Matrix isolation of fullerene-derived CO2 at ambient temperature

Article abstract of DOI:10.1016/S0009-2614(98)00979-8

Heating fullerene oxides, e.g. C₁₂₀O, C₇₀O, C₆₀O and C₆₀O₂, in a KBr matrix at 225°C under 0.2 mbar vacuum, produces a sharp IR band at 2330 cm^-1 due to matrix-isolated CO₂. The band is also obtained by heating a KBr matrix of the insoluble deposits that fullerenes form on standing in air. The matrices are extremely stable and are unchanged even by prolonged heating at 225°C under vacuum. Heating a KBr matrix of the deposit from C₈₄ produces also a sharp stable band at 2035 cm^-1 consistent with matrix-isolated C₃. Similar treatment of C₇₀F₃₈O produces matrices containing both CO₂ and CO, the latter being of lower stability.

Full text of DOI:10.1016/S0009-2614(98)00979-8

23 October 1998
Ž
.
Chemical Physics Letters 295 1998 481–486
Matrix isolation of fullerene-derived CO2 at ambient temperature
b
a
Roger Taylor a,), Alain Penicaud , Nicole J. Tower
´
a
The Chemistry Laboratory, CPES School, Sussex UniÕersity, Brighton BN1 9QJ, UK
Centre de Recherche Paul-Pascal, C.N.R.S., AÕenue Schweitzer, F-33600 Pessac, France
b
Received 29 April 1998; in final form 26 August 1998
Abstract
Heating fullerene oxides, e.g. C120O, C70O, C60O and C60O2, in a KBr matrix at 2258C under 0.2 mbar vacuum,
1
produces a sharp IR band at 2330 cmy due to matrix-isolated CO2. The band is also obtained by heating a KBr matrix of
the insoluble deposits that fullerenes form on standing in air. The matrices are extremely stable and are unchanged even by
prolonged heating at 2258C under vacuum. Heating a KBr matrix of the deposit from C84 produces also a sharp stable band
1
at 2035 cmy consistent with matrix-isolated C3. Similar treatment of C70F38O produces matrices containing both CO2 and
CO, the latter being of lower stability. q 1998 Elsevier Science B.V. All rights reserved.
1. Introduction
2340 cmy1, the shift to lower frequency compared to
the gas phase being attributed to interactions with the
w x
Various research groups have studied the IR spec-
salt surface 5 .
Ž
trum of matrix-isolated CO2 at low temperature ;
.
w
x
10 K in either hydrogen or rare gas matrices 1–4 .
Multiplicity of peaks are seen due to occupancies of
different matrix sites, various dimer orientations, and
the presence of variable-size clusters, indicated by
alteration in the relative band intensities upon an-
nealing. The gas-phase IR spectrum of carbon diox-
2. Experimental and discussion
The IR spectra were obtained on a Perkin-Elmer
1710 FTIR spectrometer at 2 cmy1 resolution; the
spectra show the usual wavenumber scale change at
ide shows antisymmetrical stretching fundamentals
2000 cmy1
.
y1
n3 at 2360 cmy1 P-branch and 2342 cm
R-
Ž
.
Ž
.
Ž
branch shoulders on the latter are due to the 13C
isotope , together with the symmetrical bending fun-
. Ž
.
2.1. Matrix isolation of CO2
y1
Ž
.
Ž
damental n2 centred at 670 cm
inset to Fig. 1
.
shows the normal gas-phase spectrum . For CO2
deposited on NaCl at y1608C, n3 appeared
as a single band of ;10 cmy1 half-height width at
During measurement of the IR spectrum of pure
C120O 6 , we observed that on heating a KBr disc
formed in a standard 10 ton press under 0.2 mbar
vacuum to eliminate moisture traces , a sharp band
w x
Ž
.
Ž
.
y1
Ž
.
;13 cm half-height width appeared in the spec-
trum at 2330 cmy1 Fig. 1 . To confirm that this was
)
Ž
.
Corresponding author. Fax: q44-1273-677-196.
0009-2614r98r$ - see front matter q 1998 Elsevier Science B.V. All rights reserved.
Ž
.
PII: S0009-2614 98 00979-8

(
)
482
R. Taylor et al.rChemical Physics Letters 295 1998 481–486
trometer shut down temporarily due to ion current
overload.
Ž
.
The loss of CO2 rather than CO from the oxides
is unexpected and indicates that a bimolecular reac-
Ž
Ž ..
tion must be involved. This predicts Eq. 1 that
Ž
.
pure C60O should produce C119 1429 amu on
heating, and this was confirmed by EI mass spec-
trometry.
2 C60O ™ CO2 qC119
.
1
Ž .
C119 has been produced previously from the reac-
tion of C60O with C60, the mechanism being as-
sumed to involve loss of CO from the epoxide to
give a carbene which then either adds across a
w
x
Ž
.
6:6-bond 7,8 , or inserts mainly into a 6:5-bond
w x
8 ; subsequent bond formation produces two iso-
Ž
.
Fig. 1. IR spectrum KBr disc of C120O, after heating at 2258C
mers of C119 one of which has been characterised
for 2h; inset shows IR spectrum of CO2 in air.
Ž
a trapped gas, the sample was dissolved in water to
.
Ž
remove the KBr and then 1,2-dichlorobenzene to
allow the gas to fully escape , and after removal of
.
the solvent, a fresh KBr disc was prepared. The band
at 2330 cmy1 was now absent, but could be regener-
Ž
.
ated with lower intensity by further strong heating
of the new disc.
Further experiments have been carried out with
HPLC purified C60O and C60O2. Neither of these
compounds shows any band in the CO2 region, but
Ž
.
on heating either of them 2258C for 2 h , an intense
band again appears at 2330 cmy1. Higher oxides of
w
x
60 fullerene, and C70O behave in the same way. In
each spectrum, n2 appears as a single band at 653"1
cmy1. Fig. 2 shows the results of heating C60O at
various timesrtemperatures and the development of
w
the main CO2 band which shows evidence of being
comprised of a mixture of molecules which are
y1
Ž
.
matrix isolated sharp band at 2330 cm
and some
Ž
which are able to rotate freely shoulder at ;2360
cmy1 . The shoulder was obtained only from the
.x
oxides which generate such large numbers of CO2
molecules that failure to isolate all of them is not
unexpected. The evolution of CO2 from C60O was
also monitored and confirmed by EI mass spectrome-
Ž
.
try of the heated matrix-free solid. Fig. 3 shows the
plot of the intensities of the total ion current and the
44 amu peak with increasing probe temperature. So
great was the CO2 generation that the mass spec-
Ž
.
Ž .
Ž . Ž .
Fig. 2. IR spectra KBr disc of C60 O:
heated for 30 min at 100, 150, 200 and 2508C, respectively.
a unheated; b – e ,

(
)
R. Taylor et al.rChemical Physics Letters 295 1998 481–486
483
Ž
.
Ž
O without
Fig. 3. Plot of intensities EI mass spectrometry of the total ion current and of the 44 amu peak on heating a sample of C60
.
matrix .
w x
w
x
Ž .
other higher fullerenes e.g., C84 behave similarly,
though with lower degradation rates e.g., 70% in 4
9 , the structure agreeing with prediction 10 . The
present Letter indicates that a second mechanism of
C119 formation exists that does not require the pres-
ence of C60 itself.
Ž
.
w x
years for C84 . Our provisional conjecture 11 that
the insoluble material might be due to graphitisation
Ž .
The sharp single lines in the spectra arise from
restriction of the molecular motions of the matrix-
trapped molecules, which could be expected to be
greater than in the material deposited on a NaCl
surface, and consequently n2 appears at a lower
frequency. This shift is not due to the different
matrix material since if NaCl is used instead of KBr,
the CO2 band again appears at 2330 cmy1. The
amazing effectiveness of the KBr matrix as a trap for
fullerene-produced CO2 is shown by the spectra
being completely unchanged after 10 months storage
of the KBr discs at room temperature. They are also
completely unchanged by heating the discs at 2258C
for 26 h.
was disproved in two ways: 1 no graphitic planes
could be detected by X-ray powder diffraction; and
Ž .
2 on heating of a KBr disc as described, it pro-
duced the intense sharp band at 2330 cmy1 Fig. 4a
Ž
.
showing that it is an oxidised material. The insoluble
degradation products of the other higher fullerenes
all give the same band. Heating a KBr matrix of
graphite does not produce the CO2 band, nor does
one of single wall nanotubes that had been hy-
drothermally purified from amorphous material.
Ž
It should be noted that C60 a pure sample stored
.
for 7 years also produces the CO2 band on heating
as described, albeit of lower intensity freshly puri-
fied material does not . Though we reported previ-
ously that allowing thin films of C60 to stand in air
Ž
.
Ž
.
Fullerenes especially the higher ones produce
insoluble material on storage. Most notable in this
for a few days, caused them to become harder to
Ž .
respect is the C2v II isomer of C78. Within 6 weeks
. w
x
of HPLC purification, ;30% of the material was
totally insoluble in all solvents. After a further 14
weeks, none of it would dissolve, and we find that

(
)
484
R. Taylor et al.rChemical Physics Letters 295 1998 481–486
produced by a gas. Given the starting materials, it
appears that the only species that could give this
Ž
.
band in this IR region is C3 see below due to cage
Ž
fragmentation, and although this CO2-isostructural
dicarbene should be reactive, the extraordinary ef-
.
fectiveness of the KBr matrix as a trap may prevent
it from undergoing further reaction. Since the yield
w
x
of 84 fullerene from graphite is only ;0.03%, con-
firmation of the identity of this band through isotopic
enrichment is precluded at present by financial con-
siderations.
There have been numerous studies of the IR
Ž
.
spectra of matrix-isolated carbon species C2–C9 in
Ž
inert gases, at ;4–10 K, during recent years lead-
ing references are to be found in Ref. 13 . C3 has
been of particular interest because it is the major
w
x.
w
x
species present in carbon vapour 14 , and nature of
the coalescence products on annealing have been
w
x
given particular attention, e.g. Ref. 15 . In the gas
phase C3 gives a sharp n3 band at 2040 cmy1, and
this becomes 2039 and 2035 cmy1 in Ar and Kr
w
x
matrices, respectively 16 . The band we observe
cannot be attributed to bonding of CO to the cage
Ž
values in this region are obtained for some s-bonded
.
inorganic complexes because it is absent in a
w
x
Ž
60 fullerene matrix-isolated spectrum of CO intro-
duced under ;10 kbar pressure 17 .
. w
x
2.3. Matrix isolation of CO
Ž .
Ž
.
Ž .
Fig. 4.
a IR spectrum KBr disc of degraded C2v II -C78 after
Ž .
Ž
.
heating at 2258C for 2h. b IR spectrum KBr disc of degraded
C84 after heating at 2258C for 2h.
We are also able to produce a matrix-isolated
spectrum of CO by heating a KBr disc of pure
w
x
C70 F38O 18 . Meier and co-workers have described
a fullerene matrix-isolated spectrum of CO which
w
x
to water-soluble derivatives 11 , these previous ob-
servations concerned highly dispersed material.
consists at 298 K of a broad band at 2120 cmy1 17 .
We obtain a broad CO band at 2108, 2107, 2105 and
2104 cmy1 after the following respective heating
regimes: 3 h at 1008C, then 2 h at 1508C, 1 h at
2008C, and a further 1.5 h at 2008C. C70 F38O is also
extremely effective at producing the CO2 band, at
w
x
2.2. EÕidence for matrix isolation of C3
Our results have an even more remarkable aspect
indicated in the spectrum of heated ‘insoluble’ C84
Fig. 4b . This shows not only the CO2 band at
2340, 2338, 2336 and 2335 cmy1 with increasing
Ž
.
Ž
Ž
2327 cmy1 in this case but also a second sharp
band at 2035 cmy1 and this is unchanged after 10
months storage. This was also reproduced from sam-
ples of different origin, and from one sample of
C60O2; like the CO2 band, it is removed by dissolv-
ing the disc and reforming, indicating that it is
intensities after the above heatingrtimes; n2 here
.
.
appears at 659 cmy1; none of these bands are pre-
sent in the pure material. The frequency shifts are
attributable to the loss of fluorine from the fullerene
precursor in the matrix on heating. The CO-contain-
ing matrix does not have the same stability of the

(
)
R. Taylor et al.rChemical Physics Letters 295 1998 481–486
485
that in Fig. 5a, because of the lower stability of the
COrKBr matrix and the extended heating.
2.4. The nature of the matrix
The similarity of the band positions when using
either KBr or NaCl appears to indicate that the
matrix is simply the fullerene embedded in the salt.
This would seem to be supported by the slightly
different band positions obtained with different
fullerene precursors. However, this is a simplifica-
tion, because as described above, heating the solids
in the absence of the salt results in release of CO2
gas. Moreover, heating to 2258C a KBr matrix of
Ž
2,4-dihydroxybenzoic acid well known to produce
.
CO2 on heating also gives an intense stable CO2
band, centred at 2337 cmy1. However, this is 2–3
times broader than the bands obtained from fullerene
y1
Ž
.
,
precursors and has a major shoulder at 2360 cm
indicating that a significant proportion of the CO2 is
able to rotate here. Overall, it seems that it is the
Ž
combination of the fullerene and salt compounded
.
by the pressure involved in forming the matrix
which gives rise to the spectra we observe.
Further investigations on the matrix isolation of
other fullerene-derived species is in hand.
Acknowledgements
We thank Dr. T. Drewello and Mr. M. Barrow
Ž
.
Warwick University for the MALDI-TOF mass
spectra of C60O and C60O2 , and Mr. R. Loutfy
Ž .
Ž
.
Fig. 5. a IR spectrum KBr disc of C70 F38O after heating for a
Ž .
total of 3h at 1008C followed by 1.5h at 1508C. b The KBr disc
Ž
.
MER for the sample of nanotubes.
from Fig. 4b, heated for a further 2.5h at 2008C.
Ž
CO2-containing matrix the difference being at-
References
.
tributed to the smaller molecular size and the CO
band is eventually lost on heating under vacuum at
2258C. Fig. 5a shows the spectrum after heating for
3 h at 1008C followed by 2 h at 1508C, and Fig. 5b
shows the same sample after further heating for a
total of 2.5 h at 2008C. The development in intensi-
ties of the n2 and n3 bands for CO2 is accompanied
w x
Ž .
E. Knoezinger, P. Beichert, J. Phys. Chem. 99 1995 4906.
1
w x
2
Ž
.
M. Falk, J. Chem. Phys. 86 1987 560.
w x
3
M.J. Irvine, J.G. Mathieson, A. David, E. Pullin, Aust. J.
Ž
.
Chem. 35 1982 1971.
w x
4
Ž
.
M.J. Irvine, A. David, E. Pullin, Aust. J. Chem. 35 1982
1961.
w x
5
J. Heidberg, S. Zehme, C.F. Chen, H. Hartmann, Ber. Bun-
by a shift to lower wavenumbers as the fluorine
Ž
.
senges. Phys. Chem. 75 1971 1009.
y1
Ž
.
C–F stretch at 1125 cm
is lost. In Fig. 5b the
w x
6
S. Lebedkin, S. Ballenweg, J. Gross, R. Taylor, W.
Ž .
Kratschmer, Tetrahedron Lett. 36 1995 4971.
¨
band for CO has decreased in intensity relative to

(
)
486
w x
R. Taylor et al.rChemical Physics Letters 295 1998 481–486
Ž
.
w
w
x
Ž
.
7
w x
8
R. Taylor, J. Chem. Soc., Chem. Commun. 1994 1629.
13 S. Tam, M. Macier, M.E. Fajardo, J. Chem. Phys. 106 1997
Ž
.
E. Albertazzi, F. Zerbetto, J. Am. Chem. Soc. 118 1996
8955.
x
14 W. Weltner, P.N. Walsh, C.L. Angell, J. Chem. Phys. 40
2734.
w x
9
Ž .
1964 1299.
A. Gromov, S. Ballenweg, S. Geisa, S. Lebedkin, W.E. Hull,
Ž
.
w
w
w
x
Ž
.
W. Kratschmer, Chem. Phys. Lett. 267 1997 460.
15 J. Sczcepanski, M. Vala, J. Phys. Chem. 95 1991 2792.
¨
w
w
x
x
x
Ž
.
10 G.B. Adams, J.B. Page, M. O’Keefe, O. Sankey, Chem.
16 J. Sczcepanski, M. Vala, J. Chem. Phys. 99 1993 7371.
17 I. Holleman, G. von Helden, E.H.T. Olthof, P.J.M. van
Ž
.
Phys. Lett. 228 1994 485.
x
Ž
.
11 R. Taylor, in: H. Kuzmany Ed. , Molecular Nanostructures
Bentum, R. Engeln, G.H. Nachtegaal, A.P.M. Kentgens,
B.H. Meier, A. van der Avoird, G.J.N. Meijer, Phys. Rev.
Ž
Proc. Int. Winterschool on Electronic Properties of Novel
.
Ž .
Materials , World Scientific, Singapore, 1998, p. 136.
Lett. 79 1997 1138.
w
x
Ž
.
w
x Ž .
18 O.V. Boltalina, J.M. Street, R. Taylor to be published .
12 R. Taylor, Interdiscip. Sci. Rev. 17 1992 161.