Structure and properties of the Fullerene Dimer C140 produced by pressure treatment of C70

Article abstract of DOI:10.1021/jp994330l

A [2+2] cycloaddition cap-to-cap C₇₀ dimer with C_2h, molecular symmetry was synthesized in high yield by pressure treatment of polycrystalline C₇₀ at 1 GPa and 200 °C. It was separated from unreacted monomers by chromatography and characterized by ¹³C NMR, Raman, and infrared spectroscopy, and other methods. Remarkably, only one isomer was produced out of the five possible [2+2] cycloaddition products which have equally low formation energies according to semiempirical modeling calculations. The dimer obtained is the one favored when C₇₀ molecules adopt an ordered packing with parallel D₅ axes. The intercage bonding in C₁₄₀, its thermal stability, and intercage vibrational modes are similar to those found for the C₆₀ dimer, C₁₂₀. Both dimers photodissociate to the monomers in solution, probably via excited triplet states. The UV absorption and fluorescence properties of C₁₄₀ are not very different from those of C₇₀, suggesting only weak electronic interactions between the two cages of C₁₄₀. In comparison, the pressure-induced dimerization of C₆₀, under the conditions used for C₇₀, results mainly in C₆₀ oligomers and polymeric chains, but the dimer C₁₂₀ could be isolated at low yield when short reaction times (≤5 min) were used.

Full text of DOI:10.1021/jp994330l

J. Phys. Chem. B 2000, 104, 4101-4110
4101
Structure and Properties of the Fullerene Dimer C140 Produced by Pressure Treatment of
C70
Sergei Lebedkin,*,† William E. Hull,‡ Alexander Soldatov,§,| Burkhard Renker,§ and
Manfred M. Kappes†,
Forschungszentrum Karlsruhe, Institut fu¨r Nanotechnologie, Postfach 3640, D-76021 Karlsruhe, Germany,
Central Spectroscopy Department, German Cancer Research Center, D-69120 Heidelberg, Germany,
Forschungszentrum Karlsruhe, Institut fu¨r Nukleare Festko¨rperphysik, D-76021 Karlsruhe, Germany,
Department of Experimental Physics, Umea˚ UniVersity, S-901 87 Umea˚, Sweden, and Institut fu¨r
Physikalische Chemie II, UniVersita¨t Karlsruhe, D-76128 Karlsruhe, Germany
ReceiVed: December 9, 1999
A [2+2] cycloaddition cap-to-cap C70 dimer with C2h molecular symmetry was synthesized in high yield by
pressure treatment of polycrystalline C70 at 1 GPa and 200 °C. It was separated from unreacted monomers by
chromatography and characterized by 13C NMR, Raman, and infrared spectroscopy, and other methods.
Remarkably, only one isomer was produced out of the five possible [2+2] cycloaddition products which
have equally low formation energies according to semiempirical modeling calculations. The dimer obtained
is the one favored when C70 molecules adopt an ordered packing with parallel D5 axes. The intercage bonding
in C140, its thermal stability, and intercage vibrational modes are similar to those found for the C60 dimer,
C
120. Both dimers photodissociate to the monomers in solution, probably via excited triplet states. The UV
absorption and fluorescence properties of C140 are not very different from those of C70, suggesting only weak
electronic interactions between the two cages of C140. In comparison, the pressure-induced dimerization of
C60, under the conditions used for C70, results mainly in C60 oligomers and polymeric chains, but the dimer
C120 could be isolated at low yield when short reaction times (e5 min) were used.
1. Introduction
al.13 observed insolubility of C70 films after long irradiation with
visible/UV light, but no changes in their vibrational spectra.
In recent years a large amount of data1 have been collected
on polymeric C60 phases obtained by applying high pressure
and temperature2-4 or by irradiation with UV-visible light.5,6
The polymerization reaction can be described formally as a
[2+2] cycloaddition across parallel double bonds (hexagon-to-
hexagon junctions) in neighboring C60 cages. At a relatively
moderate pressure of about 1 GPa and a temperature of 300
°C, linear chains of C60 molecules are formed,7,8 whereas the
application of higher pressure and temperature produces two-
dimensional networks of bonded molecules.9 The polymeric
structures are metastable at ambient pressure, reverting at
elevated temperature (∼200 °C) to monomeric C60. The dimer
C120 has been proposed as an initial intermediate in the
polymerization process.10 This “shortest-chain polymer” was
recently synthesized in a mechanochemical reaction11 and by
applying pressure to C60-[bis(ethylenedithio)tetrathiafulvalene]2
crystals.12 According to vibrational spectroscopy, C120 is also
formed in significant yields by light irradiation of pure C60 at
100 °C,6 and by a fast pressure treatment.10
The first attempt to polymerize C70 under pressure (at pressures
of 5 and 7.5 GPa and temperatures up to 1100 K) uncovered
no Raman signatures of polymerization.14 However, later, Iwasa
et al.15 and Premila et al.16 reported IR and X-ray diffraction
spectra15 showing strong structural changes in solid C70 after
treatment at 5 GPa and 200-300 °C and at 1-7.5 GPa and
750 °C, respectively. Similar to C60 polymers, these materials
reverted to C70 by heating at ambient pressure. However, the
nature of the putative polymeric structures generated in these
studies remained unclear. After treating C70 at very high
pressures and temperatures (up to 12.5 GPa and 1500 °C), Blank
et al.17 found new super-hard phases and tentatively assigned
them to three-dimensionally polymerized C70 and to a disordered
carbon state. Considering the treatment conditions used in these
previous studies, one might conclude that polymerization of C70,
in comparison with that of C60, is very inefficient at moderate
pressures and temperatures. However, Soldatov et al. recently
showed that C70 readily polymerizes similarly to C60 at pressures
and temperatures as low as 1 GPa and 200-300 °C.18
A
In principle, polymerization by cycloaddition across reactive
double bonds is also possible for C70 and other fullerenes. Up
to now, only a few studies concerning the polymerization of
C70 have appeared, in part with controversial results. Rao et
comparison of the corresponding Raman spectra with those
obtained from C120 and C60 polymers indicated that two
predominant structures are formed, depending on the applied
pressure: C70 dimers at pressure around 1 GPa and C70 chains
at 2 GPa.18 The latter hypothesis has very recently been
supported by X-ray crystal analysis.19
* To whom correspondence should be addressed. Fax, +49-7247-826368;
e-mail, lebedkin@int.fzk.de.
† Institut fu¨r Nanotechnologie.
It has been shown that C120 and other dimeric fullerenes, e.g.,
C120O, C120O2, and C130O, are soluble, in particular in o-
dichlorobenzene, and can be separated by chromatography.11,20-22
Our hypothesis was that the same approach could be applied to
‡ German Cancer Research Center.
§ Institut fu¨r Nukleare Festko¨rperphysik.
| Umea˚ University.
Universita¨t Karlsruhe.
10.1021/jp994330l CCC: $19.00 © 2000 American Chemical Society
Published on Web 04/06/2000

4102 J. Phys. Chem. B, Vol. 104, No. 17, 2000
Lebedkin et al.
was heated to the desired temperature. Then, the pressure was
increased to ca. 1 GPa within seconds; after about 1 min, the
cell was quickly cooled (initial cooling rate ca. 20 °C/min) and
the pressure released. The product comprised 15-20% unreacted
C60, ca. 3 wt % of extracted C120, and an insoluble residue of
higher C60 oligomers and polymeric chains (see Discussion).
Longer treatment resulted in a decrease in the yield of C120 and
an increase in the insoluble portion. The dimerization/polym-
erization of C60 under the above conditions appears to be
strongly promoted by interstitial gases in the starting material.
The pressure synthesis of C120 was successful with a C60 powder
sample that had been stored for a long period of time under air
or nitrogen. With a powder that had been freshly annealed at
250 °C in 10-6 mbar vacuum, practically no dimerization/
polymerization of C60 was observed. The FT-Raman, IR, and
UV-visible absorption spectra obtained from C120 extracted
from the pressure-polymerized C60 were equivalent to those
obtained from C120 obtained by a mechanochemical reaction.11
The purity of the dimeric derivatives was estimated by HPLC
and Raman spectroscopy to be >95%.
Figure 1. HPLC chromatogram of an o-dichlorobenzene solution of
polycrystalline C70 treated at ∼1 GPa and 200 °C (Cosmosil Buckyprep
column, elution with toluene, optical detection at 330 nm).
C140 as well. As described in this work, we have indeed
succeeded in dissolving and separating C140 from polycrystalline
C70 following treatment at ca. 1 GPa and ca. 200 °C. Thus, we
provide unambiguous proof for the formation of C140 at high
pressure, and, perhaps more remarkable, we demonstrate that
only a single isomer is obtained.
2.3. Spectroscopy. For 13C NMR, a saturated solution
(ODCB) of C140 (ca. 0.5 mg) enriched to ca. 7% in 13C was
prepared in a 5-mm NMR tube. Nondeuterated ODCB (0.2 mL)
contained ca. 16 mM Cr(acac)3 for paramagnetic relaxation
enhancement. The spectrum was acquired at 125.76 MHz and
30 °C with a Bruker AM-500 FT-NMR spectrometer using a
J-modulated spin-echo method for suppression of solvent CH
carbons, as described in detail previously.23 A 2-mm capillary
with acetone-d6 provided a deuterium lock signal. Data acquisi-
tion parameters were as follows: spectral width ) 25 000 Hz;
90° pulse ) 7.7 µs; half echo time delay ) 3.05 ms; relaxation
delay ) 1.9 s, acquisition time ) 0.655 s for 32 K data points;
repetition time ) 2.56 s. Successive blocks of 20 000 transients
(14.2 h) were stored separately. Identical conditions were used
to acquire a control spectrum of the ODCB + Cr(acac)3 solvent.
Data processing was performed using Bruker’s WIN NMR
software. A total of eight data blocks (114 h) were added for
the C140 spectrum, and six blocks (85 h) were added for the
solvent-control spectrum. As shown in Figure 2, the sp2 carbon
region was analyzed using a Lorentz-Gauss line shape trans-
formation for resolution enhancement (Bruker parameters: LB
) -2.5 Hz, GB ) 20%) and zero-filling to 64K, while the sp3
region was examined with sensitivity enhancement (LB ) 2.5
Hz). Chemical shifts were calibrated relative to the C1 carbon
of ODCB at 132.53 ppm, which gives a C60 shift of 142.83
ppm. The C140 and solvent spectra were baseline-corrected and
compared via a dual display routine, and the difference spectrum
was computed. In this way all signals from solvent, impurities,
and intermodulation distortion artifacts could be clearly identi-
fied. With the acquisition conditions used, the sp2 signals gave
relatively uniform intensities (integrals) per carbon atom,
whereas the sp3 carbon signals were significantly reduced,
presumably because of a longer T1 at these sites.
2. Experimental Details
2.1. Pressure Synthesis of C140. A simple piston-cylinder
pressure cell (6 mm diameter) made of hardened steel was used
for the synthesis. About 50 mg of polycrystalline C70 powder
(chromatographically purified in our lab to a level of >99%)
was loaded into the cell and compacted by applying 0.1-0.2
GPa pressure with a hand press. Loading and compaction were
carried out in an argon glovebox to minimize the possible
oxidation of C70. After loading, the cell was transferred to a
laboratory hydraulic press in which 1 GPa pressure was applied
while the temperature of the cell was increased to 200 °C and
held for 30 min. Subsequently, the cell was cooled to room
temperature within about 10 min, and the pressure was then
released.
The resulting material was dissolved in o-dichlorobenzene
(ODCB) with the aid of an ultrasonic bath to a concentration
of about 2 mg/mL, with less than 5 wt % remaining as an
insoluble residue. According to high-performance liquid chro-
matography (HPLC, Cosmosil Buckyprep column 20 × 250
mm, elution with toluene at 18 mL/min), the solution contained
predominantly unreacted C70 and C140 (Figure 1). A small
shoulder in front of the C140 peak probably was due to a minor
C140 isomer(s) with a total contribution of about 5%, as estimated
from the peak areas. No other peaks were observed up to a
retention time >120 min, and no peaks were detected for the
oxides C70Ox, x ) 1-3, which would follow the C70 peak. C140
was separated from C70 with a yield of 20-30 wt % with respect
to the starting material by two-stage flash chromatography on
alumina (activity I) using chloroform and ODCB to elute C70
and C140, respectively. Powder samples of C140 (purity >95%;
the main impurity was C70) were prepared by evaporating the
solvent and drying the residue under a vacuum.
2.2. Preparation of Other Dimeric Fullerenes. Pressure
Synthesis of C120. The C60 dimer, C120, and the dimeric oxides
IR and Raman spectra were obtained with a resolution of 2
cm-1 from powder samples (pressed in pellets with 3-mm
diameter for the far-IR or embedded in KBr pellets for the mid-
IR measurements) on a Bruker Equinox 55S spectrometer
equipped with a FRA106 FT-Raman module and an Nd:YAG
laser for excitation at 1064 nm. No decomposition of C140 due
to the intense irradiation (∼10 W/cm2) at 1064 nm was detected.
The presented Raman spectra are corrected for luminescence
background, but not for the wavelength-dependent spectrometer
response.
C120O and C120O2 were used for comparative studies. They were
prepared as described in refs 11, 20, and 21, respectively. C120
was also prepared for the first time using the conditions applied
to C70, i.e., C60 powder was subjected to pressure and temper-
ature of ca. 1 GPa and 200-220 °C for just a few minutes,
followed by ultrasonic extraction with ODCB and chromato-
graphic separation on alumina. A short treatment time appears
to be crucial. The largest yields were obtained in the following
manner. The pressure was initially set to 0.2 GPa, and the cell

Structure and Properties of C140
J. Phys. Chem. B, Vol. 104, No. 17, 2000 4103
irradiation laser intensity was measured using a Coherent 200
power meter. The values of æPD at other wavelengths were
estimated relative to æPD at 514 nm by irradiating the solutions
in the Spex Fluorolog-3 spectrometer. To determine the effect
of dissolved oxygen, the solutions were irradiated either in air
or under anaerobic conditions after careful degassing in a
vacuum. The estimated error in the determination of æPD is
(30%. All measurements, if not stated otherwise, were per-
formed at room temperature.
3. Results
Our pressure treatment of polycrystalline C70 at ∼1 GPa and
200-220 °C resulted in ∼30-50% conversion of C70 into
dimers, according to the Raman, IR, and HPLC analysis. The
rest of the material consisted mainly of unreacted C70. A small
portion (<5 wt %), insoluble in ODCB, is tentatively attributed
to higher C70 oligomers. The reaction yield did not depend
strongly on the treatment time, which was varied from several
minutes to several hours. However, the yield of C140 decreased
dramatically when the applied pressure and temperature were
lower than 0.8 GPa and 180 °C, respectively. The 30-50%
conversion of C70 achieved in this study is less than the ca.
80% conversion estimated in a previous study,18 where similar
pressure and temperature were applied. It is conceivable that
the lower efficiency was due to a relatively inhomogeneous
distribution of the pressure in our simple pressure cell. On the
other hand, residual C70 molecules may help the dissolution of
the material, acting as “spacers” between the less soluble dimer
molecules.
Figure 2. 125.76 MHz 13C NMR spectrum of a saturated solution of
C
140 (enriched to ca. 7% in 13C) in ODCB + Cr(acac)3. The sp2 region
(signals 1-35) is shown with resolution enhancement, the sp3 region
(signals 36,37) with sensitivity enhancement. The labels p1-p5 mark
the positions expected for the five signals of D5h C70 with respective
carbon counts of 10, 10, 20, 20, 10. The solvent ODCB gives three
large offscale resonances with corresponding 13C satellites, whereby
the residual signals from solvent CH groups are out of phase because
of the suppression technique used (see Experimental Details). The
signals marked “a” (artifacts) were detected in a control spectrum
obtained from the solvent alone.
3.1. Chromatography. The HPLC shows, essentially, a single
product peak at a retention time of 43 min and a peak of
unreacted C70 at 14 min (Figure 1). Under similar conditions,
the retention times for C60 and C120 are 8 and 15 min, and for
the dimeric oxides C120O and C120O2, ca. 15 and 16.5 min,
respectively. However, the elution of the heterodimeric oxide
C
130O containing both C60 and C70 cages is retarded to ∼30
min.22 This trend suggests that the peak at 43 min is a C70 dimer
and not a trimer.
Matrix-assisted laser desorption/ionization (MALDI) and
electrospray ionization (ESI) time-of-flight (TOF) mass spec-
3.2. Mass Spectrometry. Only an ion peak of C70 at m/z
840 could be observed in the negative and positive MALDI-
trometry were performed on home-built instruments.24,25
A
TOF and in the negative ESI-TOF mass spectra of C140
.
pulsed N2 laser for desorption/ionization and 9-nitroantracene
as a matrix were used for MALDI. For ESI, a solution of C140
in ODCB was sprayed with nitrogen as a nebulizing and drying
gas in an Analytica ESI source.
Evidently, there is complete dissociation of the dimer molecules/
ions under the MS conditions. The same problem was encoun-
tered for C120.
11 On the other hand, the dimeric fullerene oxides
C120O, C120O2, and C130O, with stronger cage-to-cage furanoid
links readily yield dominant parent ion peaks in the negative
MALDI-TOF mass spectra.20-22
UV-vis absorption spectra were measured on a Hitachi
U-2000 spectrometer. Fluorescence emission and excitation
spectra were recorded on a Spex Fluorolog-3 spectrometer (450
W xenon lamp, R928 photomultiplier tube) and were corrected
for the wavelength-dependent instrumental response. The pho-
tomultiplier tube operated up to ca. 850 nm, and the emission
correction was found to be reliable up to ca. 800 nm. The
fluorescence of C70 extends into the near-infrared up to ca. 1000
nm,26 and the fluorescence of C140 apparently does the same
(Figure 7); therefore, the presented emission spectra can serve
for qualitative comparisons between C140 and C70.
2.4. Photodissociation Quantum Yield. We found that in
ambient room light, C140 and C120 in dilute solutions undergo
photodissociation to C70 and C60, respectively. To estimate the
quantum yield of photodissociation, æPD, i.e., the ratio of
dissociated molecules to photons absorbed, dilute solutions in
ODCB were irradiated at 514 nm with an Ar-ion laser and
subsequently analyzed by HPLC calibrated for C70 and C60 (see
above; elution was with xylene instead of toluene in the case
of C140/C70 to shorten the retention times). The absolute
3.3. 13C NMR Measurements. C140 is only poorly soluble
in ODCB, but is even less so in CS2 or other solvents. Therefore,
a saturated solution of C140 (enriched to ca. 7% in 13C) in ODCB
was used for the NMR measurements. The 13C NMR spectrum
of C140 is shown in Figure 2; the 37 assigned resonances are
numbered and listed in Table 1. Only one (dominant) C140
isomer was found. Based on the S/N in the spectrum, if another
isomer were present, its mole fraction would have to be less
than ∼(2 × 10-3) n (where n is the number of symmetry-related
carbon sites). The only impurity detected could be assigned to
C70, which has five nonequivalent carbon sites, denoted as 1-5
in Figure 3. These sites lie on planes perpendicular to the long
axis of the “rugby ball” cage and are expected to give five 13
C
NMR signals in the ratio 1:1:2:2:1 at the positions labeled p1-
p5 in Figure 2.27 Resolved resonances for p1-p4 were detected
(Table 1), although the signal for p5 could not be observed
because of interference from the residual C3,6 signal from

4104 J. Phys. Chem. B, Vol. 104, No. 17, 2000
Lebedkin et al.
TABLE 1: 13C NMR Data for C140 in o-Dichlorobenzene at
30 °Ca
peak nr
ppm
integral
peak nr
ppm
integral
1
2
3
4
5
6
7
8
9
p1
10
11
12
13
14
15
p3
16
17
p2
18
155.733
152.224
151.519
151.267
151.201
150.865
150.732
150.494
150.326
150.291
150.218
149.718
149.313
148.978
148.637
148.518
147.750
147.468
147.263
147.047
146.973
4.06
4.01
3.85
2.20
5.05
3.83
3.90
5.04
3.30
1.52
4.27
4.23
4.53
4.93
3.63
4.24
3.07
4.27
4.19
2.18
4.33
19,20
21
22
p4
23
24
25
26
27
28
29
30
31
32
146.821
145.882
145.810
145.001
143.644
143.434
143.288
142.914
142.809
141.868
140.523
140.280
134.281
133.355
132.530
131.766
131.105
130.978
(130.5)
71.177
5.90
3.63
4.44
2.45
4.17
4.27
4.62
3.93
3.88
4.00
4.02
3.95
3.58
6.11b
Figure 4. PM3-optimized geometries for the five energetically
favorable C70 dimers. The nomenclature is defined in the text and in
Figure 3. Dimer I [(1,2)/(1,2) C2h C140] is produced by the pressure
treatment of solid C70 at ∼1 GPa and ∼200 °C.
ODCB(1,2)
33
34
35
p5
36
37
3.50
3.90
3.86
nd
0.93c
1.51c
67.826
a Chemical shifts were calibrated relative to ODCB(1,2) ) 132.53,
which corresponds to TMS ) 0 and C60 ) 142.83 ppm. Peak integrals
are relative to peak nr 28, defined as 4.00. For sp2 carbons, 33 signals
represent 4 carbons each, while signals nr 4 and 19,20 (in boldface)
represent 2 and 4+2 carbons, respectively. The signals labeled p1-p5
(italic) are from C70; the expected chemical shifts27 are 150.33, 147.03,
147.73, 145.03, and 130.53 (in benzene-d6, corrected to C60 ) 142.83
ppm ). b This integral contains a contribution from a solvent artifact
signal. c The observed integrals are less than expected (2.0) due to
partial saturation of sp3 carbons under our measurement conditions.
Figure 5. FT-IR absorption spectra of C70 (top) and its dimer C140
(bottom). The scaling of the absorption in the far-IR region to that in
the mid-IR region is only approximate.
sp2 signals with intensity 2C, and two sp3 signals with intensity
2C. The closely related isomer II, denoted (1,2)/(1,2) C2V C140
(Figure 4), has the same pattern of distinguishable carbon sites.
Thus, the observed NMR spectrum is consistent with either of
these isomers. Note that the equatorial carbons p5 are unique
in C70 and exhibit a large upfield shift compared to the other
sites. Thus, for the dimer one expects 30 signals distributed
around the positions p1-p4 and five signals with intensity 4C
near the position for p5. In Figure 2, the signals 31-35 have
been assigned to the five p5 sites after detailed comparison with
the ODCB control spectrum in which the impurity/artifact
signals at the positions labeled “a” could be identified. The two
sp2 signals numbered “4” and “20” with intensity 2C are
expected to represent carbons of types p1 and p2, which are
remote from the bridge and little perturbed by it; their chemical
shifts are in fact close to the values for p1 and p2 in C70.
Figure 3. Structure of the D5h C70 cage showing the labeling scheme
used to define sites in C70 dimers. There are five basic nonequivalent
carbon types in C70 representing “lines of latitude” from “pole” (or
cap) to “equator” in the “northern hemisphere” (1-5) and the
corresponding sites in the “southern hemisphere” (1′-4′). The number-
ing follows a zigzag course about a reference “line of longitude”. The
letters a, b, c, etc., are used to distinguish sites at a given latitude moving
from west to east. We define dimer configurations in terms of bridges
between northern hemispheres so that numbers with primes denote sites
in the opposite hemisphere remote from the bridging site.
In (3,3a)/(3,3a) C2h C140 and (3,3a)/(3a,3) C2V C140 (structures
III and IV, respectively, in Figure 4), the symmetry plane is
perpendicular to and bisects the bridge plane. The four bridging
carbons are equivalent and would give one sp3 signal of intensity
4C. Two sp2 carbons of types p1 and p2 (proximal to the bridge)
and two of types p1′ and p2′ (distal to the bridge) are on the
symmetry plane and would give four signals of intensity 2C,
whereas the remaining 32 sp2 signals would have intensity 4C.
This is clearly not consistent with the observed NMR spectrum.
Finally, the (1,2)/(3,3a) C1 C140 isomer (structure V) can be ruled
out because it has no equivalent carbons and would give four
sp3 signals and 136 sp2 signals.
ODCB. Based on the signal integrals, the sample contains C70
and C140 in a molar ratio of ca. 0.15:1.0, i.e., about 7% of the
C70 cages were present as monomers.
The five C140 isomers of lowest energy (see Discussion) are
shown in Figure 4. The nomenclature used here is (a,b)/(c,d)
C140 to denote a dimer having a-b and c-d intracage bonds
and a-c and b-d intercage bridging bonds, where a, b, c, and
d are the corresponding carbon sites on the C70 cages. In each
cage of (1,2)/(2,1) C2h C140 (structure I in Figure 4), there are
four different carbons on the σh symmetry plane: the (1,2) sp3
carbons in the bridge and two sp2 carbons from the correspond-
ing 1′, 2′ sites remote from the bridge. Altogether, there are 66
sp2 carbons that are not on the σh symmetry plane, and these
form 33 pairs of distinguishable sp2 sites in each cage. Therefore,
for isomer I we expect 33 sp2 signals with intensity 4C, two
3.4. Vibrational Spectroscopy. Figures 5 and 6 show FT-
IR and FT-Raman spectra of C140 and C70. A typical FT-Raman
pattern of a C70 powder sample after the treatment at ca. 1 GPa
and 200 °C is presented in Figure 6. This pattern and its IR
counterpart (not shown) are essentially superpositions of the
corresponding spectra of C70 and C140. As usual for fullerene
derivatives with weakly disturbed fullerene cages,28a C140
demonstrates peaks in the same regions as C70, but with
splittings and shifts due to the reduced molecular symmetry and
weak vibrational interactions between the covalently bonded
C70 cages. For instance, this results in the appearance of

Structure and Properties of C140
J. Phys. Chem. B, Vol. 104, No. 17, 2000 4105
Figure 6. FT-Raman spectra, obtained with Nd:YAG laser excitation
at 1064 nm, from a sample of C70 before (top) and after (middle)
treatment at ca. 1 GPa and 200 °C and from the purified dimer C140
(bottom).
Figure 8. FT-Raman spectra obtained from C60 (top), its dimer C120
(bottom), and samples of C60 after treatment for a few minutes (“short”)
and for 1 h (“long”) at ca. 1 GPa and 220 °C.
118, and 129 for C140 vs 96, 127, and 139 cm-1 for C120) are
proportional to M-1/2, where M is the cage mass, as expected
in the harmonic oscillator approximation. This indicates that
both dimers have the same type of the intercage bonding.
As for C120,
32 we observed slight differences in the vibrational
spectra of different samples of C140, which may be attributed
to differences in local molecular orientations and intermolecular
interactions in those samples. For instance, we observed a shift
of the stretching-like intercage vibration, to 87 cm-1, in the
Raman spectrum of a well-crystallized C140 powder that
probably contained a substantial number of interstitial solvent
molecules.
Figure 7. Electronic absorption and fluorescence (FL) spectra of C70
(bottom) and its dimer C140 (top) in ODCB. The absorption spectra
were measured with 10 µM solutions. The FL emission spectra (450-
nm excitation) were obtained with solutions with optical density 0.2.
The FL excitation spectrum (dotted line, shifted for clarity) was obtained
3.5. Electronic Absorption and Fluorescence. The light
absorption spectrum of C140 resembles that of C70 (Figure 7),
but the bands are broadened and shifted, and the onset is at a
longer wavelength (ca. 690 nm) compared to C70 (ca. 640 nm).
This difference can be attributed to the reduced molecular
symmetry and a smaller HOMO-LUMO gap for the dimer.13
The molar absorptivity of C140 is larger than that of C70 by
approximately a factor of 2 (Figure 7). Both the monomer33
and the dimer exhibit weak fluorescence (FL) which is
somewhat red-shifted and structureless for the dimer (Figure
7). The FL yield of C70 is 5.9 × 10-4 in toluene, not very
different in other solvents and independent of the excitation
wavelength.33 Using a solution of C70 in toluene as a reference,
approximately the same FL yield was estimated for C70 in
ODCB. The FL yield of the dimer is about a factor of 2 higher
in ODCB (Figure 7). It is interesting to note that the FL yield
of C120 in different solvents is also larger than that of C60 by a
factor of 3.34 The FL spectrum of C140 is essentially independent
of the excitation wavelength (320-650 nm) and practically the
same in ODCB, 1-methylnaphthalene, and CS2. The above
similarities between C140 and C70 suggest that there is only
moderate perturbation of the ground and excited electronic states
in C70 upon dimerization.
with a 1 µM solution of C140
.
originally silent modes or an increase in weak modes in the
far-IR spectrum (Figure 5).
Compared to C70, new low-frequency Raman peaks appear
in the spectrum of C140 at 89, 118, and 129 cm-1. The peaks
are attributed to intercage vibrational modes where the heavy
fullerene cages move more or less as rigid units. Six such modes,
which originate from the coupled translational and rotational
motions of the two cages, are a common feature for the dimeric
fullerene molecules and have been studied in detail, both
experimentally and theoretically.22,29-32 The frequencies and
relative spectral intensities of these modes depend on the exact
nature of the bridge between two fullerene cages. The C60 dimer
C120 (D2h molecular symmetry) exhibits three Raman-active
intercage modes32 at 96 (the strongest peak due to the stretching-
like vibration), 127, and 139 cm-1 (Figure 8). Three further
intercage modes of C120 (two IR-active and one silent) are
expected below 50 cm-1. The corresponding three remaining
vibrations in C140 should also appear at very low frequencies,
and are therefore inaccessible in our measurements (see Discus-
sion). Note that the relative peak intensities of the intercage
vibrations in the Raman spectra of C140 and C120 are similar
(Figures 6 and 8), and that their respective frequencies ν (89,
3.6. Photodissociation. A characteristic process involving
excited electronic states was observed for C140, namely its
photodissociation (PD) to C70 upon irradiation in the range of
400-600 nm. The quantum yield of PD, æPD, is quite low: 5
× 10-5 for a 10 µM solution of C140 in ODCB. It does not

4106 J. Phys. Chem. B, Vol. 104, No. 17, 2000
Lebedkin et al.
TABLE 2: Semi-Empirical Modeling (PM3) of C70 Dimers
Formed by [2+2] Cycloaddition Containing a Cyclobutane
Intercage Bridge
depend significantly on the irradiation wavelength in the range
probed, indicating that PD occurs from a low-lying excited state.
Dissolved oxygen strongly reduces æPD, which increased to ca.
2-3 × 10-3 when the C140 solution was degassed. This result
suggests that C140 dissociates via the excited triplet state.
Assuming that the latter is quenched to the ground state in about
1 µs by dissolved oxygen (ca. 1 mM in air-saturated aromatic
solvents35) at the diffusion-limited rate constant of ca. 109
relative energyb
(kJ/mol)
bridge bond lengthsc
inter-/intracage (Å)
C140 isomera symmetry
(1,2)/(2,1)
(1,2)/(1,2)
C2h
C2V
C2h
C2V
C1
0
0.0
-0.5
-0.3
-0.2
1.547/1.601
1.547/1.601
1.555/1.578
1.554/1.579
(3,3a)/(3,3a)
(3,3a)/(3a,3)
(1,2)/(3,3a)
(1,2)/(3a,3)
(3,3a)/(4,5)
(3a,3)/(4,5)
(4,4i)/(4,4i)
(1,2)/(5,5a)
(1,2)/(5a,5)
(2,3)/(3,2)
M-1 -1
s , one can estimate from the increase in æPD for degassed
1.551;1.551/1.600;1.579
solutions that the intrinsic lifetime of a triplet state of C140 in
ODCB is about 50 µs. For comparison, the intrinsic lifetime of
a triplet state of C70 is g12 ms in toluene.36 Despite the low
value of æPD in air, prolonged photolysis or exposure of C140
solutions to daylight resulted in complete dissociation of the
dimer to C70 (confirmed by HPLC).
C1
73
1.553;1.563/1.576;1.610
C2h
Cs
74
93
1.561/1.636
1.555;1.554/1.592;1.692
C2
D2h
160
550
1.549/1.630
1.523/1.624
An interesting question is whether the other dimeric fullerenes
(5,5a)/(5,5a)
are light-sensitive. In disagreement with a previous report about
a The nomenclature is described in the text and in Figure 3; the first
five isomers are those depicted in Figure 4; isomers listed as pairs are
enantiomers. b Calculated from PM3 heats of formation; the relative
energy for two moles of C70 is 159 kJ. c For C120, the results are 1.549/
1.598 Å.
34
the photochemical stability of C120
,
we found that the C60
dimer also photodissociates to monomeric C60 with approxi-
mately the same quantum efficiency as C140 when irradiated at
400 nm. The æPD values of 6 × 10-5 and 7 × 10-4 were
estimated for, respectively, air-saturated and degassed C120
solutions in ODCB (ca. 5 µM). The decrease in æPD upon
admittance of air likely is due to the same triplet-quenching
mechanism proposed for C140. The estimated intrinsic lifetime
for the triplet state of C120 in ODCB is about 10 µs. This agrees
well with the triplet-state lifetime for C120 in toluene of g10
µs, as obtained from transient absorption measurements.34 For
comparison, a significantly longer lifetime of 133 µs has been
reported for the triplet state of C60 in toluene.37
TABLE 3: Bond Types and Lengths in D5h C70
length
approx
bond order
bond
(Å) (PM3)a
code
general type
1-1
1-2
2-3
3-3
3-4
4-4
4-5
5-5
1.457
1.386
1.453
1.374
1.463
1.426
1.412
1.463
hp-hh
hh-pp
hp-hh
hh-pp
hp-hh
hp-hh
hh-hp
hh-hh
single
double
single
double
single
aromatic
aromatic
single
1
2
1
2
1
1.5
1.5
1
In contrast to C140 and C120, we found no HPLC or
spectroscopic evidence for the photodecomposition of the
dimeric fullerene oxides C120O and C120O2 over the 400-600
nm range.
a The PM3 values of bond lengths in C70 presented here for
illustrative purposes are in good agreement with experimental and ab
initio theoretical data.28b
3.7. Thermal Stability. In general, fullerene polymers revert
back to the monomeric state at ambient pressure and elevated
temperature, and, as expected, C140 decomposes thermally to
C70. A detailed study of the depolymerization kinetics for
different polymeric C70 materials will be published else-
where.18,38 Here we only wish to emphasize that C140 and C120
show similar behavior. For example, a powder sample of C140
(in KBr pellet) decomposes by first-order kinetics with τ ) 65
( 5 min at 150 °C, as determined from intensity changes of
the characteristic IR bands of C140 and C70. At the same
temperature, C120 decomposes in the solid state to C60 by first-
order kinetics with τ ) 43 min, according to dilatometric
measurements.38
rings, hexagon or pentagon (h or p), and connects the vertexes
of two other rings. The two types of bonds can be defined in
terms of a junction-vertexes letter code, i.e., hp-hh and hh-
pp (or simply as hp and hh type) with bond lengths of 1.457
and 1.384 Å, respectively (PM3 calculations). The hh-type
double bonds are the reactive centers for C60 dimerization,
oxidation, and many other reactions.40 Thus, for C60 there is
only one possible [2+2] reaction across hh-type bonds, leading
to the D2h C120 dimer.
The D5h C70 molecule has a total of eight distinguishable C-C
bonds28b which can be classified as shown in Table 3. Only
two of these, 1-2 and 3-3, are “double” bonds analogous to
those in C60, and these sites in the cap region are almost
exclusively the ones involved in cycloadditions.40 In contrast,
the equatorial region contains “single” bonds (3-4 and 5-5)
and “aromatic” bonds of intermediate length (4-4 and 4-5);
this region is relatively inert to addition reactions, although
minor cycloadducts of benzyne across the 4-5 and 4-4 bonds
of C70 have been reported.41 Therefore, the number of chemically
accessible [2+2] C140 isomers is expected to be limited to those
involving 1-2 or 3-3 bonds, i.e., the five structures I-V shown
in Figure 4 and Table 2. In agreement with previous semiem-
pirical MNDO results,42 our semiempirical PM3 calculations
(Gaussian 98W PC software package43) gave essentially equal
heats of formation for these five isomers (Table 2). We expanded
the calculations to include several other [2+2] C140 isomers
involving addition across single or aromatic type bonds and
obtained substantially higher energies (Table 2). The belt-to-
belt (5,5)/(5,5) D2h isomer was found to be the most energetically
4. Discussion
4.1. Structure of C140. Our 13C NMR, Raman, HPLC, and
other results show that the main product of the pressure-
temperature treatment of solid C70 at 1 GPa and 200 °C is the
dimer C140. It is reasonable to expect that C140 is formed by a
[2+2] cycloaddition across double bonds to give a four-
membered ring as a cage-to-cage bridging unit, in complete
analogy with the reaction of C60 to form C120 and polymeric
C60.39 Indeed, remarkable similarities are observed in those
properties of C140 and C120 that are primarily determined by
the nature of the intercage bonding. In particular, the photo-
and thermal dissociation properties and intercage vibrational
modes in the Raman spectra are similar for the two dimers.
In Ih C60, there are only two types of bonds, generally
represented as “single” and “double” bonds in modeled
structures.28b Each bond serves as the junction between two

Structure and Properties of C140
J. Phys. Chem. B, Vol. 104, No. 17, 2000 4107
unfavorable dimer, with a PM3 energy of +550 kJ/mol relative
to I (+940 kJ/mol, according to tight-binding molecular
dynamics calculations13).
for dimerization. Such close and favorable intermolecular
contacts are not available for two (3,3) bonds or a (3,3) and
(1,2) combination. It was for this reason that Rao et al.13
proposed a cap-to-cap (1,2)/(2,1) C2h C140 isomer as the single
possible dimer in the room-temperature photopolymerization of
C70. This unique structure would also be consistent with our
results. However, above T01 ∼ 350 K, tumbling rotations of
the C70 molecules could bring all reactive bonds into reaction
contact, and the formation of all five energetically favorable
C140 isomers would be expected.
Remarkably, our 13C NMR and HPLC results indicate that
only one C140 isomer (g95%) is produced by the high-pressure
synthesis. The NMR spectrum is consistent only with the (1,2)/
(2,1) C2h or the (1,2)/(1,2) C2V isomer (structures I and II in
Figure 4). These two isomers should have the same basic pattern
of carbon chemical shifts, with only minor differences in the
actual shift values (based on PM3 atomic charges) so that it is
not possible to assign structure I or II on an a priori basis. Some
of the features of the 13C NMR spectrum can be explained on
the basis of the PM3 modeling results. The atomic (Mulliken)
charges on the bridging sp3 carbons were found to be q ) 0.0636
and 0.0616 for carbons 1 and 2 in isomer I, 0.0661 and 0.0591
for carbons 1 and 2 in isomer II, and 0.0636 for C120. The
corresponding chemical shift for C120 is 76.22 ppm, which is
close to the shifts measured for C140 (71.18 and 67.83 ppm).
The “center of gravity” for sp2 chemical shifts is shifted
downfield in both C140 and C120, consistent with the negative
average charge on sp2 carbons. For p5 type carbons in C70, q )
+0.0289, and the values for I and II are in the range 0.0232-
0.0315; this is consistent with the appearance of NMR signals
31-35 in the relatively narrow range of 131-135 ppm near
the position of p5 in C70. On the basis of atomic charge, stronger
perturbations occur in the dimer for carbons of types 3 and 4,
consistent with the broad range of chemical shifts observed from
140 to 152 ppm. In each cage of C140 two sp2 carbons of type
1 and two of type 3 are attached to the bridging sp3 carbons,
and their charges show the strongest perturbations compared
to C70 (q ) ca. -0.033 vs +0.0052 and -0.029 vs +0.0062,
respectively). For C120 the corresponding carbons have q )
-0.0366, resulting in one isolated resonance at low field (151.43
ppm). For C140 there is an analogous isolated resonance at
155.73 ppm (5.5 ppm downfield of the position p1) which can
be tentatively assigned to the type 1 sp2 carbons attached to the
bridge. Similarly, resonance 2 may correspond to the type 3
carbons attached to the bridge (downfield shift of 4.5 ppm
compared to p3). Finally, the PM3 calculations predict that the
1′ and 2′ carbons located on the symmetry plane of the dimer
have minor but opposite perturbations in charge (<0.001 units);
this is consistent with the small and opposite shifts in the two-
carbon resonances 4 and 20 relative to p1 and p2.
The formation of a single C140 isomer by high-pressure
treatment at ∼200 °C (470 K) suggests that the orientational
ordering favorable for (1,2)/(2,1) bonding is preserved at 1 GPa
pressure even at elevated temperatures. Although very little is
yet known about the structure and dynamics of C70 under high
pressure, this suggestion is consistent with the strong effects of
pressure on the orientational phase transitions of C70 observed
by Kawamura et al.45 and Samara et al.46 Despite quantitative
discrepancies, their results agree with regard to a large shift of
the rh-to-fcc transition temperature under pressure (g300
K/GPa). This implies that the application of ∼1 GPa pressure
would preserve the orientational ordering in C70 up to much
higher temperatures than were applied in this work. According
to the data of Soldatov and Sundqvist,47 the molecules rotate
uniaxially at p,T conditions of our experiment, i.e., they possess
sufficient mobility for the spatial adjustment of (1,2)/(2,1) bonds
required for dimerization.
Strong support for the (1,2)/(2,1) C2h C140 isomer is also
provided by a very recent study of the polymerization of a C70
single crystal at 2.0 GPa and 570 K.19 X-ray diffraction analysis
established the formation of “zigzag” chains of C70 molecules
having the expected (1,2)/(2,1) intermolecular bonding.19 This
result also suggests that the orientational anisotropy and uniaxial
rotations favorable for (1,2)/(2,1) bonding can be preserved in
pressurized crystalline C70 even at 2.0 GPa and at temperatures
much higher than T01 ∼ 350 K.
Another approach to isomer assignment involves comparing
experimental and theoretical vibrational spectra. This approach
is feasible for fullerenes because the van der Waals intermo-
lecular forces in bulk samples are weak and result in relatively
small perturbations of the molecular vibrations.28a For example,
quantum molecular dynamics simulations of Raman spectra have
been successfully applied to isomers of the odd-numbered
dimeric fullerene C119.
48 However, the vibrational modes on the
The appearance of a single C140 isomer out of the five possible
and energetically equivalent structures I-V (equivalent for
isolated molecules in a vacuum) must be due to some additional
kinetic and/ or energetic constraints imposed in the solid state
under high pressure. The situation would be clear if dimerization
of C70 were to occur under ambient conditions. Indeed, the
general features of the phase diagram of C70 at normal pressure
are fairly well understood. There are two orientational ordering
transitions in C70 on cooling from the high-temperature face-
centered cubic/hexagonal close-packed (fcc/hcp) phase that have
freely rotating molecules: the freezing of the tumbling motion
at T01 ∼ 350 K and the freezing of the uniaxial rotations of the
molecules around their long axes and complete ordering at T02
∼ 280 K.44 The intermediate phase is a rhombohedral (rh) or
deformed hcp phase, with the long axes of the molecules aligned
in parallel along [001] hcp/[111] fcc. In the intermediate and
low-temperature phase, short intermolecular contacts between
reactive double bonds of C70 are realized only for the (1,2) bonds
of neighboring C70 molecules in stacked layers.13,44 The uniaxial
rotations of the molecules in the intermediate phase can readily
bring these bonds into a parallel orientation, which is favorable
two cages in dimeric fullerenes interact only weakly with each
other.31 Consequently, very similar in-cage modes are expected
for the C140 isomers I-V, especially for the closely similar pairs
I, II or III, IV.
The intercage vibrations might be expected to be more
isomer-specific. In any event, the corresponding Raman spectra
are much simpler to analyze in the region below 200 cm-1 than
above 200 cm-1, where there is a large and increasing density
of the on-cage vibrations. It has been shown that the PM3
method quite accurately predicts the low-frequency vibrations
in C120 and other dimeric fullerenes.22 Therefore, we have
calculated the intercage vibrational frequencies for the PM3-
optimized C140 isomers I-V and (5,5)/(5,5) D2h C140 by this
method (Table 4). The Raman-active intercage vibrations for
(1,2)/(2,1) C2h C140 are calculated to occur at 87 (the stretching-
like vibration), 122, and 134 cm-1 and closely match the Raman
features observed at 89, 118, and 129 cm-1. The other three
intercage modes of this dimer are IR-active and are predicted
to occur at 20, 22, and 32 cm-1. However, such low frequencies
were inaccessible for our Raman/IR measurements. Unfortu-

4108 J. Phys. Chem. B, Vol. 104, No. 17, 2000
Lebedkin et al.
TABLE 4: PM3 Modeling of C140 Isomers: Calculated Intercage Vibrations
C140 isomer
sym
calculated intercage vibrations (cm-1
)
(1,2)/(2,1)
(1,2)/(1,2)
(1,2)/(3,3a)
(3,3a)/(3,3a)
(3,3a)/(3a,3)
(5,5)/(5,5)
C2h
C2V
C1
C2h
C2V
D2h
20 (Au)
21 (A2)
20
20 (Au)
20 (A2)
9 (Au)
22 (Au)
22 (B1)
22
22 (Au)
22 (B1)
24 (B1u)
32 (Bu)
32 (A1)
32
33 (Bu)
33 (A1)
34 (B2u)
87 (Ag)
88 (A1)
87
85 (Ag)
88 (A1)
88 (Ag)
123 (Bg)
123 (A1)
120
120 (Bg)
117 (A1)
121 (B1g)
134 (Ag)
133 (B2)
134
134 (Ag)
134 (B2)
127 (B2g)
nately, no significant differences were predicted for the intercage
vibration frequencies of C140 isomers, including the energetically
unfavorable (5,5)/(5,5) D2h isomer (Table 3). This probably is
due to the closely similar geometries of the four-membered
intercage ring in isomers I-V. Thus, for C140 an isomer
assignment on the basis of calculated intercage vibration
frequencies is not possible.
The two NMR-consistent dimers I and II have different
symmetries and should, therefore, have different Raman vs IR
activities for the intercage modes. For isomer I ,with a center
of symmetry in the bridge, the vibrations observed in the Raman
spectrum at 89, 118, and 129 cm-1 should be IR-inactive,
whereas for isomer II these vibrations, belonging to the A1 and
B1 symmetry class (Table 3), should be IR-active. We found
no IR features in this frequency range for C140 (ca. 4 mg pressed
in a 3-mm-diameter pellet), and this result would be consistent
with the C2h isomer. However, the expected (calculated) IR
intensities for the intercage modes of the C140 isomers are very
low, and the poor sensitivity of our spectrometer in the far-IR
region does not allow us to make an unambiguous structure
assignment on the basis of the vibrational spectra alone.
The experimental spectroscopic results and calculations
presented here and discussed above provide firm evidence that
us to estimate the intrinsic triplet-state lifetimes of about 50 µs
for C140 and 10 µs for C120. If one assumes that the efficiency
of intersystem crossing to the triplet state is high (about unity
for monomeric C60 and C70),49 then dissociation must be only
a minor decay channel for the triplet state of the dimers. The
main decay may be due to intercage excitation quenching.
Although it substantially decreases the triplet-state lifetime of
C140 relative to C70 (g12 ms in toluene36) and of C120 relative
to C60 (133 µs in toluene37), such intercage quenching with kq
e 105 s-1 would be quite slow on a photophysical time scale.34
Again, this suggests moderate electronic interactions between
the two linked cages in the dimers. The nature of these
interactions remains to be understood. The results for the C60
dimer have shown no evidence for the formation of a singlet or
triplet excimer-like state.34 In other words, electronic (photo)-
excitations seem to be essentially localized on one of the
fullerene cages. Etheridge and Weisman have reported the
formation of a triplet C70 excimer through encounters between
triplet- and ground-state C70 molecules in toluene.36 However,
the structure of the C70 excimer may be different from that of
C140, and the triplet state of the dimer may be of a localized
character like that of C120. To clarify this and other aspects of
the photophysics of the C70 dimer, further time-resolved
absorption and fluorescence studies would be very helpful.
The triplet-state dissociation of the C60 and C70 dimers is
expected to be the reverse of [2+2] cycloaddition of the
monomers via the excited triplet state. The latter reaction has
been proposed for the photopolymerization of solid C60 and
C70.1,5,13 The observation of dimer dissociation is important
because it gives additional support to this proposal. Obviously,
the overall efficiencies of the photoinduced [2+2] ring formation
and decomposition can strongly depend on the reaction condi-
tions. In general, however, these efficiencies seem quite low
for the fullerenes both in the solid state and in solution. Our
results show that the [2+2] C60 and C70 dimers photodissociate
in (deoxygenated) solution with a quantum yield of the order
of 10-3. No reverse photocycloaddition has been observed in
deoxygenated C60 solutions,50 probably because of its very low
yield. The photophysics is evidently much more complicated
in solid fullerenes, where not only intramolecular, but also
intermolecular electronic processes determine the photoexcita-
tion relaxation.51 The cycloaddition of C60 molecules seems to
be only a minor relaxation channel. Nevertheless, a long and
intense photoirradiation of C60 films yields a well-polymerized
material.5 Apparently, photocycloaddition prevails over possible
photodissociation in solid C60. In contrast to C60, there is little
evidence for photopolymerization in solid C70. Its efficiency
appears to be very low.13 This difference between C60 and C70
has been attributed to the smaller number of reactive double
bonds in C70.13 We believe that, apart from structural differences,
the photophysics behind photopolymerization may also play an
important role.
C140 produced at high pressure has either structure I or II. The
evidence available concerning phase transitions and molecular
packing in the solid state allow us to conclude that the C140
produced in our experiments is most likely the (1,2)/(2,1) C2h
isomer (structure I).
4.2. Photodissociation of C140 and C120. We have frequently
mentioned that C140 and C120 have very similar properties due
to the structural similarities of the intercage bonding. Remark-
able parallels can be found in the photophysical properties of
the dimers relative to the monomers. The photophysics of C120
have been studied in detail by Ma et al.,34 who concluded that
the C60 dimer can be thought of as a pair of C60 derivatives,
each with one functionalized double bond. In other words, there
is no significant electronic interaction between the two cages
in either the ground or the excited state. The similarities in the
absorption and fluorescence properties of C140 and C70 indicate
that the model of independent cages applies in this case as well.
Especially interesting is the observation of the photodisso-
ciation of C120 and C140 into monomeric C60 and C70. We found
essentially the same, relatively low photodissociation quantum
yield for both dimers (about 5 × 10-5 for the air-saturated dilute
ODCB solutions irradiated at 400-600 nm). In contrast, Ma et
al.34 reported no effects of photoirradiation on the absorption
and fluorescence spectra of C120, i.e., no photodissociation, using
approximately the same wavelengths reported here. The reason
for this disagreement with our results is not clear. Obviously,
our HPLC analysis is much more sensitive and specific for
monitoring low-yield photodissociation. However, the conver-
sion of a substantial part of the dimers into monomers was also
accompanied by notable changes in our absorption spectra. We
believe that photodissociation proceeds via an excited triplet
state for both dimers. The photodissociation yield increased in
deoxygenated solutions for both C140 and C120, and this allowed
4.3. Dimerization of C70 and C60 under Pressure. In the
pressure-induced dimerization of solid C60 and C70, we observed
considerable analogy but also significant differences between
these two fullerenes. Both C60 and C70 dimerize at 1 GPa and

Structure and Properties of C140
J. Phys. Chem. B, Vol. 104, No. 17, 2000 4109
200-220 °C, and the resulting dimers have been isolated and
characterized as individual compounds. However, in contrast
to C140, the yield of C120 dramatically decreased with increasing
treatment time, as illustrated in Figure 8. C120 shows charac-
teristic peaks at 1463 cm-1 (shifted pentagonal pinch mode,
which occurs at 1468 cm-1 in C60) and the strong intercage
vibration peak at 96 cm-1. Polymerization of C60 and C120 to
higher C60 oligomers and polymeric chains7 (characteristic
Raman peak at 1458 cm-1) is relatively fast and after an hour
depletes most of the C60 and C120. In fact, the hypothesis that
under high pressure/temperature C60 initially forms dimers which
then readily transform into higher oligomers or polymeric chains
has already received strong support based on vibrational
spectroscopy studies.1,6,10 According to the Raman and IR
spectra (the latter not shown), the short-time treated C60 contains
a high fraction of C120 (∼30 wt %). However, we could dissolve
only ∼3 wt % of C120 by sonication in ODCB. Conceivably,
most dimer molecules were intercalated in a matrix of higher
oligomers and could not be easily released to the solvent.
Treatment of solid C70 at 1 GPa and 200-220 °C yields 30-
50% C140 and only a small amount of insoluble (in ODCB)
residue, which is tentatively assigned to higher C70 oligomers.
The product composition does not depend significantly on the
treatment time. Thus, in contrast to C60, the polymerization of
this ordering is preserved under 1 GPa pressure even at
temperatures well above 200 °C, in agreement with previous
results.45,46 In this ordered phase, the uniaxial rotations of C70
monomers along D5 axes47 place the (1,2) double bonds in
antiparallel juxtaposition so that the cap-to-cap (1,2)/(2,1) dimer
with C2h symmetry (isomer I in Figure 4) can be readily formed.
This structure, which is consistent with all spectroscopic data,
was proposed earlier for the room-temperature photopolymer-
ization of C70.13 The only other structure consistent with the
NMR data is the (1,2)/(1,2) C2V isomer II, but this geometry
would require a parallel juxtaposition of (1,2) double bonds with
the symmetry axes of neighboring C70 monomers at angles of
ca. 120°, which in turn requires a disordered state.
The C70 dimer is structurally analogous to the [2+2]
cycloaddition C60 dimer C120.
11 Vibrational spectroscopy studies
have suggested that C60 dimers can initially form under high
pressure and temperature, but that these species are rapidly
transformed into higher C60 oligomers and polymers.1,10 We
were able to isolate C120 from solid C60 treated under the same
conditions used to prepare C140 when the reaction time was
limited to e5 min. However, even short treatment resulted in
an insoluble residue as the major product, attributed to higher
C60 oligomers and polymeric chains, whose yield increased at
the expense of C120 as treatment time was increased.
C
70 appears to be kinetically “frozen” at the dimer stage under
C140 and C120 feature the same intercage bonding (four-
membered ring) and exhibit close correspondence in many
properties. Thus, C120 exhibits characteristic Raman peaks at
96, 127, and 139 cm-1 for intercage vibrations, which are shifted
to 89, 118, and 129 cm-1, respectively, in C140 according to
the factor (60/70)1/2, exactly as expected for the effect of cage
mass. Both dimers thermally decompose to the monomers with
similar kinetic parameters, and both show little interaction
between the two fullerene cages in the ground and excited
electronic states. Finally, both dimers photodissociate to the
respective monomers with a quantum yield of the order of 10-3
in deoxygenated solutions (400-600 nm irradiation). It is likely
that photodissociation occurs via the excited triplet state of the
dimers and is, therefore, the reverse of the triplet-state [2+2]
cycloaddition of the monomerssa reaction proposed for the
photopolymerization of solid C60 and C70.
our conditions. The efficient formation of C70 polymeric chains
appears to require substantially higher pressures of g2.0 GPa
and temperatures of 250-300 °C.18,19 One possible explanation
for this behavior is that the formation of extended and ordered
polymeric structures is simply more difficult in C70 because of
its lower symmetry and the lower number of reactive double
bonds. However, this does not adequately explain why there
are few C70 trimers and other oligomers (insoluble residue) in
the C70/C140 material.
An additional complication for fullerene polymerization may
arise from the presence of intercalated (interstitial) gases. We
found that the dimerization/polymerization of C60 under short
pressure treatment at 1 GPa and 200-220 °C proceeds ef-
ficiently for samples stored for a long time in air or nitrogen,
i.e., for samples containing some amount of intercalated gas,
but is inefficient for samples freshly annealed in a vacuum and
containing only a small, if any, amount of intercalated argon
due to handling in a glovebox. In contrast to C60, the dimer-
ization of C70 did not depend in this fashion on the sample
history. It is well-established that interstitial gases, even at low
occupation numbers, can notably affect the phase diagram,
molecular rotations, and other physical properties of C60 and
C70.1,46 However, their influence on the polymerization reaction
has not yet been studied. In general, interstitial gases would be
expected to hinder the interactions between fullerene molecules
and the polymerization, but our results suggest that there are
more interesting effects which deserve further investigation.
Acknowledgment. This work was supported by the Deutsche
Forschungsgemeinschaft under SFB 551 “Kohlenstoff aus der
Gasphase: Elementarreaktionen, Strukturen, Werkstoffe”. A.S.
acknowledges a Fellowship for Visiting Researchers from the
Forschungszentrum Karlsruhe. The authors thank Jochen Friedrich
and Carsten Stoermer for the mass spectrometry measurements
and Patrick Weis and Frank Hennrich for the help with the
photodissociation experiments.
References and Notes
(1) Sundqvist, B. AdV. Phys. 1999, 48, 1; and references therein.
(2) 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.; Dabbagh, G.; Krajewski, J. J.; Thomas, G. A.; Yagi, T.
Science 1994, 264, 1570.
(3) Bashkin, I. O.; Rashchupkin, V. I.; Gurov, A. F.; Moravsky, A. P.;
Rybchenko, O. G.; Kobelev, N. P.; Soifer, Ya. M.; Ponyatovsky, E. G. J.
Phys.: Condens. Matter 1994, 6, 7491.
(4) Rao, A. M.; Eklund, P. C.; Hodeau, J.-L.; Marques, L.; Nunez-
Regueiro, M. Phys. ReV. 1997, B55, 4766.
(5) Rao, A. M.; Zhou, P.; Wang, K.; Hager, G. T.; Holden, J. M.; Wang,
Y.; Lee, W.; Bi, X.; Eklund, P. C.; Cornett, D. S.; Duncan, M. A.; Amster,
I. J. Science 1993, 259, 955.
Summary
We have shown that at a pressure of 1 GPa and a temperature
of 200 °C, solid C70 reacts to form a single dimeric species
with high yield. The dimer was isolated by HPLC and
characterized by 13C NMR, vibrational, UV-visible absorption,
and fluorescence spectroscopy. Semiempirical PM3 calculations
predicted that the five possible isomers involving [2+2]
cycloaddition across the short double bonds in C70 should have
essentially equal energy. The preferential formation of a single
C140 isomer indicates that orientational ordering of C70 mono-
mers with parallel D5 axes is the determining factor and that
(6) Burger, B.; Winter, J.; Kuzmany, H. Z. Phys. 1996, 101, 227.
(7) Persson, P.-A.; Edlund, U.; Jacobsson, P.; Johnels, D.; Soldatov,
A.; Sundqvist, B. Chem. Phys. Lett. 1996, 258, 540.

4110 J. Phys. Chem. B, Vol. 104, No. 17, 2000
Lebedkin et al.
(8) Moret, R.; Launois, P.; Persson, P.-A.; Sundqvist, B. Europhys.
Lett. 1997, 40, 55.
(9) Marques, L.; Hodeau, J.-L.; Nunez-Regueiro, M.; Perroux, M. Phys.
ReV. 1996, 54, R12633.
(10) Davydov, V. A.; Kashevarova, L. S.; Rakhmanina, A. V.; Senyavin,
V. M.; Agafonov, V.; Ceolin, R.; Szwarc, H. JETP Lett. 1998, 68, 928.
(11) Wang, G.-W.; Komatsu, K.; Murata, Y.; Shiro, M. Nature 1997,
387, 583.
(12) Iwasa, Y.; Tanoue, K.; Mitani, T.; Izuoka, A.; Sugawara, T.; Yagi,
T. Chem. Commun. 1998, 1411.
(13) Rao, A. M.; Menon, M.; Wang, K.-A.; Eklund, P. C.; Subbaswamy,
K. R.; Cornett, D. S.; Duncan, M. A.; Amster, I. J. Chem. Phys. Lett. 1994,
224, 106. Menon, M.; Rao, A. M.; Subbaswamy, K. R.; Eklund, P. C. Phys.
ReV. 1995, B51, 800.
(14) Sundar, C. S.; Sahu, P. C.; Sastry, V. S.; Rao, G. V. N.; Sridharan,
V.; Premila, M.; Bharathi, A.; Hariharan, Y.; Radhakrishnan, T. S.; Muthu,
D. V. S.; Sood, A. K. Phys. ReV. 1996, B53, 8180.
(15) Iwasa, Y.; Furudate, T.; Fukawa, T.; Ozaki, T.; Mitani, T.; Yagi,
T.; Arima, T. Appl. Phys. 1997, A64, 251.
(31) Krause, M.; Dunsch, L.; Seifert, G.; Fowler, P. W.; Gromov, A.;
Kra¨tschmer, W.; Gutierez, R.; Porezag, D.; Fraunheim, T. J. Chem. Soc.,
Faraday Trans. 1998, 94, 2287.
(32) Lebedkin, S.; Gromov, A.; Giesa, S.; Gleiter, R.; Renker, B.;
Rietschel, H.; Kra¨tschmer, W. Chem. Phys. Lett. 1998, 285, 210.
(33) Sun, Y.-P.; Bunker, C. E. J. Phys. Chem. 1993, 97, 6770.
(34) Ma, B.; Riggs, J. E.; Sun, Y.-P. J. Phys. Chem. 1998, B102, 5999.
(35) Solubility Data Series; Battino, R., Ed.; Pergamon Press: Oxford,
1981; Vol. 7, Chapter 1.8.
(36) Etheridge, H. T., III.; Weisman, R. B. J. Phys. Chem. 1995, 99,
2782.
(37) Fraelich, M. R.; Weisman, R. B. J. Phys. Chem. 1993, 97, 11145.
(38) Nagel, P.; Pasler, V.; Lebedkin, S.; Soldatov, A.; Meingast, C.;
Sundqvist, B.; Persson, P.-A.; Tanaka, T.; Komatsu, K.; Buga, S.; Inaba,
A. Phys. ReV. 1999, B60, 16920.
(39) In difference to uncharged [2+2] bonded dimers, single-bonded
dimers have been proposed for charged fullerene molecules as for Rb1C60:
Oszla´nyi, G.; Bortel, G.; Faigel, G.; Gra´na´sy, L.; Bendele, G. M.; Stephens,
P. W.; Forro, L. Phys. ReV. 1996, B54, 11849. Ku¨rti, J.; Ne´meth, K. Chem.
Phys. Lett. 1996, 256, 119. Schober, H.; To¨lle, A.; Renker, B.; Heid, R.;
Gompf, F. Phys. ReV. 1997, B56, 5937.
(40) The Chemistry of Fullerenes; Taylor, R., Ed.; World Scientific:
River Edge, NJ, 1995.
(41) Meier, M. S.; Wang, G.-W.; Haddon, R. C.; Brock, C. P.; Lloyd,
M. A.; Selegue, J. P. J. Am. Chem. Soc. 1998, 120, 2337.
(42) Fowler, P. W.; Mitchell, D.; Taylor, R.; Seifert, G. J. Chem. Soc.,
Perkin Trans. 2 1997, 1901.
(16) Premila, M.; Sundar, C. S.; Sahu, P. C.; Bharathi, A.; Hariharan,
Y.; Muthu, D. V. S.; Sood, A. K. Solid State Commun. 1997, 104, 237.
(17) Blank, V. D.; Serebryanaya, N. R.; Dubitsky, G. A.; Buga, S. G.;
Denisov, V. N.; Mavrin, B. N.; Ivlev, A. N.; Sulyanov, S. N.; Lvova, N.
A. Phys. Lett. 1998, A248, 415.
(18) Soldatov, A.; Nagel, P.; Pasler, V.; Lebedkin, S.; Meingast, C.;
Roth, G.; Sundqvist, B. In Fullerenes and Fullerene Nanostructures.
Proceedings of the International Winterschool on Electronic Properties of
NoVel Materials 1999; Kuzmany, H., Fink, J., Mehring, M., Roth, S., Eds.;
AIP Conference Proceedings 486, Melville, NY, 1999, p 12.
(19) Soldatov, A.; Roth, G.; Lebedkin, S.; Meingast, C.; Dzyabchenko,
A.; Johnels, D.; Haluska, M.; Kuzmany, H.; in preparation.
(20) Lebedkin, S.; Ballenweg, S.; Gross, J.; Taylor, R.; Kra¨tschmer, W.
Tetrahedron Lett. 1995, 36, 4971. Smith, A. B.; Tokuyama, H.; Strongin,
M.; Furst, G. T.; Romanov, W. J. J. Am. Chem. Soc. 1995, 117, 9359.
(21) Gromov, A.; Lebedkin, S.; Ballenweg, S.; Avent, A. G.; Taylor,
R.; Kra¨tschmer, W. J. Chem. Soc., Chem. Commun. 1997, 209.
(22) Eisler, H.-J.; Hennrich, F. H.; Werner, E.; Hertwig, A.; Stoermer,
C.; Kappes, M. M. J. Phys. Chem. 1998, A102, 3889.
(23) Gromov, A.; Ballenweg, S.; Giesa, S.; Lebedkin, S.; Hull, W. E.;
Kra¨tschmer, W. Chem. Phys. Lett. 1997, 267, 460.
(24) Stoermer, C. Ph.D. Thesis, University of Karlsruhe, 1999.
(25) Friedrich, J.; Weis, P.; Kaller, J.; Whetten, R.; Kappes, M. M. Z.
Phys. D.; in press.
(26) Foote, C. S. In Topics in Current Chemistry: Electron-Transfer I;
Mattay, J., Ed.; Springer-Verlag: Berlin, 1994; p 347.
(43) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B. et al. Gaussian 98;
Gaussian Inc.: Pittsburgh, PA, 1998.
(44) Verheijen, M. A.; Meeks, H.; Meyer, G.; Bennema, P.; de Boer, J.
P.; van Smaalen, S.; van Tendeloo, G.; Amelinckx, S.; Muto, S.; van
Landuyt, J. Chem. Phys. Lett. 1992, 166, 287. Vaughan, G. B. M.; Heiney,
P.; Cox, D. E.; Fischer, J. E.; McGhie, A. R.; Smith, A. L.; Strongin, R.
M.; Cichy, M. A.; Smith, A. B., III. Chem. Phys. 1993, 178, 599. McGhie,
A. R.; Fisher, J. E.; Stephens, P. W.; Cappelletti, R. L.; Neumann, D. A.;
Mueller, W. H.; Mohn, H.; ter Meer, H.-U. Phys. ReV. 1994, B49, 12614.
(45) Kawamura, H.; Akahama, Y.; Kobayashi, M.; Shinohara, H.; Sato,
H.; Saito, Y.; Kikegawa, T.; Shimomura, O.; Aoki, K. J. Phys. Chem. Solids
1993, 54, 1675.
(46) Samara, G. A.; Hansen, L. V.; Morosin, B.; Schirber, J. E. Phys.
ReV. 1996, B53, 5211.
(47) Soldatov, A.; Sundqvist, B. J. Phys. Chem. Solids 1996, 57, 1371.
(48) Lebedkin, S.; Rietschel, H.; Adams, G. B.; Page, J. B.; Hull, W.
E.; Hennrich, F. H.; Eisler, H.-J.; Kappes, M. M. J. Chem. Phys. 1999,
110, 11768.
(49) Arbogast, J. W.; Darmanyan, A. P.; Foote, C. S.; Rubin, Y.;
Diederich, F. N.; Alvarez, M. M.; Anz, S. J.; Whetten, R. L. J. Phys. Chem.
1991, 95, 11. Arbogast, J. W.; Foote, C. S. J. Am. Chem. Soc. 1991, 113,
8886.
(50) Sun, Y.-P. In Molecular and Supramolecular Photochemistry;
Ramamurthy, V., Schanze, K. S., Eds.; Marcel Dekker: New York, 1997;
Vol. 1, p 325.
(27) Ajie, H.; Alvarez, M.; Anz, S. J.; Beck, R. D.; Diederich, F.;
Fostiropoulos, K.; Huffman, D. R.; Kra¨tschmer, W.; Rubin, Y.; Schriver,
K. E.; Sensharma, D.; Whetten, R. L. J. Phys. Chem. 1990, 94, 8630.
(28) Dresselhaus, M. S.; Dresselhaus, G.; Eklund, P. C. Science of
Fullerenes and Carbon Nanotubes; Academic Press: San Diego, 1996; and
references therein. (a) Chapter 11. (b) Chapter 3.
(29) Adams, G. B.; Page, J. B.; Sankey, O. F.; O’Keeffe, M. Phys. ReV.
1994, B50, 17471.
(30) Porezag, D.; Pederson, M. R.; Frauenheim, T.; Ko¨hler, T. Phys.
ReV. 1995, B52, 14963.
(51) Kazaoui, S.; Minami, N.; Tanabe, Y.; Byrne, H. J.; Eilmes, A.;
Petelenz, P. Phys. ReV. 1998, B58, 7689.