 262] treated at somewhat higher pressures
and temperatures have clearly shown that the initially cubic lattice is transformed
into an orthorhombic structure by the formation of linear chains along the initial
cubic (110) directions (the face diagonals). Because all such directions are equivalent,
the original single crystals contained a number of di€ erently oriented domains after
polymerization, but, since each domain is rather large (much greater than 20 nm) and
has one of only 12 possible orientations relative to the original lattice, a detailed
structural analysis is possible. The existence of an orthorhombic phase at higher
pressures was reported earlier by Nu  nÄ ez-Regueiro et al. [246](to be discussed further
in next section) but, surprisingly, the phase formed at 1± 1.5GPa does not seem to
have the same structure as the high-pressure phase. To distinguish between the two
Â
phases, I shall denote the low-pressure phase orh and the high-pressure variety orh,
Â
in analogy with the terminology O and O used by Agafonov and co-workers [261,
262]. The study of the powder samples [261]indicated that the lattice parameters in
the orh phase were di€ erent from those in the previously identi® ed orh phase, with a
Â
small in-chain intermolecular distance of about a = 9.1 A Ê and with b = 9.8 A Ê and
c = 14.7 AÊ , but could not determine the orientational angle of the chains around the
axis. A later theoretical calculation [262]gave a most probable value of near 61ë for
this angle relative to the b axis. The single-crystal studies [259, 260] gave similar
Ê , b = 9.90A
lattice constants, a = 9.14 A Ê and c = 14.66AÊ , corresponding to a volume
decrease by 6.5%, and could resolve the structure in detail. Polymerization conserves
one of the original close-packed cubic (111) planes and one [1110] Å direction, which
implies that the change in structure occurs through a small shift (0.82A Ê ) of the (111)
planes and a small reduction (0.23A Ê ) in the interplane distance. (The energy cost of
the shift in the (111) plane is low in the fcc structure, resulting in a high density of
stacking faults in both C60 and C70. ) The results veri® ed that the plane of the
intermolecular bonds (four-membered rings) was not parallel to any side of the unit
cell and that the angles of neighbouring chains were related by mirror planes, giving
a Pmnn symmetry. However, it was not possible to identify ® nally the angles since
the data could be equally well described by both a 45ë angle and the angle suggested
by Davydov et al. [262], corresponding to 29ë from the c direction.
Like the photopolymerized material, pressure-polymerized C60 has been exten-
sively studied by spectroscopic methods. The formation of covalent intermolecular
bonds breaks the molecular symmetry and allows a large number of new IR and
Raman lines or bands to appear [228, 241, 248, 251, 257, 258, 263± 265], for example
in the range 800± 1000cm- 1 . This is shown very clearly in the IR data in ® gure 32
[228, 263]. Although Raman studies were also carried out by Iwasa et al. [241] and
Kozlov et al. [228, 263], I have chosen to show in ® gure 33 the results of Persson et
al. [248] which have a better signal-to-noise ratio. These data were obtained on
pressure-polymerized material using near-IR (Nd-doped yttrium aluminium garnet
laser) excitation to avoid unintentional photopolymerization of the material. Similar
data were also shown by Szwarc et al. [264] and more recently by Rao and co-
workers [251, 257, 258], Davydov et al. [265] and Persson et al. [266]. Some of the
Fullerenes under high pressures 59

Figure 32. IR spectra at zero pressure for pristine C60 and for C60 treated at 3 GPa and the
temperatures (in kelvins) indicated. ( Reprinted with permission from Kozlov et al. [228].)

Figure 33. Raman spectra for pristine and pressure-treated C60. ( Reprinted from Persson et
al. [248].)

new lines appearing in the spectra for the polymers are previously forbidden C60
lines, now permitted by the decrease in symmetry, while others are truly new and
characteristic for the newly formed intermolecular bonds, dimers, chains or other
types of cluster. No complete identi® cation or analysis of all lines observed has yet
been carried out, but Rao et al. [257]have given a rather extensive analysis of both
IR and Raman spectra of several di€ erent types of C60 polymer. Because poly-
merization occurs by breaking double bonds on the molecules, which leads to a
lower average bond sti€ ness, most high-frequency (intramolecular) peaks shift to
60 B. Sundqvist

lower frequencies on polymerization. New bands appearing at very low frequencies


below 200cm- 1 are of particular interest, since they correspond to intermolecular
vibrations in polymeric chains. Persson et al. [248, 266]reported observations of such
modes at approximately 97, 119 and 173cm- 1 and Szwarc et al. at 118cm- 1 ,
frequencies in very good agreement with results from theoretical calculations [267,
268]of the Raman frequencies of C60 polymers. Also, it is now well known that the
Ag (2) `pentagonal pinch’ mode shifts downwards from 1469 to 1459cm- 1 on
polymerization, whether by radiation or pressure treatment, and the relative
intensities of these two components can be used to estimate the fraction of
polymerized material which for samples polymerized in this pressure range is rarely
100% even after very long times or for samples seemingly completely insoluble in
common solvents. (Note in table 6 above that the application of pressure to pristine
C60 instead shifts the frequency of this mode upwards.)
In section 5.1 it was suggested that under certain circumstances a dimer phase
could be obtained through photopolymerization. It is, therefore, interesting to note
that the Ag (2) mode of partially polymerized samples often shows a strong
component with a smaller shift, by - 6 cm- 1 to 1463cm- 1 [248], and that the
intensity of this new line then decreases again with increasing reaction time until
the `normal’ polymeric line at 1459cm- 1 becomes dominant after long times. A
comparison with the calculations by Porezag et al. [268]suggests that this low-shift
component may be due to the initial formation of C60 dimers, which later coalesce to
form longer chains or larger cross-linked clusters. This view has recently received
strong support. Burger et al. [233] used the theoretical results mentioned above to
correlate the presence of dimers in photopolymerized C60 with the appearance of a
new Raman line at about 96 cm- 1 and a shift of the Ag (2) mode to 1462cm- 1 , while
longer polymer chains cause a peak to appear near 119cm- 1 . Our studies of
pressure-polymerized materials [248, 266, 269] showed the same features. Samples
reacted for short times show lines at 96 and 1463cm- 1 , indicating the probable
presence of dimers, while samples reacted for longer times and thus presumably
containing longer chains or larger clusters show instead lines at 119 and 1459cm- 1 .
In another very recent study, Davydov et al. [265]found the same 96 cm- 1 Raman
line in samples reacted at 1.2 GPa and 623± 723K for a very short time (10 min). The
structure of these samples can be indexed as either fcc or orh, in excellent agreement
with the data of Persson et al. [248], Moravsky et al. [249]and others, and they also
concluded that the sample contains a high concentration of dimers. Finally, very
recently, Y. Iwasa (1997, private communication) has used a template synthesis
method to produce high-pressure-reacted material with an IR spectrum identical
with that for pure C60 dimers [235]. Again, the observed characteristics of the Raman
spectrum agrees with those reported above for dimers.
UV and visible-light absorption studies and luminescence studies have also been
carried out on the polymer [228, 251, 258, 263], even in single-crystal form [270], at
ambient pressure. These studies show a semicontinuous evolution with reaction
temperature that indicates a gradual increase in average molecular deformation with
increasing degree of polymerization, as might be expected. Photoluminescence and
photoconductivity studies [251, 258, 263, 270] also indicated that the bandgap of
polymerized material is slightly smaller than for the pristine material. As already
discussed in section 4.2.5, resistivity studies by Saito et al. [218] indicated that the
bandgap of orh C60 depends little on pressure, possibly re¯ ecting a much reduced
Fullerenes under high pressures 61

compressibility compared with monomeric C60 . However, we must remember that it


is still uncertain whether resistivity studies actually re¯ ect the true bandgap or not.
The change in character and strength of the intermolecular bonds from basically
van der Waals (or p orbital overlap) type to covalent is re¯ ected in signi® cant
changes in all thermophysical properties determined by the intermolecular potential
[250, 271]. The thermal expansion coe cient a [250, 271, 272]decreases by a factor
of about two compared with pristine C60 and B increases signi® cantly [147] to
32.9GPa (see table 2). The high-temperature thermal GruÈneisen parameter
g = a B /dcp thus increases slightly compared with that of pristine C60 . Adiabatic
calorimetry showed [250, 271] that the low-temperature speci® c heat capacity cp
dominated by long-wavelength acoustic phonons decreases, indicating an increase in
e€ ective Debye temperature by about 30%, but except for the obvious absence of a
transition peak near 260K the speci® c heat is otherwise very similar to that of
pristine C60 . Data from other sources [147], however, reported much larger
di€ erences. Surprisingly, in spite of the increase in intermolecular bond strength
the magnitude of the thermal conductivity does not di€ er much from that of pristine
C60 [273± 275], as shown in ® gure 34. The reason seems to be very strong scattering of
phonons in the polymerized material, which is due either to a randomly disordered
chain structure or to phonon scattering on remaining unreacted C60 molecules, such
that the material behaves like a very much disordered material or glass. This is
re¯ ected in the positive slope of the thermal conductivity against temperature for the
polymer shown in the ® gure. Very recently, several groups have also carried out
inelastic neutron scattering studies on polymeric C60 [275± 278]. Since the material is
polycrystalline, only the density of phonon states can be obtained. Results for
pressure-polymerized C60 and KC60 [277] were almost identical, and all groups
agreed that the data for the former di€ ered from data for pristine C60 mainly in the
appearance of a new band of translational and librational intermolecular phonons

Figure 34. Thermal conductivity of pristine ( , ) and polymeric (s ) C60 near 0.3GPa.
(Reprinted from Sundqvist et al. [273].)
62 B. Sundqvist

between 10 and 22meV and small shifts of the lowest phonon and libron bands
towards higher energies. A splitting of the radial intramolecular modes was also
observed [275, 276, 278].
ESR studies [279] showed the material to be very similar to monomeric C60
except for an increase in signal intensity two to three orders of magnitude which was
very di cult to reconcile with covalent intermolecular bonds. It might be that a large
fraction of the ESR signal observed in pristine C60 was actually due to covalent
surface bonds created inadvertently while handling the sample in a well illuminated
environment, a view supported by the fact that a much lower signal level was found
for heat-treated (reverted) polymer than in nominally pristine samples.
As mentioned above, the low-pressure-polymerized and photopolymerized
phases revert to `normal’ monomeric C60 on heating (see ® gure 30). Several studies
of the high-temperature breakdown of polymeric C60 have been carried out in order
to estimate the activation energy involved. For photopolymerized C60 , Wang et al.
[280] used Raman scattering to estimate the temperature dependence of the
polymerized fraction and found an activation energy of 1.25 eV for the dissociation
of dimers, while Robert et al. [281] found an activation energy of 1.1 eV for the
breakdown of the linear polymer KC60 from ESR studies. For the pressure-
polymerized material the results di€ ered signi® cantly between di€ erent groups or
di€ erent methods. Moravsky et al. [249]found a value of 2.3eV from measurements
of cp, while Nagel et al. [272]measured the thermal expansion and deduced an initial
activation energy of 0.75eV near 400K, increasing to 1.9 eV near 500K. WaÊ gberg et
al. [282]studied the breakdown of the polymeric phase using Raman scattering and
found an initial energy of 0.75eV, assumed to correspond to the initial large clusters,
increasing to about 0.9eV for the ® nal small clusters or dimers in each run. However,
the recent report by Wang et al. [235]that the pure dimer C120 dissociates in a fairly
well de® ned temperature range between 425 and 445K suggests instead that the
initial e€ ect observed is the breakdown of dimers near 430K with an activation
energy of 0.75 eV, while the larger values for the activation energy are those for the
more stable longer chains which need long-time (1h) treatment at 470± 500K in order
to break down completely [272]. An inverse study has been carried out by Soldatov
and co-workers [250, 266, 274] who studied instead the polymerization in situ
through thermal conductivity measurements at 0.8 GPa, where the intermolecular
interaction should be much stronger than at zero pressure. Preliminary data
indicated that the activation energy is then much lower, about 0.4 eV [266, 274].
It is obvious from the discussion above that the zero-pressure-photopolymerized
state and the orthorhombic polymeric phase(s) of pressure-polymerized C60 must be
very similar. C60 doped with alkali metal A, namely AC60 (section 8), also forms an
orthorhombic structure which is very similar to these two, although because of the
presence of alkali metal ions there can be no detailed agreement between the lattice
structures of AC60 and pure C60. However, Raman studies by Winter et al. [283]
showed that the Raman spectra of AC60 and pressure-polymerized C60 were very
similar between 200 and 1600cm- 1 , indicating that the basic C60 polymer structure
of the two materials were very similar and that orh AC60 can be considered as doped
orh C60 . More recently, Wa Ê gberg and co-workers [266, 282] compared the Raman
spectra of nominally identical thin C60 ® lms, polymerized using either pressure or
radiation treatment. None of the ® lms was completely polymerized, as deduced from
the shift of the Ag (2) mode, but ® lms selected to have similar fractions of
polymerized material showed almost identical Raman spectra between 200 and
Fullerenes under high pressures 63

Figure 35. Raman spectra from 200 to 1600cm- 1 for photopolymerized C60 (curves (a))
pressure-polymerized C60 (curves (b)) and pristine C60 (curves (c)). (Reprinted from
WaÊ gberg et al. [282].)

1600cm- 1 , as shown in ® gure 35. Below 200cm- 1 the relative intensities di€ ered, but
even here the same lines could be observed in both experiments. Similar results were
obtained by Rao and co-workers [251, 257, 258]although no information was given
on the polymerized fractions. The preliminary conclusion must be that the pressure-
polymerized orh phase (or low-pressure-polymerized mixed phase, see above) is
Â
basically identical with photopolymerized C60 , but that the average cluster size or
orientation di€ ers because of the di€ erent routes used in the polymerization
processes. Marques et al. [284]argued that photopolymerized material must always
be more disordered than high-pressure-polymerized material since photopolymeriza-
tion depends on the random rotational molecular motion to bring double bonds into
contact, while in high-pressure polymerization the molecules are usually brought
together in or close to the (H) orientationally ordered phase and polymerization is
carried out by exciting intense librational motion by heating. However, while this
model may be correct at pressures well above 1GPa, it is probably not applicable to
the polymers formed in the range discussed in this section.
A large number of theoretical calculations of the properties of polymeric C60,
formed by either photopolymerization or pressure polymerization, have also been
carried out, and only a small number of recent examples will be mentioned. As
already mentioned brie¯ y above, Adams et al. [267] calculated IR and Raman
frequencies for both dimers and an in® nite linear polymer chain and found
reasonable agreement with experimental data. Harigaya [285, 286] and Tanaka et
al. [287] carried out band-structure calculations for linear chains and found a
bandgap near 1.1eV, about two thirds that in monomeric C60 , in reasonable
agreement with the optical studies mentioned above. Harigaya also suggested that
the application of pressure would result in a decreasing bandgap, as indeed observed
for monomeric C60 (section 4.2.4.1), followed by an insulator-to-metal and a metal-
to-insulator transition at some high pressures. However, no such transitions have yet
been observed. Doping [286, 288] by one electron per molecule is found always to
give a metal, while doping by two electrons per molecule gives an insulator. Harigaya
[286]also studied the magnetic properties of linear chains. StafstroÈm and co-workers
have calculated the electronic levels of dimers [289, 290] and trimers [291] and
64 B. Sundqvist

obtained both theoretical absorption spectra in qualitative agreement with experi-


ments and a good understanding of the stability and polymerization process of C60 .
As already mentioned above, Porezag et al. [268, 292] calculated IR and Raman
vibrational frequencies of both dimers and higher oligomers, in both straight and
bent con® gurations as well as in a four-molecule square cluster. Both low- and high-
frequency modes were in good agreement with experiments. Finally, Springborg
[293] carried out an extensive calculation of electronic, structural and vibrational
properties and gave references to earlier work.
Â
As discussed above it is not yet clear why most orh samples produced at
pressures below 2 GPa are strongly disordered in spite of the fact that large single
crystals can sometimes polymerize into an ordered state [250, 259, 260]. A possible
explanation given by Marques et al. [284]in terms of molecular rotational disorder
during polymerization has already been discussed above. Further polymerization
experiments are in progress to correlate the structures obtained with the reaction
conditions used, but little progress has been made so far [269]. As mentioned above,
Â
an orthorhombic phase probably structurally di€ erent from the orh phase discussed
here can also be produced at higher pressures and further studies of this phase will be
discussed below in next section. Before closing this section, however, it should be
noted that other structural phases have also been observed to form at pressures
below 2 GPa. The low-pressure studies of Bashkin et al. [147, 242]did not extend to
temperatures higher than 700K (® gure 30). As will be discussed below, it is well
known that treatment at higher temperatures above 2GPa results in the formation of
two-dimensionally polymerized material with a rh structure, usually containing a
signi® cant fraction of a phase with a tg structure [284]. Sundqvist et al. recently
treated C60 at 1.5 GPa and 800± 900K and showed that at 800K a disordered, mainly
rh phase with all the spectroscopic characteristics expected for a two-dimensionally
polymerized material was obtained [294, 295]. Treatment at 900K, on the other
hand, did not result in polymerization but kept the pristine fcc structure unchanged.
These results are in excellent agreement with the phase diagram in ® gure 30 since the
two points discussed straddle an extrapolation of the upper phase line in this ® gure.
The phase diagram in ® gure 30 can thus be extrapolated to temperatures near 900K.
Further experiments were carried out near 2 GPa with the intention of producing
larger amounts of the disordered rh phase for studies of its physical properties, but
these will be discussed below. In order to test whether C60 can be polymerized under
completely hydrostatic conditions, powder samples were also treated in both N2 and
He gas near 1.2 GPa and 600K. Preliminary results from these studies [294, 295]
showed that, while polymerization in N2 gave results very similar to those obtained
in oil or solid media, treatment in He gas resulted in the formation of a cubic
structure whose Raman spectrum showed the characteristic Ag (2) peak of a
completely polymerized material. These results are not yet understood.
Finally, for completeness, I note that attempts have been made [296] to
intercalate the polymeric phase obtained in this pressure range with alkali metals.
However, all studies so far have shown that on intercalation the polymer breaks
down into intercalation compounds of monomeric C60.

5.4. Intermediate-pressure range (2± 8 GPa)


The orthorhombic phase discussed in the last section is usually produced by
heating C60 to well above RT at pressures above 0.8GPa, or by radiation with visible
light. However, several groups have also reported the formation of a polymer at RT
Fullerenes under high pressures 65

at higher pressures, especially under non-hydrostatic conditions. The compression


studies of Duclos et al. [49] (® gure 15) gave very di€ erent results under hydrostatic
and non-hydrostatic pressures, probably because of the formation of a polymeric
phase. Because an increase in volume during polymerization is thermodynamically
unlikely, we can extrapolate their results under non-hydrostatic conditions to lower
pressures and assume that a polymeric phase started to form where the extrapolated
curves cross the results under hydrostatic conditions. This procedure gives a
polymerization pressure of 6 6 1 GPa. As already discussed in section 4.2.1, the
slopes of the extrapolated curves indicate a very uncertain value B > 200GPa (table
1) for the average bulk modulus between 5 and 20 GPa in this phase, agreeing to
within the order of magnitude with the predictions of Ruo€ and Ruo€ [135, 136].
Yamawaki et al. [229, 297] reported the appearance of new spectral features in the
range 600± 800cm- 1 after submitting C60 to 4± 6 GPa under hydrostatic conditions,
and also that these features disappear after annealing of the samples at 473K. Mass
spectrometry [297] of the pressure-treated material also showed clearly a large
fraction of (C60 )n clusters with n< 6. As will be discussed below, the observed
depolymerization temperature indicates that the polymerized material was in the orh
phase.
Very non-hydrostatic conditions can lead to a transformation in C60 at RT at
even lower pressures. Kokorevics et al. [298]carried out combined compression and
rotational shear deformation studies on C60 at 3.3 GPa at RT and found a partial
transformation into a hard phase that scratched the tungsten carbide anvils used.
Similar results were obtained by Lundin et al. [90] who, instead of the expected
lubricating action of C60 , observed scratches in an extremely hard steel cylinder after
compression of C60 to only 1GPa without gaskets, probably because of transforma-
tion into a hard phase by shear in the space between pistons and cylinder. UR, IR
and Raman studies by Kokorevics et al. showed that the hard phase created under
shear transformation was probably amorphous, while material treated to 6 GPa in
the same anvil device without rotational shear was almost unchanged.
Several groups have treated C60 samples at elevated temperatures in this pressure
range. Iwasa et al. [241, 253, 299] reacted samples at temperatures from 300 to
1000ë C (from 573 to 1273K) at 5 GPa and found that all materials treated at or
below 800ë C were insoluble after treatment but reverted to normal after high-
temperature vacuum annealing. Material reacted at 1000ë C was found to consist of
amorphous C. The X-ray di€ raction patterns found after the reaction at 300 and
400ë C were very similar to those observed by Persson et al. [248]and discussed above
for partially reacted orthorhombic C60 , although Iwasa et al. preferred to index the
broadened peaks as fcc with a lattice constant 3.7% lower than at normal pressure
and used the term fcc (pC60 ) for this material. IR and Raman spectra were also very
Â
similar to those for the orh, orh and phototransformed phases [299], showing new
bands in the range 700± 800cm- 1 and a downward shift in the high-frequency modes.
Although the reported NMR spectrum di€ ers from that obtained in most other
studies in showing a broad peak near 125ppm, there is now no doubt that the
material produced had an orthorhombic structure, as identi® ed by Nu  nÄ ez-Regueiro
and co-workers [246, 284]. Reaction at 500± 800ë C, on the other hand, resulted in a
denser rhombohedral (rh) structure (rhpC60) suggested to originate in an elongation
of the fcc lattice along the [111] direction, probably because of complete
polymerization in the close-packed k 111 l planes (see ® gure 36 below). The formation
of six covalent in-plane bonds from every molecule results in a decrease in the
66 B. Sundqvist

molecular dimension by about 5% perpendicular to the polymerized planes [300]and


a nearest-neighbour distance near 9.2A Ê . As might be expected from the large
deformation of the molecules in this phase, new IR and Raman peaks appear and
the remaining original peaks shift even farther from their positions in pristine
material than for the phases discussed in the last section. As one example, the
Ag (2) mode is reported to shift to 1447cm- 1 , compared with 1459cm- 1 in the orh Â
phase and 1468cm- 1 in the pristine material. At zero pressure the rh phase was
found to be stable to higher temperatures than the orh phase, and to break down to
monomeric C60 with a well de® ned DSC peak at 270ë C (543K) [299].
The phase diagram and the structure of the polymerized phases observed to form
in the pressure range 3± 8 GPa have been discussed in some detail by Nu nÄ ez-Regueiro
and co-workers [12, 246, 301, 302]. A rh majority phase is obtained over large ranges
in temperature and pressure [284], but simulations [246]also showed that the X-ray
di€ raction results could often be best described if it was assumed that a tg minority
phase (15± 35%) was present at high temperatures. Both phases reverted to normal
after long-time anneal above 200ë C. The structural models used are shown in ® gure
36, where it is obvious that the orh, rh and tg structures can be derived from the
initial fcc or sc lattice by polymerization along suitable directions. As discussed
Â
above, the orh phase probably di€ ers from the orh phase discussed above by a
di€ erent rotational angle of the chains. Also, it has been suggested that the tg-
structure might not, in fact, exist as a stable structure but may be simulated by
intertwined, locally orthorhombic lattices [302]. High-temperature studies revealed a
transition to a graphite-like phase above 750ë C at 2.5 GPa [237]. A very systematic
study of the phase diagram was carried out by Marques et al. [284], who submitted

Figure 36. Overview of intermediate-pressure polymer phases of C60 and their relationship
to the fcc structure. (Reprinted with permission from Marques et al. [302]. )
Fullerenes under high pressures 67

Figure 37. Phase diagram of C60 between 3.5 and 8.5GPa. (Reprinted with permission from
Marques et al. [284].)

about 50 samples to various pressures between 3.7 and 8 GPa and temperatures in
the range 20± 900ë C for 1 h and found the phase diagram shown in ® gure 37. It is
obvious that the structure obtained depends more on the reaction temperature than
on the pressure, with basically unchanged sc C60 with some orthorhombic distortion
at room temperature, orh C60 between 150 and 350ë C, and a rh phase above.
Interestingly, the structural evolution seems to be almost continuous in spite of the
di€ erence in topology between the orh phase, built up from one-dimensional (1D)
chains, and the rh phase, built up from two-dimensional (2D) planes. Such a
continuous evolution was also found in the optical spectra of orh and rh C60 by
Kozlov et al. [228, 263]. Marques et al. reported that the structural disorder was
always larger in the orh phase than in rh C60, probably because defects form more
easily in a chain-based polymer than in a planar polymer. In addition to the rh phase
a tg phase is formed at high temperatures near 4 GPa, and above 800ë C an
amorphous graphite-like (sp2 type) but very hard phase is formed. This amorphous
phase will be discussed further near the end of this section. The main features of the
phase diagram in ® gure 37 agree with the results from most other recent studies [244,
246, 264, 303± 305] and the range of stability of the rh and rh + tg phases can be
extrapolated to lower pressures, as shown recently by Sundqvist et al. [295] and
Davydov et al. [265].
A number of questions regarding the phase diagram in ® gure 37 remain,
however. Davydov et al. [265] recently reported observations of both hexagonal
and cubic structures, the former at very high temperatures, and the latter at low
temperatures, at both low and very high pressures. It is possible that the hexagonal
phase is related to graphite (see section 5.6.1) and the high-temperature cubic
structure to the cubic polymeric phases observed above 9 GPa (see section 5.5).
The cubic phases observed at low pressures and temperatures probably indicate that
the samples are not fully reacted after 10 min; as discussed above, Persson et al. [248]
reported that treatment times of several hours are often necessary at low pressures.
68 B. Sundqvist

On the other hand, Sundar et al. [306] reported that all their samples reacted near
800ë C and up to 7.5 GPa were orh, in spite of reaction times of several hours.
Returning to the low-pressure end of ® gure 37, Sundqvist et al. [294, 295, 307]
recently called attention to the fact that the boundary between the `pure’ rh and the
mixed rh + tg phases agreed extremely well with an extrapolation of the fcc-to-sc
phase line discussed in section 4.1.2. (The broken line in ® gure 37 was unfortunately
drawn using [284] the slope characteristic for helium-intercalated C60 ; see table 1).
This indicates that free molecular rotation might be a prerequisite for the formation
of the tg phase. Sundqvist [307]suggested that the tg phase might form in the range
of stability of the rh phase if molecular rotation induced a random formation of
bonds in the [110]directions rather than in the close-packed planes, and that the tg
phase could thus be seen as a disorder-induced phase rather than as an equilibrium
phase. In their response to Sundqvist’s [307]comment, Marques et al. [308]correctly
pointed out that molecular rotation in the fcc phase does not hinder the formation of
Â
a well ordered orh phase [261, 262], even in single crystals [250, 259, 260]. However,
in these cases the samples were always heated slowly to the reaction temperature at
constant pressure and there is a possibility that the reaction actually started while the
material was still in the sc phase with well H-oriented molecules (section 4.1.3) and
thus well de® ned crystal directions. In such a model, well oriented crystal `seeds’
already exist when the material enters the fcc range, where the polymerization
reaction rate becomes large, and there should a be high probability of obtaining a
well oriented polymer. As mentioned above, Marques et al. [284] also argued that
low-pressure-polymerized material must always be less well ordered than that treated
at high pressures since the former process depends on random rotational molecular
motion to bring double bonds into contact, while at high pressures the molecules are
brought together in the H orientationally ordered phase and polymerization initiated
by librational motion. Another question, which has already been discussed above, is
Â
whether the orh phase identi® ed at low pressures [259± 262] is identical with or
di€ erent from the orh phase [246, 284]discussed here and, if they are di€ erent, where
the `phase boundary’ between the phases is and what the exact structure of the orh
phase is. (For example, it has been noted that the di€ raction patterns of the orh
phase can actually often be indexed also as a rh lattice [241, 261, 262].) The orh-orh Â
question also has a wider signi® cance. Marques et al. [284] reported that the
transformation between the orh and rh phases was continuous and suggested that
it occurred through a simple translation of the chains in the orh structure. However,
it has been pointed out that, while this is in principle possible, such a mechanism
Â
would be impossible for transformation between the orh and rh phases because of
the di€ erent rotational angles of the chains [260, 262]. A transition from the orh to Â
the tg phase is unlikely but in principle possible [261, 262]. If the two structures are
indeed di€ erent, this might give an alternative explanation why high-temperature
treatment gives a rh phase (from orh) at high pressures and a tg phase (from orh ) at
Â
lower pressures. Davydov et al. [265] made a simple test of this hypothesis by ® rst
treating a sample at 6 GPa and 873K to obtain the rh phase and then treating the
same sample again at 2.5 GPa and 873K. After the second pressure run the sample
also contained a fraction of the tg phase which could be taken as a sign that a rh-to-
tg transformation is possible. However, the sample may also have contained
unreacted C60 which formed the tg phase in the second run.
The two-dimensionally polymerized phases formed in this pressure range have
also been studied theoretically. Xu and Scuseria [309]found intermolecular distances
Fullerenes under high pressures 69

for both the rh and the orh phases in good agreement with experiments. The
calculations also showed that the energy barrier for breaking the intermolecular
bonds in the rh phase should be about 1.6 eV and that the 2D in-plane bulk modulus
was 110GPa. The rh phase was found to be a 3D semiconductor with an electronic
bandgap of about 1eV, while Okada and Saito [310] found a bandgap of only
0.35eV. However, Okada and Saito commented that their method is known to be
uncertain also for C60 and that a more accurate statement is that the bandgap of rh
C60 is about 0.7 eV smaller than that of pristine monomeric C60 . Their calculation
also suggests that it might be possible to dope the rh phase into a metallic state using
both alkali and alkaline-earth metals. Harigaya [311] ® nds a re-entrant insulator±
metal± insulator behaviour against pressure for the tg phase, and possibly also for
the rh.
Signi® cantly less is known about the physical properties of the rh, orh and tg
phases of C60 than about the properties of material polymerized at lower pressures.
The main reason is that the high-pressure phases can only be produced using belt- or
anvil-type devices, and in most cases only quite small amounts can be produced in
each pressure run. Still, many properties have by now been reasonably well studied.
Both 1D and 2D polymers have been investigated by NMR. The results of Persson
Â
and co-workers [248, 254]for disordered orh C60 were mentioned above, and very
similar data were later found by Rachdi et al. [255, 256]for the orh phase produced
at higher pressures. For the rh phase, Rachdi and coworkers [255, 312] found
basically very similar results with a major sp2 peak near 146ppm and a smaller sp3
peak at 73.5 ppm, indicating the presence of covalent intermolecular bonds, as
shown in ® gure 38. However, for this material the sp2 peak could be resolved into six
closely spaced lines. One of these corresponds to a small amount of non-reacted C60

Figure 38. NMR spectra of rh C60, showing the highly split main peak near 145ppm and a
large sp3 peak near 77ppm: ( w ) spinning side bands. Details of the experimental
conditions have been given in the original paper. ( Reprinted with permission from
Goze et al. [312].)
70 B. Sundqvist

while the other ® ve correspond to the inequivalence of the carbon atoms on the
slightly deformed molecules in this phase [300, 309]. Also, the sp3 peak is
signi® cantly stronger for the rh phase than for the orh (or orh ) phase, re¯ ecting
Â
the nominally threefold increase in the average number of intermolecular bonds per
molecule. The static NMR spectra of pristine, orh (`fcc’ ) and rh C60 as well as of
nanocrystalline diamond, carbon nanotubes, graphite and the hard amorphous
phase mentioned above, were also recently compared by Maniwa et al. [313]. The
results agree rather well with those already presented.
Most studies carried out on the orh, rh and tg phases so far have used
spectroscopic methods. Kozlov and co-workers carried out IR and Raman studies
on the orh and rh phases produced at 3 GPa by heat treatment at temperatures from
100 to 500ë C [228, 263], and more recently on mixed rh + tg material produced near
900ë C at the same pressure [314, 315]. The results for the orh phase have already
been discussed above. For the rh phase, the IR data (top curve in ® gure 32) di€ er
very little from those for high-temperature-processed orh material, which itself
shows a rapid evolution with reaction temperature. The low-resolution Raman data
of Kozlov et al. for the rh and orh phases showed only very broad peaks and were
dominated by a giant peak near 1300cm- 1 , while the more detailed data of Iwasa et
al. [241]showed a ® ne structure with several new peaks in the range 1100± 1700cm- 1 ,
resulting again from the reduced symmetry of the distorted covalently bound
molecules. Detailed IR and Raman data for the orh, rh and tg phases were recently
presented and discussed in some detail by Rao et al. [251, 257, 258], while Blank et al.
[305] showed IR and Raman spectra for the rh and orh phases. Very recently, IR
spectra for the rh phase were also discussed in detail by Kamara  s et al. [316]. As
might be expected, the data for the rh and tg phases showed an even richer structure
than that of the orh (or orh ) phase because the increased number of bonds distorts
Â
the molecule further, giving a further reduction in molecular symmetry. This
reduction in symmetry is also re¯ ected in changes in the UV and visible absorption
spectra [228, 263, 314] and in luminescence spectra [228, 251, 257, 263, 314, 315].
These studies also show that the bandgap of the pressure-polymerized material is
smaller than for pristine C60, in agreement with theoretical studies (see above). The
photoluminescence spectra have been studied as functions of both temperature [251,
257]and pressure [257]. At lower temperatures an increased structure is found in the
spectra, while the peaks observed disappear under pressure. The bandgap decreases
with increasing pressure at a rate of only 0.03 eV GPa- 1 , signi® cantly smaller than
the rate of 0.14 eV GPa- 1 found for pristine C60 (see section 4.2.4.1), probably
because of the increase in bulk modulus (see below). ESR studies [279] revealed a
very large di€ erence between the orh and rh phases. While the signal of orh C60 was
almost identical with that of pristine material, the rh material showed an intense
narrow ESR line obeying Curie’s law for localized spins. This signal is probably
connected with multiple structural defects arising in the basically random polymer-
ization process. Surprisingly, in contrast with other studies, this ESR study shows
that the rh structure does not revert completely to monomeric C60 even after 4 h at
700K.
In addition to these spectroscopic data a small number of studies have been
carried out on other physical properties of the rh, orh and tg phases. The electrical
resistance R was measured as a function of reaction temperature near 5 GPa by Ma
and Zou [230], as already mentioned brie¯ y in section 4.2.5. Over the range 300±
850K the logarithm of R decreased almost linearly with increasing T, with a total
Fullerenes under high pressures 71

decrease in R by four orders of magnitude. Small anomalies, perhaps indicating


phase transitions into the orh state, are observed near 400 and 550K. R has a
minimum at 850K, where we would expect the rh phase to be formed and then
increases up to 950K where there is a sharp drop by over two orders of magnitude.
Ma and Zou also presented Raman spectra and X-ray di€ raction diagrams that
agree well with the results presented above. The resistance of a sample polymerized
at 5.4 GPa and 673K for several hours and thus probably mixed orh± rh was also
studied over the range 500± 680K at zero pressure by Takahashi et al. [317] who
found that the initial resistance of the polymerized material was higher than the
resistance obtained after cooling. This is surprising since after heating to 680K the
material should be in the `normal’ monomeric phase which according to the
theoretical calculations discussed above should have a larger bandgap. Takahashi
et al. also reported that samples polymerized near 473K revert to monomeric C60
near 500K, while samples prepared at 673K are stable to near 540K. A much more
complete di€ erential scanning calorimetry (DSC) study by Dworkin et al. [318] on
samples treated for shorter times (10± 15 min) in the ranges 373± 973K and 2± 7GPa
con® rmed that all samples which according to the phase diagram in ® gure 37 should
be in the orh phase transform back to normal in the range 450± 510K, usually with
depolymerization enthalpies of 10± 12kJ mol- 1, while samples that should be in the
rh phase are stable to 510± 540K and have a much higher enthalpy change of 20±
30kJ mol- 1 . These data agree with the model discussed in section 5.3 above in
connection with the breakdown of dimers and linear chains. Among the polymerized
forms of C60 discussed so far the two-dimensionally polymerized forms (rh and tg)
are thus the most stable with breakdown temperatures near 540K. Linear chains
break down at 470± 500K, while the dimer is the least stable form and is stable only
up to 420K. These numbers are, of course, only approximate since the breakdown of
the polymer is a thermally activated process. Further studies on the thermophysical
properties of the mixed (rh + tg) phase are now in progress, and Sundqvist et al.
[295]recently reported preliminary results showing that for this material the thermal
expansion coe cient is between 30 and 50% of that of pristine C60 while the bulk
modulus is about 45 GPa, values which are in general agreement with a simple model
in which the material can expand and contract or be compressed in one single
direction only (compared with the 3D pristine material).
It was mentioned above that a hard, disordered but graphite-like phase could be
obtained when heating C60 to very high temperature (® gure 37). The drop in
resistance observed by Ma and Zou near 950K may be connected with the results
of an early experiment by Kozlov et al. [245, 319, 320], in which C60 was reacted for
2 h under non-hydrostatic pressures from 2.6 to 3 GPa at temperatures from 700ë C
(973K) to 900ë C, sometimes mixed with Al powder which was expected to act as a
catalyst in the formation of high-pressure phases. Treatment near 700ë C resulted in a
black phase with a density of near 1. 9gmcm3 , an X-ray di€ raction pattern with few,
broad peaks indicating a very disordered structure, an almost featureless Raman
spectrum and a low, almost temperature-independent electrical resistivity near
10mV cm typical for a semimetal. Electron-energy-loss spectroscopy (EELS) showed
a spectrum intermediate between that of graphite and diamond [320± 322]. The
material was very hard with a measured microhardness of 400kgf mm- 2 , about two-
thirds that of diamond, and easily scratched sapphire. This phase is probably
identical with the `amorphous sp2 ’ phase observed by Hodeau et al. [323] in
mixed-phase samples and by Marques et al. [284], and which showed a single very
72 B. Sundqvist

Figure 39. Thermal conductivity of amorphous high-temperature phase of C60 and


crystalline arti® cial diamond (a.d.). ( Reprinted with permission from Biljakovic et al. [325].)

broad line in NMR [312, 313]. Blank et al. [305]also recently showed that treatment
temperatures between 1200 and 1800K at 8 GPa resulted in the formation of very
hard materials with hardness between 30 and 40 GPa and gave IR and Raman
spectra for these materials. Kozlov et al. suggested that the high reaction
temperature partially destroyed the molecules, resulting in a disordered phase with
a large proportion of sp3 -type bonds between molecular remnants. The strongly
disordered structure also results in the very unusual temperature dependence of the
thermal conductivity shown in ® gure 39 [324, 325], that is an almost perfect linearity
in temperature from 30 to 330K. Except for a factor of ten di€ erence in magnitude,
probably because of a larger number of covalent bonds per molecule, these results
are similar to those for orh C60 [273, 274](® gure 34). The material also has a much
lower resistivity than the crystalline phases, with zero-pressure values between
10 V cm for semiconducting material treated at 1000K and 0.01 V cm for semi-
metallic materials obtained above 1100K. In contrast with the crystalline phases
discussed above, this material does not revert to normal C60 on heating. However,
Dworkin et al. [318] reported the observation of a glass transition during the ® rst
heating of the material in a DSC study. Later, the material was laser ablated onto
NaCl substrates using an ArF excimer laser [320± 322]. The resulting yellowish
transparent ® lms were investigated by Raman, EELS and electron di€ raction in a
transmission electron microscope, and the results showed clear evidence for the
presence of small diamond crystallites in an amorphous matrix. Treatment at still
higher temperatures seems to break down the molecules even further into softer,
more graphite-like materials, as will be discussed in section 5.6
Fullerenes under high pressures 73

5.5. High-pressure range ( p > 8 GPa)


While there has been a reasonable consensus for some time on the phase diagram
at low and intermediate pressures, the very-high-pressure phase diagram above
8 GPa has only recently been investigated in detail and a rather extensive review of
the structures and phases found in this region was recently given by Blank et al.
[247]. As mentioned above, many early RT studies [49, 186, 188, 217, 236] found
structural phase changes in C60 at pressures of 10 GPa and higher. In-situ optical
studies, as well as X-ray and spectroscopic studies on recovered material at zero
pressure, showed that the high-pressure phases were amorphous and insulating. A
comparison with the phase diagram of graphite in ® gure 14 shows that these phases
form in the stability range of diamond, and it was generally assumed that they were
formed when the C60 molecules broke down into a preferentially sp3-bonded
diamond-like amorphous structure. Nu  nÄ ez-Regueiro et al. [236] even reported the
recovery of crystalline diamond after explosive pressure quenching from very high
pressures. (Diamond production from fullerenes will be brie¯ y discussed in section
5.6.) Later, Blank et al. showed [114, 243, 326, 327]that the hardnesses of some high-
pressure phases exceeded that of diamond, and recent studies have shown that a
large number of phases can be produced if C60 is heated to high temperatures under a
high pressure. However, it is still uncertain which of the phases observed are
equilibrium states and which are only transient intermediate states, or whether
mixtures of some phases mimic new lattice structures in intermediate ranges. For
technical reasons, most studies in this range have been carried out by rapid heating,
keeping samples for 1min or less at the maximum temperature before rapidly
quenching to RT and decreasing the pressure to zero for structural and other
studies. The problem is thus twofold: the material produced may revert from the true
high-pressure high-temperature state to some other metastable or stable phase
during the temperature and pressure decrease, or the sample may not have time to
transform fully to its stable equilibrium state during the brief period at a high
temperature.
A large number of studies have been carried out even in the pressure range above
8 GPa. Two groups have carried out systematic studies over wide ranges in
temperature and pressure of both the structures of samples quenched to ambient
conditions from high temperatures and pressures and of the physical properties of
the reacted materials, and several groups have carried out studies over smaller
ranges. Figures 40 and 41 show two versions of the phase diagram in the high-
pressure range. Figure 40 is taken from Davydov et al. [303, 304] and covers the
range 0± 10 GPa while 41 is taken from Blank et al. [244] and ranges from 5 to
13GPa. A diagram similar to that of Davydov et al. was also shown by Szwarc et al.
[264]. A comparison shows that both these diagrams agree well with that shown in
® gure 37 up to about 8 GPa, above which pressure new phases appear.
Both Blank et al. [328]and Davydov et al. [329]have made extensive studies of
the temperature evolution of the structure under non-hydrostatic conditions near
9.5 GPa, and Blank et al. [244] have recently extended this work to 13 GPa (® gure
42). The fcc phase is reported to be stable at RT to above 10 GPa. However, as
discussed above, it is possible that the true structure may be instead a disordered orh
phase, and band-structure calculations on fcc C60 at a nominal pressure of 13GPa
also show the presence of covalent intermolecular bonds [162]. On heating at 9.5 or
11GPa the fcc phase is stable up to almost 600K, above which the measured unit-
cell volume starts to decrease rapidly with increasing temperature without any
74 B. Sundqvist

Figure 40. High-pressure phase diagram of C60. ( Reprinted with permission from Davydov
et al. [303].)

Figure 41. Very-high-pressure phase diagram of C60 . Reprinted with permission from Blank
et al. [244].)
Fullerenes under high pressures 75

Figure 42. X-ray di€ raction diagrams for C60 samples heated to the temperatures indicated
at 13GPa. (Reprinted with permission from Blank et al. [244].)

observable change in structure except for a broadening of the di€ raction peaks,
probably because of increased disorder. Between about 600 and 700K, both groups
reported a strongly disordered fcc structure with a much reduced lattice constant
a < 13. 6AÊ (`hard fcc’ ). For this phase, Blank et al. [328] reported a continuous
increase in hardness from similar to that of sapphire near 600K to `ultrahard’, that is
harder than diamond, near 700K, and showed images of scratched diamond
surfaces. Davydov et al. [329] also reported that the Raman spectrum near 623K
is identical with that of the `ultrahard’ phase reported earlier [114]. The decrease in
lattice constant and increase in hardness suggest a continuous formation of
intermolecular bonds in random directions (to retain a fcc structure). Brazhkin et
al. [330]described the increase in hardness in terms of a rigidity percolation model, in
which small clusters form at low temperatures and gradually grow until a randomly
bonded 3D, still basically fcc polymeric structure is formed when the average number
of bonds per molecule exceeds a critical value near 2.4 [331]. With increasing
temperature there is then a rapid increase in the number of intermolecular bonds
and thus also in the bulk modulus, rigidity and hardness, and a decrease in lattice
constant. At the highest temperatures the molecules also become increasingly
distorted, as evidenced by the appearance of strong (111) re¯ ections which should
be absent for spherical molecules because of the molecular form factor. Above about
76 B. Sundqvist

800K the structure gradually becomes amorphous (`amorphous 1’ ). Although the


molecules seem to remain basically intact, the increasing number of random
intermolecular bonds probably becomes su ciently high to distort the molecules
even further and/or to break up the periodicity of the lattice. Davydov et al. [303,
304, 332] have suggested a `polycondensated’ structure in which the C60 molecules
partially penetrate each other and coalesce, each losing ® ve to six carbon atoms in
the process. Although super® cially similar to the schwarzite-type structure [333± 335]
suggested to exist by several groups, the high-pressure phase has a much higher
density. AFM images [303, 332]of surfaces in this phase show quasi-oriented worm-
like polymeric structures with intermolecular distances of 6.5± 7.7 A Ê , signi® cantly
smaller than in normal (or even crystalline polymerized) C60. However, carbon-
based materials are notoriously di cult to study by AFM and it is not clear how the
pictures should be analysed. For example, Surja n et al. [336] recently showed
scanning tunnelling microscopy (STM) results for a photopolymerized ® lm in which
a strong periodic modulation with a period of 4.5A Ê was observed along linear
molecular chains. In that case the short period probably arises because of a similarly
high charge density both on the molecular `belly’ and on the four-membered ring
linking the molecules.
Above 1023K the molecules themselves begin to collapse [329] and material
produced under these conditions was reported [264]to be in many respects similar to
that produced by Kozlov et al. [245, 319]. With increasing temperature there is then a
graphitization transformation from the `ultrahard’ amorphous state to what seems
to be turbostratic graphite near 1800K [328]. However, Blank et al. [244] claimed
that all graphite-like phases in the range up to 1800K are `superhard’, that is harder
than cubic BN, and very recently showed hardness data to verify this [305]. Both
groups also reported Raman spectra which show a corresponding development from
that of pristine C60 , over intermediate phases rich in structure and showing a large
number of new lines, typical for polymeric C60 , to almost featureless spectra for the
amorphous phase. The basic results of the two groups are thus in reasonable
agreement.
At high pressures, structural data have been obtained by Brazhkin et al. [330,
337, 338] at 12.5 GPa and by Blank et al. [151, 244]at 13 GPa (® gures 41 and 42).
The orh or fcc are still the stable phases at RT. On heating, a lattice collapse similar
to that at 9.5 GPa is observed near 600K [339]. Brazhkin et al. [330, 337]reported an
fcc structure, as observed at 9.5GPa, while Blank et al. found at least one body-
centred structure at slightly higher pressures. Blank et al. reported the coexistence of
a fcc phase between 500 and 900K with a range of basically body-centred cubic (bcc)
structures with distortions ranging from tg to orthorhombic. The latter is about 20%
denser than the fcc phase. Further structures, such as a hexagonal modi® cation, were
reported in other publications [339± 341]. Above about 800K the structure gradually
changes to a monoclinic (mc) lattice which can be detected up to 1300K. A mc
distortion of the fcc lattice was also reported by Brazhkin et al., but only near 725K.
On further heating the di€ raction lines become increasingly broad and di€ use and
above about 1100K a second `ultrahard’ amorphous phase (amorphous 2), di€ erent
from that observed at lower pressures, is observed (® gure 42). This phase has a
density of 3.1± 3.4 gcm- 3 , only slightly less than that of diamond (3.5 g cm- 3 ), and
Blank et al. suggested that its disordered structure is similar to that found in the
theoretical calculation of Zhang et al. [342] to be discussed brie¯ y below. Above
2000K, ® nally, very small diamond crystallites are produced. As before, both groups
Fullerenes under high pressures 77

also showed Raman data which basically follow the same trend as for the material
treated at 9.5 GPa.
The structural evolution with temperature and pressure above 8 GPa is still a very
active ® eld of research, and a large number of papers have appeared very recently,
for example [305, 330, 343± 345]. It is therefore still impossible to discuss the structure
observed in detail. The current state of the art was, however, recently reviewed in
some detail by Blank et al. [248], especially regarding the many disordered and
amorphous phases observed in this range.
A number of physical properties have been measured for several of these phases.
However, since few of the samples brought back to zero pressure have a pure single
phase and most contain a signi® cant amount of disordered material, the values
reported for the various physical properties should be considered as preliminary
only. As discussed above, all groups presented Raman spectra for most of the phases
observed. Blank et al. [328, 346] used Raman scattering studies to deduce an
activation energy for defects of 0.24 eV in the `hard’ fcc phase. All groups also
reported values for the macroscopic density of their samples at zero pressure after
high-pressure treatment. These densities range from 1.67g cm- 3 as for pristine C60 to
values near 3.3 g cm- 3 for the high-pressure amorphous phase 2 [248]. Both Brazhkin
et al. [337] and Blank et al. [151] also reported values for various mechanical
properties, such as Young’s modulus, adiabatic bulk modulus Bs, Poisson’s ratio,
hardness, yield stress and coe cient of crack resistance, for their samples. The
phases formed under very-high-pressure conditions are also very stable on heating in
vacuo or on boiling in acids [151]but burn in air above 900K. As mentioned above,
several of the phases are reported to be `harder than diamond’, as predicted for
compressed C60 by Ruo€ and Ruo€ [135, 136]. This statement is usually based on
experiments [114, 151, 243, 328, 347] in which the surfaces of diamonds have been
scratched using small sharp pieces of the new materials, while attempts to scratch the
latter using diamond styli have failed when the diamond tips have broken under
load. However, Blank et al. also reported an experimental value BS = 540GPa [151]
for the amorphous phase 2, signi® cantly larger than the value of 441GPa [137]for
normal cubic diamond and comparable with the molecular bulk modulus calculated
to be between 717GPa [176]and 900GPa [177]. Measurement of extreme hardnesses
such as those observed for these materials have been discussed in some detail by
Blank et al. [248], who also reported electrical properties of several `ultrahard’
phases. The amorphous 1 phase seems to be metallic or semimetallic [348] with a
resistivity of about 20 mV cm while the transparent amorphous phase 2 is semi-
conducting with a resistivity greater than 105 V cm [151, 328, 348]. Also, the speci® c
heat capacity per mole of the amorphous 2 phase seems to be among the lowest
recorded for any solid material in the 400± 600K range [151, 347], indicating that the
Debye temperature is even higher than for diamond and that the intermolecular or
interatomic bonds are extremely strong. Available data for various physical proper-
ties have recently been reviewed in some detail by Blank et al. [248].
Going on to even higher pressures, static studies have so far only been carried out
at RT. As mentioned above, several groups reported phase transitions above 10 GPa.
Yoo and Nellis [188] reported the reversible formation of a semitransparent
disordered diamond-like phase consisting of `agglomerates’ of molecules and thus
probably polymerized, above 15± 25 GPa. This phase is probably identical with the
`amorphous 2’ phase discussed above. Application of pressures above 27GPa made
the transition irreversible. Optical studies by Moshary et al. [201] and others [185,
78 B. Sundqvist

186, 202, 239]also showed the presence of transparent or black amorphous phases in
this pressure range, and for material retained to zero pressure, Kosowsky et al. [239]
failed to ® nd any structure by either X-ray di€ raction or Raman methods. However,
using electron di€ raction, Hodeau et al. [323]found that the transparent phase has a
structure that varies continuously from that of polycrystalline diamond to amor-
phous, and synchrotron studies revealed a small amount of diamond even in the
amorphous grains. The structure of the amorphous phase was studied in detail, and
extensive simulations were carried out, showing that the best agreement between the
observed X-ray di€ raction diagrams was obtained for a pure sp3 -bonded structure
containing clusters structurally similar to hexagonal and cubic diamond. Other
possible structural models have been given by Zeger and Kaxiras [349]and Zhang et
al. [342]. The former group suggested the formation of triple-layer C100 clusters built
on a C20 core and bound mainly by sp3 bonds, while Zhang et al. made molecular
dynamics simulations on four C60 molecules at extreme pressures and temperatures
(2500K) quenched to ambient conditions. Their ® nal disordered structure is
reproduced in ® gure 43 and is found to contain about 80% sp3-bonded atoms,
rather close to the value found by Hodeau et al. The calculated bandgap against
pressure and optical absorption spectrum agreed well with the experimental data
obtained by Moshary et al. [201].
Other groups have carried out X-ray di€ raction studies of the structure in situ
under pressure at RT. Haines and Le ger [149], Ludwig et al. [141], Nguyen et al.
[152] and Oszlanyi et al. [238] all found that the fcc phase is stable up to 10 GPa.
Above this an increasing broadening of the peaks is observed in the data and Haines
et al. stated that the diagrams are no longer consistent with a fcc structure. These

Figure 43. Theoretical structure for C60 treated at very high pressures and temperatures.
( Reprinted with permission from Zhang et al. [342]. )
Fullerenes under high pressures 79

observations are all in good agreement with the formation of the orthorhombic
polymer phase above 6± 10 GPa at RT as discussed above. Haines and Le ger and
Nguyen et al. also observed a second transition into a transparent disordered phase
above 20 GPa (in the latter case only on decreasing pressure). Again, this disordered
phase must be identi® ed as the `amorphous 2’ phase discussed.
Finally, Blank et al. [114, 243, 326] carried out a series of experiments up to
40GPa, using a specially designed DAC with one rotatable anvil. Measurements of
the pressure distribution in their cell by the standard ruby ¯ uorescence method
revealed sharp transformations at 2.3 (reversible), 6 and 18 GPa, probably corre-
sponding to the orientational transition discussed in section 4.1.3, transformation
into the polymerized orh phase, and into the transparent amorphous phase
(amorphous 2) respectively. The most interesting observation was the fact already
mentioned above that after the experiments the diamond anvils showed circular
grooves caused by the transformed fullerene phases. This was the ® rst indication
that fullerenes could indeed be transformed into phases which were at least as hard
as the diamond surface. The materials produced were characterized by optical
studies and Raman scattering [114] with results in general agreement with those
given above.
To complete this section, two studies by STM at zero pressure may be mentioned.
Lang et al. [350, 351]observed that C60 ® lms can be transformed into an amorphous
state by operating the scanning tunnelling microscope under extreme force condi-
tions. They calculated e€ ective pressures in the 20 GPa range and observed that
under these conditions a C60 ® lm transforms into an array of rings, containing ® ve to
eight carbon atoms, interpreted as the amorphous high-pressure phase discussed
above. The second study, by Rao et al. [352], compared the surfaces of photo-
polymerized ® lms of C60 and C70. Thin ® lms were submitted to UV light and the
di€ erence between the resulting structures for C60 and C70 was studied. The results
showed that the C60 lattice contracted by 13% (probably by volume), while the C70
lattice expanded by 10%. A possible reason for the latter is molecular reorientation
from an `upright’ position for unreacted C70 to a `recumbent’ state for polymerized
C70 molecules in a ® lm, because the reactive bonds on the molecules are all close to
the polar caps. Such widely di€ erent behaviours would of course explain why
polymerization in C60 is promoted by pressure while that in C70 is not, but the
extent to which the behaviour of thin ® lms can be applied to bulk materials is
uncertain.

5.6. Transformations into other forms of carbon


5.6.1. High static pressures and/or temperatures
Although fullerenes are surprisingly stable from a thermodynamic point of view
[197] the molecules seem to collapse under extreme conditions, that is at very high
temperatures, very high pressures, or a combination of both, forming amorphous
phases where the basic fullerene cages no longer can be observed. A comparison with
the phase diagram of carbon in ® gure 14 shows that the resulting material should
prefer to revert to the graphite structure at high temperatures and to the diamond
structure at high pressures. Several groups have reported the formation of
`amorphous graphite’ or `amorphous diamond’ under extreme conditions, while
other groups preferred other descriptions of the materials and structures observed.
Strictly speaking the terms `graphite’ and `diamond’ refer to two particular crystal-
line structures, hexagonal (Bernal) graphite and cubic diamond, but in a more
80 B. Sundqvist

relaxed sense the terms are usually also applied to related crystalline structures such
as rhombohedral graphite and hexagonal diamond (lonsdaleite) [353]. When
discussing more disordered materials the term graphite is usually changed to
graphitic carbons, meaning materials consisting of carbon with the graphite
structure but a large number of structural defects, and when the c-axis order of
graphite is lacking in a material consisting of the planar hexagonal in-plane network
of graphite the material is usually called non-graphitic carbon. The terms `amor-
phous diamond’ or `amorphous graphite’ are thus not correct, although amorphous
materials could be labelled graphite like or diamond like by reference to the position
of the most intense Raman lines which indicate a majority of sp2 or sp3 bonds
between nearest neighbours in the material.
When C60 is submitted to a high temperature at or below atmospheric pressure, it
sublimes, as discussed in section 4.1.1 above, but it has been reported [64] that
heating above 700ë C results in destruction of the C60 molecules themselves and the
formation of an amorphous graphite-like carbon phase. The low temperature
required is somewhat surprising, since simulations predict the C60 molecule to be
stable to about 3000K [125]. However, it should also be noted that C60 produced by
the standard solvent method always leaves behind an amorphous residue even if
sublimed under milder conditions. The amount left behind seems to depend on many
factors, such as the heating rate and the pressure. On the other hand, if already
sublimed material is reheated, it usually sublimes almost completely, indicating that
the origin of the residue is mainly reactions between the C60 and residual solvent
molecules.
High-pressure studies show that the high-temperature limit for the stability of the
C60 molecule does not change much with pressure. Several studies [12, 244, 245]have
shown that C60 breaks down into amorphous carbon if heated much above 1000K.
Raman studies revealed that the bonds in this phase are mainly of the sp2 type, that
is the material is basically `graphite like’ but usually disordered and mechanically
soft [245]. Near the upper temperature boundary, however, Kozlov et al. [245] and
others found a very hard phase as discussed above, while further increases in
temperature gave softer, more graphitic materials. These data can thus be used to
® nd an approximate upper limit for the stability of the C60 molecule. However,
considering the possibility that the materials studied all contain traces of residual
solvent the true stability line for pure C60 may be shifted to higher temperatures.
At higher pressures, several groups have reported the presence of crystalline
diamond and/or `amorphous diamond’, i.e. preferentially sp3 bound amorphous
carbon, among the reaction products. Nu  nÄ ez-Regueiro et al. [12, 236] identi® ed
crystalline diamond in samples submitted to highly non-hydrostatic very high
pressures (greater than 20 GPa), but other studies have not been able to reproduce
these results. Both Blank et al. [244]and Ma and Zou [230]reported that crystalline
diamond is produced only if very high pressures are combined with extreme
temperatures (e.g. 13 GPa and 2100K [244]). A di€ erent road to diamond produc-
tion was found by Kozlov et al. [245, 319, 321, 322] who found that diamond
crystallites were produced under laser desorption or evaporation of the very hard
amorphous phase produced by heating to near the stability range of the molecules at
pressures of 2± 5 GPa.
The transformation of C60 into graphite, diamond and various other carbon
phases has also been discussed in more detail by Blank et al. [247].
Fullerenes under high pressures 81

5.6.2. Catalytic conversion to diamond


It should be noted that, starting from C60, direct synthesis of normal cubic
crystalline diamond, identi® ed by Raman spectroscopy and X-ray di€ raction analy-
sis, under static pressure has indeed been reported [230, 244, 354, 355]. In these cases
the standard catalyst method for diamond synthesis [119] was used except for the
replacement of graphite by C60 . Bocquillon et al. [354] reported that diamond is
produced near 6.7 GPa between 1200 and 1850ë C when using Co as catalyst, while
Ma et al. [230, 355]used low pressures and temperatures (5.5 GPa; 1400ë C) and a Ni±
Mn± Co catalyst, and Blank et al. [244]a higher pressure of 8 GPa at 1800K, V. A.
Davydov et al. (1997, private communication) also reported similar results obtained
when a Ni alloy was in unintentional contact with a C60 sample at very high
pressures and temperatures. The reverse process, that is conversion of diamond to
fullerenes under intense laser irradiation, has also been reported [356] but is
obviously of less interest.

5.6.3. Shock wave experiments


Very-high-pressure shock wave experiments on C60 result in the formation of
transparent high-pressure phases if the material is rapidly quenched in temperature
after the shock. These phases are usually reported to be diamond or `amorphous
diamond’, which as indicated above usually means that analysis of the materials
reveals a basic tetrahedral atomic coordination and a majority fraction of sp3 bonds
in the material. Yoo and Nellis [240, 357, 358] ® rst submitted C60 to high shock
pressures and reported a transition to well ordered graphite above 17 GPa and 800K
and then a second transition into an amorphous carbon state near 50GPa and
1600K. These workers speculated that diamond formed at the highest pressures and
temperatures but reverted to amorphous carbon during the relatively slow cooling to
ambient conditions. At the highest pressure of 69 GPa and 2200K, an unidenti® ed
transparent amorphous phase was produced. This phase was barely metastable and
transformed over several days into black disordered carbon and small amounts of
the high-pressure phase of diamond known as n-diamond [359]. In a similar
experiment, Sekine et al. [360]found that, above 20± 24 GPa and 2400K, di€ raction
peaks due to C60 disappeared and the (111) peak of (crystalline) diamond appeared.
The diamond yield increased with pressure up to at least 37 GPa, depending on
sample purity. The theoretical possibilities for a transformation from C60 to
crystalline graphite or diamond on a short time scale have been studied by SandreÂ
and Cyrot-Lackmann [361], who found that such a transformation could indeed
occur under non-hydrostatic conditions through a topological transformation with-
out long-range atomic di€ usion. The resulting ® nal structure, however, should
contain an appreciable fraction (about 7%) of atomic vacancies.
In another series of papers, Hirai and co-workers [362± 369] presented results
from a number of high-shock-pressure experiments using an extremely rapid post-
shock temperature quench technique with a cooling rate of 1013 Ks- 1 and special
precautions to keep the shock temperature low (less than 1000K). In their ® rst
papers [362, 363] they reported complete conversion of C60 to crystalline cubic
diamond at the highest shock pressures, 54 GPa at 815K, with a decreasing diamond
fraction and an increasing amount of amorphous carbon with decreasing pressure
down to 16 GPa. Below this, no diamond was observed. No trace of graphite was
observed in any experiment. The materials produced were studied by electron
di€ raction and EELS [365]and these researchers found a development from pristine
82 B. Sundqvist

C60 over a modi® ed structure (polymer?) to a disordered structure rich in sp3 bonds.
In a later experiment, lower cooling rates and higher peak temperatures were used
[364, 366] and an amorphous transparent phase was observed after submitting the
sample to 55 GPa and 2000K. This material was found to lack long-range order and
to be preferentially sp3 bonded, and Hirai and co-workers chose to use the term
`amorphous diamond’ in their description. The transition behaviour from C60 over
the modi® ed C60 phases to diamond was studied in more detail in [367, 368]and the
structure of the `amorphous diamond’ phase was studied by electron di€ raction
[369]. The latter study showed that the material has a very well de® ned short-range
order giving three observable peaks in the radial distribution function. Whether this
phase is identical with the `amorphous 2’ phase or not is still not clear.
Although a large number of studies have been carried out, it is di cult to relate
the results from the shock compression studies to those obtained by static methods.
However, a comparison between the shock pressure coordinates above and the phase
diagram of graphite and diamond (® gure 14) shows that C60 transforms into
diamond or sp3 phases at shock pressures and temperatures in the range of `static’
diamond stability, that is above 30 GPa, but into graphite, diamond or various C60
phases if the maximum shock pressure is below 30 GPa.

5.6.4. Other studies


To complete this section, direct conversion of giant fullerenes (`carbon onions’ )
to crystalline diamond has recently been reported by Banhart and Ajayan [370] to
occur under the combined action of electron irradiation and high temperature in a
high-voltage transmission electron microscope. Heating carbon onions to between
925 and 1025K on the stage of the electron microscope during high-beam-intensity
irradiation was observed to cause a shrinking of the onions because of the formation
of lattice defects due to the irradiation. Under the combined in¯ uence of the high
pressure due to the shrinking of the outer layers and the high temperature, crystalline
diamond grains, identi® ed from their cubic structure and their lattice spacing, were
observed to grow at the centre of the onions. In e€ ect, the outer layers of the onions
acted as miniature (`nanoscopic’ ) pressure cells, and applied pressures up to 50±
100GPa were calculated from the measured lattice spacing in the shells. The
formation of the much denser diamond phase relieves the pressure in the onions,
resulting in an observed increase in the onion shell spacing. Banhart and Ajayan also
suggested that other encapsulated materials could be submitted to high pressures in
this way. Similar results have recently been obtained by irradiation of carbon onions
by megaelectronvolt Ne+ ions [371], and Pasqualini [372] has veri® ed the growth
mechanism by theoretical calculations.

5.7. Pressure± temperature phase diagram of C6 0


The information collected in sections 2± 5 can be summarized in the preliminary
`phase diagram’ shown in ® gure 44, which can be considered a modi® ed version of
® gure 1 of [248]. It is important to note that over most of the pressure± temperature
plane this is not an equilibrium phase diagram in the standard sense, but rather a
map showing which phases and structures are retained as metastable material to zero
pressure if a C60 specimen is treated for a su ciently long time at the reaction
coordinates indicated and then quenched to RT. In many cases it is thus still
uncertain whether a particular structure is a real reaction product, characteristic of
these pressure± temperature conditions, or whether it is produced by the collapse of a
Fullerenes under high pressures 83

Figure 44. Tentative `phase diagram’ for C60, showing phases or structures normally
observed at zero pressure after temperature quenching from the pressures and
temperatures indicated. Phase boundaries have been interpolated between references
given in this section and those shown shaded may either be uncertain or form mixed-
phase regions. In some areas of the diagram other phases have also been reported, as
discussed in the text.

primary high-pressure phase during quenching in temperature or during the pressure


decrease. The phase boundaries in the ® gure have been interpolated between the
results of di€ erent groups. Many boundaries are shown as shaded areas, indicating
that mixed phases are often obtained and that structures may di€ er between di€ erent
experiments. The low-pressure part of the diagram, up to about 8 GPa, is now rather
well established but, as discussed above, the very-high-pressure range has only been
studied by transient heating experiments and there are still many questions regarding
which are the stable equilibrium phases in di€ erent parts of the pressure± temperature
plane. Because only metastable samples have so far been studied at zero pressure,
there is also a great need for in-situ studies at high temperatures and pressures. In
addition to the phases shown, other phases with a di€ erent structure, such as the
hexagonal phase reported by Blank et al. [340, 341]and Serebryanaya et al. [339]can
be produced by adding strong shear stress in the experiment. Such shear stress can
also lead to the formation of phases normally produced only at much higher
pressures under hydrostatic or quasihydrostatic conditions, explaining observations
such as the scratching of steel vessels near 1 GPa [90] and tungsten carbide anvils
near 3GPa [298].

6. Structures and properties of C70


6.1. Phase diagram
While previous sections have treated the physical properties of C60 , in this section
the discussion will turn to the next higher fullerene, C70. In comparison with C60 ,
very little is known about the high-pressure properties of C70, and nothing is yet
84 B. Sundqvist

known about the higher fullerenes C76 , C80 , etc. One reason for this is of course that
C70 is not as widely available in the pure form as C60 and that its price is signi® cantly
higher. As discussed in section 2, a second, equally important reason is that C70 is a
much more complicated material from every aspect, both static and dynamic, and
regarding molecular, translational, orientational and rotational properties. For the
higher fullerenes there is also the additional problem that all have several isomers
which may also have slightly di€ erent properties.
Regarding the phase diagram, it should be stated from the beginning that there is
still no consensus on the high-pressure phase boundaries of C70 , nor is the structural
evolution of C70 with pressure known with any certainty. As discussed in section 2
the structural properties of C70 are not well known even at zero pressure, and it is
possible, and indeed probable, that all real C70 crystals contain mixtures of several
structural phases. To recapitulate, there are two basic possibilities for the transla-
tional structure at zero pressure. The high temperature structure is either fcc, as in
C60 , or hcp. The two types of crystal evolve in di€ erent ways when cooled and go
through reasonably well de® ned structural sequences which unfortunately often
show strong e€ ects of hysteresis and metastability. Much of this behaviour is
probably aggravated by the e€ ect of structural disorder since the next-nearest-
neighbour interaction is probably not strong enough to give well de® ned stacking
sequences ABCABC . . . . (fcc) or ABAB . . . (hcp) over any large range. Some of the
di€ erences might also be due to the presence of impurities, since it was shown by
Kawamura et al. [373]that the presence of 10% of C60 made the fcc phase stable to
much lower temperatures (or higher pressures) than in pure C70 . Like C60, C70 can be
transformed into an insoluble polymeric state by irradiation with visible light [352,
374], but the polymerization process is considerably slower because only the double
bonds close to the `poles’ of the anisotropic molecules take part in polymerization
[374].
Studies of the low-pressure phase diagram up to a few gigapascals have been
carried out by thermal, structural and spectroscopic methods but, as already
mentioned, the results obtained sometimes di€ er strongly between di€ erent groups.
The most direct results are of course those obtained by structural studies. Since the
fcc lattice is the most stable structure at high temperatures, most groups have used
fcc crystals as their starting materials. As discussed above, such crystals may be
either fcc or rh below about 350K, depending on the degree of orientational order,
and fcc, rh or mc below about 280K. Kawamura et al. [88, 373, 375, 376] have
carried out a series of X-ray di€ raction studies on the structure of C70 under pressure
at and above RT. In their ® rst study [375] they found that a crystal which initially
was in the fcc phase at RT and zero pressure transformed into the rh structure under
high pressure, with the long axes of the molecules oriented along the (111) axis of the
lattice, as expected. In a later, more extensive study [88]they made X-ray di€ raction
studies at a large number of temperatures during heating from 290 to 1170K at 0.4,
0.9 and 1.5 GPa in a multiple-anvil device with a boron-epoxy pressure medium. At
RT all samples were found to have a rh structure under an applied pressure, and a
phase change from rh to fcc was observed at all pressures on heating as shown in
® gure 45. The slope of the phase line deduced from the transition temperatures
observed was about 300KGPa- 1 between 0.4 and 1.5 GPa, and an even steeper
slope can be deduced at lower pressures if literature data for the zero-pressure
transition temperature are used. An interesting observation is that the volume
di€ erence D V across the transition decreases with increasing pressure such that it
Fullerenes under high pressures 85

Figure 45. Phase diagram of C70 as observed by X-ray di€ raction: (n ) fcc; ( s ), rh; (d ),
amorphous phases. (Reprinted with permission from Kawamura et al. [88].)

is almost zero near 1.5 GPa, a behaviour also observed by some researchers at the
fcc-to-sc transition in C60 (see section 4.1.2). From the Clausius± Clapeyron relation
dTc /dp = D V /D S this implies either that D S also decreases sharply, implying a
change from a ® rst-order towards a higher-order transition at high pressures, or that
the slope decreases at higher pressure. Such a decrease in slope is compatible with the
experimental data at high pressures (see ® gure 45). At RT the rh angle of the lattice
decreased with increasing pressure but saturated near 1 GPa, and it was speculated
that this might signify a transition to the mc phase. Indications that a transition
might occur to a phase with a lower symmetry than rh had also been observed in the
® rst experiment [375]. At the highest temperature investigated (1173K), an
amorphous phase was observed at both 0.9 and 1.5 GPa and Raman studies showed
that this phase had a mixed sp2 ± sp3 bond structure. However, little is yet known
about this phase. As for C60, several studies have also been carried out on the vapour
pressure [57, 60, 377], and in principle the low-pressure high-temperature phase
diagram can be deduced from these studies.
Qualitatively similar results were obtained by Christides et al. [378], who carried
out an X-ray di€ raction study under hydrostatic pressures up to 25 GPa at RT. They
found that an initially fcc sample transformed to the rh phase under pressure,
starting near 0.35GPa and with the transformation being completed near 1 GPa. The
transformation was associated with a signi® cant decrease in volume. On returning to
zero pressure the rh phase remained stable and, if subjected to a higher pressure, the
sample remained in the rh state until an amorphous phase was formed in the range
11± 18 GPa.
Compression studies have also been carried out on nominally fcc C70 by Lundin
et al. [379, 380] using the piston-and-cylinder method. Below 236K no phase
86 B. Sundqvist

transitions were observed up to 1 GPa, although a large volume decrease was


observed during the ® rst pressure increase at 150K indicating a possible orienta-
tional ordering. However, at 296 and 343K a very large initial drop in volume was
observed which Lundin et al. suggested might be associated with orientational
ordering of an initial zero-pressure fcc phase to the rh structure. The same group
also carried out thermal conductivity studies under pressure [380± 382]. These studies
showed sharp anomalies in ¸ against temperature near RT, similar to that observed
at the fcc-to-sc transition in C60 . The anomaly, which shifted to higher temperatures
under pressure at a rate dT /dp = 70KGPa- 1 , was assumed to correspond to the
arrest of molecular rotation in the material. Although the anomaly might possibly be
linked with the boundary for free rotation a comparison with data for linear alkanes
[383]showed that it must instead correspond to the ® nal complete arrest of uniaxial
rotation.
The phase diagram has also been investigated by DTA by Samara et al. [75, 384]
up to 0.8 GPa. These studies were carried out on both fcc and hcp C70 and in both
helium and N2 and, as in the case of C60 (sections 4.12 and 7.4), large di€ erences
were found between experiments because of gas intercalation e€ ects. For all samples,
e€ ects due to transformations in both fcc- and hcp-type structures were observed and
discussed. The results of most interest here are those for fcc C70 obtained in N2 .
Samara et al. [384] found phase line slopes larger than those observed in other
experiments, estimating the slope of the rh-to-fcc phase line to be greater than
700KGPa- 1 and that of the rh-to-mc transition line 320KGPa- 1 .
Finally, a number of other types of study have shown anomalies that may be used
to identify the various phases and phase boundaries of C70. Resistance studies on
mixed fcc± hcp C70 by Ramasesha et al. [385] showed three not very well de® ned
anomalies that shift with pressure at rates from 53 to 84KGPa- 1 , in general
agreement with the results of Soldatov and Sundqvist [381, 382]. Huang et al.
[118] reported a splitting of the IR mode near 535cm- 1 above 1 GPa, indicating a
transition into a less symmetric state. Maksimov and co-workers [386, 387]observed
anomalies in the Raman spectra of C70 which indicated transitions near 2 and
5.5 GPa at 400K, and at least that at 2 GPa was found to be reversible. In a similar
experiment closer to RT, Sood et al. [388]found anomalies in dx /dp between 1.2 and
2 GPa for several Raman lines, and for the strong 1567cm- 1 mode a large increase in
linewidth is reported near 1.2 GPa. Both groups associated these transitions with the
ordering transitions in C70 at atmospheric pressure. Snoke et al. [186] reported a
large anomaly in the slope of the mode at 1188cm- 1 at higher pressures near 4 GPa,
in general agreement with the data of Maksimov and co-workers.
Most of the results from the studies discussed above agree reasonably well with
the tentative low-pressure phase diagram shown in ® gure 46, which is based on our
previous results for the thermophysical properties of C70 under pressure [379, 382].
The two solid curves, drawn through the transition temperatures 280 and 340K
observed at zero pressure [3], should be understood to denote phase lines
corresponding to rotational rather than to structural states. At high temperatures,
quasi-isotropic molecular rotation leads to a fcc structure, and the upper phase
boundary corresponds to that given by Samara et al. [384]at low pressures and by
Kawamura et al. [88]above 0.4 GPa. In the intermediate-temperature range the low-
pressure structure (shaded area) is either fcc or rh, depending on the previous
history, but compression of the lattice forces the molecules to line up in parallel,
improving the orientational order such that the high-pressure phase is rh. The
Fullerenes under high pressures 87

Figure 46. Tentative pressure-temperature phase diagram of C70. The slopes of the full lines
are taken from [75]and [384], [379± 382]and [88], as discussed in the text. The broken
line is the transition line given by Samara et al. [75, 384]. In the shaded region,
Sundqvist and co-workers [379± 382] deduced a continuous orientational ordering
with increasing pressure. Symbols denote positions of anomalies observed by
Christides et al. [378] ( d ), Huang et al. [118] ( n ), Sood et al. [388] ( s ), Maksimov
and co-workers [386, 387] ( h ), Snoke et al. [186] ( , ) and Iwasa et al. [392] ( j ).

detailed low-pressure structure is uncertain and might also involve a spinning of the
long molecular axis around the [111] direction, as discussed in section 2. Below the
lower curve, molecular rotation has probably ceased and the structure is metastable
fcc (at zero pressure) or rh, or mc, again depending on the thermal history of the
sample. Again, the application of pressure should force an orientational ordering of
the molecules such that the high-pressure structure should be rh or mc. It should be
noted that the di€ erences between the X-ray di€ raction spectra of the rh and mc
phases are small and the two structures cannot easily be distinguished in a DAC
experiment. Two low-temperature phase boundaries are given. The solid curve is
that given by Soldatov and Sundqvist [382] while the broken line is that given by
Samara et al. [384]. The di€ erence between the slopes is surprisingly large, and more
experimental work is clearly needed to clarify this part of the phase diagram for C70.
A recent theoretical calculation within the Landau theory by Fradkin [389]
demonstrated the e€ ect of pressure on the fcc-to-rh transformation in C70 and gave
an explicit formula for the critical temperature as a function of hydrostatic pressure.
Unfortunately, su cient experimental data are not yet available to calculate
numerical results from this formula.
The only data available for the phase diagram of hcp C70 are those obtained in a
helium environment by Samara et al. [384]. These data show boundary slopes similar
to those for the nominal mc-to-rh transition in fcc C70 under a helium pressure.
However, no data are available for this phase diagram of non-intercalated hcp C70.
As will be discussed below, hcp C70 has also been reported to transform to the rh
phase under high pressure [306]and thence to the fcc structure at zero pressure.
Turning now to the very-high-pressure range, few studies have been made on C70
88 B. Sundqvist

in the pressure range above 3 GPa, apart from the Raman studies discussed above
which showed transition anomalies near 4 GPa at 300K [185, 186]and at 5 GPa near
400K [386, 387]. Chandrabhas and co-workers [388, 390] reported that the
intramolecular Raman modes of C70 broadened with increasing pressure but could
be observed up to 12 GPa. Above 20 GPa the original Raman lines disappeared, and
a new line at 1650cm- 1 characteristic of amorphous carbon appeared, in agreement
with the earlier report based on structural studies by Christides et al. [378] that
amorphization occurs at 18 GPa. Surprisingly, Chandrabhas and co-workers also
reported that this amorphization is reversible, that is normal Raman lines reappear
on decreasing the pressure to below 10GPa. Sood et al. [388]compared the pressure
dependences of the intermolecular distances in C60 and C70 with the interplanar
distance in graphite and concluded that C70 and graphite are very similar and
di€ erent from C60. Graphite has also been reported to show a reversible transition
into a possibly amorphous state [186]. Since the amorphization observed could be a
signature of polymerization and C70 is known [374, 391] to polymerize under the
action of light, Sundar et al. [306]recently subjected hcp C70 to temperatures up to
1100K for up to 6 h at 5 and 7.5 GPa in an attempt to ® nd indications of
polymerized structures similar to those seen in C60 under high pressures. However,
no evidence for such a behaviour was found. X-ray di€ raction studies at zero
pressure after the experiment showed that all samples had transformed from the hcp
to the rh phase, which on annealing transformed into the normal fcc structure. This
tends to verify theoretical predictions from molecular dynamics calculations [38]that
the fcc structure and its low-temperature counterparts are thermodynamically more
stable than the hcp structural sequence and also indicates that the energy di€ erence
probably increases under pressure. Very recently, however, Iwasa et al. [392]
subjected both hcp and fcc C70 to temperatures in the range up to 800ë C for 1 h
at 5 GPa. In contrast with Sundar et al., they observed clear signs of polymerization
after the experiment by X-ray di€ raction, IR spectroscopy and UV± visible absorp-
tion spectroscopy. The same new (insoluble) polymerized structure was always
obtained on heating to 2± 300ë C for both initial zero-pressure structures, and the
observed IR frequencies were in good agreement with the results of a theoretical
calculation for C70 dimers. From 400ë C and upwards an increasing irreversible
amorphization into a conducting graphite-like phase was observed, which Iwasa et
al. concluded is a sign that the C70 cages have collapsed under pressure, while heating
at 100ë C resulted in partial polymerization only. Like C60, the polymerized material
reverted to monomeric C70 on heating to 300ë C at zero pressure. Again, the structure
after this treatment was always fcc, irrespectively of the initial pre-polymerization
lattice structure.
The phase diagram above about 3 GPa must clearly be investigated further. It
seems reasonable to assume that the anomalies observed in spectroscopic studies
near 4± 5 GPa are connected with the polymerization observed by Iwasa et al. [392].
The polymerization pressure thus seems to be higher than for C60 . (However, recent
measurements of thermal conductivity of C70 under applied pressure by A. Soldatov
(1997, private communication) suggest that very slow polymerization may occur also
at pressures close to 1 GPa.) The reluctance to form polymeric phases may arise from
the molecular structure. C70 molecules are known to be much less reactive than those
of C60 because the double bonds assumed to take part in polymerization are
localized in the `polar’ parts of the molecules only. This might lead not only to a
reluctance to form polymers but also to a lower strength of the intermolecular bonds,
Fullerenes under high pressures 89

once formed. Questions that remain are why the experiments of Sundar et al. [306]
failed to show polymerization in the same range as studied by Iwasa et al., and why a
reversible amorphization is observed [390]at pressures above 10 GPa. One possibility
is that a second type of polymerization occurs at pressures between 10 and 20 GPa,
but that this polymerized structure is not stable below some lower pressure limit
between 0.1 and 10 GPa [390]where the covalent intermolecular bonds break and the
material reverts into molecular C70 again. To investigate this possibility it would be
necessary to study the lattice structure in situ at high temperatures and pressures, or
possibly to quench the high-pressure phase into liquid N2 or helium before returning
to zero pressure.

6.2. Physical properties of C7 0


6.2.1. Thermophysical properties
As discussed in the previous section the bulk modulus of C70 has been measured
by several groups. To present the data in alphabetical order by authors, the X-ray
di€ raction measurements of Christides et al. [378]have given the highest values for
the bulk modulus. A ® t of their data for the lattice volume in the rh phase to the
Murnaghan equation of state gave an initial bulk modulus B(0) = 25 GPa with a
slope B = dB /dp = 10. 6. However, as noted in the section above the initial slope up
Â
to 1 GPa is signi® cantly larger and a ® t to the data in this range gives an average
slope near 10.5 GPa [380]. Kawamura et al. [376]also made X-ray di€ raction studies
as a function of temperature at 0.4, 0.9 and 1.5 GPa. From these data the average
bulk modulus between 0.4 and 1.5 GPa at 290K, that is nominally in the rh phase,
was found to be 11 GPa. The average linear thermal expansion coe cient in the rh
and fcc phases at 0.9 GPa can also be deduced from the data given, and is about
2.2 ´ 10- 5 K- 1 in the rh phase between 300 and 650K and slightly larger in the fcc
phase at higher temperatures. Finally, Lundin et al. [379, 380] measured the
compression behaviour over a large temperature range up to 1 GPa using the
piston-and-cylinder method. Referring to the phase diagram in ® gure 46 above,
stable and repeatable data were obtained in the `rotation-free’ phase below 280K
and in the high-temperature fcc phase above 350K. However, as mentioned above, a
very large and reversible initial drop in volume was always observed in the
intermediate-temperature range, a drop which might be associated with orientational
ordering of an initially more disordered zero-pressure phase and which gave rise to a
large low-pressure anomaly in the measured bulk modulus B, as shown in ® gure 47.
Available experimental data for the bulk modulus and its pressure dependence are
collected in table 7, which also shows over which ranges in temperature and pressure
the data have been measured. The relevant phases are tentatively identi® ed from the
phase diagram in ® gure 46 above. The bulk moduli of C70 have values very similar to
those of C60 (table 2) but a look at the table indicates that further measurements are
needed, particularly in order to understand the structural behaviour of C70 near RT.
Table 7 also shows theoretical values for the bulk modulus for the fcc phase from
three sources, all of which present data for both the thermal expansion coe cient a
and the bulk modulus over wide ranges in temperature. Kniaz et al. [393] used a
spherically symmetric potential originally developed for use with C60 and gave the
relative volume as a function of pressure as well as calculated data for the
compressibility from 400 to above 1000K. They found that C70 is slightly less
compressible than C60, in reasonable agreement with experiment. Similar results
were more recently found by a similar method by Barrio et al. [394], who also
90 B. Sundqvist

Figure 47. Bulk modulus of C70 against pressure at 296K ( s ) and 343K ( h ). (Reprinted
from Lundin et al. [379]. )

Table 7. Data for the bulk modulus B(0) of C70 at zero pressure and its pressure dependence
Â
B = dB /dp, from several investigations. The phase identi® cations refer to the phase
diagram in ® gure 46 above. Phase II denotes the intermediate phase with uniaxial
rotation and a nominal rh structure and III the low-temperature phase.

Pressure
T range B( 0)
(K) (GPa) (GPa) B Â Phase Reference

400 0 13.2a Ð fcc Kniaz et al. [393]


365 0± 1 7.9 16 fcc Lundin et al. [380]
300 0 15.5± 16.6a Ð fcc Barrio et al. [394]
297 0 18.6a Ð fcc Abramo and Caccamo [124]
343 0.2± 1 8.4 14 II Lundin et al. [380]
296 0.2± 1 8.5 14 II Lundin et al. [380]
293 0± 1 10.5 Ð II Christides et al. [378]
293 0.5± 1.5 11 Ð II Kawamura et al. [88]
293 1± 10 25 10.6 III? Christides et al. [378]
236 0± 1 10.1 15 III Lundin et al. [380]
185 0± 1 13.1 14 III Lundin et al. [380]
a
Theoretical calculation.

presented calculated data for a number of other properties such as the speci® c heat
capacity, the sound velocities, and the elastic constants from 300 to 1900K. Finally,
Abramo and Caccamo [124] used molecular dynamics simulations to ® nd the
molecular volume against pressure and temperature up to 15GPa at RT and up
to 2400K at zero pressure.
The speci® c heat of C70 per unit volume has also been measured under pressure
[379, 381, 382], but only with a rather low accuracy for the purpose of identifying
phase transition lines. More data are available for the thermal conductivity [379, 381,
Fullerenes under high pressures 91

382], data for which were also used in deriving the phase diagram above. The data
for the thermal conductivity ¸ of C70 show a behaviour similar to that of C60 , with a
constant ¸ above RT, an increase with decreasing temperature below this followed
by a peak and ® nally a decrease at low temperatures. The anomaly near RT could
also be observed in the measurements of ¸ against pressure. At RT, the Bridgman
parameter g = d(ln ¸) /d(ln V ) increased with increasing pressure from a value near
4.4, typical for a disordered material, to a value near 8.8 at 0.7 GPa typical for an
ordered crystal. This behaviour agrees with the ordering e€ ect of pressure discussed
in section 6.1.

6.2.2. L attice vibrations


A small number of IR and Raman spectroscopy studies have been carried out,
often on mixtures of C60 and C70 instead of pure C70 . Because of the reduced
symmetry of the molecule a much larger number of allowed IR- and Raman-active
modes exists in C70 than in C60 [3, 179]. Huang et al. [118] measured the pressure
shift of 11 IR-active modes up to at most 3 GPa. Several modes disappeared already
at low pressures, and one mode near 535cm- 1 showed a splitting above 1 GPa that
might be connected with the fcc-to-rh transition discussed above. The data given are
listed in table 8. As expected, the frequencies of most modes increase with increasing
pressure, but for three modes a frequency decrease is observed.
Raman data are given by several groups as shown in table 9. Data from several
groups show strong anomalies near 1GPa or at 4± 5 GPa, or both, as discussed in the
section on the phase diagram above. The most dramatic of these anomalies is the
appearance of a very strong negative slope dx /dp in the mode near 1188cm- 1 above
4 GPa observed by Syassen’s group [185, 186]. Again, a few modes have negative
slopes, but most frequencies increase with increasing pressure. No theoretical data
yet exist with which to compare the results for the pressure dependence.
As in the case of C60, no neutron scattering data exist from which to deduce the
pressure dependence of the low-energy modes. Again, we can deduce a thermal

Table 8. Pressure dependence of the IR-active modes of C70, as measured by


Huang et al. [118]and Yamawaki et al. [229].

dx /dp ( cm- 1 GPa- 1) from the following references


x (cm- 1) [118] [229]

1460 1.3
1415 - 7.6
1289 10.2
1260 0.7
1134 3.9
796 - 0.6 - 0.44
674 - 0.5 - 0.39
642 0.3
578 - 0.2
566 3.4
535 2.7a
458 2.8
a
Splits above 1 GPa.
92 B. Sundqvist

Table 9. Pressure dependence of some Raman modes in C7 0 at RT, as measured by Snoke and
co-workers [185, 186], Meletov and co-workers [386, 387]and Sood et al. [388].

dx /dp (cm- 1 GPa- 1) from the following references


x (cm- 1) [185, 186] [386, 387]a [388]

1567 2.73 3; 7; 2.8b 5.15; 3.9c


1513 4.5 3.5; 8; 2b 4.1; 1.65c
1470 8.1; 1.65c
1449 5.5; 1.6c
1370 1.1
1261 4.5; Ð c
1224 3.2 5; 5.3c
1182 4; - 10.3d 5; 8; 3b,e 12; Ð c
1065 1.1 4.2; 1.6c
740 0.12
564 - 0.06
517 0.38
256 1.65 4; 2.5; 1.5b 3.4; 1.4c
220 Ð ; 3.5; - 0. 3
a
At 400 K.
b
In the pressure ranges 0± 2 GPa; 2± 5 GPa; above 5 GPa.
c
Below 1.2 GPa; above 1.2 GPa.
d
Below 4 GPa; above 4 GPa.
e
Line splitting observed above 2 GPa; data for higher-frequency component given.

average of the volume dependence of these modes using the GruÈneisen parameter g
(see section 4.2.2). Kniaz et al. [393]have calculated a theoretical value g = 9.5 and a
Debye temperature of 73 K, very similar to the values found for C60 .

6.2.3. Electronic properties


Like C60, pure C70 is a semiconductor with a rather large bandgap. The electrical
resistivity of C70 was measured as a function of pressure over a large range in
temperature by Nu  nÄ ez-Regueiro et al. [219]. As already discussed in section 4.2.5,
Nu nÄ ez-Regueiro et al. suggested that the experimental results are best explained by a
variable-range hopping model and a good linearity is obtained when their data are
plotted as log R against T 1 /4 . Unfortunately, experimental data were only obtained
over the range 13.5± 24.5 GPa and showed little change with pressure except at the
highest pressure where a semiconductor-to-insulator transition was observed to
start. Ramasesha et al. [385] measured the resistivity over a signi® cantly smaller
range in temperature and pressure (p < 1.2 GPa) using a solid medium, primarily in
order to map the phase diagram. Their sample was a mixture of hcp and fcc and
showed three not very well de® ned anomalies in the resistance over the range 290±
425K. Again, no data were given that allow a determination of band-structure
changes with pressure.
Optical absorption spectra have been measured in C70 under hydrostatic
pressures to 10GPa by Meletov et al. [205, 395] and have a pressure dependence
very close to those of C60 . The absorption edge at zero pressure is 1.78 eV and the
absorption edge position shifts towards lower energies under increasing pressure at
approximately the same rate as for C60 . The initial slope was found to be
Fullerenes under high pressures 93

dE /dp = - 0.1 eV GPa- 1, rapidly decreasing to about - 0.03eV GPa- 1 at the highest
pressures studied. The measurements suggested that no metallization transition
should occur in C70 below at least 50 GPa. The band-gap has also been studied
through photoluminescence studies up to 2.8 GPa by Sood and co-workers [388,
390], who found a very similar energy shift of dE /dp = - 0.09 eV GPa- 1 . However,
after compression to greater than 20GPa the bandgap is observed to be about 0.1eV
larger than in the initial sample, possibly indicating a small fraction of polymerized
material. Finally, Lang et al. [396] recently reported studies of the luminescence
spectrum of C70 up to 10 GPa, showing that the luminescence line sharpened and
shifted to lower energies. However, these studies were carried out on molecular C70
dissolved in poly(methylmethacrylate) and the extent to which they re¯ ect the
properties of solid C70 is uncertain. Clearly, less information is available for C70
than for C60 , and further measurements on well characterized materials are needed to
improve our understanding of the band structure under high pressures.

7. Other fullerenes and fullerene compounds


7.1. Nanotubes
In addition to the small quasispherical fullerenes C60 and C70 discussed above,
the fullerene family in a wide sense also includes the heavy carbon `onions’ brie¯ y
discussed in section 5.6.4 and the very long thin nanotubes of buckytubes which can
be of either a single-walled or a multiple-walled nature. The nanotube ® eld, which
was very recently reviewed in detail by Ajayan and Ebbesen [397], is at present in
very rapid development. Methods have recently been found to produce well de® ned
material, both pure single-walled nanotubes and metal-® lled tubes, to dope them by
intercalation of alkali metals, and to measure the resistance of individual nanotubes,
even of the single-walled variety. However, no real high-pressure studies have yet
been reported, except for a study showing that the closely related BN nanotubes can
be formed by laser heating under N2 pressures of 5± 15 GPa [398]. While the
properties of pure nanotubes will probably not be signi® cantly a€ ected by pressures
in the gigapascal range, because the bulk modulus will be similar to that of the C60
molecule (see section 4.2.1), those of doped nanotubes probably will.
Although no high-pressure studies have been carried out so far, some informa-
tion is available on the behaviours of nanotubes under stress. Yakobson et al. [399]
® rst used molecular dynamic simulations to study buckling of nanotubes under
uniaxial (longitudinal) stress. The results showed that the microscopic properties of
nanotubes were very similar to those of a rolled-up graphite sheet and Yakobson et
al. calculated Young’s modulus Y = 5. 5 TPa, a very high value indeed. (As a
comparison, the in-plane value for single-crystal graphite is about 1 TPa.) Interest-
ingly, interatomic bonds were shown to be stable even under extreme conditions, and
tubes could be deformed elastically to extreme shapes or elongated by up to 40%
before rupture occurred. On the experimental side, qualitative studies of nanotubes
stressed by the shrinking of a polymeric substrate were carried out by Ruo€ and co-
workers [400± 402]. Later, Treacy et al [403]measured Young’s moduli of individual
carbon nanotubes by observing the amplitude of their thermal vibrations in a
transmission electron microscope. The measured data varied signi® cantly between
di€ erent samples, and numerical results varied between 0.40 and 3.7TPa, with an
average near 1.8 TPa. Treacy et al. argued, however, that all reasonable experimental
errors tend to underestimate the true modulus, which should thus fall in the upper
94 B. Sundqvist

end of the measured interval, that is very close to the value obtained from the
molecular dynamics simulations [399]. Very recent data obtained by direct bending
of nanotubes with an atomic force microscope gave a modulus of 1.28 TPa [404],
which agrees with the results of Treacy et al. within the experimental error. However,
the most surprising property of nanotubes is their enormous toughness and
resilience, and it has been shown both experimentally and theoretically that they
can be bent with a very small radius of curvature without breaking and show a very
interesting buckling behaviour [399± 405]. Recent theoretical calculations also tend to
verify that Young’s modulus is close to 1 TPa [406, 407]. Lu [407]also calculated that
the bulk modulus should be about 190GPa, which veri® es the assumption made
above that pressure should have a very small e€ ect on the properties of these
interesting materials.

7.2. C60O, C61H2, C60Hx and (C59 N) 2


Although a very large number of fullerene derivatives and compounds have been
produced, only a small number of these have been investigated under high pressures.
The compounds C60 O and C61 H2 have very small side groups attached to the basic
C60 soccerball. Both have been quite well studied at zero pressure, both experi-
mentally [408± 412] and theoretically [413], and these experiments have shown that
their rotational properties are surprisingly similar to those of C60 although the
addition of the side group changes the rotational properties of the molecules. The
results may thus have a direct bearing on the understanding of the rotational and
orientational properties of C60 and C70 . Both C60 O and C61 H2 have the same
structural evolution with temperature as C60, with a high-temperature fcc phase,
turning into a low-temperature orientationally ordered phase and ® nally into an
orientational glass at low temperatures. The rotational dynamics are somewhat more
complicated than for C60 since the molecular rotation consists of a rapid rotation
around the long axis of the molecule, while the axis itself points towards any of the
neighbouring interstitial voids in the lattice and jumps much more slowly between
these directions. Because of the increased intermolecular interaction due to the side
groups the transition temperatures are somewhat higher than for C60, the relative
energies of the minority and majority orientational states are di€ erent from those in
C60 and, at least in C61 H2 [410], there are three di€ erent orientational states in the sc
phase with temperature-independent population factors.
A high-pressure compressibility study of C61 D2 (deuterated C61 H2) has been
carried out by Lundin et al. [414, 415] in an attempt to map the pressure±
temperature phase diagram. These studies showed that C61D2 was slightly more
compressible than fcc C60 at all temperatures, but the bulk modulus (table 10) was in
fact very similar to what would be expected for hypothetical expanded C60 with the

Table 10. Zero-pressure bulk modulus B( 0) and its pressure deriva-


tive B = dB /dp for C61D2 at selected temperatures (from [414]).
Â
T B(0)
(K) (GPa) B Â
343 5.2 13.9
307 6.4 16.0
175 6.7 17.7
Fullerenes under high pressures 95

Figure 48. Volume against pressure for C61D2 at (from the top curve down) 175, 307 and
343K. The inset shows deviations from the ® tted equation state at 175K. (Reprinted
from Lundin et al. [414].)

same lattice constant as C61 D2 . However, the experiment failed to give a reasonable
map of the phase diagram. The zero-pressure value for the glass transition
temperature Tg is about 140K [411, 412] but our low-temperature studies failed to
detect any signi® cant anomaly that could be related to Tg . However, the volume
expansion anomaly at Tg is very small [411, 412], and a very small deviation from a
smooth compression behaviour was indeed observed near 0.3 GPa at 175K (® gure
48). If this anomaly corresponds to the glass transition, the pressure slope must be
unusually large with dTg /dp = 90± 120KGPa- 1 . More surprisingly, near and above
RT no trace was found of the expected orientational transition which at zero
pressure occurs at 290K. Instead, a temperature-independent behaviour with a very
large reversible initial compression step up to 0.2 GPa was observed in this range,
similar to that found in the uniaxially rotating phase of C70 by Lundin et al. [379,
380] (® gure 47). For C70 , this behaviour was attributed to molecular reorientation
with increasing pressure, but the extent to which the C61 H2 lattice would gain energy
by an orientational ordering under an applied pressure is uncertain. Clearly, further
structural studies of C61H2 under pressure are needed in order to explain this
interesting behaviour.
For C60O no high-pressure studies have yet been carried out. However, Soldatov
and Andersson [274] observed by mass spectrometry that their high-pressure-
polymerized C60 sample contained a small amount of C60 O, possibly created by
reaction between the sample and O2 gas intercalated while ® lling the pressure cell.
They correlated the presence of C60 O with the appearance of previous unexplained,
well de® ned anomalies in the measured thermal conductivity data. The temperature
at which these anomalies occurred was pressure dependent and extrapolated to the
known orientational ordering transition temperature 285± 290K [408] for C60 O at
96 B. Sundqvist

normal pressure. If this interpretation is correct, the orientational transition


temperature shifts with pressure at a rate of dT /dp = 130KGPa- 1 , not very far
from that of the corresponding transition in C60 (section 4.1.2).
Kolesnikov et al. [416, 417] have recently subjected C60 to high hydrogen
pressures at high temperatures and observed the formation of `hydrofullerites’
(fullerene hydrides) C60Hx with x > 24. Pure C60 was treated at 620K in H2 gas
at 0.6 GPa for 24 h and then quenched to 80 K before decompression. As-treated
samples stored in liquid N2 were found to contain both C60 H24 and some interstitial
H2 , which di€ used out of the lattice on annealing at 300K for 35h. The lattice
structure of the as-treated and annealed C60 H24 was bcc with lattice constants of
12.00A Ê and 11.72A Ê respectively. (The formation of C60Hx, with 2 < x < 18, had
previously been observed at 620K under H2 pressures of 50± 85 MPa [418].)
Kolesnikov et al. also investigated both quenched and annealed samples by inelastic
neutron scattering. The addition of hydrogen atoms on the surface of the C60 cages
does not change signi® cantly the density of states for low-frequency (intermolecular)
libration and phonon modes up to 8meV in the lattice, in spite of the change from a
sc to a bcc lattice, and thus a decrease in the number of nearest neighbours from 12
to eight, and changes in the electron band structure. For the intramolecular modes,
however, signi® cant di€ erences can be observed between C60 and C60H24 , probably
because the addition of hydrogen atoms decreases the average strength of the bonds
on the molecule. Also, near 155meV a very strong peak attributed to C± H bending
modes appears for C60H24 .
Fullerenes may be modi® ed on a molecular scale not only by the addition of side
groups, but also by substitution of atoms in the molecule. Hummelen et al. [419]
recently substituted one carbon atom per molecule by nitrogen to form the
azafullerene compound C59 N, which spontaneously transforms into the more stable
dimer (C59 N) 2 . In the solid state, the most stable structure of this material seems to
be a hexagonal lattice with a c /a ratio of 1.623, close to the ideal value of 1.633,
although less well ordered structures consisting of mesoscopic hollow spherical
particles have also been reported. Brown et al. [420] have recently studied the RT
equation of state of the hexagonal phase up to 22 GPa by X-ray di€ raction. Because
the dimer axis is not parallel to any of the crystallographic axes, the lattice is
compressed almost isotropically under pressures such that the c /a ratio approaches
the ideal value near 4.5 GPa and reaches a stable value of 1.634 at all pressures
greater than 6.5 GPa. In this high-pressure state the interdimer distance should be
approximately identical with the distance of 9.4A Ê between the C59N units in the
dimer, indicating that the intermolecular bonding has become almost isotropic. It is
therefore uncertain whether actual dimers still exist above 6.5 GPa or whether a new
phase with free monomers has been created. Because the molecule itself is again
almost incompressible compared with the lattice, interference between the molecular
form factor and lattice scattering causes several di€ raction peaks to disappear in
certain pressure ranges. Fitting data for V against pressure to the Murnaghan
equation of state, Brown et al. found an isothermal bulk modulus B(0) = 21.6 GPa,
with B = 4.2. A comparison with the data for the bulk modulus of C60 given in table
Â
2 and with the discussion in section 4.2.1 shows that (C59N) 2 has a bulk modulus
about 50% larger than that of C60 , while the linear chain polymer (C60 ) n (orh C60)
has a bulk modulus 50% larger than (C59 N) 2 , in agreement with intuitive qualitative
expectations.
Fullerenes under high pressures 97

7.3. Other C60 complexes and compounds


Some scattered data also exist for various complexes and acceptor intercalation
compounds of C60 (data on donor compounds are discussed in section 8). The
resistivity of the acceptor intercalation compound C60I4 [421] was measured by
Nu  nÄ ez-Regueiro et al. [219]over a wide range in temperatures and between 1.4 and
23GPa. As discussed in section 4.2.5, the best ® t to the data was obtained in a
variable-range hopping model. Although no charge-transfer e€ ects have been
reported for this compound [422], the sample resistance was somewhat lower than
for C60 possibly for geometric reasons. Up to about 7 GPa the resistance decreased
exponentially by four orders of magnitude, that is a behaviour not very di€ erent
from that of C60 (section 4.2.5), but above 7 GPa a very much weaker pressure
dependence was observed. However, the model used indicates that the density of
hopping states increases continuously with increasing pressure without any large
anomalies between 5 and 10GPa. Akahama and co-workers [423, 424] studied the
resistivity of the similar iodine-doped compounds C60 Ix up to 36 GPa. Values of
x = 1.6 and 2 were deduced from chemical analysis and structural studies respec-
tively. The logarithm of the resistance decreased linearly with increasing pressure
between 6 and 13GPa, in general agreement with the data of Nu  nÄ ez-Regueiro et al.
[219]. Above 13 GPa the decrease is less rapid, and after a minimum near 27GPa the
resistance is observed to increase again, indicating a transition to a less conducting
phase. The apparent bandgap deduced from the temperature dependence is about
0.3 eV, similar to that observed for pure C60 (see section 4.2.5).
Matsuzaki et al. investigated the optical absorption [425]and IR spectra of C60
complexes with the organic metal bis(ethylenedithio)tetrathiafulvalene (BEDT-TTF)
and with ferrocene at RT under pressures up to 6GPa. The materials were produced
by co-crystallization of the components by evaporation of the solvents used and were
found to be electrically insulating at zero pressure. The absorption band(s) shifted
slightly with increasing pressure while in both compounds a new absorption band
appeared near a wavelength of 100nm under pressures of 0.8 and 3 GPa in the
ferrocene and BEDT-TTF compounds respectively. Matsuzaki et al. concluded that
a high pressure causes strong changes in the electronic structure, and that the new
absorption band arises from a change in ionicity such that the band is connected
with the formation of C60 - ions. The IR frequencies observed [425] from the C60
molecules are identical with those for pure C60 and have the same pressure
dependence as given in table 5 (approximately - 1 cm- 1 GPa- 1, 2.9 cm- 1 GPa- 1
and 3 cm- 1 GPa- 1 for the modes at 526cm- 1 , 576cm- 1 and 1183cm- 1 respectively)
at low pressures, showing that no strong covalent bonds are formed under an applied
pressure. However, above the transition pressures given above, changes are observed
in dx /dp for many modes, showing a change in bond character. For the ferrocene
complex a new mode also appears near 4 GPa, indicating a further change in the
structure of the compound.
Another interesting and by now quite well investigated material is the C60
complex with tetrakis(dimethylamino)ethylene (TDAE), C2N4 (CH3 ) 8 , which is
reported to be ferromagnetic below about 16 K [426]. The magnetic properties of
C60 ± TDAE are extremely sensitive to pressure [427, 428] such that no trace of any
magnetic transitions could be observed even at 0.16 GPa. The transition temperature
seems to decrease under pressure at an average rate dT /dp < - 100KGPa- 1 while
the magnetic moment decreases even more rapidly. As also discussed by Wudl and
Thompson [429], this very strong pressure dependence can be explained under the
98 B. Sundqvist

assumption that C60± TDAE is a metal and an itinerant-electron ferromagnet. In this


model the ferromagnetism might arise from a band anisotropy that is rapidly
destroyed on compression of the soft lattice.

7.4. Intercalation of gases into C60


One of the most common problems in high-pressure studies is the choice of
pressure-transmitting medium. The ideal medium should not react with the sample
under investigation and should be as hydrostatic as possible up to the highest
possible pressure, and for optical studies it should also be transparent over the
maximum possible range in wavelength. The ¯ uid that usually ® lls these require-
ments best is helium gas (or liquid), and helium is thus generally considered as the
`best’ pressure medium available although it also has large practical disadvantages
regarding handling and leakage. The ® rst studies of the phase diagram of C60 [69, 70]
were therefore carried out using helium gas as the pressure-transmitting medium.
However, later studies have shown that this was an unfortunate choice since helium
and several other light gases and ¯ uids are able to di€ use into (and out of ) the
interstitials in the C60 lattice. The resulting intercalation compounds are less
compressible than the original C60 lattice since the foreign atoms or molecules ® ll
out the interstitials and resist compression, and the properties of the new materials
are thus di€ erent from those of pure C60 . This section discusses reversible
intercalation of noble and other gases into C60 under pressure, while the formation
of other more or less stable intercalation compounds of C60 are brie¯ y discussed in
section 7.3 (acceptor compounds) and section 8 (donor compounds), and the
formation of endohedral compounds in section 7.5.
For obvious reasons, the gases that have been most studied are the noble gases
and the main atmospheric components oxygen and nitrogen. The former are of
interest because they are often used as pressure media and to protect samples during
handling and storage, and the latter because samples are often exposed to the
atmosphere during handling and experiments. However, a few studies of other gases
have also been carried out. I mentioned in section 7.2 that hydrogen easily penetrates
the C60 lattice, and that heat treatment of C60 under a rather low H2 pressure results
in the formation of fullerene hydrides, C60Hx , with a simultaneous intercalation of
H2 molecules into the intermolecular voids in the C60 lattice. The latter e€ ect also
appears on pressurization at RT. Kolesnikov et al. [416, 417] showed that two to
three H2 molecules could be intercalated per C60 molecule at 0.6GPa and also
carried out neutron scattering studies showing that the intercalated H2 molecules did
not form bonds with the C60 cages but were free to rotate. Transitions between
rotational states in the H2 molecules could be clearly observed at 85K. Assink et al.
[430] also observed intercalation of hydrogen into C60 and C70 , but at much lower
pressures (14MPa).
The intercalation of noble gases with C60 has been studied by Schirber and co-
workers [72, 431], primarily in an attempt to explain the results from an earlier study
of the phase diagram of C60 [71]and to reconcile the widely scattered available data
on the compressibility (see section 4.2.1). The results from these investigations have
recently been reviewed by Kwei and co-workers [432]and can be brie¯ y summarized
as follows. Argon, which has an atomic radius of 1.54 A Ê , does not intercalate into
C60 to a measurable extent even when the sample is kept under argon gas at 0.6 GPa
for 6 days, and measurements in argon can thus be expected to re¯ ect the intrinsic
properties of C60. Helium on the other hand, which has an atomic radius of 0.933A Ê
Fullerenes under high pressures 99

is found to di€ use into (or out of ) the interstitials in the C60 lattice on a time scale of
minutes or shorter, too short to be measurable on the time scale (hours) of the
neutron scattering experiment. Neon, ® nally, with its intermediate atomic radius of
1.12A Ê , takes up an intermediate position, di€ using into the lattice with a time
constant of about 1 h in fcc C60 and 5 h in sc C60 [431, 432]. Strangely enough, neon
was observed [72, 432] to di€ use out of the lattice much more rapidly, a very
interesting phenomenon that is still unexplained. One possibility is that the C60
lattice is simply unable to contain the tension from the internal neon pressure and
breaks up on a microscopic scale to release the gas. Both neon and helium have
strong e€ ects on both the compression behaviour (section 4.2.1) and the phase
diagram of C60 . For fcc C60 , the `best’ value for the zero-pressure bulk modulus was
given above as 9.6GPa, but intercalation results in a signi® cant increase in B(0) to
12.5GPa in helium and to 12 GPa in neon [72]. A similar e€ ect was noted in the sc
phase. In all noble gases the fcc-to-sc transformation occurs at a lattice constant of
14.065A Ê at RT [72] but, since the presence of di€ erent molecules gives di€ erent
lattice expansions and bulk moduli, this means that the fcc-to-sc transformation
occurs at about 0.255GPa, 0.305GPa and 0.325GPa in argon, neon and helium
respectively at RT. The change in the transformation pressure was veri® ed in quite
an elegant experiment where a sample was rapidly compressed in neon until it had
transformed into the sc phase near 0.275GPa [72]. With time, neon di€ used into the
lattice, increasing the lattice parameter and transforming the sample back into the
fcc phase. A later more detailed study by Morosin and co-workers [431]showed that
the neon atoms occupied the octahedral sites only, as might be expected from the
radii of the octahedral (greater than 2 A Ê ) and tetrahedral (1.1 AÊ ) sites. This study also
showed that the equilibrium interstitial site occupancy, which is basically zero for
argon and one for helium even at 1 atm varied rapidly from 0.2 at zero pressure to
one above 0.25 GPa, and that the lattice parameter, rather than the lattice volume,
increased linearly with increasing neon content. While neon thus penetrates the
lattice of pure C60 relatively easily, intercalation with alkali metals ® lls the
interstitials in the lattice and completely stops neon intercalation [433, 434]. Tests
show that no signi® cant amounts of neon penetrate into Na2CsC60 even after several
weeks at 0.5 GPa. The intercalation and release rates also depend on the tempera-
ture, such that the release of neon is slow enough to be easily observed at
temperatures below 200K. Grube [73] showed that even helium di€ used only very
slowly into C60 at low temperatures. Below about 180K, helium intercalation
became too slow to be observed over the time scale of a typical experiment (about
24h) and by cooling samples to su ciently low temperatures before the pressure cell
was ® lled with gas, Pintschovius et al. [74]were able to use helium gas as the pressure
medium in their low-temperature compressibility studies. It is thus possible that gas
di€ usion is mediated by molecular rotation or ratcheting in the C60 lattice.
Above RT, gas di€ usion into C60 becomes even easier and, at or above 473K,
argon easily di€ uses into the C60 lattice even at 0.17 GPa to form the ordered
compound ArC60 [435± 437] which is subsequently stable for a long time under
ambient conditions. Because of the weak interaction between the argon atoms and
the host lattice the properties of ArC60 are very similar to those of C60. The IR
spectra are almost identical [435], and ArC60 has a lattice parameter only slightly
larger than that of C60 [437]. Since the fcc-to-sc transition occurs at the same lattice
parameter as in C60 (cf. above), the transition temperature is somewhat lower, 250K,
100 B. Sundqvist

for ArC60 . The thermal pressure of the argon atoms also makes the thermal
expansion somewhat higher in ArC60 .
Regarding atmospheric gases, oxygen readily intercalates into the C60 interstitials
while the nitrogen molecule is too large and does not. The e€ ect of intercalated
oxygen on C60 has also been studied by several methods. Assink et al. [430, 438]
showed that NMR studies on C60 stored in air showed a small extra resonance in
addition to the normal C60 line at 143.7ppm. The amplitude of this extra line
decreased when the sample was stored in a vacuum or in nitrogen but increased in
oxygen and its temperature dependence showed a paramagnetic behaviour. Assink et
al. concluded that the line was due to the presence of intercalated oxygen. After the
sample had been submitted to oxygen at 100MPa for 134 h several resonances were
observed, and the number of resonances was found to correlate with the fraction of
octahedral intermolecular sites ® lled with oxygen. At atmospheric pressure, only a
very small fraction (about 1%) of the available sites are ® lled in equilibrium while, at
100MPa, about 50% are ® lled. The di€ usion of oxygen in the C60 crystal is very
slow, however, and only in ® ne powder can equilibrium be reached within reasonable
times. Later, more detailed studies have veri® ed this model [439]. The presence of
oxygen in the lattice has a large e€ ect on the resistance (see references in section
4.2.5) and has also been shown to modify signi® cantly the thermodynamic properties
and even the phase diagram by depressing the fcc-to-sc transition temperature by
several kelvins [440± 443], and prolonged exposure to oxygen under pressure,
especially at elevated temperatures, has been reported to cause amorphization of
the original material [444]. The mechanism for these e€ ects has not yet been
identi® ed. It has also been suggested that the double fcc-to-sc transition sometimes
observed is due to partial intercalation of oxygen [443]. Studies have also been
carried out on the intercalation of nitrogen [440, 442] and, very recently, carbon
monoxide [445]into C60 under pressure.

7.5. Endohedral fullerenes


Very soon after the discovery of C60 it was realized that the hollow molecule
might be used as a container for other atoms or molecules, and that such endohedral
compounds could have interesting physical or chemical properties [3]. In principle,
metal-® lled fullerenes might show a very interesting doping behaviour and should be
metallic because of charge transfer from the captured metal atom to its cage. While a
number of compounds of this type have been produced, mostly ® lled with one or
more metal atoms such as yttrium or lanthanum these materials have not yet been
available in su cient amounts that high-pressure investigations have been made
possible. This type of material is usually produced by evaporating metal-® lled
graphite rods in a standard fullerene reactor [3] but quasibulk materials can also
be built up by successively irradiating thin C60 layers by high-energy metal ions [446].
Separation is then usually carried out by multiple-step liquid chromatography, but
the yield of endohedral compounds is very small. It should also be noted that giant
fullerene `onions’ can be produced by the arc method, and under suitable conditions
these can be ® lled with signi® cant amounts of foreign materials, for example metal
carbides. The carbon walls protect the ® ller compounds from oxidation, potentially
making possible the observation of the properties of new exotic compounds.
Another type of endohedral compound is that in which cooling gas molecules
such as helium become trapped in a fullerene cage during production or are inserted
into the cage through ion bombardment [3]. The detection of such noble gas
Fullerenes under high pressures 101

compounds is not easy but can be carried out using either mass spectroscopy or
NMR [447], and it has been shown [448]that during normal production conditions
about one in every 106 fullerene molecules produced contains a helium atom. These
compounds are stable at RT but release the trapped atoms on heating, presumably
by brie¯ y opening a `window’ in the molecular wall [448]. The latter observation
prompted experiments [449± 451] in which fullerene solids were submitted to high
noble-gas pressures near 0.3 GPa at temperatures near 925K, after which quite a
high proportion, of the order of 0.1%, of the fullerene molecules were found to
contain noble-gas guest atoms. This e€ ect of a high pressure is of course completely
di€ erent from the `normal’ exohedral intercalation of noble gases observed by
Samara et al. [71] and discussed in section 7.4. Experiments showed that helium,
neon, argon and krypton all formed endohedral compounds in approximately the
same amounts, while the larger xenon molecule evidently did not ® t into the C60
cage. A discussion of the properties of these materials and their chemical derivatives,
as well as their possible applications, has been given by Saunders et al. [447]but is
outside the scope of this paper. Finally, it has very recently been shown [452] that
argon gas can penetrate into the axial voids of multiple-walled carbon nanotubes at
650ë C under a pressure of 0.17 GPa, and to remain trapped under ambient
conditions for several months, suggesting a potential for these materials to be used
for gas storage.

8. Alkali-metal- and alkaline-earth-metal-doped C60


The early discovery by Haddon et al. [453] that C60 , like graphite, could be
intercalated with alkali-metal atoms to form metallic compounds immediately
sparked much interest. This interest then increased signi® cantly when some of these
compounds turned out to be not only metallic but also superconducting [454], with
critical temperatures Tc higher than any other known superconductor except the
ceramic cuprates and thus in the `high-Tc ’ range. The study of donor intercalation
compounds of C60 has already developed into a separate research sub® eld with a
large number of papers appearing annually, and a number of general and specialized
reviews have appeared (e.g. [3, 16, 18, 179, 455± 465]. The intercalation compounds
discussed in this paper have compositions AxC60 with 0< x< 6, where A can be any
alkali metal, alkali-metal alloy or alkaline-earth metal (barium, calcium, etc. ). Most
studies in this ® eld have been carried out on the metallic superconducting phases
such as A3 C60 , but some studies have also been carried out on AC60 and A4 C60 . The
latter is normally a semiconducting solid while the former usually has a very
complicated phase diagram even at zero pressure and can be obtained in the form
of a metallic monomer, an insulating dimer or a conducting air-stable polymer,
depending on temperature and treatment conditions [462, 463]. The polymer is of
particular interest, since it has an orthorhombic structure believed to be very similar
to that of the orthorhombic photopolymerized and pressure-polymerized materials,
except for the presence of alkali-metal ions between the linear C60 chains. More
information can be found in the references above. A large number of other metal-
doped materials with interesting properties exist but have not yet been studied under
high pressures.
High pressure can be used as a tool not only to study the physical properties of
doped C60 but also to produce doped materials. This process has been used in several
cases to obtain highly dense intercalation compounds of graphite, but so far only a
102 B. Sundqvist

very small number of preliminary experiments has been reported for fullerenes [466,
467]. Menu et al. [467]reported that high-temperature reactions of C60 with lithium
or sodium at pressures above 4.5 GPa resulted in the formation of compounds
AxC60, with x = 21 for sodium and x = 50 for lithium deduced from NMR
measurements. The successful intercalation of large amounts of lithium (two atoms
of lithium per atom of carbon) into multiple-walled nanotubes has also been
reported [466, 467], and results from X-ray di€ raction, NMR and IR spectroscopy
indicate that the lithium atoms mainly take up positions between the walls of the
concentric nanotubes. However, the presence of a small amount of metallic lithium
inside the tubes cannot be excluded.

8.1. Normal-state properties


8.1.1. Bulk modulus
Starting again with the most basic of the high-pressure properties, the compress-
ibility of alkali-metal-intercalated C60 is lower than that of pure C60 because of the
interaction between the fullerene molecules and the alkali-metal ions in the lattice
interstitials. A similar but smaller e€ ect was noted above for C60 intercalated by
noble-gas atoms. The compression properties of K3 C60 and Rb3 C60 were ® rst
measured by X-ray di€ raction under hydrostatic pressures up to 2.8 GPa in a
DAC at 300K by Zhou et al. [468]and were very important in verifying that there
exists a strong quasilinear correlation between Tc and the lattice parameter (see
below). The decrease in lattice parameter with pressure was linear up to near 1.3 GPa
and the rubidium-intercalated material was more compressible, as might be expected
from its larger lattice parameter at zero pressure. The compressibility of Rb3C60 was
later remeasured by Ludwig et al. [469]up to 6 GPa using a similar technique. These
data gave a somewhat larger bulk modulus in the range above 1 GPa. However, as
discussed in section 4.2.1, Ludwig et al. found that standard equations of state are
unable to describe the compression properties of fullerene materials, and they
analysed their data in terms of a modi® ed equation of state where the volume of
the incompressible molecules had been subtracted. Using their modi® ed equation of
state they found an initial bulk modulus B(0) = 20.5 GPa for the compressible part
of the crystal. However, for the crystal itself a graphic analysis of their raw data gives
an initial bulk modulus near 27GPa. Finally, Diederichs et al. [470] have recently
carried out neutron di€ raction studies of the compressibility of the same material at
both a low temperature (14 K) and RT up to 1GPa using Fluorinert as a hydrostatic
pressure medium and NaCl as an internal pressure standard. A comparison between
data from di€ erent groups is shown in ® gure 49 (reprinted from Diederichs et al.
[470]), and indicates that the data of Ludwig et al. give slightly larger values for the
bulk modulus than other data, which are in rather good agreement. (Such an e€ ect
was also observed in the data for pure C60 (section 4.2.1.) Diederichs et al. also gave
theoretical values for the bulk modulus as functions of temperature from 0 to 400K.
As will be shown below, compression data are rather important for the under-
standing of both superconductivity and the electrical resistivity in these materials.
Table 11 collects all available data in this ® eld, and already from the small number of
entries it is clear that much further work is needed before complete understanding
can be reached. Good data exists only for the two materials already discussed, for
Na2CsC60 [434], which has been suspected to polymerize under pressure [474] and
will be discussed further in section 8.2, and for the linear-chain polymers KC60 and
RbC60 [471]. A comparison with the data in table 2 shows that the bulk moduli of the
Fullerenes under high pressures 103

Figure 49. Measured relative decrease in volume against pressure for Rb3C60. (Reprinted
with permission from Diederichs et al. [470].)

Table 11. Measured and calculated bulk moduli B for metal-intercalated C60

T B(0) pmax
Material (K) (GPa) B Â (GPa) Reference

KC60 300 40a Ð 5 Khazeni et al. [471]


RbC60 300 58a Ð 5 Khazeni et al. [471]
K3C60 300 28 Ð 2.8 Zhou et al. [468]
K3C60 300 24b 9b > 25 Martins and Troullier [472]
K3C60 300 28. 4b 3. 41b >5 Huang et al. [473]
Rb3C60 300 22 Ð 2.8 Zhou et al. [468]
Rb3C60 300 20. 5c 4. 5c 6 Ludwig et al. [469]
Rb3C60 300 28d Ð 6 Ludwig et al. [469]
Rb3C60 295 17.4 3.9 0.91 Diederichs et al. [470]
Rb3C60 14 18.3 3.9 0.73 Diederichs et al. [470]
Na2CsC60 300 21. 2e Ð 0.45 Morosin et al. [434]
Na2CsC60 82 21. 2e Ð 0.35 Morosin et al. [434]
Na2CsC60 83 28. 9 f Ð 0.35 Morosin et al. [434]
a
Polymer phase.
b
Theoretical value.
c
From a modi® ed equation of state [141, 469].
d
From the measured curve of V against p.
e
Cubic phase.
f
Orthorhombic (polymerized?) phase.

two latter materials are much higher than that of C60, and even signi® cantly larger
than that for the orh C60 phase containing linear polymer chains, re¯ ecting the fact
that the alkali-metal atoms increase the repulsive interaction between the chains.
The bulk modulus has also been calculated theoretically. Martins and Troullier
[472] carried out a local-density pseudopotential calculation of the volume against
104 B. Sundqvist

pressure. They ® tted their data to three di€ erent equations of state and found zero-
Â
pressure bulk moduli between 22 and 24 GPa, with values for B between 9 and 14.
The results obtained using the Murnaghan [140]equation of state are shown in table
11 in order to compare the results with both experiments and with the local-density
calculation by Huang et al. [473], which gave B(0) = 28. 4 GPa with B = 3. 41.
Â
8.1.2. Electrical resistivity
The normal-state transport properties of alkali-metal-doped C60 at both atmos-
pheric and high pressure have recently been reviewed by Zettl [463]. Starting with the
metallic A3 C60 materials, early studies on alkali-metal-intercalated powders and
® lms showed their resistivities q to be low and in the metallic range, but it was not
until single-crystal K3 C60 became available that it was proved that q had the positive
temperature coe cient typical for metals [475]. The magnitude of q near RT was
found to about 5 mV cm, rather high for a metallic material, but the temperature
dependence was super® cially very similar to that for normal metals with a constant
value near 0K and an increasing slope at higher temperatures. However, instead of a
linear dependence on temperature as in standard metals, q conformed instead to a
function
q = a + cT 2 (8.1)
from below 100 to at least 600K [476], unless saturation intervened, indicating that q
might be completely dominated by electron± electron scattering. This would also
indicate that the electron± phonon interaction would be small enough to rule out the
standard Bardeen± Cooper± Schrie€ er (BCS) mechanism for superconductivity, and
various attempts have therefore been made to reconcile theory with experiments by
introducing unusual scattering mechanisms for the electrons or additional scattering
by high-frequency optical phonons.
Later high-pressure studies on Rb3 C60, however, have revealed that pressure has
a very large e€ ect on the resistivity of these soft materials [463, 476± 478]. Since the
thermal expansivity is also large, there is a signi® cant di€ erence between the
resistivities at constant pressure (which is usually measured) and that at constant
volume (which is what theory usually predicts), as also independently predicted by
Sundqvist and Nilsson [479]. Vareka and co-workers [476± 478] measured the
resistivity of a single crystal of Rb3 C60 up to hydrostatic pressures above 0.8 GPa
over a rather large interval in temperature. The resistivity proved to be very strongly
and nonlinearly dependent on pressure, and both the temperature-dependent
(`phonon’ ) and the temperature-independent (`impurity’ ) terms decreased rapidly
under pressure [476, 478]. At RT, the total resistance decrease was about 50% to
800MPa. Using data for the thermal expansion and the bulk modulus from the
literature they were able to show that within the experimental error the resistivity at
constant volume was, in fact, a linear function of temperature from 100 to 350K, as
would be expected for a metallic material with a Debye temperature H D well below
100K such as Rb3 C60. Their results are shown in ® gure 50, reprinted from [477]. The
large di€ erence between the behaviour at constant volume and at constant pressure
is not an unusual phenomenon; it is a well known fact, strangely neglected in most
textbooks, that the measured resistivity of the alkali metals is not linear in T as
predicted by theory but proportional to T 1.1- 1.3 [480]because of thermal expansion
e€ ects, as discussed by Mott and Jones [481]. (Note that this is not connected with
the small change in geometric factors, such as the cross-sectional area. The question
Fullerenes under high pressures 105

Figure 50. Normal-state resistivity of Rb3 C60 at constant pressure and at constant volume.
(Reprinted with permission from Vareka et al. [477]. )

is simply: what pressure must be applied at high temperatures to decrease the volume
to its initial value?) Similar e€ ects also occur in many other materials, such as
graphite intercalation compounds and high-transition-temperature superconductors
[479, 482].
The results are of great importance to the theoretical analysis of the electron±
phonon interaction in the material. At high temperatures T > H D the resistivity
should be a linear function of T and [483]:
8p 2¸trkB T
(T) = (8.2)
x 2ph .
q

From the constant-pressure data of Vareka and Zettl [478] a very small electron±
phonon interaction factor ¸tr is deduced. However, using the data obtained for the
constant-volume case [478] together with experimental values for the plasma
frequency x p it is possible to deduce values ¸tr = 0.66± 0.8, values which indicate
quite high transition temperatures and agree well with estimates from the measured
Tc. This illustrates the importance of always correcting experimental data to
constant-volume conditions before a comparison is made with theoretical models
or calculations.
A theoretical calculation of the pressure dependence of several normal-state
transport properties of K3 C60 has also been carried out from ® rst principles by Erwin
and Pickett [484]. The found that q should be a linear function of temperature, in
excellent agreement with the constant-volume data given above for Rb3 C60 . The
calculated pressure coe cient at RT was d(ln q ) /dp = - 0. 41 GPa- 1, in surprisingly
good agreement with the experimental result for Rb3 C60 . Erwin and Pickett also
calculated the pressure coe cient of the Hall coe cient RH as d(ln RH) /dp =
- 0.03 GPa- 1, and the-pressure derivative of the thermopower (Seebeck coe cient)
S as dS /dp = 10 m V K 1 GPa- 1. No data yet exist for either S or RH under pressure,
but Zettl [463]showed that RH is the same linear function of lattice constant in both
K3 C60 and Rb3C60. These data indicated that the true pressure dependence of RH
may be larger than suggested by Erwin and Pickett.
106 B. Sundqvist

Figure 51. Relative resistance of KC60 as a function of temperature at the pressures


indicated. ( Reprinted with permission from Hone et al. [486]. )

Recently, Hone and co-workers [471, 485± 487] have also carried out resistance
measurements on the orthorhombic polymeric form of KC60 under an applied
pressure. This material is air stable and shows a metallic conductivity from 50 to
400K at zero pressure. Again, the resistivity against temperature shows a strong
curvature but a correction to constant volume gives a basically linear dependence on
temperature [471, 485, 487]. Figure 51, taken from [486], shows the relative resistance
against temperature at several pressures. At zero pressure the slope dq /dT changes
sign near 50K from positive above to negative below, indicating some type of
transition in the material. This transition has also been observed in other types of
experiment and it has been speculated that the anomalies observed are caused by the
formation of a spin- or charge-density wave on the molecular chains, opening a gap
at the Fermi surface. However, for KC60 the observed resistivity below 50K is linear
in temperature and does not show the low-temperature exponential behaviour
expected in a material with such a gap. The application of a pressure at RT results
in a rapid nonlinear decrease in the resistivity, with an initial pressure coe cient
d(ln q ) /dp < - 1.5 GPa- 1 which very rapidly decreases to a much lower value above
0.2 GPa [471, 485, 487]. Near 1.5 GPa the resistivity is almost linear in temperature,
as shown in ® gure 51, and the anomaly below 50 K seems to have disappeared. The
decrease in transition temperature can be followed from the resistance curves at
intermediate pressures and extrapolates to zero near 0.45 GPa. For orthorhombic
RbC60 the resistivity shows a `semiconductor-like’ behaviour at zero pressure with a
transition anomaly near 200K, but the application of a pressure rapidly leads to a
transition into a metallic state [471, 486, 487]. Again, in this state the resistivity at
constant volume is linear in temperature [487]. At zero pressure, RbC60 has been
shown to have an antiferromagnetic phase with a quasi-1D spin-density wave below
35K [488] but this is suppressed in the metallic state under an applied pressure.
Fullerenes under high pressures 107

Surprisingly, the 200K anomaly is reported to appear at all pressures and divides
two insulating phases at zero pressure, an insulator and a metal at intermediate
pressures, and two metallic states at high pressures, with little change in the
transition temperature. The origin of this behaviour is not known.

8.1.3. Other studies


The electronic density of states, N(EF) , and Tc in Rb3C60 have been investigated
under an applied pressure by Diederichs et al. [489] in parallel experiments on the
same samples. The total magnetic susceptibility c decreased rather rapidly with
increasing pressure as d(ln c ) /dp = - 0.64 GPa- 1 at 300K. At 50 K, c was quite
nonlinear in p up to about 200MPa, but the average d(ln c ) /dp was similar to that
found at 300K. From these data, N(EF) was calculated as a function of pressure at
both 50 and 300K and was found to decrease under increasing pressure with a
[ ]
pressure coe cient d{ ln N(EF) } /dp = - 0.145GPa- 1. The results for Tc are
reported in the next section. The observed decreases in N(EF) and Tc were found
to be consistent with both weak-coupling electron± phonon models for the super-
conductivity and, for example, small-polaron theory, provided that the coupling is
with intramolecular (high-frequency) phonons. The density of states has also been
investigated by NMR experiments under pressure, since N(EF) is usually taken to be
proportional to (1 /T1 T ) 1 /2 where T1 is the spin± lattice relaxation time. Using this
model, Quirion et al. [490± 492] and Auban-Senzier et al. [493] found that N(EF)
decreases by 11%GPa- 1 in K3 C60 at 20, 80 and 293K, in reasonable agreement with
the value given above for Rb3C60 . However, the initial slope at RT is signi® cantly
larger. The same slope, 11%GPa- 1 , was deduced from the Knight shift K [76]. A
later study [493, 494]gave a pressure dependence in better agreement with the very
strong initial slope of the ® rst experiment, or - 0.56GPa- 1 . For K3C60, the density of
states has also been calculated theoretically as a function of pressure by two groups.
Martins and Troullier [472]calculated the band structure at 0 and 3 GPa and found
that intercalation ® lled rigid C60 bands at zero pressure. Under increasing pressure
the band dispersion increased and the density of states decreased with a pressure
coe cient of about - 0. 22 GPa- 1, intermediate between the experimental results
given above. Huang et al. [473]also found an increase in bandwidth under increasing
[ ]
pressure and a decrease in the density of states by d{ ln N(EF) } /dp = - 0.20 GPa- 1 .
They discussed the electron band structure in detail and also calculated the real and
imaginary dielectric functions as functions of pressure. The static dielectric constant
²0 is found to decrease under increasing pressure at a rate d (ln ²0 ) /dp =
- 0.17 GPa- 1, in contrast with that of C60 which was found to increase with
increasing pressure (section 4.2.4.3). An early study by Fleming et al. [495] also
found a decrease in the density of states under increasing pressure of about
[ ]
d{ ln N(EF) } /dp = - 0.15GPa- 1 .
A Raman study has also been carried out on ammoniated K3C60 showing that
the Ag mode shifts upwards from its zero-pressure frequency under increasing
pressure at a rate of 3.1 cm- 1 GPa - 1 [496]. A comparison with table 6 shows that
this rate is well within the observed range of variation for pristine C60 .
The AC60 -type metallic polymers have also been studied. Forro  et al. [497]
carried out ESR studies under pressures up to 0.5 GPa on CsC60 , RbC60 and KC60.
The motivation behind this work was to ® nd out whether these materials are 1D
metals, as might be inferred from the linear structure of the polymeric chains, or 3D
metals as suggested by recent band-structure calculations [498]. ESR should give a
108 B. Sundqvist

signi® cantly smaller signal for 1D than for 3D materials [497], the linewidth of a 1D
material should increase rapidly with increasing pressure, and a 1D material should
have a stronger pressure dependence for the spin susceptibility than a 3D material.
These features are also observed for the materials studied. The pressure coe cients
[ ]
of the linewidths D H is d ln (D H) /dp = 1.2GPa- 1 for CsC60, 0.9 GPa- 1 for RbC60 ,
and 0.33GPa- 1 for KC60, while the spin susceptibility c s has a pressure coe cient
of d(ln c s) /dp = - 0.2 GPa- 1 for K3 C60 and - 0.44 GPa- 1 , - 0. 55 GPa- 1 and
- 0.9 GPa- 1, for KC60, RbC60 and CsC60 respectively. The data for K3C60 can be
compared with the data for Rb3C60 obtained by Diederichs et al. [489](see above).
Assuming that c s can be obtained by subtraction of the (pressure-independent)
susceptibilities of Rb and C60 from the measured total c , as done by Diederichs et
al., we found d(ln c s) /dp = - 0.42 GPa- 1 for Rb3C60, about twice the value
obtained from K3 C60. The data also agreed with the suggestion that the caesium
compound should be closest to the 1D case while KC60, with the smallest interchain
distance, should be most 3D like. Below 50K an anomaly was found in the ESR
signal for RbC60 . This anomaly was interpreted as arising from a charge-density
wave transition, and a complicated shift in transition temperature with pressure was
interpreted in terms of increasing interchain coupling.
Finally, the little-known semiconducting Rb4 C60 phase has also been studied by
NMR under high pressures. Auban-Senzier and co-workers [493, 499± 501] found
that the initially semiconducting material undergoes a transition to a semimetallic
state through a continuous closing of the bandgap, and above 0.8GPa a non-zero
density of states was observed. Since, at 1.2 GPa, N(EF) is only slightly lower than
for Rb3C60, the samples were tested for superconductivity but no trace of such
behaviour was observed above 0.4K. Kluthe et al. [502]found that T1 increases by
about 75% GPa- 1 while Auban-Senzier and co-workers [499, 501] reported an
increase by 130%GPa- 1.
As already stated above, many other alkali-metal-doped systems are known but
have never been studied under an applied pressure. As one example, a mc 2D doped
polymer Na4 C60 , with C60 molecules linked by single instead of double C± C bonds
has recently been discovered [503], and a comparison with the 2D rh and tg
polymeric phases would be very interesting.

8.2. Superconducting-state properties


Superconductivity in doped fullerenes has been discussed in a large number of
recent reviews [3, 455± 465], and a particularly detailed, complete and up-to-date
review has been given by Gunnarsson [464]. Regarding high-pressure studies it must
be stated that the title of this section is somewhat misleading, since the only
superconducting-state property measured so far under an applied pressure is the
critical temperature Tc . Already the ® rst study [504]of the pressure dependence of Tc
showed that pressure had a signi® cant e€ ect on superconductivity in intercalated
fullerenes as might be expected for such comparatively soft materials. The ® rst
studies were carried out on K3 C60 , for which both Schirber et al. [504]and Sparn et
al. [505]reported pressure derivatives strong enough to wipe out superconductivity
completely within a few gigapascals. In fact, the pressure dependences observed were
among the strongest observed for any superconducting system. Since then a large
number of studies have been carried out and the pressure dependence of Tc is now
known for many, but far from all, fullerene-based superconductors. Available data
have been collected in table 12. Again, measurements carried out using helium as a
Fullerenes under high pressures 109

Table 12. Pressure dependence of Tc for a number of fullerene-based superconductors

Tc ( 0) dTc /dp pmax Pressure


Material (K) (KGPa- 1) (GPa) medium Reference

K3C60 18.0 - 6.3 0.6 Solid He Schirber et al. [504, 506]


K3C60 19.3 - 7.8 2.1 Fluorinert Sparn et al. [505]
K3C60 19 - 6.2 1.5 Oil Zhou et al. [496]
Rb3C60 29.6 - 9.7 1.9 Fluorinert Sparn et al. [507]
Rb3C60 29? - 9.4 0.85 Fluorinert Vareka and Zettl [478]
Rb3C60 28.6 -7 0.6 Solid He Schirber et al. [506]
Rb3C60 29.5 - 8.7 0.65 Fluorinert Diederichs et al. [489]
Rb2CsC60 31.1 -7 0.6 Solid He Schirber et al. [506]
Rb0.5 Cs2.5C60a 30.5 - 9.8 1.6 Fluorinert Movshovicz et al. [508]
Na2CsC60 12.5 - 12.5 0.6 Fluorinert Mizuki et al. [509]
Na2CsC60 10.6 - 12 0.3 Solid Ar, N2 Schirber et al. [510]
Na2CsC60 10.6 - 8.3 0.6 Solid He Schirber et al. [506, 510]
Na2CsC60 10.6 - 13 0.3 Solid Ar, N2 Schirber et al. [433]
Na2Csc60 11.5 -7 0.6 Solid He Schirber and co-workers
[433, 434]
Na2CsC60 10.7 - 12 0.4 Solid Ne Schirber and co-workers
[433, 434]
Na2CsC60 10.6b -3 0.4 Solid Ne Schirber and co-workers
[433, 434]
Na2CsC60 7.2b -3 0.4 Solid Ne Schirber and co-workers
[433, 434]
(NH3) K3C60 Ð c 5. 5 1.5 Oil Zhou et al. [496]
(NH3) 4Na2CsC60 12.5 - 10 0.6 Solid He Schirber et al. [510]
Cs3C60 18 15? 1.4 Mineral oil Palstra et al. [511, 512]
Ca5C60 8.4 1.1 0.6 Solid He Schirber et al. [506, 513]
YbxC60 d 6.25 0.3 0.6 Solid He Schirber et al. [506]
a
Nominal composition; superconducting phase believed to be Rb2 CsC60.
b
Orh (polymeric?) phase.
c
Non-superconducting at p = 0; Tc < 22 K at 0.39 GPa.
d
Probably x < 3.

pressure-transmitting medium di€ er signi® cantly from other studies because the
helium atoms easily co-intercalate with the alkali-metal ions [502, 510], changing the
lattice compression properties [510] and giving always a smaller change in Tc with
pressure.
To a ® rst approximation the results for alkali-metal- and alkali-metal-alloy-
intercalated C60 can be described in a surprisingly simple way, namely that a plot of
Tc against the lattice parameter falls very close to one of the two curves shown in
® gure 52 [448, 514]. The ® rst of these lines is valid for fcc, orientationally disordered
A3 C60 materials such as K3 C60 and Rb3 C60 and the second steeper line for trianionic
materials such as Na2 RbC60 which have the same sc structure as undoped C60 [458].
It was suggested already by Sparn et al. [505, 507] that this behaviour could be
interpreted in a remarkably simple way in a `semiclassical’ model for phonon-
mediated superconductivity. The electrons donated by the alkali-metal ions enter the
C60 electron bands, and the resulting material is to a good approximation metallic
C60 with an N(EF) value at the Fermi surface which depends mainly on the
intermolecular distance and the doping level. On compressing the lattice the electron
bands broaden and N(EF) decreases (see section 8.1.3), and the usual BCS-type
110 B. Sundqvist

Figure 52. Critical temperature Tc as a function of lattice parameter for alkali-metal-


intercalated fullerides. (Reprinted with permission from Yildirim et al. [514]. )

expression
1
Tc ~ x D exp - (8.3)
VN(EF)

shows that Tc must decrease. The alkali-metal ions thus act only as spacers and
electron donors, and the model explains why Rb3C60 , with its larger lattice constant,
has a higher Tc than K3 C60 at zero pressure but the same Tc as K3C60 if compressed
to the same lattice constant. The existence of two separate types of behaviour has
been rationalized [514± 516] on the grounds that both the electron band structure
[211] and the intermolecular phonon spectrum [167] of C60 are sensitive to
orientational order. The latter factor should not, however, be very important for Tc .
However, this simple model has lately been questioned since an increasing
number of materials are found which do not conform to the model. First, Cs3C60
[511, 512]shows a very rapid increase in Tc with increasing pressure until it reaches
40K near 1.4 GPa, the highest value reported for any fullerene superconductor.
Palstra et al. [511, 512] tentatively attributed this behaviour to either a pressure-
induced suppression of ¯ uctuations in the superconducting phase at grain bound-
aries or to a continuous pressure-induced transition from a Mott insulator state to a
metallic state. Ammoniation should increase the lattice constant and increase Tc but,
for NH3 K3 C60 , Zhou et al. [496]found both an absence of superconductivity at zero
pressure and, after superconductivity appears near 0.4 GPa, a positive dTc /dp, again
attributed to either a Mott transition under pressure or simply to a pressure induced
sintering of the grains which made possible the detection of intrinsic superconduc-
tivity existing but undetectable at zero pressure. Iwasa and co-workers [517± 519]
have shown that other ammoniated compounds such as (NH3 ) xNaA2 C60 with
Fullerenes under high pressures 111

Figure 53. Relation between Tc and unit-cell volume for ammonia complex fullerides: ( d ),
(s ), various alkali-metal fullerides. (Reprinted with permission from Iwasa et al.
[517].)

A = Rb or K (® gure 53) also have an inverse dependence of Tc on lattice constant. It


is possible that these results indicate a parabolic dependence of Tc on lattice
parameter similar to that for Tc against doping in ceramic superconductors. A
positive pressure dependence has also been observed by Schirber et al. for alkaline-
earth-doped C60 and was attributed to the di€ erent band-structure changes with
pressure in a material with up to ten extra electrons per C60 [506, 513]. (It might be
noted that pure calcium also shows a positive dTc /dp [520].) Diederichs et al. [470]
recently showed that a pressure of less than 1GPa shifted Tc for Rb3 C60 well o€ the
`main line’ in ® gure 52, and ® nally Yildirim et al. [460, 521]recently showed (® gure
54) that tuning N(EF) through the optimum value for superconductivity in two
families of doped fullerenes resulted in a very rapid variation in Tc that could not be
explained in any reasonable weak- or strong-coupling theory for superconductivity.
Although the simple theoretical model given above might thus still be correct in the
sense that this e€ ect produces most of the observed change in Tc with pressure, other
mechanisms might also be active and strongly modify the total dTc /dp for many
materials. (However, it has been noted by Crespi and Cohen [522] that dTc /dp is
actually surprisingly similar for all `standard’ alkali-metal-doped fulleride super-
conductors, considering the very wide range of variation in Tc . This can be clearly
seen in table 12.) As might be expected, a large number of other theories have been
proposed in addition to the `semiclassical’ theory of phonon-mediated super-
conductivity. These will not be discussed here but the interested reader is referred
to the reviews mentioned above. At the moment, even high-pressure studies a€ ord
no possibility to choose between the various models since almost all researchers
claim to ® nd excellent agreement between experimental and calculated values for
both Tc and dTc /dp, irrespective of whether the mechanism studied is a variety of the
`semiclassical’ phonon coupling (see above), small-polaron-mediated coupling [523],
electron± electron resonating-valence-bond ( RVB) coupling [524], indirect coupling
over carbon± carbon double bonds [525], etc.
112 B. Sundqvist

Figure 54. Relative critical temperature Tc as a function of doping for two families of doped
fullerides. (Reprinted with permission from Yildirim et al. [521].)

In view of the extensive discussion in previous parts of this paper, a particularly


interesting feature of some trianionic alkali metal± C60 compounds is that they can be
polymerized by the application of pressure. X-ray di€ raction studies by Zhu (474,
526]showed that Na2 RbC60 , Na2 CsC60 and Na2 C60 [474]transformed from a cubic
structure with partial orientational order at zero pressure to an orthorhombic phase
containing linear C60 chains at very modest pressures below 350MPa at RT. Only
the ® rst of these compounds became su ciently well ordered for structural studies,
however. While a high pressure is needed to polymerize undoped C60, the smaller
lattice constant of KC60 and RbC60 leads to the formation of polymeric phases even
at RT and zero pressure, and the trianionic sodium-based compounds are believed to
fall in between these two cases. Since the superconducting critical temperature of
Na2 RbC60 has been measured to pressures signi® cantly in excess of the RT
polymerization pressure (see table 12), Zhu [474] suggested that these data were
actually valid for a polymeric fulleride superconductor, which might be 1D in
character. However, very recent data obtained by Prassides et al. [527] (discussed
below) showed that this is probably not the case. The pressure dependence of both
the structure and Tc was investigated further by Morosin et al. [434]and Schirber et
al. [433]. Structural studies [434] revealed that, under a neon pressure, Na2CsC60
transformed near 0.5 GPa into an orthorhombic phase, but the data could not be
® tted to the polymer chain model suggested by Zhu. The orthorhombic phase was
stable on cooling under an applied pressure, and on reheating near zero pressure the
sample reverted to the cubic phase near 210K. Cooling at low pressures (50MPa)
did not result in any phase change. The pressure dependence of Tc was found to be
very sensitive to the experimental conditions. If the pressure was changed near RT
and the sample cooled through Tc at each pressure, a large pressure shift of
- 13K GPa- -1 1 was obtained in argon or N2 (and, because of gas intercalation e€ ects,
- 7 KGPa in helium). However, if the sample was pressurized to 0.5 GPa at RT
Fullerenes under high pressures 113

and cooled, and then the pressure was released at 77 K, a much smaller pressure
slope was observed, resulting in a zero-pressure Tc near 7.2K. Reheating to about
200K recovered the original Tc near 10.5 K but the very small dTc /dp near
- 3 KGPa- 1 remained. This puzzling and complicated behaviour can probably be
explained by the recent results of Prassides et al. [527]who showed that Na2 RbC60
can be polymerized into an orthorhombic linear chain phase by very slow cooling
even at zero pressure. The intermolecular distances on the chains in this phase are
unusually large, suggesting that the molecules are linked by single C± C bonds, as in
the KC60 dimer phase or in (C59N) 2 rather than by four-membered rings as in the
pure C60 linear polymers. Low-temperature experiments showed that the super-
conducting fraction of the sample decreased with an increasing degree of polymer-
ization, and Prassides et al. concluded that the polymer phase is not superconducting
above 2 K. The changes in Tc and dTc /dp observed in the high-pressure experiments
are thus probably a result of the break-up of the superconducting phase into a ® nely
divided percolating network between the polymer chains and/or grains.

9. Comments, speculations and conclusions


In this paper I have shown that a very large amount of information has already
been collected on the behaviour and properties of C60 and other fullerenes under
high pressures, but also that there are still very large gaps in our knowledge. The
phase diagram of C60 is extremely rich, with several crystalline and amorphous
structural phases and with a very interesting continuous orientational reordering
with pressure and temperature in the low-pressure region, and we have only just
started to study the properties of the known high-pressure phases. For C70 , even the
zero-pressure phase diagram is not yet understood, and the phase diagram given
above is very tentative and will probably be signi® cantly modi® ed when more
information has been obtained. Since all higher fullerenes have even more aniso-
tropic molecules, it might be expected that their phase diagrams will have an
increasing complexity with increasing molecular weight.
The phase diagram of C60 has now been reasonably well investigated over very
large ranges in pressure and temperature, and in the range up to 8 GPa it is probably
rather well understood at most temperatures. However, even in this range there are
still several questions that remain to be answered. The physical properties of the
`glassy’ phase below 90K have only been probed brie¯ y by thermal conductivity [84]
and Raman [101, 189]studies under high pressures, and it would be very interesting
to learn whether the glass transition still exists and, if so, whether it has the same
slope dTg /dp in the range above 1GPa where almost all molecules should be in the
H-oriented state. This question is of course also connected with the anomalies
observed near 2 GPa at RT, which I have tentatively identi® ed above with the
reorientational transformation into an H-oriented structure but which might poss-
ibly instead be connected with orientational freezing, that is the glass transition. In
the high-temperature range there is still some controversy over the actual upper limit
of stability of the C60 molecule itself, at both zero and higher pressures, since some
experiments show the molecules of the solid to break down above 970K [64], and
others suggest a signi® cantly higher breakdown temperature for free molecules [65],
and since theory predicts a very high breakdown temperature. I suggested above that
one possible reason for the low stability limit in the solid state was the presence of
solvents in the lattice, since it was recently shown by Moro [528] that signi® cant
114 B. Sundqvist

concentrations of solvent could remain in the C60 lattice even after prolonged
annealing under very low dynamic pressures, and that solvents could only be
`eliminated’ by (repeated) sublimation of the material. If the molecule is stable, it
might also be possible to carry out rapid heating experiments to search for a liquid
phase under an applied pressure. Also, the high-temperature low-pressure phase
boundaries for the formation of the orh and rh polymeric forms of C60 have not yet
been investigated.
At higher pressures the most serious problem is that very few of the known
structural phases have actually been studied in situ. Most studies have instead been
carried out at zero pressure on metastable material produced by rapidly cooling
samples from the reaction temperature. As discussed above, it is therefore uncertain
which of these phases are true equilibrium phases, and which are only transient
phases formed during cooling. The resistance studies of Saito [218], for example,
could be interpreted to show that the transition between the rh and orh polymeric
phases is reversible when a pressure is applied. The details of the phase diagram can
thus probably only be revealed by spectroscopic and di€ raction studies under
equilibrium conditions. Very little is yet known about the physical properties of
the high-pressure phases, but what little is known is very interesting and suggests
that the study of these materials will be scienti® cally very rewarding. Most of the
very-high-pressure phases have been reported to be harder than diamond and, as
discussed above, recent studies verify that the speci® c heat is lower than for diamond
(indicating a very high Debye temperature because of strong interatomic bonds) and
the adiabatic bulk modulus higher.
For C70 , higher fullerenes and doped fullerenes, very much less is known. Judging
from available information on C70 , sorting out the phase diagram of the higher
fullerenes will probably prove to be rather di cult, partly because of the many
rotational degrees of freedom and the many possible structural and orientational
phases and partly because of the sluggishness and hysteresis observed in most
transitions. For the superconducting fullerenes, only the most basic types of study
have been carried out and very little is known about the phase diagrams under an
applied pressure, the physical properties in the normal state, and the superconduct-
ing properties other than Tc (i.e. critical ® elds, critical currents, coherence lengths,
penetration depths, etc. ). High-pressure studies on other doped fullerenes would also
be of great interest. In particular, the AC60 materials have a very complicated and
rich structural phase behaviour even at zero pressure, and unravelling the corre-
sponding phase diagram under higher pressures would be a very interesting task. The
similarities and di€ erences between the various polymeric phases observed at low
pressures (photopolymerized and pressure-polymerized orh C60 , the polymeric form
of AC60 and polymerized Na2 AC60 ) are also very interesting subjects that merit
further study.
To conclude, I have tried to outline what has been done in this ® eld up to now
and to suggest a small number of interesting problems for study. However, I am sure
that future work in this ® eld will turn up a large number of surprising discoveries
that I cannot even try to predict, and I am convinced that high-pressure studies of
fullerenes will continue to be a rewarding ® eld of study for many years to come and
that the results of future research in this ® eld will be of great value from a basic
scienti® c point of view and possibly also from a commercial and technical point of
view.
Fullerenes under high pressures 115

Acknowledgments
I would like to thank ® rst the many persons who have been kind enough to send
me preprints of their work prior to publication, helping me to make this review as up
to date as possible, and also the many persons with whom I have discussed the high-
pressure properties of fullerenes at conferences, meetings and visits. In particular, of
course, I would like to thank my colleagues here in UmeaÊ who have made me aware
of a number of important papers in the ® eld and always helped me whenever I had a
di cult question. Finally, this work would have been impossible without ® nancial
support from Naturvetenskapliga forskningsra Ê det and Teknikvetenskapliga for
research on fullerenes under high pressures.

References

[1] Kroto, H. W., Heath, J. R., O’ Brien, S. C., Curl, R. F., and Smalley, R. E., 1985,
Nature, 318, 162.
[2] Kraï tschmer, W., Lamb, L. D., Fostiropoulos, K., and Huffman, D. R., 1990,
Nature, 347, 354.
[3] Dresselhaus , M. S., Dresselhaus, G., and Eklund, P. C., 1996, Science of Fullerenes
and Carbon Nanotubes (San Diego, California: Academic Press).
[4] Krishnamurthy, H. R., and Sood , A. K., 1991, Rev. solid st. Sci., 5, 587.
[5] Hariharan, Y., Bharathi, A., Sundar, C. S., Sastry, V. S., Yousuf, M.,
Radhakrishnan, T. S., Rao, G. V. N., Kumary, T. G., Subramanian, N., Sahu ,
P. Ch., Raghunathan, V. S., and Valsakumar , M. C., 1992, Curr. Sci., 63, 25.
[6] Dresselhaus , M. S., Dresselhaus, G., and Eklund, P. C., 1993, J. Mater. Res., 8,
2054.
[7] Rao, C. N. R., Seshadri, R., Govindaraj , A., and Sen, R., 1995, Mater. Sci. Engng,
R15, 209.
[8] Hammond, G. S., and Kuck , V. J., 1992, Fullerenes, ACS Symposium Series, vol. 481
(Washington, DC: American Chemical Society).
[9] Ehrenreich, H., and Spaepen, F. (editors), 1994, Solid. St. Phys., 48.
[10] Billups, W. E., and Ciufolini, M. A., 1993, Buckminsterfullerenes (Weinheim: VCH).
[11] Nuíñez -Regueiro , M., 1992, Mod. Phys. Lett., B, 6, 1153.
[12] Nuíñez -Regueiro , M., Beíthoux , O., Hodeau , J.-L., Marques, L., Tonnerre, J. M.,
and Bouchet-Fabre, B., 1994, Fullerenes: Recent Advances in the Physics and
Chemistry of Fullerenes, edited by K. M. Kadish and R. S. Ruo€ (Pennington, New
Jersey: Electrochemical Society), pp. 519± 529.
[13] Schirber , J. E., Samara , G. A., Morosin, B., Assink , R., Loy, D., Wang , H.,
Williams, J., Murphy, D., Kortan, A. R., Rosseinsky , M., Zhou , O., Zhu, Q.,
Kniaz , K., and Fischer , J. E., 1994, High-Pressure Science and TechnologyÐ 1993,
edited by S. C. Schmidt, J. W. Shaner, G. A. Samara and M. Ross (New York:
American Institute of Physics), pp. 639± 642.
[14] Sundqvist , B., Soldatov , A., Andersson, O., Lundin, A., and Persson, P.-A.,
1995, Fullerenes: Recent Advances in the Chemistry and Physics of Fullerenes, Vol. 2,
edited by R. S. Ruo€ and K. M. Kadish (Pennington, New Jersey: Electrochemical
Society), pp. 891± 905.
[15] Sundqvist , B., Andersson, O., Lundin, A., Persson, P.-A., and Soldatov , A., 1996,
High Pressure Science and Technology, edited by W. A. Trzeciakowski (World Scienti® c:
Singapore), pp. 697± 701.
[16] Fischer , J. E., and Heiney, P. A., 1993, J. Phys. Chem. Solids, 54, 1725.
[17] Fischer , J. E., Heiney, P. A., Luz zi, D. E., and Cox , D. E., 1992, Fullerenes, ACS
Symposium Series, Vol. 481, edited by G. S. Hammond and V. J. Kuck (Washington,
DC: American Chemical Society), pp. 55± 69.
[18] Prassides, K., Kroto, H. W., Taylor , R., Walton, D. R. M., David , W. I. F.,
Tomkinson, J., Haddon, R. C., Rosseinsky, M. J., and Murphy, D. W., 1992,
Carbon, 30, 1277.
116 B. Sundqvist

[19] Axe, J. D., Moss, S. C., and Neumann, D. A., 1994, Solid St. Phys., 48, 149.
[20] Pickett, W. E., 1994, Solid St. Phys., 48, 225.
[21] Pintschovius, L., 1996, Rep. Progr. Phys., 57, 473.
[22] Haufler , R. E., Conceicao , J. J., Chibante, L. P. F., Chai, Y., Byrne, N. E.,
Flanagan, S., Haley, M. M., O’ Brien, S. C., Pan, C., Xiao, Z., Billups, W. E.,
Ciufolini, M. A., Hauge, R. H., Margrave, J. L., Wilson, L. J., Curl , R. F., and
Smalley, R. E., 1990, J. phys. Chem., 94, 9634.
[23] Howard, J. B., Mc Kinnon, J.Th., Makarovsky, Y., Lafleur , A. L., and Johnson,
M. E., 1991, Nature, 352, 139.
[24] Buseck , P. R., Tsipursky, S. J., and Hettich, R., 1992, Science, 257, 215.
[25] Daly, T. K., Buseck , P. R., Williams, P., and Lewis, C. F., 1993, Science, 259, 1599.
[26] Heymann, D., Chibante, L. P. F., Brooks, P. R., Wolbach, W. S., and Smalley,
R. E., 1994, Science, 265, 645.
[27] Di Broz olo, F. R., Bunch, T. E., Fleming , R. H., and Macklin, J., 1994, Nature,
369, 37.
[28] Chow , P. C., Jiang , X., Reiter , G., Wochner , P., Moss, S. C., Axe, J. D., Hanson,
J. C., Mc Mullan, R. K., Meng , R. L., and Chu, C. W., 1992, Phys. Rev. Lett., 69,
2943.
[29] David , W. I. F., Ibberson, R. M., and Matsuo, T., 1993, Proc. Royal Soc. A, 442,
129.
[30] Pintschovius, L., Chaplot, S. L., Roth, G., and Heger , G., 1995, Physics and
Chemistry of Fullerenes and Derivatives, edited by H. Kuzmany, J. Fink, M. Mehring,
and S. Roth (Singapore: World Scienti® c), pp. 39± 43.
[31] Pintschovius, L., Chaplot, S. L., Roth, G., and Heger , G., 1995, Phys. Rev. Lett.,
75, 2843.
[32] Heiney, P. A., Fischer , J. E., Mc Ghie, A. R., Romanow , W. J., Denenstein, A. M.,
Mc Cauley , J. P., Jr , Smith, A. B., III, and Cox , D. E., 1991, Phys. Rev. Lett., 66,
2911.
[33] David , W. I. F., Ibberson, R. M., Dennis, T. J. S., Hare, J. P., and Prassides, K.,
1992, Europhys. Lett., 18, 219.
[34] Blaschko, O., Krexner , G., Maier , Ch., and Karawatz ki, R., 1997, Phys. Rev. B,
56, 2288.
[35] Gugenberger , F., Heid , R., Meingast , C., Adelmann, P., Braun, M., Wuï hl , H.,
Haluska , M., and Kuz many, H., 1992, Phys. Rev. Lett., 69, 3774.
[36] Rao, A. M., Zhou , P., Wang , K.-A., Hager , G. T., Holden, J. M., Wang , Y., Lee,
W.-T., Bi, X.-X., Eklund , P. C., Cornett, D. S., Duncan, M. A., and Amster ,
I. J., 1993, Science, 259, 955.
[37] Cheng , A., Klein, M. L., Parrinello , M., and Sprik, M., 1992, Phil. Trans. R. Soc.
A, 341, 133.
[38] Cheng , A., and Klein, M. L., 1992, Phys. Rev. B, 46, 4958.
[39] Pickholz , M., and Gamba , Z., 1996, Phys. Rev. B, 53, 2159.
[40] Sprik, M., Cheng , A., and Klein, M. L., 1992, Phys. Rev. Lett., 69, 1660.
[41] Blinc , R., DolinSÏek , J., Seliger , J., and Arcon, D., 1993, Solid St. Commun., 88, 9.
[42] Verheijen, M. A., Meekes, H., Meijer , G., Bennema, P., de Boer , J. L., van
Smaalen, S., van Tendeloo, G., Amelinckx , S., Muto, S., and van Landuyt, J.,
1992, Chem. Phys., 166, 287.
[43] Dennis, T. J. S., Prassides, K., Roduner , E., Cristofolini, L., and DeRenz i, R.,
1993, J. phys. Chem., 97, 8553.
[44] Misof, K., Vogl, G., Wende, L., and Sielemann, R., 1994, Progress in Fullerene
Research, edited by H. Kuzmany, J. Fink, M. Mehring and S. Roth (Singapore:
World Scienti® c), pp. 87± 90.
[45] Binninger , U., Roduner , E., Reid , I. D., Bernhard, C., Hofer , A., Recknagel , E.,
Erxmeyer , J., and Niedermayer , Ch., 1994, Progress in Fullerene Research, edited
by H. Kuzmany, J. Fink, M. Mehring and S. Roth (Singapore: World Scienti® c),
pp. 207± 210.
[46] Tanokura , Y., Nishimura , M., Sekine, T., and Takeuchi, T., 1996, Physica B, 219±
220, 157.
Fullerenes under high pressures 117

[47] Nagel , P., Meingast , C., Verheijen, M. A., and Meijer , G., 1996, Fullerenes and
Fullerene Nanostructures, edited by H. Kuzmany, J. Fink, M. Mehring and S. Roth
(Singapore: World Scienti® c), pp. 68± 71.
[48] Ossipyan, Yu. A., Bobrov , V. S., Grushko, Yu . S., Dilanyan, R. A., Zharikov ,
O. V., Lebyodkin, M. A., and Sheckhtman, V. Sh., 1993, Appl. Phys. A, 56, 413.
[49] Duclos, S. J., Brister , K., Haddon, R. C., Kortan, A. R., and Thiel , F. A., 1991,
Nature, 351, 380.
[50] Piermarini, G. J., Block , S., and Barnett, J. D., 1973, J. appl. Phys., 44, 5377.
[51] Besson, J. M., and Pinceaux , J. P., 1979, Science, 206, 1073.
[52] Jayaraman, A., 1983, Rev. mod. Phys., 55, 65.
[53] Sherman, W. F., and Stadtmuller , A. A., 1987, Experimental Techniques in High-
Pressure Research (Chichester, West Sussex: Wiley); Eremets, M., 1997, High Pressure
Experimental Methods (Oxford University Press); Holz apfel , W. B., and Isaacs, N. S.,
1997, High-Pressure Techniques in Chemistry and Physics (Oxford University Press).
[54] Bundy, F. P., 1963, J. chem. Phys., 38, 631.
[55] Andersson, G., Sundqvist , B., and Baï ckstroï m, G., 1989, J. appl. Phys., 65, 3943.
[56] Poirier , D. M., Owens, D. W., and Weaver , J. H., 1995, Phys. Rev. B, 51, 1830.
[57] Abrefah, J., Olander , D. R., Balooch, M., and Siekhaus, W. J., 1992, Appl. Phys.
Lett., 60, 1313.
[58] Mathews, C. K., Sai Baba , M., Lakshmi Narasimhan, T. S., Balasubramanian,
R., Sivaraman, N., Srinivasan , T. G., and Vasudeva Rao, P. R., 1992, J. phys.
Chem., 96, 3566.
[59] Pan, C., Chandrasekharaiah, M. S., Agan, D., Hauge, R. H., and Margrave,
J. L., 1992, J. phys. Chem., 96, 6752.
[60] Sai Baba , M., Narasimhan, T. S. L., Balasubramanian, R., Sivaraman, N., and
Mathews, C. K., 1994, J. phys. Chem., 98, 1333.
[61] Vorob’ ev , V. S., and Eletskii, A. V., 1995, Teplo® z. Vys. Temp., 33, 862 (Engl.
Transl., 1995, High Temp., 33, 858).
[62] Vorob’ ev , V. S., and Eletskii, A. V., 1996, Chem. Phys. Lett., 254, 263.
[63] Martyna , G. J., Tobias, D. J., and Klein, M. L., 1994, J. chem Phys., 101, 4177.
[64] Sundar, C. S., Bharathi, A., Hariharan, Y., Janaki, J., Sastry, V. S., and
Radhakrishnan, T. S., 1992, Solid St. Commun., 84, 823.
[65] Kolodney, E., Tsipinyuk , B., and Budrevich, A., 1994, J. chem. Phys., 100, 8542.
[66] Heiney, P. A., Vaughan, G. B. M., Fischer , J. E., Coustel , N., Cox , D. E., Copley,
J. R. D., Neumann, D. A., Kamitakahara , W. A., Creegan, K. M., Cox , D. M.,
Mc Cauley, J. P., Jr , and Smith, A. B., III, 1992, Phys. Rev. B, 45, 4544.
[67] Atake, T., Tanaka , T., Kawaji, H., Kikuchi, K., Saito, K., Suz uki, S., Ikemoto,
I., and Achiba , Y., 1991, Physica C, 185± 189, 427.
[68] Fischer , J. E., Heiney, P. A., McGhie, A. R., Romanow , W. J., Denenstein , A. M.,
Mc Cauley, J. P., and Smith, A. B., III, 1991, Science, 252, 1288.
[69] Samara , G. A., Schirber , J. E., Morosin, B., Hansen, L. V., Loy, D., and
Sylwester , A. P., 1991, Phys. Rev. Lett., 67, 3136.
[70] Kriz a, G., Ameline, J.-C., Jeírome, D., Dworkin, A., Sz warc, H., Fabre, C.,
Schuï tz , D., Rassat , A., Bernier , P., and Zahab, A., 1991, J. Phys. Paris, I, 1,
1361.
[71] Samara , G. A., Hansen, L. V., Assink , R. A., Morosin, B., Schirber , J. E., and
Loy, D., 1993, Phys. Rev. B, 47, 4756.
[72] Schirber , J. E., Kwei, G. H., Jorgensen, J. D., Hitterman, R. L., and Morosin, B.,
1995, Phys. Rev. B, 51, 12014.
[73] Grube, K., 1995, Thesis, Karlsruhe; 1995, Report FZKA 5611, Karlsruhe.
[74] Pintschovius, L., Blaschko, O., and Pyka , N., 1998 (to be published).
[75] Samara , G. A., hansen, L. V., Morosin, B., and Schirber , J. E., 1994, High-Pressure
Science and TechnologyÐ 1993, edited by S. C. Schmidt, J. W. Shaner, G. A. Samara and
M. Ross (New York: American Institute of Physics), pp. 643± 646.
[76] Auban-Senz ier , P., Kerkoud , R., Godard, J., Jeírome, D, Lambert, J. M., Zahab,
A., Rachdi, F., and Bernier , P., 1994, Progress in Fullerene Research edited by H.
Kuzmany, J. Fink, M. Mehring and S. Roth (Singapore: World Scienti® c), pp. 451± 455.
118 B. Sundqvist

[77] Kerkoud , R., Auban-Senz ier , P., Godard, J., Jeírome, D., Lambert, J.-M., and
Bernier , P., 1994, Adv. Mater., 6, 782.
[78] Matsuura , S., Ishiguro, T., Kikuchi, K., Achiba , Y., and Ikemoto, I., 1995,
Fullerene Sci. Technol., 3, 437.
[79] Lundin, A., and Sundqvist , B., 1994, Europhys. Lett., 27, 463.
[80] Lundin, A., and Sundqvist , B., 1996, Phys. Rev. B, 53, 8329.
[81] Meletov , K. P., Christofilos, D., Kourouklis, G. A., and Ves, S., 1995, Chem.
Phys. Lett., 236, 265.
[82] Meletov , K. P., Christofilos, D., Ves, S., and Kourouklis, G. A., 1995, Phys. Rev.
B, 52, 10090.
[83] Meletov , K. P., Kourouklis, G. A., Christofilos, D., and Ves, S., 1995, Zh.
eÂksp. teor. Fiz., 108, 1456 (Engl. Transl., 1995, J. exp. theoret. Physics, 81,
798).
[84] Andersson, O., Soldatov , A., and Sundqvist , B., 1996, Phys. Rev. B, 54, 3093.
[85] Ramasesha , S. K., and Singh, A. K., 1994, Solid St. Commun., 91, 25.
[86] Jean, Y. C., Lu, X., Lou , Y., Bharathi, A., Sundar , C. S., Lyu , Y., Hor, P. H., and
Chu , C. W., 1992, Phys. Rev. B, 45, 12126.
[87] Kempinski, W., Stankowski, J., Trybula , Z., Hoffman, S., Hilcz er, W., Czyz ak,
B., Krupski, M., Loís, S., Andrzejewski, B., Kraï tschmer , W., and Bysz ewski, P.,
1994, Progress in Fullerene Research, edited by H. Kuzmany, J. Fink, M. Mehring and S.
Roth (Singapore: World Scienti® c), pp. 188± 192.
[88] Kawamura, H., Akahama , Y., Kobayashi, M., Shinohara , H., Sato, H., Saito, Y.,
Kikegawa , T., Shimomura , O., and Aoki, K., 1993, J. Phys. Chem. Solids., 54, 1675.
[89] Garland , C. W., and Weiner , B. B., 1971, Phys. Rev. B, 3, 1634.
[90] Lundin, A., Sundqvist , B., Skoglund , P., Fransson, Å., and Pettersson , S., 1992,
Solid St. Commun., 84, 879.
[91] Bao, Z., Liu , C., Li, Y., and Zhu, D., 1995, Chin. Sci. Bull., 40, 898.
[92] Blaschko, O., Rom, W., and Goncharenko, I. N., 1996, J. Phys.: condens. Matter., 8,
4235.
[93] Burgos, E., Halac , E., and Bonadeo, H., 1994, Phys. Rev. B, 49, 15544.
[94] Lu , J. P., Li, X.-P., and Martin, R. M., 1992, Phys. Rev. Lett., 68, 1551.
[95] Lamoen, D., and Michel , K. H., 1993, Phys. Rev. B, 48, 807.
[96] Rasolt , M., 1992, Phys. Rev. B, 46, 1944.
[97] Fischer , J. E., Mc Ghie, A. R., Estrada , J. K., Haluska , M., Kuz many, H., and
ter Meer , H.-U., 1996, Phys. Rev. B, 53, 11418.
[98] Goldoni, A., Cepek , C., and Modesti, S., 1996, Phys. Rev. B, 54, 2890.
[99] David , W. I. F., and Ibberson, R. M., 1993, J. Phys.: condens. Matter, 5, 7923.
[100] Andersson, O., Soldatov , A., and Sundqvist , B., 1995, Phys. Lett. A, 206, 260.
[101] Wolk , J. A., Horoyski, P. J., and Thewalt, M. L. W., 1995, Phys. Rev. Lett., 74,
3483.
[102] Sundqvist , B., Andersson, O., Lundin, A., and Soldatov , A., 1994, High Pressure
in Materials Science and Geoscience, edited by J. Kamara d, Z. Arnold and A. Kapicka
(Prague: Prometheus Press), pp. 109± 112.
[103] Sundqvist , B., Andersson, O., Lundin, A., and Soldatov , A., 1995, Solid St.
Commun., 93, 109.
[104] Andersson, O., Soldatov , A., and Sundqvist , B., 1995, Science and Technology of
Fullerene Materials, edited by P. Bernier, D. S. Bethune, L. Y., Chiang, T. W. Ebbesen,
R. M. Metzger and J. W. Mintmire (Pittsburgh, Pennsylvania: Materials Research
Society), pp. 549± 554.
[105] Saito, R., Dresselhaus , G., and Dresselhaus , M. S., 1994, Phys. Rev. B, 49, 2143.
[106] Fomenko, L. S., Natsik , V. D., Lubenets, S. V., Lirtsman, V. G., Aksenova , N. A.,
Isakina , A. P., Prokhvatilov , A. I., Strzhemechny, M. A., and Ruoff, R. S.,
1995, Fullerenes: Recent Advances in the Chemistry and Physics of Fullerenes, Vol. 2,
edited by R. S. Ruo€ and K. M. Kadish (Pennington, New Jersey: Electrochemical
Society), pp. 926± 934.
[107] Schranz , W., Fuith, A., Dolinar , P., Warhanek , H., Halu SÏka, M., and
Kuz many, H., 1993, Electronic Properties of Fullerenes, edited by H. Kuzmany, J.
Fink, M. Mehring and S. Roth (Berlin: Springer), pp. 177± 181.
Fullerenes under high pressures 119

[108] Moret, R., 1993, Phys. Rev. B, 48, 17619.


[109] Shi, X. D., Kortan, A. R., Williams, J. M., Kini, A. M., Savall , B. M., and
Chaikin, P. M., 1992, Phys. Rev. Lett., 68, 827.
[110] Alers, G. B., Golding , B., Kortan, A. R., Haddon, R. C., and Thiel , F. A., 1992,
Science, 257, 511.
[111] Nordheim, L., 1931, Ann. d. Physik (Lpz), 9, 607.
[112] Jaï ckle, J., 1986, Rep. Progr. Phys., 49, 171.
[113] Sundqvist , B., Andersson, O., and Soldatov , A., 1995, Phys. Rev. Lett., 75, 2906;
Wolk , J. A., Horoyski, P. J. and Thewalt, M. L. W., 1995, Phys. Rev. Lett., 75,
2907.
[114] Blank , V., Popov , M., Buga , S., Davydov , V., Denisov , V. N., Ivlev , A. N.,
Mavrin, B. N., Agafonov , V., Ceolin, R. Sz warc, H., and Rassat , A., 1994,
Phys. Lett. A, 188, 281.
[115] Meletov , K. P., Dolganov , V. K, Zharikov , O. V., Kremeenskaya , I. N., and
Ossipyan, Yu . A., 1992, J. Phys. Paris, 2, 2097.
[116] Jephcoat , A. P., Hriljac , J. A., Finger , L. W., and Cox , D. E., 1994, Europhys.
Lett., 25, 429.
[117] Jeon, S.-J., Kim, D., Kim, S. K., and Jeon, I. C., 1992, J. Raman Spectrosc., 23, 311.
[118] Huang , Y., Gilson, D. F. R., and Butler , I. S., 1991, J. phys. Chem., 95, 5723.
[119] Bundy, F. P., Bassett , W. A., Weathers, M. S., Hemley, R. J., Mao, H. K., and
Goncharov , A. F., 1996, Carbon, 34, 141.
[120] Ashcroft, N. W., 1991, Europhys. Lett., 16, 355.
[121] Girifalco, L. A., 1992, J. phys. Chem., 96, 858.
[122] Hagen, M. H. J., Meijer , E. J., Mooij , G. C. A. M., Frenkel , D., and
Lekkerkerker , H. N. W., 1993, Nature, 365, 425.
[123] Cheng , A., Klein, M. L., and Caccamo, C., 1993, Phys. Rev. Lett., 71, 1200.
[124] Abramo, M. C., and Caccamo, C., 1996, J. Phys. Chem. Solids., 57, 1751.
[125] Kim, S. G., and Tomaínek , D., 1994, Phys. Rev. Lett., 72, 2418.
[126] Mederos, L., and Navascues, G., 1994, Phys. Rev. B, 50, 1301.
[127] Serra , S., Sanguinetti, S., and Colombo, L., 1994, Chem. Phys. Lett., 225, 191.
[128] Shchelkacheva , T. I., 1996, Phys. Lett. A, 214, 95.
[129] Tau , M., Parola , A., Pini, D., and Reatto, L., 1995, Phys. Rev. E, 52, 2644.
[130] Rascoín, C., Navascues, G., and Mederos, L., 1995, Phys. Rev. B, 51, 14899.
[131] Yakub, L. N., 1993, Fiz. Nizk. Temp., 19, 726 (Engl. Transl., 1993, Low Temp. Phys.,
19, 522).
[132] Gschneider , K. A., Jr., 1964, Solid St. Phys., 16, 275.
[133] Zubov , V. I., Tretiakov , N. P., Sanchez , J. F., and Caparicia , A. A., 1996, Phys.
Rev. B, 53, 12080.
[134] Lundin, A., Ross, R. G., and Baï ckstroï m, G., 1994, High Temp. ± High Pressures, 26,
477.
[135] Ruoff, R. S., and Ruoff, A. L., 1991, Nature, 350, 663.
[136] Ruoff, R. S., and Ruoff, A. L., 1991, Appl. Phys. Lett., 59, 1553.
[137] McSkimin, H. J., and Bond, W. L., 1957, Phys. Rev., 105, 116.
[138] Hanfland, M., Beister , H., and Syassen, K., 1989, Phys. Rev. B, 39, 12598.
[139] Vinet, P., Ferrante, J., Smith, J. R., and Rose, J. H., 1986, J. Phys: condens. Matter,
19, L467.
[140] Murnaghan, F. D., 1994, Proc. Natn. Acad. Sci. USA, 30, 244.
[141] Ludwig , H. A., Fietz , W. H., Hornung , F. W., Grube, K., Wagner , B., and
Burkhart, G. J., 1994, Z. Phys. B, 96, 179.
[142] Coufal, H., Meyer , K., Grygier , R. K., de Vries, M., Jenrich, D., and Hess, P.,
1994, Appl. Phys. A, 59, 83.
[143] Kobelev , N. P., Moravskii, A. P., Soifer , Ya. M., Bashkin, I. O., and Rybchenko,
O. G., 1994, Fiz, tverd. Tela., 36, 2732 (Engl. Transl., 1994, Phys. Solid St., 36,
1491.
[144] Kobelev , N. P., Soifer , Ya. M., Bashkin, I. O., Gurov , A. F., Moravskii, A. P., and
Rybchenko, O. G., 1995, Phys. stat. sol. (b), 190, 157.
[145] Fioretto, D., Carlotti, G., Socino, G., Modesti, S., Cepek , C., Giovanni, L.,
Donz elli, O., and Niz zoli, F., 1995, Phys. Rev. B, 52, R8707.
120 B. Sundqvist

[146] Komori, R., and Miyamoto, Y., 1995, J. Phys. Chem. Solids., 56, 535.
[147] Bashkin, I. O., Rashchupkin, V. I., Gurov , A. F., Moravsky, A. P., Rybchenko,
O. G., Kobelev , N. P., Soifer , Ya . M., and Ponyatovskii, E. G., 1994, J. Phys.:
condens. Matter, 6, 7491.
[148] Wang , J., Wang , L., Chen, L., Chen, H., Wang , R., Zhang , Z., Che, R., and Zhou ,
L., 1993, Chin. Phys. Lett., 10 (3), 159.
[149] Haines, J., and Leíger , J. M., 1994, Solid St. Commun., 90, 361.
[150] Shimomura , S., Fujii, Y., Noz awa , S., Kikuchi, K., Achiba , Y., and Ikemoto, I.,
1993, Solid St. Commun., 85, 471.
[151] Blank , V. D., Buga , S. G., Serebryanaya , N. R., Dubitsky , G. A., Bagramov ,
R. H., Popov , M. Yu ., Prochorov , V. M., and Sulyanov , S. A., 1997, Appl. Phys.
A, 64, 247.
[152] Nguyen, J. H., Kruger , M. B., and Jeanloz , R., 1993, Solid St. Commun., 88, 719.
[153] Singh, A. K., 1993, Phil. Mag. Lett., 67, 379.
[154] Girifalco, L. A., 1995, Phys. Rev. B, 52, 9910.
[155] Hoen, S., Chopra , N. G., Xiang , X.-D., Mostovoy, R., Huy, J., Vareka , W. A.,
and Zettl , A., 1992, Phys. Rev. B, 46, 12737.
[156] Yan, F., Wang , Y. N., and Gu , M., 1997, Phys. Rev. B, 55, R4918.
[157] Eom, C. B., Hebard , A. F., Trimble, L. E., Celler , G. K., and Haddon, R. C., 1993,
Science, 259, 1887.
[158] Wang , Y., Tomaínek, D., and Bertsch, G. F., 1991, Phys. Rev. B, 44, 6562.
[159] Cheng , A., and Klein, M. L., 1991, J. phys. Chem., 95, 6750.
[160] Cheng , A., and Klein, M. L., 1992, Phys. Rev. B, 45, 1889.
[161] Bortel , G., Faigel , G., Osz laínyi, G., Pekker , S., and Tegz e, M., 1993, Electronic
Properties of Fullerenes, edited by H. Kuzmany, J. Fink, M. Mehring and S. Roth
(Berlin: Springer), pp. 207± 210.
[162] Xu, Y.-N., Huang , M.-Z., and Ching , W. Y., 1992, Phys. Rev. B, 46, 4241.
[163] Sprik, M., Cheng , A., and Klein, M. L., 1992, J. phys. Chem., 96, 2027.
[164] Burgos, E., Halac , E., and Bonadeo, H., 1993, Phys. Rev. B, 47, 13903.
[165] Yildirim, T., and Harris, A. B., 1992, Phys. Rev. B, 46, 7878.
[166] Li, X.-P., Lu , J. P., and Martin, R. M., 1992, Phys. Rev. B, 46, 4301.
[167] Pintschovius, L., and Chaplot, S. L., 1995, Z. Phys. B, 98, 527.
[168] La Rocca , G. C., 1994, Europhys. Lett., 25, 5.
[169] Prilutski, Yu. I., and Shapovalov , G. G., 1997, Phys. stat. sol. (b), 201, 361.
[170] Dzyabchenko, A. V., D’ Yachkov , P. N., and Agafonov , V. N., 1995, Izv. Akad.
Nauk, Ser. Khim, 1466 (Engl. Transl., 1995, Russ. Chem. Bull., 44, 1408); Launois, P.,
Ravy, S. and Moret, R., 1997, Phys. Rev. B, 55, 2651; Erratum, ibid. 56, 7019;
Michel , K. H., and Copley, J. R. D., 1997, Z. Phys. B, 103, 369; Savin, S., Harris,
A. B., and Yildirim, T., 1997, Phys. Rev. B, 55, 14182.
[171] Zubov , V. I., Tretiakov , N. P., Teixeira Rabelo , J. N., and Sanchez Ortiz , J. F.,
1995, Science and Technology of Fullerene Materials, edited by P. Bernier, D. S. Bethune,
L. Y. Chiang, T. W. Ebbesen, R. M. Metzger and J. W. Mintmire (Pittsburgh,
Pennsylvania: Materials Research Society), pp. 253± 258.
[172] Zubov , V. I., Sanchez , J. F., Tretiakov , N. P., Caparica , A. A., and Zubov , I. V.,
1997, Carbon, 35, 729.
[173] Venkatesh, R., and Gopala Rao, R. V., 1997, Phys. Rev. B, 55, 15.
[174] Yu, J., Bi, L., Kalia , R. K., and Vashishta , P., 1994, Phys. Rev. B, 49, 5008.
[175] Yu, J., Kalia , R. K., and Vashishta , P., 1993, Appl. Phys. Lett., 63, 3152.
[176] Woo, S. J., Lee, S. H., Kim, E., Lee, K. H., Lee, Y. H., Hwang, S. Y., and Jeon, I. C.,
1992, Phys. Lett. A, 162, 501.
[177] Jishi, R. A., Mirie, R. M., and Dresselhaus , M. S., 1992, Novel Forms of Carbon,
edited by C. L. Renschler, J. J. Pouch and D. M. Cox (Pittsburgh, Pennsylvania:
Materials Research Society), pp. 197± 202.
[178] Jishi, R. A., Mirie, R. M., and Dresselhaus , M S., 1992, Phys. Rev. B, 45, 13685.
[179] Dresselhaus , M. S., Dresselhaus , G., and Eklund, P. C., 1996, J. Raman
Spectrosc., 27, 351.
[180] Martin, M. C., Du , X., Kwon, J., and Mihaly, L., 1994, Phys. Rev. B, 50, 173.
Fullerenes under high pressures 121

[181] Aoki, K., Yamawaki, H., Kakudate, Y., Yoshida , M., Usuba , S., Yokoi, H.,
Fujiwara , S., Bae, Y., Malhotra , R., and Lorents, D., 1991, J. phys. Chem., 95,
9037.
[182] Klug , D. D., Howard, J. A., and Wilkinson, D. A., 1992, Chem. Phys. Lett., 188,
168.
[183] Yamawaki, H., Aoki, K., Kakudate, Y., Yoshida , M., Usuba , S., Yokoi, H.,
Fujiwara , S., Bae, Y., Malhotra , R., and Lorents, D., 1992, Recent Trends in
High Pressure Research, edited by A. K. Singh (New Delhi: Oxford & IBH
Publishing), pp. 125± 127.
[184] Chandrabhas, N., Shashikala , M. N., Muthu , D. V. S., Sood , A. K., and Rao,
C. N. R., 1992, Chem. Phys. Lett., 197, 319.
[185] Raptis, Y. S., Snoke, D. W., Syassen, K., Roth, S., Bernier , P., and Zahab, A.,
1992, High Pressure Res., 9, 41.
[186] Snoke, D. W., Raptis, Y. S., and Syassen , K., 1992, Phys. Rev. B, 45, 14 419.
[187] Tolbert, S. H., Alivisatos , A. P., Lorenz ana , H. E., Kruger , M. B., and Jeanloz ,
R., 1992, Chem. Phys. Lett., 188, 163.
[188] Yoo, C. S., and Nellis, W. J., 1992, Chem. Phys. Lett., 198, 379.
[189] Horoyski, P. J., Wolk , J. A., and Thewalt, M. L. W., 1995, Solid St. Commun., 93,
575.
[190] Meletov , K. P., Christofilos, D., Ves, S., and Kourouklis, G. A., 1996, Phys. st.
sol. (b), 198, 553.
[191] Nishikawa , K., and Miyamoto, Y., 1995, J. Phys. Chem. Solids, 56, 577.
[192] Yu, J., Kalia , R. K., and Vashishta , P., 1993, J. chem. Phys., 99, 10001.
[193] Vashishta , P., Kalia , R. K., Nakano, A., and Yu, J., 1994, Int. J. mod. Phys. C, 5,
281.
[194] Prassides, K., 1993, Physica Scripta, T, 49, 735.
[195] Sundqvist , B., 1993, Phys. Rev. B, 48, 14712.
[196] White, M. A., Meingast , C., David , W. I. F., and Matsuo, T., 1995, Solid. St.
Commun., 94, 481.
[197] Yakovlev , E. N., and Voronov , O. A., 1994, High Temp. ± High Pressures, 26, 639.
[198] HÅ kansson, B., Andersson, P., and Baï ckstroï m, G., 1988, Rev. scient. Instrum., 59,
2269.
[199] Yu, R. C., Tea , N., Salamon, M. B., Lorents, D., and Malhotra , R., 1992, Phys.
Rev. Lett., 68, 2050.
[200] Dolganov , V. K., Zharikov , O. V., Kremenskaja , I. N., Meletov , K. P., and
Ossipyan, Yu . A., 1992, Solid St. Commun., 83, 63.
[201] Moshary, F., Chen, N. H., Silvera , I. F., Brown, C. A., Dorn, H. C., de Vries,
M. S., and Bethune, D. S., 1992, Phys. Rev. Lett., 69, 466.
[202] Snoke, D. W., Syassen, K., and Mittelbach, A., 1993, Phys. Rev. B, 47, 4146.
[203] Hess, B. C., Forgy, E. A., Frolov , S., Dick , D. D., and Vardeny, Z. V., 1994,
Molec. Crytals liq. Crystals., 256, 775.
[204] Hess, B. C., Forgy, E. A., Frolov , S., Dick , D. D., and Vardeny, Z. V., 1994, Phys.
Rev. B, 50, 4871.
[205] Meletov , K. P., Dolganov , V. K., and Ossipyan , Yu . A., 1994, Molec. Crystals. liq.
Crystals., 256, 915.
[206] Harmann, C., Zigone, M., Martinez , G., Shirley, E. L., Benedict, L. X., Louie,
S. G., Fuhrer , M. S., and Zettl , A., 1995, Phys. Rev. B, 52, R5550.
[207] Sood , A. K., 1991, Recent Trends in High Pressure Research, edited by A. K. Singh
(New Delhi: Oxford & IBH Publishing), pp. 377± 382.
[208] Sood , A. K., Chandrabhas, N., Muthu , D. V. S., Jayaraman, A., Kumar, N.,
Krishnamurthy, H. R., Pradeep, T., and Rao, C. N. R., 1992, Solid St. Commun.,
81, 89.
[209] Ching , W. Y., Huang , M.-Z., Xu , Y.-N., and Gan, F., 1992, Mod. Phys. Lett., 6, 309.
[210] Gu, B.-L., Maruyama , Y., Yu , J.-Z., Ohno, K., and Kawazoe, Y., 1994, Novel
Forms of Carbon II, edited by C. L. Renschler, D. M. Cox, J. J. Pouch and Y.
Achiba (Pittsburgh, Pennsylvania: Materials Research Society), pp. 289± 294.
[211] Laouini, N., Andersen, O. K., and Gunnarsson, O., 1995, Phys. Rev. B, 51, 17446.
[212] Shirley, E. L., Benedict, L. X., and Louie, S. G., 1996, Phys. Rev. B, 54, 10970.
122 B. Sundqvist

[213] Sundar, C. S., Premila , M., Gopalan, P., Hariharan, Y., and Bharathi, A., 1995,
Fullerene Sci. Technol., 3, 661.
[214] Schaefer , H.-E., Forster , M., Wuï rschum, R., Kraï tschmer, W., Fostiropoulos ,
K., and Huffman, D. R., 1992, Mater. Sci. Forum., 105± 110, 815.
[215] Schaefer , H. E., Forster , M., Wuï rschum, R., and Banhart, F., 1992, Novel
Forms of Carbon, edited by C. L. Renschler, J. J. Pouch and D. M. Cox (Pittsburgh,
Pennsylvania: Materials Research Society), pp. 311± 314.
[216] Schaefer , H.-E., Forster , M., Wuï rschum, R., Kraï tschmer, W., and Huffman,
D. R., 1992, Phys. Rev. B, 45, 12164.
[217] Nuíñez -Regueiro , M., Monceau , P., Rassat , A., Bernier , P., and Zahab, A, 1991,
Nature, 354, 289.
[218] Saito, Y., Shinohara , H., Kato, M., Nagashima , H., Ohkohchi, M., and Ando, Y.,
1992, Chem. Phys. Lett., 189, 236.
[219] Nuìñez -Regueiro , M., Bethoux , O., Mignot, J.-M., Monceau , P., Bernier , P.,
Fabre, C., and Rassat , A., 1993, Europhys. Lett., 21, 49.
[220] Bao, Z.-X., Gu , H.-C., Wang , J.-F., Chen, H., Li, Y.-L., Yao, Y.-X., and Zhu,
D.-B., 1993, Chin. Sci. Bull., 38, 1079.
[221] Arai, T., Murakami, Y., Suematsu , H., Kikuchi, K., Achiba , Y., and Ikemoto, I.,
1992, Solid St. Commun., 84, 827.
[222] Hamed , A., Sun, Y. Y., Tao, Y. K., Meng , R. L., and Hor, P. H., 1993, Phys. Rev. B,
47, 10873.
[223] Fujimori, S., Hoshimono, K., Fujita , S., and Fujita , S., 1994, Solid. St. Commun.,
89, 437.
[224] Nemchuk, N. I., Makarova , T. L., and Vul ’ , A. Ya ., 1996, Fullerenes: Recent
Advances in the Physics and Chemistry of Fullerenes, Vol. 3, edited by K. M. Kadish
and R. S. Ruo€ (Pennington, New Jersey: Electrochemical Society), pp. 464± 473.
[225] Sherman, A. B., Stotskii, Yu. A., and Shakin, O. V., 1996, Fiz. tverd. Tela., 38, 1742
(Engl. Transl., 1996, Phys. Solid St., 38, 961).
[226] Gudaev , O. A., Malinovskii, V. K., Maz alov , L. N., Okotrub, O. V., Paul , Eí. Eí.,
Chuvilin , A. L., and Shevtsov , Yu . V., 1995, Pis’ma Zh. tekh. Fiz., 21 (8), 15 (Engl.
Transl., 1995, Tech. Phys. Lett., 21, 589.)
[227] Mott, N. F., and Davis, E. A., 1979, Electronic Processes in Non-Crystalline Materials
(Oxford: Clarendon).
[228] Koz lov , M. E., and Yakushi, K., 1995, J. Phys.: condens. Matter., 7, L209.
[229] Yamawaki, H., Yoshida , M., Kakudate, Y., Usuba , S., Yokoi, H., Fujiwara , S.,
Aoki, K., Ruoff, R., Malhotra , R., and Lorents, D., 1993, J. phys. Chem., 97,
11161.
[230] Ma, Y., and Zou , G., 1996, High Pressure Science and Technology, edited by W. A.
Trzeciakowski (World Scienti® c: Singapore), pp. 702± 706.
[231] Pennington, C. H., and Stenger , V. A., 1996, Rev. mod. Phys., 68, 855.
[232] Yannoni, C. S., Bernier , P. P., Bethune, D. S., Meijer , G., and Salem, J. R., 1991,
J. Am. chem. Soc., 113, 3190.
[233] Burger , B., Winter , J., and Kuz many, H., 1996, Z. Phys. B, 101, 227; 1997, Synth.
Metals, 86, 2329.
[234] Hassanien, A., Gasperic , J., Demsar, J., Mu SÏevic , I., and Mihailovic , D., 1997,
Appl. Phys. Lett., 70, 417.
[235] Wang , G.-W., Komatsu , K., Murata , Y., and Shiro, M., 1997, Nature, 387, 583.
[236] Nuíñez -Regueiro , M., Monceau , P., and Hodeau , J.-L., 1992, Nature, 355, 237.
[237] Nuíñez -Regueiro , M., Abello , L., Lucaz eau , G., and Hodeau , J.-L., 1992, Phys.
Rev. B, 46, 9903.
[238] Osz lanyi, G., Allan, D. R., Cernik , R. J., Bushnell -Wye, G., Faigel , G.,
Pekker , S., and Tegz e, M., 1992, Solid St. Commun., 84, 1081.
[239] Kosowsky, S. D., Hsu , C.-H., Chen, N. H., Moshary, F., Pershan, P. S., and
Silvera , I. F., 1993, Phys. Rev. B, 48, 8474.
[240] Yoo, C. S., and Nellis, W. J., 1991, Science, 254, 1489
[241] Iwasa , Y., Arima, T., Fleming , R. M., Siegrist , T., Zhou , O., Haddon, R. C.,
Rothberg , L. J., Lyons, K. B., Carter , H. L., Jr ., Hebard , A. F., Tycko, R.,
Fullerenes under high pressures 123

Dabbagh, G., Krajewski, J. J., Thomas, G. A., and Yagi, T., 1994, Science, 264,
1570.
[242] Bashkin, I. O., Rashchupkin, V. I., Kobelev , N. P., Soifer , Ya . M., Ponyatovski,
E. G., and Moravskii, A. P., 1994, Pis’ma Zh. eksp. teor. Fiz., 59, 258 (Engl.
Transl., 1994, JETP Lett., 59, 279).
[243] Blank , V. D., Buga , S. G., Popov , M.Yu, Davydov , V. A., Agafonov , V., Shvark,
A., Seolya , R., Rassa , A., and Fabre, K., 1994, Tech. Phys., 39, 828.
[244] Blank , V. D., Buga , S. G., Serebryanaya , N. R., Dubitsky, G. A., Sulyanov ,
S. N., Popov , M. Yu ., Denisov , V. N., Ivlev , A. N., and Mavrin, B. N., 1996,
Phys. Lett. A, 220, 149.
[245] Koz lov , M. E., Hirabayashi, M., Noz aki, K., Tokumoto, M., and Ihara, H., 1995,
Appl. Phys. Lett., 66, 1199.
[246] Nuíñez -Regueiro , M., Marques, L., Hodeau , J.-L., Beíthoux , O., and Perroux , M.,
1995, Phys. Rev. Lett., 74, 278.
[247] Blank , V. D., Buga , S. G., Dubitsky, G. A., Serebryanaya , N. R., Popov , M. Yu.,
and Sundqvist , B., 1998, Carbon (to be published).
[248] Persson, P.-A., Edlund, U., Jacobsson, P., Johnels, D., Soldatov , A., and
Sundqvist , B., 1996, Chem. Phys. Lett., 258, 540.
[249] Moravsky, A. P., Abrosimova , G. E., Bashkin, I. O., Dilanian, R. A., Gurov , A.
F., Kobelev , N. P., Rashchupkin, V. I., Rybchenko, O. G., Soifer , Ya . M.,
Shekhtman, V. Sh., and Ponyatovsky , E. G., 1995, Fullerenes: Recent Advances in
the Chemistry and Physics of Fullerenes and Related Materials, Vol. 2, edited by R. S.
Ruo€ and K. M. Kadish (Pennington, New Jersey: Electrochemical Society), pp. 952±
963.
[250] Sundqvist , B., Andersson, O., Edlund, U., Fransson, Å, Inaba , A., Jacobsson, P.,
Johnels, D., Launois, P., Meingast, C., Moret, R., Moritz , T., Persson, P.-A.,
Soldatov , A., and WÅ gberg , T., 1996, Fullerenes: Recent Advances in the Chemistry
and Physics of Fullerenes, Vol. 3, edited by K. M. Kadish and R. S. Ruo€ (Pennington,
New Jersey: Electrochemical Society), pp. 1014± 1028.
[251] Rao, A. M., Eklund , P. C., Venkateswaran, U. D., Tucker , J., Duncan, M. A.,
Bendele, G. M., Stephens, P. W., Hodeau , J.-L., Marques, L., Nuíñez -Regueiro,
M., Bashkin, I. O., Ponyatovsky , E. G., and Morovsky, A. P., 1997, Appl. Phys.
A, 64, 231.
[252] Kuz many, H., Winter , J., and Burger , B., 1997, Synth. Metals, 85, 1173.
[253] Iwasa , Y., 1996, Optical Properties of Low-Dimensional Materials, edited by T. Ogawa
and Y. Kanemitsu (Singapore: World Scienti® c), pp. 340± 386.
[254] Soldatov , A., Jacobsson, P., Persson, P.-A., Sundqvist , B., Edlund, U., and
Johnels, D., 1996, Fullerenes and Fullerene Nanostructures, edited by H. Kuzmany,
J. Fink, M. Mehring and S. Roth (Singapore: World Scienti® c), pp. 344± 348.
[255] Rachdi, F., Goz e, C., Nuíñez -Regueiro, M., Marques, L., Hodeau , J.-L., and
Mehring , M., 1996, Fullerenes and Fullerene Nanostructures, edited by H. Kuzmany,
J. Fink, M. Mehring and S. Roth (Singapore: World Scienti® c), pp. 337± 343.
[256] Rachdi, F., Goz e, C., Hajji, L., Thier , K. F., Zimmer, G., Mehring , M., and
Nuíñez -Regueiro , M., 1997, Appl. Phys. A, 64, 295.
[257] Rao, A. M., Eklund , P. C., Hodeau , J.-L., Marques, L., and Nuíñez -Regueiro, M.,
1997, Phys. Rev. B, 55, 4766.
[258] Venkateswaran, U. D., Sanzi, D., Krishnappa , J., Marques, L., Hodeau , J.-L,
Nuíñez -Regueiro , M., Rao, A. M., and Eklund , P. C., 1997, Phys. stat. sol. (b),
198, 545.
[259] Launois, P., Moret, R., Persson, P.-A., and Sundqvist , B., 1998, Molecular
Nanostructures, edited by H. Kuzmany, J. Fink, M. Mehring and S. Roth (Singapore:
World Scienti® c), pp. 348± 352.
[260] Moret, R., Launois, P., Persson, P.-A., and Sundqvist, B., 1997, Europhys. Lett.,
40, 55.
[261] Agafonov , V., Davydov , V. A., Kashevarova , L. S., Rakhmanina, A. V., Kahn-
Harari, A., Dubois, P., Ceíolin, R., and Sz warc , H., 1997, Chem. Phys. Lett., 267,
193.
124 B. Sundqvist

[262] Davydov , V. A., Kashevarova , L. S., Rakhmanina , A. V., Dzyabchenko, A. V.,


Agafonov , V., Dubois, P., Ceíolin, R., and Szwarc, H., 1997, Pis’ma Zh. eksp.
teor. Fiz., 66, 110 (Engl. Transl., 1997, JETP Lett., 66, 120.
[263] Koz lov , M. E., Kaz aoui, S., Minami, N., Yakushi, K., and Tokumoto, M., 1995,
Fullerenes: Recent Advances in the Chemistry and Physics of Fullerenes and Related
materials, vol. 2, edited by R. S. Ruo€ and K. M., Kadish (Pennington, New Jersey:
Electrochemical Society), pp. 90± 98.
[264] Sz warc , H., Davydov , V. A., Plotianskaya , S. A., Kashevarova , L. S.,
Agafonov , V., and Ceíolin, R., 1996, Synth. Metals., 77, 265.
[265] Davydov , V. A., Kashevarova , L. S., Rakhmanina , A. V., Agafonov , V., Ceíolin,
R., and Sz warc, H., 1997, Carbon, 35, 735.
[266] Persson, P.-A., Andersson, O., Jacobsson , P., Soldatov , A., Sundqvist , B., and
WÅ gberg , T., 1997, J. Phys. Chem. Solids, 58, 1881.
[267] Adams, G. B., Page, J. B., Sankey, O. F., and O’ Keeffe, 1994, Phys. Rev. B, 50,
17471.
[268] Porez ag , D., Pedersson, M. R., Frauenheim, T., and Koï hler , T., 1995, Phys. Rev.
B, 52, 14963.
[269] WÅ gberg , T., Persson, P.-A., Sundqvist , B., and Jacobsson , P., 1997, Fullerenes:
Recent Advances in the Chemistry and Physics of Fullerenes and Related Materials, Vol.
5, edited by K. M. Kadish and R. S. Ruo€ (Pennington, New Jersey: Electrochemical
Society), pp. 674± 679.
[270] Bashkin, I. O., Izotov , A. N., Moravsky, A. P., Negrii, V. D., Nikolaev , R. K.,
Ossipyan, Yu . A., Ponyatovsky , E. G., and Steinman, E. A., 1997, Chem. Phys.
Lett., 272, 32.
[271] Persson, P.-A., Edlund, U., Fransson, Å., Inaba , A., Jacobsson, P., Johnels, D.,
Meingast , C., Soldatov , A., and Sundqvist , B., 1996, High Pressure Science and
Technology, edited by W. A. Trzeciakowski (World Scienti® c: Singapore), pp. 716± 718.
[272] Nagel , P., Meingast , C., Persson, P.-A., and Sundqvist, B., 1998, Molecular
Nanostructures, edited by H. Kuzmany, J. Fink, M. Mehring and S. Roth (Singapore:
World Scienti® c), pp. 365± 368.
[273] Sundqvist , B., Soldatov , A., Lundin, A., and Andersson, O., 1997, High Temp. ±
High Pressures, 29, 119.
[274] Soldatov , A., and Andersson, O., 1997, Appl. Phys. A, 64, 227.
[275] Soldatov , A., Andersson, O., Sundqvist , B., and Prassides, K., 1998, Molec.
Mater, 10, 271.
[276] Kolesnikov , A. I., Bashkin, I. O., Moravsky, A. P., Adams, M. A., Prager , M.,
and Ponyatovsky, E. G., 1996, J. Phys.: condens. Matter, 8, 10939.
[277] Renker , B., Heid , R., and Schober , H., 1998, Molecular Nanostructures, edited by H.
Kuzmany, J. Fink, M. Mehring and S. Roth (Singapore: World Scienti® c), pp. 353± 356.
[278] Soldatov , A., Prassides, K., and Ward , A., 1998, (to be published).
[279] Koz lov , M. E., Zakhidov , A. A., Yakushi, K., and Tokumoto, M., 1996, Phys. stat.
sol. (b), 197, 187.
[280] Wang , Y., Holden, J. M., Bi, X.-X., and Eklund , P. C., 1994, Chem. Phys. Lett.,
217, 413.
[281] Robert, J., Petit, P., and Fischer , J. E., 1996, Physica C, 262, 27.
[282] WÅ gberg , T., Persson, P.-A., Sundqvist , B., and Jacobsson, P., 1997, Appl. Phys. A,
64, 223.
[283] Winter , J., Kuz many, H., Soldatov , A., Persson, P.-A., Jacobsson, P., and
Sundqvist , B., 1996, Phys. Rev. B, 54, 17486.
[284] Marques, L., Hodeau , J.-L., Nuíñez -Regueiro, M., and Perroux , M., 1996, Phys.
Rev. B, 54, R12633.
[285] Harigaya , K., 1995, Phys. Rev. B, 52, 7968.
[286] Harigaya , K., 1996, Fullerenes: Recent Advances in the Chemistry and Physics of
Fullerenes and Related Materials, Vol. 3, edited by K. M. Kadish and R. S. Ruo€
(Pennington, New Jersey: Electrochemical Society), pp. 1102± 1114.
[287] Tanaka , K., Matsuura , Y., Oshima, Y., Yamabe, T., Asai, Y., and Tokumoto, M.,
1995, Solid St. Commun., 93, 163.
[288] Harigaya , K., 1996, Chem. Phys. Lett., 253, 420.
Fullerenes under high pressures 125

[289] Fagerstroï m, J., and Stafstroï m, S., 1996, Phys. Rev. B, 53, 13150.
[290] Stafstroï m, S., and Fagerstroï m, J., 1997, Appl. Phys. A, 64, 307.
[291] Stafstroï m, S., Boman, M., and Fagerstroï m, J., 1995, Europhys. Lett., 30, 295.
[292] Porez ag , D., Jungnickel , G., Frauenheim, Th., Seifert, G., Ayuela , A., and
Pederson, M. R., 1997, Appl. Phys. A, 64, 321.
[293] Springborg , M., 1995, Phys. Rev. B, 52, 2935.
[294] Sundqvist , B., Edlund , U., Jacobsson , P., Johnels, D., Jun, J., Launois, P.,
Moret, R., Persson, P.-A., Soldatov , A., and WÅ gberg , T., 1998, Carbon (to be
published).
[295] Sundqvist , B., Jacobsson, P., Jun, J., Launois, P., Moret, R., Soldatov , A., and
WÅ gberg , T., 1997, Fullerenes: Recent Advances in the Chemistry and Physics of
Fullerenes and Related Materials, Vol. 5, edited by K. M. Kadish and R. S. Ruo€
(Pennington, New Jersey: Electrochemical Society), pp. 439± 449.
[296] Winter , J., Burger , B., Hulman, M., Kuz many, H., and Soldatov , A., 1997, Appl.
Phys. A, 64, 257.
[297] Yamawaki, H., Togashi, H., Yoshida , M., Kakudate, Y., Usuba , S., Yokoi, H.,
Fujiwara , S., Aoki, K., Ruoff, R., Malhotra , R., and Lorents, D., 1994,
Fullerenes: Recent Advances in the Chemistry and Physics of Fullerenes, edited by R. S.
Ruo€ and K. M. Kadish (Pennington, New Jersey: Electrochemical Society),
pp. 594± 602.
[298] Kokorevics, A., Gravitis, J., and Kalnacs, J., 1995, Chem. Phys. Lett., 243, 205.
[299] Iwasa , Y., Arima, T., Rothberg , L. J., Fleming , R. M., Zhou , O., Lyons, K. B.,
Cheong , S.-W., Haddon, R. C., Hebard , A. F., and Thomas, G. A., 1995, Synth.
Metals, 70, 1407.
[300] Osz lanyi, G., and Forro, L., 1995, Solid St. Commun., 93, 265.
[301] Marques, L., Hodeau , J.-L., Perroux , M., Beíthoux , O., and Nuíñ ez -Regueiro, M.,
1995, Fullerenes: Recent Advances in the Chemistry and Physics of Fullerenes and Related
Materials, Vol. 2, edited by R. S. Ruo€ and K. M. Kadish (Pennington, New Jersey:
Electrochemical Society), pp. 906± 913.
[302] Marques, L., Hodeau , J.-L., and Nuíñ ez -Regueiro , M., 1996, Molec. Mater., 8, 49.
[303] Davydov , V. A., Kashevarova , L. S., Rakhmanina , A. V., Agafonov , V. N.,
Ceolin, R., and Szwarc, H., 1996, Pis’ma Zh. eksp. teor. Fiz., 63, 778 (Engl. Transl.
1996, JETP Lett., 63, 818).
[304] Davydov , V. A., Kashevarova , L. S., Rakhmanina , A. V., Agafonov , V., Ceolin,
R., Dubois, P., Sz warc, H., Keita , B., and Nadjo, L., 1996, Fullerenes: Recent
Advances in the Chemistry and Physics of Fullerenes, Vol. 3, edited by K. M. Kadish
and R. S. Ruo€ (Pennington, New Jersey: Electrochemical Society), pp. 1143± 1155.
[305] Blank ,V. D., Denisov , V. N., Ivlev , I. N., Mavrin, B. N., Serebryanaya , N. R.,
Dubitsky, G. A., Sulyanov , S. N., Popov , M. Yu ., Lvova , N. A., Buga , S. G., and
Kremkova , G. N., 1998, Carbon (to be published).
[306] Sundar, C. S., Sahu , P. Ch., Sastry, V. S., Rao, G. V. N., Sridharan, V., Premila ,
M., Bharathi, A., Hariharan, Y., Radhakrishnan, T. S., Muthu , D. V. S., and
Sood , A. K., 1996, Phys. Rev. B, 53, 8180.
[307] Sundqvist , B., 1998, Phys. Rev. B, 57, 3164.
[308] Marques, L., Hodeau , J.-L., Nuíñ ez -Regueiro, M., and Perroux , M., 1998, Phys.
Rev. B, 57, 3166.
[309] Xu, C. H., and Scuseria , G. E., 1995, Phys. Rev. Lett., 74, 274.
[310] Okada, S., and Saito, S., 1997, Phys. Rev. B, 55, 4039.
[311] Harigaya , K., 1995, Chem. Phys. Lett., 242, 585.
[312] Goz e, C., Rachdi, F., Hajji, L., Nuíñ ez -Regueiro, M., Marques, L., Hodeau ,
J.-L., and Mehring , M., 1996, Phys. Rev. B, 54, R3676.
[313] Maniwa, Y., Sato, M., Kume, K., Koz lov , M. E., and Tokumoto, M., 1996,
Carbon, 34, 1287.
[314] Koz lov , M. E., Tokumoto, M., and Yakushi, K., 1997, Appl. Phys. A, 64, 241; 1998,
Molecular Nanostructures, edited by H. Kuzmany, J. Fink, M. Mehring and S. Roth
(Singapore: World Scienti® c), pp. 361± 364.
[315] Koz lov , M. E., Tokumoto, M., and Yakushi, K., 1997, Synth. Metals, 86, 2349.
[316] Kamaraís, K., Iwasa , Y., and Forroí, L., 1997, Phys. Rev. B, 55, 10999.
126 B. Sundqvist

[317] Takahashi, Y., Takada , Y., Kotake, S., Matsumuro, A., and Senoo, M., 1996,
J. Jpn Inst. Metals, 60, 700.
[318] Dworkin, A., Szwarc, H., Davydov , V. A., Kashevarova , L. S., Rakhmanina ,
A. V., Agafonov , V., and Ceíolin, R., 1997, Carbon, 35, 745.
[319] Koz lov , M. E., Hirabayashi, M., Noz aki, K., Tokumoto, M., and Ihara, H., 1995,
Synth. Metals., 70, 1411.
[320] Koz lov , M. E., Yase, K., Minami, N., Fons, P., Durand, H.-A., Hirabayashi, M.,
Noz aki, K., Tokumoto, M., and Ihara, H., 1995, Fullerenes: Recent Advances in the
Chemistry and Physics of Fullerenes and Related Materials, Vol. 2, edited by R. S. Ruo€
and K. M. Kadish (Pennington, New Jersey: Electrochemical Society), pp. 99± 107.
[321] Koz lov , M. E., Fons, P., Durand, H.-A., Noz aki, K., Tokumoto, M., Yase, K.,
and Minami, N., 1996, J. appl. Phys., 80, 1182.
[322] Koz lov , M. E., Yase, K., Minami, N., Fons, P., Durand, H.-A., Obraz tsov , A. N.,
Noz aki, K., and Tokumoto, M., 1996, J. Phys. D, 29, 929.
[323] Hodeau , J. L., Tonnerre, J. M., Bouchet-Fabre, B., Nuí ñ ez -Regueiro, M.,
Capponi, J. J., and Perroux , M., 1994, Phys. Rev. B, 50, 10311.
[324] Smontara , A., Biljakovic , K., Staresinic , D., Pajic , D., Koz lov , M. E.,
Hirabayashi, M., Tokumoto, M., and Ihara, H., 1996, Physica B, 219± 220, 160.
[325] Biljakovic , K., Smontara , A., Staresinic , D., Pajic , D., Koz lov , M. E.,
Hirabayashi, M., Tokumoto, M., and Ihara , H., 1996, J. Phys.: condens. Matter, 8,
L27.
[326] Blank , V., Buga , S., Popov , M., Davydov , V., Kulnitsky, B., Tatyanin, E.,
Agafonov , V., Ceolin, R., Szwarc, H., Rassat, A., and Fabre, C., 1994, Molec.
Mater., 4, 149.
[327] Blank , V. D., Buga , S. G., Popov , M.Yu., Davydov , V. A., Agafonov , V., Ceíolin,
R., Szwarc, H., Rassat , A., and Fabre, C., 1995, New J. Chem., 19, 253.
[328] Blank , V. D., Buga , S. G., Serebryanaya , N. R., Denisov , V. N., Dubitsky, G. A.,
Ivlev , A. N., Mavrin, B. N., and Popov , M.Yu ., 1995, Phys. Lett. A, 205, 208.
[329] Davydov , V. A., Dubitsky, G. A., Kashevarova , L. S., Plotyanskaya , S. A.,
Korobov , M. V., Agafonov , V., Ceolin, R., and Szwarc, H., 1995, Fullerenes:
Recent Advances in the Chemistry and Physics of Fullerenes, Vol. 2, edited by R. S.
Ruo€ and K. M. Kadish (Pennington, New Jersey: Electrochemical Society), pp. 964±
972.
[330] Braz hkin, V. V., Lyapin, A. G., and Popova , S. V., 1997, Pis’ma Zh. eksp. teor. Fiz.,
64, 755 (Engl. Transl., 1997, JETP Lett., 64, 802).
[331] Thorpe, M. F., 1983, J. non-crystalline Solids, 57, 355; He, H. and Thorpe, M. F.,
1985, Phys. Rev. Lett., 54, 2107; Djordjevic , B. R., and Thorpe, M. F., 1997, J. Phys.:
condens. Matter, 9, 1983.
[332] Keita , B., Nadjo, L., Davydov , V. A., Agafonov , V., Ceíolin, R., and Szwarc, H.,
1995, New J. Chem., 19, 769.
[333] Lenosky, T., Gonze, X., Teter , M., and Elser , V., 1992, Nature, 355, 333.
[334] Townsend, S. J., Lenosky, T. J., Muller , D. A., Nichols, C. S., and Elser , V.,
1992, Phys. Rev. Lett., 69, 921.
[335] Vanderbilt , D., and Tersoff, J., 1992, Phys. Rev. Lett., 68, 511.
[336] Surjaín, P. R., Aí ngyaín, J. G., Laízaír , A., Neímeth, K., and Biíroí, L. P., 1996,
Fullerenes and Fullerene Nanostructures, edited by H. Kuzmany, J. Fink, M. Mehring
and S. Roth (Singapore: World Scienti® c), pp. 319± 322.
[337] Braz hkin, V. V., Lyapin, A. G., Antonov , Yu. V., Popova , S. V., Klyuev , Yu . A.,
Naletov , A. M., and Mel ’ nik, N. N., 1995, Pis’ma Zh. eksp. teor. Fiz., 62, 328
(Engl. Transl., 1995, JETP Lett., 62, 350).
[338] Braz hkin, V. V., Lyapin, A. G., Stalgorova , O. V., Gromnitskaya , E. L., Popova ,
S. V., and Tsiok , O. B., 1996, High Pressure Science and Technology, edited by W. A.
Trzeciakowski (Singapore: World Scienti® c), pp. 285± 288.
[339] Serebryanaya , N. R., Blank , V. D., Dubitsky, G. A., Buga , S. G., Aksenenkov ,
V. V., and Popov , M. Yu ., 1996, High Pressure Science and Technology, edited by
W. A. Trzeciakowski (Singapore: World Scienti® c), pp. 719± 721.
[340] Blank , V. D., Kulnitskiy, B. A., and Tatyanin, Ye. V., 1995, Phys. Lett. A, 204,
151.
Fullerenes under high pressures 127

[341] Blank , V. D., Kulnitskiy , B. A., and Tatyanin, Ye. V., 1996, High Pressure Science
and Technology, edited by W. A. Trzeciakowski (Singapore: World Scienti® c), pp. 710±
712.
[342] Zhang, B. L., Wang , C. Z., Ho, K. M., and Chan, C. T., 1994, Europhys. Lett., 28,
219.
[343] Blank , V. D., Buga , S. G., Dubitsky, G. A., Serebryanaya , N. R., Denisov , V. N.,
Ivlev , A. V., Mavrin, B. N., and Popov , M. Yu ., 1996, Molec. Mater., 7, 251.
[344] Blank , V. D., Tatyanin, Ye. V., and Kulnitskiy , B. A., 1997, Phys. Lett. A, 225,
121.
[345] Blank , V. D., Buga , S. G., Dubitsky , G. A., Serebryanaya , N. R., Sulyanov ,
S. N., and Popov , M. Yu ., 1998, Molecular Nanostructures, edited by H. Kuzmany,
J. Fink, M. Mehring and S. Roth (Singapore: World Scienti® c), pp. 506± 510.
[346] Blank , V. D., Buga , S. G., Dubitsky, G. A., Denisov , V. N., Ivlev , A. N., Mavrin,
B. N., and Popov , M. Yu ., 1996, High Pressure Science and Technology, edited by
W. A. Trzeciakowski (Singapore: World Scienti® c), pp. 713± 715.
[347] Blank , V. D., Buga , S. G., Dubitsky , G. A., Serebryanaya , N. R., Bagramov ,
R. H., and Popov , M. Yu ., 1996, Fullerenes and Fullerene Nanostructures, edited by
H. Kuzmany, J. Fink, M. Mehring and S. Roth (Singapore: World Scienti® c), pp. 613±
617.
[348] Blank , V. D., Buga , S. G., Dubitsky, G. A., Popov , M. Yu., and Serebryanaya ,
N. R., 1996, High Pressure Science and Technology, edited by W. A. Trzeciakowski
(Singapore: World Scienti® c), pp. 707± 709.
[349] Zeger , L., and Kaxiras, E., 1993, Phys. Rev. Lett., 70, 2920.
[350] Lang , H. P., Wiesendanger , R., Thommen-Geiser , V., Hofer , R., and
Guï ntherodt, H.-J., 1994, J. Vac. Sci. Technol B, 12, 2136.
[351] Lang , H. P., Thommen-Geiser , V., and Guï ntherodt, H.-J., 1996, Synth. Metals., 77,
161.
[352] Rao, C. N. R., Govindaraj , A., Aiyer , H. N., and Seshadri, R., 1995, J. phys.
Chem., 99, 16814.
[353] Pierson, H. O., 1993, Handbook of Carbon, Graphite, Diamond and Fullerenes (Park
Ridge, New Jersey: Noyes Publications).
[354] Bocquillon, G., Bogicevic , C., Clerc , F., Leíger , J. M., Fabre, C., and Rassat, A.,
1994, High-Pressure Science and TechnologyÐ 1993, edited by S. C. Schmidt, J. W.
Shaner, G. A. Samara and M. Ross (New York: American Institute of Physics),
pp. 647± 649.
[355] May, Y., Zou , G., Yang , H., and Meng , J., 1994, Appl. Phys. Lett., 65, 822.
[356] Bassett , W. A., and Weathers, M. S., 1994, High Pressure Science and TechnologyÐ
1993, edited by S. C. Schmidt, J. W. Shaner, G. A. Samara and M. Ross (New York:
American Institute of Physics), pp. 651± 653.
[357] Yoo, C. S., Nellis, W. J., Sattler , M. L., Musket, R. G., Hinsey, N., and
Brocious, W., 1992, Novel Forms of Carbon, edited by C. L. Renschler, J. J. Pouch
and D. E. Cox (Pittsburgh, Pennsylvania: Materials Research Society), pp. 155± 160.
[358] Yoo, C. S., Nellis, W. J., Sattler , M. L., and Musket, R. G., 1992, Appl. Phys.
Lett., 61, 273.
[359] Hirai, H., and Kondo, K., 1991, Science, 253, 772.
[360] Sekine, T., Maruyama , Y., Nagata , M., Miz utani, N., Kitagawa , H., and
Inoguchi, H., 1994, High-Pressure Science and TechnologyÐ 1993, edited by S. C.
Schmidt, J. W. Shaner, G. A. Samara and M. Ross (New York: American Institute
of Physics), pp. 655± 658.
[361] Sandreí, E., and Cyrot-Lackmann, F., 1994, Solid St. Commun., 90, 431.
[362] Hirai, H., Kondo, K., and Ohwada , T., 1993, Fullerene Sci. Technol., 1, 379.
[363] Hirai, H., Kondo, K., and Ohwada , T., 1993, Carbon, 31, 1095.
[364] Hirai, H., and Kondo, K., 1994, High-Pressure Science and TechnologyÐ 1993, edited
by S. C. Schmidt, J. W. Shaner, G. A. Samara and M. Ross (New York: American
Institute of Physics), pp. 659± 662.
[365] Hirai, H., Kondo, K., Yoshiz awa , N., and Shiraishi, M., 1994, Chem. Phys. Lett.,
226, 595.
128 B. Sundqvist

[366] Hirai, H., Kondo, K., Yoshiz awa, N., and Shiraishi, M., 1994, Appl Phys. Lett., 64,
1797.
[367] Hirai, H., and Kondo, K., 1995, Fullerene Sci. Technol., 3, 369.
[368] Hirai, H., and Kondo, K., 1995, Phys. Rev. B, 51, 15555.
[369] Hirai, H., Tabira , Y., Kondo, K., Oikawa , T., and Ishiz awa , N., 1995, Phys. Rev.
B, 52, 6162.
[370] Banhart, F., and Ajayan, P. M., 1996, Nature, 382, 433.
[371] Wesolowski, P., Lyutovicz , Y., Banhart, F., Carstanjen, H. D., and
Kronmuï ller , H., 1997, Appl. Phys. Lett., 71, 1948.
[372] Pasqualini, E., 1997, Phys. Rev. B, 56, 7751.
[373] Kawamura, H., Akahama, Y., Kobayashi, M., Shinohara , H., and Saito, Y., 1994,
J. Phys. Soc. Japan, 63, 2445.
[374] Rao, A. M., Menon, M., Wang , K.-A., Eklund, P. C., Subbaswamy, K. R.,
Cornett, D. S., Duncan, M. A., and Amster , I. J., 1994, Chem. Phys. Lett., 224, 106.
[375] Kawamura, H., Kobayashi, M., Akahama , Y., Shinohara , H., Sato, H., and
Saito, Y., 1992, Solid St. Commun., 83, 563.
[376] Kawamura, H., Akahama , Y., Kobayashi, M., Hasegawa , Y., Shinohara , H.,
Sato, H., and Saito, Y., 1993, Jap. J. appl. Phys., 32, L101.
[377] Piacente, V., Gigli, G., Scardala , P., Giustini, A., and Bardi, G., 1996, J. phys.
Chem., 100, 9815.
[378] Christides, C., Thomas, I. M., Dennis, T. J. S., and Prassides, K., 1993, Europhys.
Lett., 22, 611.
[379] Lundin, A., Soldatov , A., and Sundqvist , B., 1995, Europhys. Lett., 30, 469.
[380] Lundin, A., Soldatov , A., and Sundqvist , B., 1995, Science and Technology of
Fullerene Materials, edited by P. Bernier, D. S. Bethune, L. Y. Chiang, T. W.
Ebbesen, R. M., Metzger and J. W. Mintmire (Pittsburgh, Pennsylvania: Materials
Research Society), pp. 555± 560.
[381] Soldatov , A., and Sundqvist , B., 1995, Fullerenes: Recent Advances in the Physics and
Chemistry of Fullerenes, Vol. 2, edited by R. S. Ruo€ and K. M. Kadish (Pennington,
New Jersey: Electrochemical Society), pp. 881± 890.
[382] Soldatov , A., and Sundqvist , B., 1996, J. Phys. Chem. Solids., 57, 1371.
[383] Forsman, H., and Andersson, P., 1984, J. chem. Phys., 80, 2804.
[384] Samara , G. A., Hansen, L. V., Morosin, B., and Schirber , J. E., 1996, Phys. Rev. B,
53, 5211.
[385] Ramasesha , S. K., Singh, A. K., Seshadri, R., Sood , A. K., and Rao, C. N. R., 1994,
Chem. Phys. Lett., 220, 203.
[386] Maksimov , A. A., Meletov , K. P., Osip’ yan, Yu.A., Tartakovskii, I. I., Artemov ,
Yu . V., and Nudel ’ man, M. A., 1993, Pis’ma Zh. eksp. teor. Fiz., 57, 801 (Engl.
Transl., 1993, JETP Lett., 57, 816).
[387] Meletov , K. P., Maksimov , A., Ossipyan, Yu., and Tartakovskii, I., 1994, Molec.
Crystals liq. Crystals., 256, 909.
[388] Sood , A. K., Chandrabhas, N., Muthu , D. V. S., Hariharan, Y., Bharathi, A.,
and Sundar, C. S., 1994, Phil. Mag. B, 70, 347.
[389] Fradkin, M. A., 1996, Pis’ma Zh. eksp. teor. Fiz., 63, 594 (Engl. Transl., 1996, JETP
Lett., 62, 628).
[390] Chandrabhas, N., Sood , A. K., Muthu , D. V. S., Sundar , C. S., Bharathi, A.,
Hariharan, Y., and Rao, C. N. R., 1994, Phys. Rev. Lett., 73, 3411.
[391] Menon, M., Rao, A. M., Subbaswamy, K. R., and Eklund , P. C., 1995, Phys. Rev. B,
51, 800.
[392] Iwasa , Y., Furudate, T., Fukawa , T., Ozaki, T., Mitani, T., Yagi, T., and Arima,
T., 1997, Appl. Phys. A, 64, 251.
[393] Kniaz , K., Girifalco, L. A., and Fischer , J. E., 1995, J. phys. Chem., 99, 16804.
[394] Barrio, J. B. M., Tretiakov , N. P., and Zubov , V. I., 1997, Phys. Lett. A, 234, 69.
[395] Meletov , K. P., Dolganov , V. K., and Ossipyan, Yu . A., 1993, Solid St. Commun.,
87, 639.
[396] Lang , J. M., Dreger , Z. A., and Drickamer, H. G., 1996, J. phys. Chem., 100, 8064.
[397] Ajayan, P. M., and Ebbesen, T. W., 1997, Rep. Prog. Phys., 60, 1025.
Fullerenes under high pressures 129

[398] Golberg , D., Bando, Y., Eremets, M., Takemura , K., Kurashima , K., and Yusa ,
H., 1996, Appl. Phys. Lett., 69, 2045.
[399] Yakobson, B. I., Brabec , C. J., and Bernholc , J., 1996, Phys. Rev. Lett., 76, 2511.
[400] Ruoff, R. S., Lorents, D. C., Laduca , R., Awadalla , S., Weathersby , S., Parvin,
K., and Subramoney, S., 1995, Fullerenes: Recent Advances in the Physics and
Chemistry of Fullerenes, Vol. 2, edited by R. S. Ruo€ and K. M. Kadish (Pennington,
New Jersey: Electrochemical Society), pp. 557± 562.
[401] Subramoney, S., Ruoff, R. S., Laduca , R., Awadalla , S., and Parvin, K., 1995,
Fullerenes: Recent Advances in the Physics and Chemistry of Fullerenes, Vol. 2, edited by
R. S. Ruo€ and K. M. Kadish (Pennington, New Jersey: Electrochemical Society),
pp. 563± 569.
[402] Subramoney, S., Ruoff, R. S., Laduca , R., and Parvin, K., 1996, Fullerenes: Recent
Advances in the Physics and Chemistry of Fullerenes, Vol. 3, edited by K. M. Kadish and
R. S. Ruo€ (Pennington, New Jersey: Electrochemical Society), pp. 728± 740.
[403] Treacy, M. M. J., Ebbesen, T. W., and Gibson, J. M., 1996, Nature, 381, 678.
[404] Wong , E. W., Sheehan, P. E., and Lieber , C. M., 1997, Science, 277, 1971.
[405] Falvo, M. R., Clary, G. J., Taylor , R. M., II, Chi, V., Brooks, F. P., Jr,
Washburn, S., and Superfine, R., 1997, Nature, 389, 582.
[406] Gamaly, E. G., 1998, Molecular Nanostructures, edited by H. Kuzmany, J. Fink, M.
Mehring and S. Roth (Singapore: World Scienti® c), pp. 482± 488.
[407] Lu, J. P., 1997, Phys. Rev. Lett., 79, 1297.
[408] Vaughan, G. B. M., Heiney, P. A., Cox , D. E., Mc Ghie, A. R., Jones, D. R.,
Strongin, R. M., Cichy, M. A., and Smith, A. B., III, 1992, Chem. Phys., 168, 185.
[409] Lommen, A. N., Heiney, P. A., Vaughan, G. B. M., Stephens, P. W., Liu , D., Li, D.,
Smith, A. L., Mc Ghie, A. R., Strongin, R. M., Brard, L., and Smith, A. B., III,
1994, Phys. Rev. B, 49, 12572.
[410] Neumann, D. A., Huang , Q., Copley , J. R. D., Fischer , J. E., Heiney, P. A.,
Strongin, R. M., Brard , L., and Smith, A. B., III, 1995, Fullerenes: Recent
Advances in the Physics and Chemistry of Fullerenes, Vol. 2, edited by R. S. Ruo€ and
K. M. Kadish (Pennington, New Jersey: Electrochemical Society), pp. 791± 797.
[411] Meingast , C., Roth, G., Kappes, M. M., Michel , R. H., Stoermer , C., Brard , L.,
Heiney, P. A. Fischer , J. E., Smith, A. B., III, and Strongin, R. M., 1995, Physics
and Chemistry of Fullerenes and Derivatives, edited by H. Kuzmany, J. Fink, M.
Mehring and S. Roth (Singapore: World Scienti® c), pp. 167± 170.
[412] Meingast , C., Roth, G., Pintschovius, L., Michel , R. H., Stoermer , C., Kappes,
M. M., Heiney, P. A., Brard, L., Strongin, R. M., and Smith, A. B., III, 1996,
Phys. Rev. B, 54, 124.
[413] Cheng , A., and Klein, M. L., 1992, J. Chem. Soc., Faraday Trans., 88, 1949.
[414] Lundin, A., Soldatov , A., Sundqvist , B., Strongin, R. M., Brard, L., Fischer ,
J. E., and Smith, A. B., III, 1996, Carbon, 34, 1119.
[415] Lundin, A., Soldatov , A., and Sundqvist, B., 1996, Fullerenes: Recent Advances in
the Chemistry and Physics of Fullerenes and Related Materials, Vol. 3, edited by K. M.
Kadish and R. S. Ruo€ (Pennington, New Jersey: Electrochemical Society), pp. 1138±
1142.
[416] Kolesnikov , A. I., Antonov , V. E., Bashkin, I. O., Ponyatovsky, E. G.,
Muz ychka , A. Yu ., Moravsky, A. P., Grosse, G., and Wagner , F. E., 1997,
Physica B, 234± 236, 10.
[417] Kolesnikov , A. I., Antonov , V. E., Bashkin, I. O., Grosse, G., Moravsky, A. P.,
Muz ychka , A. Yu ., Ponyatovsky , E. G., and Wagner , F. E., 1997, J. Phys.: condens.
Matter, 9, 2831.
[418] Jin, C., Hettich, R., Compton, R., Joyce, D., Blencoe, J., and Burch, T., 1994,
J. phys. Chem., 98, 4215.
[419] Hummelen, J. C., Knight, B., Pavlovich, J., Gonzaílez , R., and Wudl, F., 1995,
Science, 269, 1554.
[420] Brown, C. M., Beer , E., Bellavia , C., Cristofolini, L., Gonzaílez , R., Hanfland,
M., Haï usermann, D., Keshavarz -K, M., Kordatos, K., Prassides, K., and
Wudl , F., 1996, J. Am. Chem. Soc., 118, 8715.
130 B. Sundqvist

[421] Zhu, Q., Cox , D. E., Fischer , J. E., Kniaz , K., McGhie, A. R., and Zhou , O., 1992,
Nature, 355, 712.
[422] Werner , H., Wesemann, M., and Schloï gl , R., 1992, Europhys. Lett., 20, 107.
[423] Akahama, Y., Kobayashi, M., Kawamura , H., Shinohara , H., Sato, H., and
Saito, Y., 1992, Solid St. Commun., 82, 605.
[424] Kobayashi, M., Akahama, Y., Kawamura , H., Shinohara , H., Sato, H., and
Saito, Y., 1993, Mater. Sci. Engng, B19, 100.
[425] Matsuz aki, S., Hashimoto, K., Tomiku, T., Fujimoto, H., and Ichimura , K., 1995,
Fullerene Sci. Technol., 3, 59.
[426] Allemand , P.-M., Khemani, K. C., Koch, A., Wudl , F., Holcz er , K., Donovan,
S., Gruï ner , G., and Thompson, J. D., 1991, Science, 253, 1154.
[427] Sparn, G., Thompson, J. D., Allemand, P.-M., Li, Q., Wudl , F., Holcz er, K., and
Stephens, P. W., 1992, Solid St. Commun., 82, 779.
[428] Thompson, J. D., Sparn, G., Diederich, F., Gruï ner, G., Holcz er , K., Kaner ,
R. B., Whetten, R. L., Allemand , P.-M., Li, Q., and Wudl , F., 1992, Electrical,
Optical, and Magnetic Properties of Organic Solid State Materials, edited by L. Y.
Chiang, A. F. Garito and D. J. Sandman (Pittsburgh, Pennsylvania: Materials
Research Society), pp. 315± 320.
[429] Wudl , F., and Thompson, J. D., 1992, J. Phys. Chem. Solids, 53, 1449.
[430] Assink , R. A., Schirber , J. E., Loy, D. A., Morosin, B., and Carlsson, G. A., 1992,
J. Mater. Res., 7, 2136.
[431] Morosin, B., Jorgensen, J. D., Short, S., Kwei, G. H., and Schirber , J. E., 1996,
Phys. Rev. B, 53, 1675.
[432] Kwei, G. J., Jorgensen, J. D., Schirber , J. E., and Morosin, B., 1997, Fullerene Sci.
Technol., 5, 243.
[433] Schirber , J. E., Hansen, L., Morosin, B., Fischer , J. E., Jorgensen, J. D., and
Kwei, G. H., 1996, Physica C, 260, 173.
[434] Morosin, B., Schirber , J. E., Jorgensen, J. D., Kwei, G. H., Yildirim, T., and
Fischer , J. E., 1996, Fullerenes: Recent Advances in the Chemistry and Physics of
Fullerenes and Related Materials, Vol. 3, edited by K. M. Kadish and R. S. Ruo€
(Pennington, New Jersey: Electrochemical Society), pp. 446± 456.
[435] Gadd , G. E., James, M., Moricca , S., Evans, P. J., and Davis, R. L., 1996, Fullerene
Sci. Technol., 4, 853.
[436] Gadd , G. E., Evans, P. J., Hurwood, D. J., Wood , J., Moricca , S., Blackford ,
M., Elcombe, M., Kennedy, S., and James, M., 1996, Chem. Phys. Lett., 261, 221.
[437] Gadd , G. E., Kennedy, S. J., Moricca , S., Howard, C. J., Elcombe, M. M., Evans,
P. J., and James, M., 1997, Phys. Rev. B, 55, 14794.
[438] Assink , R. A., Loy, D. A., Schirber , J. E., and Morosin, B., 1992, Novel Forms of
Carbon, edited by C. L. Renschler, J. J. Pouch and D. M. Cox (Pittsburgh,
Pennsylvania: Materials Research Society), pp. 255± 260.
[439] Bernier , P., Luk’ yanchuk , I., Belahmer , Z., Ribet, M., and Firlej , L., 1996, Phys.
Rev. B, 53, 7535.
[440] Dworkin, A., Szwarc, H., and Ceíolin, R., 1993, Europhys. Lett., 22, 35.
[441] Myers, S. A., Assink , R. A., Schirber , J. E., and Loy, D. A., 1995, Science and
Technology of Fullerene Materials, edited by P. Bernier, D. S. Bethune, L. Y. Chiang,
T. W. Ebbesen, R. M. Metzger and J. W. Mintmire (Pittsburgh, Pennsylvania; Materials
Research Society), pp. 505± 509.
[442] Schirber , J. E., Assink , R. A., Samara , G. A., Morosin, B., and Loy, D., 1995,
Phys. Rev. B, 51, 15552.
[443] Gu, M., Wang , Y., Tang , T. B., Zhang, W., Hu, C., Yan, F., and Feng , D., 1996,
Phys. Lett. A, 223, 273.
[444] Morris, D. E., Singh, K. K., and Sinha , A. P. B., 1993, J. Mater. Res., 8, 2273.
[445] Holleman, I., von Helden, G., Olthof, E. H. T., van Bentum, P. J. M., Engeln,
R., Nachtegaal , G. H., Kentgens, A. P. M., Meier , B. H., van der Avoird , A.,
and Meijer , G., 1997, Phys. Rev. Lett., 79, 1138.
[446] Tellgmann, R., Krawez , N., Lin, S.-H., Hertel , I. V., and Campbell , E. E. B.,
1996, Nature, 382, 407.
Fullerenes under high pressures 131

[447] Saunders, M., Cross, R. J., Jimeínez -Vaízquez , H. A., Shimsi, R., and Khong, A.,
1996, Science, 271, 1693.
[448] Saunders, M., Jimeínez -Vaz quez , H. A., Cross, R. J., and Poreda , R. J., 1993,
Science, 259, 1428.
[449] Saunders, M., Jimeínez -Vaíz quez , H. A., Cross, R. J., Mrocz kowski, S., Gross,
M. L., Giblin, D. E., and Poreda , R. J., 1994, J. Am. chem. Soc., 116, 2193.
[450] Cross, R. J., Jimeínez -Vaíz quez , H. A., Khong , A., Shimshi, R., and Saunders, M.,
1996, Fullerenes: Recent Advances in the Chemistry and Physics of Fullerenes, Vol. 3,
edited by K. M. Kadish and R. S. Ruo€ (Pennington, New Jersey: Electrochemical
Society), pp. 586± 592.
[451] DiCamillo, B. A., Hettich, R. L., Guiochon, G., Compton, R. N., Saunders, M.,
Jimenez -Vaíz quez , H. A., Khong, A., and Cross, R. J., 1996, J. phys. Chem., 100,
9197.
[452] Gadd , G. E., Blackford , M., Moricca , S., Webb, N., Evans, P. J., Smith, A. M.,
Jacobsen, G., Leung , S., Day, A., and Hua , Q., 1997, Science, 277, 933.
[453] Haddon, R. C., Hebard , A. F., Rosseinsky, M. J., Murphy, D. W., Duclos, S. J.,
Lyons, K. B., Miller , B., Rosamilia , J. M., Fleming , R. M., Kortan, A. R.,
Glarum, S. H., Makhija , A. V., Muller , A. J., Eick , R. H., Zahurak , S. M.,
Tycko, R., Dabbagh, G., and Thiel , F. A., 1991, Nature, 350, 320.
[454] Hebard , A. F., Rosseinsky, M. J., Haddon, R. C., Murphy, D. W., Glarum, S. H.,
Palstra , T. T. M., Ramirez , A. P., and Kortan, A. R., 1991, Nature, 350, 600.
[455] Haddon, R. C., Hebard , A. F., Rosseinsky , M. J., Murphy, D. W., Glarum, S. H.,
Palstra , T. T. M., Ramirez , A. P., Duclos, S. J., Fleming , R. M., Siegrist , T.,
and Tycko, R., 1992, Fullerenes, ACS Symposium Series, Vol. 481, edited by G. S.
Hammond and V. J. Kuck (Washington, DC: American Chemical Society), pp. 71± 89.
[456] Holcz er , K., and Whetten, R. L., 1992, Carbon, 30, 1261.
[457] Holcz er , K., 1992, Int. J. mod. Phys. B, 6, 3967.
[458] Tanigaki, K., 1993, Mater. Sci. Engng, B19, 135.
[459] Fischer , J. E., 1994, Proceedings of the NATO Advanced Workshop on the Physics and
Chemistry of the Fullerenes, edited by K. Prassides (Dordrecht: Kluwer), pp. 223± 244.
[460] Yildirim, T., 1996, Fullerenes: Recent Advances in the Chemistry and Physics of
Fullerenes and Related Materials, Vol. 3, edited by K. M. Kadish and R S. Ruo€
(Pennington, New Jersey: Electrochemical Society), pp. 1155± 1169.
[461] Buntar , V., and Weber , H. W., 1996, Supercond. Sci. Technol., 9, 599.
[462] Stephens, P. W., 1995, Physics and Chemistry of Fullerenes and Derivatives, edited by
H. Kuzmany, J. Fink, M. Mehring and S. Roth (Singapore: World Scienti® c), pp. 291±
296.
[463] Zettl , A., 1996, Fullerenes and Fullerene Nanostructures, edited by H. Kuzmany,
J. Fink, M. Mehring and S. Roth (Singapore: World Scienti® c), pp. 14± 33.
[464] Gunnarsson, O., 1997, Rev. mod. Phys., 69, 575.
[465] Schilling , J. S., Diederichs, J., and Gangyopadhyay, A., 1997, Fullerenes: Recent
Advances in the Chemistry and Physics of Fullerenes and Related Materials, Vol. 4, edited
by R. S. Ruo€ and K. M. Kadish (Pennington, New Jersey: Electrochemical Society),
pp. 980± 995.
[466] Nalimova , V. A., Sklovsky, D. E., Bondarenko, G. N., Alvergnat-Gaucher , H.,
Bonnamy, S., and Beíguin, F., 1997, Synth. Metals, 88, 89.
[467] Menu , S., Gaucher , H., Conard, J., Lauginie, P., Noz hov , A., and Nalimova ,
V. A., 1998, Molecular Nanostructures, edited by H. Kuzmany, J. Fink, M. Mehring
and S. Roth (Singapore: World Scienti® c), pp. 262± 265.
[468] Zhou , O., Vaughan, G. B. M., Zhu, Q., Fischer , J. E., Heiney, P. A., Coustel , N.,
Mc Cauley, J. P., Jr , and Smith, A. B., III, 1992, Science, 255, 833.
[469] Ludwig , H. A., Fietz , W. H., Hornung , F. W., Grube, K., Renker , B., and
Burkhart, G. J., 1994, Physica C, 234, 45.
[470] Diederichs, J., Schilling , J. S., Herwig , K. W., and Yelon, W. B., 1997, J. Phys.
Chem. Solids, 58, 123.
[471] Khaz eni, K., Hone, J., Chopra , N. G., Zettl , A., Nguyen, J., and Jeanloz , R.,
1997, Appl. Phys. A, 64, 263.
[472] Martins, J. L., and Troullier , N., 1992, Phys. Rev. B, 46, 1766.
132 B. Sundqvist

[473] Huang , M.-Z., Xu , Y.-N., and Ching , W. Y., 1993, Phys. Rev. B, 47, 8249.
[474] Zhu, Q., 1995, Phys. Rev. B, 52, R723.
[475] Xiang , X.-D., Hou , J. G., Briceño, G., Vareka , W. A., Mostovoy, R., Zettl , A.,
Crespi, V. H., and Cohen, M. L., 1992, Science, 256, 1190.
[476] Zettl , A., Lu, L., Xiang , X.-D., Hou , J. G., Vareka , W. A., and Fuhrer, M. S.,
1994, J. Supercond., 7, 639.
[477] Vareka , W. A., Fuhrer , M. S., and Zettl , A., 1994, Physica C, 235± 240, 2507.
[478] Vareka , W. A., and Zettl , A., 1994, Phys. Rev. Lett., 72, 4121.
[479] Sundqvist , B., and Nilsson, E. M. C., 1994, Physica C, 235± 240, 1407.
[480] Dyos, G. T., and Farrell , T., 1992, Electrial Resistivity Handbook (London: Peter
Peregrinus).
[481] Mott, N. F., and Jones, H., 1936, The Theory of the Properties of Metals and Alloys
(Oxford: Clarendon) p. 268; 1958 reprinted (New York: Dover Publications).
[482] Sundqvist , B., 1993, Mod. Phys. Lett. B, 7, 491.
[483] Grimvall , G., 1981, The Electron± Phonon Interaction in Metals (Amsterdam: North-
Holland), p. 216.
[484] Erwin, S. C., and Pickett, W. E., Phys. Rev. B, 46, 14257.
[485] Hone, J., Fuhrer, M. S., Khazeni, K., and Zettl , A., 1995, Phys. Rev. B, 52, R8700.
[486] Hone, J., Khaz eni, K., and Zettl , A., 1996, Fullerenes and Fullerene Nanostructures,
edited by H. Kuzmany, J. Fink, M. Mehring and S. Roth (Singapore: World Scienti® c),
pp. 115± 118.
[487] Khaz eni, K., Crespi, V. H., Hone, J., Zettl, A., and Cohen, M. L., 1997, Phys. Rev.
B, 56, 6627.
[488] Jaínossy, A., Nemes, N., Feheír, T., Osz laínyi, G., Baumgartner , G., and Forroí, L.,
1997, Phys. Rev. Lett., 79, 2718.
[489] Diederichs, J., Gangopadhyay, A. K., and Schilling , J. S., 1996, Phys. Rev. B, 54,
R9662.
[490] Quirion, G., Bourbonnais, C., Barthel , E., Auban, P., Jeírome, D., Lambert, J. M.,
Zahab, A., Bernier , P., Fabre, C., and Rassat , A., 1993, Europhys. Lett., 21, 233.
[491] Quirion, G., Bourbonnais, C., Kerkoud , R., Barthel , E., Auban, P., Jeírome, D.,
Lambert, J. M., Zahab, A., Bernier , P., Fabre, C., and Rassat , A., 1993,
Electronic Properties of Fullerenes, edited by H. Kuzmany, J. Fink, M. Mehring and
S. Roth (Berlin: Springer), pp. 334± 338.
[492] Quirion, G., Bourbonnais, C., Barthel, E., Auban, P., Wzietek , P., Jeírome, D.,
Lambert, J. M., Zahab, A., Bernier , P., Fabre, C., and Rassat, A., 1993, Synth.
Metals, 55± 57, 3154.
[493] Auban-Senz ier , P., Kerkoud , R., Jeírome, D., Rachdi, F., and Bernier , P., 1995,
Science and Technology of Fullerene Materials, edited by P. Bernier, D. S. Bethune, L. Y.
Chiang, T. W. Ebbesen, R. M. Metzger and J. W. Mintmire (Pittsburgh, Pennsylvania:
Materials Research Society), pp. 261± 271.
[494] Kerkoud , R., Auban-Senz ier , P., Jeírome, D., Lambert, J. M., Zahab, A., and
Bernier , P., 1994, Europhys. Lett., 25, 379.
[495] Fleming , R. M., Ramirez , A. P., Rosseinsky, M. J., Murphy, D. W., Haddon,
R. C., Zahurak , S. M., and Makhija , A. V., 1991, Nature, 352, 787.
[496] Zhou , O., Palstra , T. T. M., Iwasa , Y., Fleming , R. M., Hebard , A. F., Sulewski,
P. E., Murphy, D. W., and Zegarski, B. R., 1995, Phys. Rev. B, 52, 483.
[497] Forroí, L., Baumgartner , G., Sienkiewicz , A., Pekker , S., Chauvet, O., Beuneu ,
F., and Alloul , H., 1996, Fullerenes and Fullerene Nanostructures, edited by H.
Kuzmany, J. Fink, M. Mehring and S. Roth (Singapore: World Scienti® c), pp. 102± 109.
[498] Erwin, S. C., Krishna , G. V., and Mele, E. J., 1995, Phys. Rev. B, 51, 7345.
[499] Rachdi, F., Goz e, C., Kirova , N., Luckyanchuk , I., Kerkoud , R., Auban-
Senz ier , P., Jeírome, D., Brazovskii, S., Zimmer , G., and Mehring , M., 1995,
Fullerenes: Recent Advances in the Chemistry and Physics of Fullerenes and Related
Materials, Vol. 2, edited by R. S. Ruo€ and K. M. Kadish (Pennington, New Jersey:
Electrochemical Society), pp. 1059± 1070.
[500] Kerkoud , R., Auban-Senz ier , P., Jeírome, D., Braz ovskii, S., Kirova , N.,
Luk’ yanchuk , I., Rachdi, F., and Goz e, C., 1996, Synth. Metals, 77, 205.
Fullerenes under high pressures 133

[501] Kerkoud , R., Auban-Senz ier , P., Jeírome, D., Braz ovskii, S., Luk’ yanchuk , I.,
Kirova , N., Rachdi, F., and Goz e, C., 1996, J. Phys. Chem. Solids., 57, 143.
[502] Kluthe, S., Roos, J., Mali, M., Brinkmann, D., Lang , H. P., Thommen-Geiser , V.,
and Guï ntherodt, H.-J., 1994, Progress in Fullerene Research, edited by H.
Kuzmany, J. Fink, M. Mehring and S. Roth (Singapore: World Scienti® c), pp. 466± 469.
[503] Osz laínyi, G., Baumgartner , G., Faigel , G., and Forroí, L., 1997, Phys. Rev. Lett.,
78, 4438.
[504] Schirber , J. E., Overmyer , D. L., Wang , H. H., Williams, J. M., Carlson, K. D.,
Kini, A. M., Welp, U., and Kwok , W.-K., 1991, Physica C, 178, 137.
[505] Sparn, G., Thompson, J. D., Huang , S.-M., Kaner , R. B., Diederich, F., Whetten,
R. L., Gruï ner, G., and Holcz er , K., 1991, Science, 252, 1829.
[506] Schirber , J. E., Bayless, W. R., Kortan, A. R., Rosseinsky, M. J., Oz das, E.,
Zhou , O., Fleming , R. M., and Murphy, D., 1994, Fullerenes: Recent Advances in
the Chemistry and Physics of Fullerenes and Related Materials, edited by K. M. Kadish
and R. S. Ruo€ (Pennington, New Jersey: Electrochemical Society), pp. 556± 559.
[507] Sparn, G., Thompson, J. D., Whetten, R. L., Huang , S.-M., Kaner , R. B.,
Diederich, F., Gruï ner, G., and Holcz er , K., 1992, Phys. Rev. Lett., 68, 1228.
[508] Movshovich , R., Thompson, J. D., Chen, C.-C., and Lieber , C. M., 1994, Phys. Rev.
B, 49, 3619.
[509] Miz uki, J., Takai, M., Takahashi, H., Moîri, N., Tanigaki, K., Hirosawa , I., and
Prassides, K., 1994, Phys. Rev. B, 50, 3466.
[510] Schirber , J. E., Bayless, W. R., Rosseinsky, M. J., Zhou , O., Fleming , R. M.,
Murphy, D., and Fischer , J. E., 1995, Science and Technology of Fullerene
Materials, edited by P. Bernier, D. S. Bethune, L. Y. Chiang, T. W. Ebbesen, R. M.
Metzger and J. W. Mintmire (Pittsburgh, Pennsylvania: Material Research Society),
pp. 289± 294.
[511] Palstra , T. T. M., Zhou , O., Iwasa , Y., Sulewski, P. E., Fleming , R. M., and
Zegarski, B. R., 1995, Solid. St. Commun., 93, 327.
[512] Palstra , T. T. M., Zhou , O., Iwasa , Y., Suleski, P. E., Fleming , R. M., and
Zegarski, B. R., 1995, Science and Technology of Fullerene Materials, edited by P.
Bernier, D. S. Bethune, L. Y. Chiang, T. W. Ebbesen, R. M. Metzger and J. W.
Mintmire (Pittsburgh, Pennsylvania: Material Research Society), pp. 285± 288.
[513] Schirber , J. E., Bayless, W. R., Kortan, A. R., and Kopylov , N., 1993, Physica C,
213, 190.
[514] Yildirim, T., Fischer , J. E., Dinnebier , R., Stephens, P. W., and Lin, C. L., 1995,
Solid St. Commun., 93, 269.
[515] Yildirim, T., Fischer , J. E., Stephens, P. W., and Mc Ghie, A. R., 1994, Progress in
Fullerene Research, edited by H. Kuzmany, J. Fink, M. Mehring and S. Roth
(Singapore: World Scienti® c), pp. 235± 238.
[516] Prassides, K., Tanigaki, K., Miz uki, J., and Hirosawa , I., 1994, Progress in
Fullerene Research, edited by H. Kuzmany, J. Fink, M. Mehring and S. Roth
(Singapore: World Scienti® c), pp. 495± 500.
[517] Iwasa , Y., Shimoda, H., Miyamoto, Y., Hayashi, H., Furudate, T., Mitani, T.,
and Maniwa, Y., 1996, Fullerenes: Recent Advances in the Chemistry and Physics of
Fullerenes and Related Materials, Vol. 3, edited by K. M. Kadish and R. S. Ruo€
(Pennington, New Jersey: Electrochemical Society), pp. 1054± 1062.
[518] Shimoda , H., Iwasa , Y., Miyamoto, Y., Maniwa , Y., and Mitani, T., 1996, Phys.
Rev. B, 54, R15653.
[519] Shimoda , H., Iwasa , Y., and Mitani, T., 1997, Synth. Metals, 85, 1593.
[520] Okada, S., Shimiz u, K., Kobayashi, T. C., Amaya , K., and Endo, S., 1996, J. phys.
Soc. Japan., 65, 1924.
[521] Yildirim, T., Barbedette, L., Fischer , J. E., Lin, C. L., Robert, J., Petit, P., and
Palstra , T. T. M., 1996, Phys. Rev. Lett., 77, 167.
[522] Crespi, V. H., and Cohen, M. L., 1996, Phys. Rev. B, 53, 56.
[523] Alexandrov , A. S., and Kabanov , V. V., 1996, Phys. Rev. B, 54, 3655.
[524] Chakravarty, S., Gelfand , M. P., and Kivelson, S., 1991, Science, 254, 970.
[525] Jansen, L., Chandran, L., and Block R., 1993, Chem. Phys., 176, 1.
134 Fullerenes under high pressures

[526] Zhu, Q., 1995, Fullerenes: Recent Advances in the Chemistry and Physics of Fullerenes
and Related Materials, Vol. 2, edited by R. S. Ruo€ and K. M. Kadish (Pennington,
New Jersey: Electrochemical Society), pp. 798± 806.
[527] Prassides, K., Vavekis, K., Kordatos, K., Tanigaki, K., Bendele, G. M., and
Stephens, P. W., 1997, J. Am. chem. Soc., 119, 834.
[528] Moro, L., 1995, Symposium on Fullerenes: Chemistry, Physics, and New Directions
VII, Reno, Nevada, unpublished.

Anda mungkin juga menyukai