Detection of cyclic carbon clusters. I. Isotopic study of the v4(e′) mode of cyclic C6 in solid Ar

Article abstract of DOI:10.1063/1.475316

The cyclic C₆ cluster has been identified for the first time in Fourier transform infrared spectra of the products from the laser evaporation of graphite rods trapped in Ar at ～ 10 K. Measurements on spectra produced using both ¹²C-

Full text of DOI:10.1063/1.475316

Detection of cyclic carbon clusters. I. Isotopic study of the ν 4 (e′) mode of cyclic C 6 in
solid Ar
S. L. Wang, C. M. L. Rittby, and W. R. M. Graham
Citation: The Journal of Chemical Physics 107, 6032 (1997); doi: 10.1063/1.475316
View online: http://dx.doi.org/10.1063/1.475316
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/107/16?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Vibrational spectra of germanium-carbon clusters in solid Ar: Identification of the ν 4 ( σ u ) mode of linear Ge C 5
Ge
J. Chem. Phys. 125, 044504 (2006); 10.1063/1.2227398
Isomers of S 2 O : Infrared absorption spectra of cyclic S 2 O in solid Ar
J. Chem. Phys. 117, 6655 (2002); 10.1063/1.1506155
Fourier transform infrared isotopic study of the C 12 chain trapped in solid Ar
J. Chem. Phys. 112, 5113 (2000); 10.1063/1.481067
Fourier-transform infrared observation of SiC n chains. I. The ν 4 (σ) mode of linear SiC 9 in Ar at 10 K
J. Chem. Phys. 110, 11214 (1999); 10.1063/1.479062
Detection of cyclic carbon clusters. II. Isotopic study of the ν 12 (e u ) mode of cyclic C 8 in solid Ar
J. Chem. Phys. 107, 7025 (1997); 10.1063/1.474945
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
138.38.151.151 On: Sun, 23 Nov 2014 16:01:57

Detection of cyclic carbon clusters. I. Isotopic study of the ␯ „eЈ… mode
4
of cyclic C in solid Ar
6
S. L. Wang, C. M. L. Rittby, and W. R. M. Graham
Department of Physics, Texas Christian University, Fort Worth, Texas 76129
͑
Received 28 April 1997; accepted 21 July 1997͒
The cyclic C cluster has been identified for the first time in Fourier transform infrared spectra of
6
the products from the laser evaporation of graphite rods trapped in Ar at ϳ10 K. Measurements on
spectra produced using both 12C- and C-enriched rods combined with the results of new density
functional theory calculations performed in the present work as well as previous calculations by
13
Martin and Taylor, have resulted in the assignment of the most intense infrared active mode,
Ϫ1
␯4
(eЈ)ϭ1694.9 cm of the cyclic C isomer with D symmetry. This assignment is based on
6 3h
13
excellent agreement of the frequency, C isotopic shifts, and relative intensities with the theoretical
predictions. © 1997 American Institute of Physics. ͓S0021-9606͑97͒02540-3͔
I. INTRODUCTION
coupled cluster calculations with single, double, and triple
excitations ͓CCSD͑T͒/TZ2P͔. They found that the linear and
Although carbon clusters have long been the subject of
extensive theoretical and experimental studies ͑see Refs. 1
and 2 for reviews͒, activity in this area of research has now
reached an unprecedented level of intensity with the ongoing
development of fullerene chemistry. Linear C with a D
cyclic forms of C were very close in energy and concluded
6
that both structures might be experimentally observable.
15
In experimental work, Szczepanski and Vala reported
Ϫ1
that the intensity of an infrared band located at 1695 cm
,
6
ϱh
which was first assigned to a pure carbon molecule in an Ar
cumulenic form has been identified and its fundamental vi-
brations have been assigned.3 Similarly, vibrational funda-
mentals have been firmly established by matrix and gas
phase measurements for all the other small linear carbon
clusters up to C , with the possible exception of C . More
3
matrix in Weltner’s classic study, grew dramatically upon
–8
annealing and speculated that it belonged to a cyclic cluster
containing six or more carbons. In this laboratory, compari-
son of the results of density functional theory ͑DFT͒
16
9
8
calculations at the BVWN5 level ͑Becke–Vosko–Wilk–
recently laser diode spectroscopic measurement of a mode of
Nusair functionals͒ using a standard 6-31G* basis set, with
9
linear C13 has also been reported. No identification of vibra-
preliminary data obtained from the laser evaporation of
tional modes of cyclic carbon clusters has been reported,
however, despite the predictions of extensive ab initio calcu-
lations that cyclic geometries are generally preferred as the
lowest energy structures for small carbon clusters with an
even number of atoms (nр8) and for all larger clusters (n
у10).
Ϫ1
graphite suggested that the 1694.9 cm band, also seen in
17
earlier thermal evaporation experiments, was a potential
candidate for the cyclic C isomer.
6
18
Martin and Taylor showed on the basis of calculations
at the CCSD͑T͒/cc-pVTZ level, that the cyclic form of C is
6
slightly favored energetically over the linear form. Earlier,
In the case of C ͑and C ͒, because of an apparently
19
6
4
Martin and co-workers predicted a frequency for the most
small difference in energy between the linear and cyclic iso-
mers, some uncertainty surrounds the lowest energy struc-
ture. A cyclic C isomer with a D symmetry, which is
intense band of this cyclic structure using the B3LYP ͑Becke
3
-parameter-Lee–Yang–Parr͒ functionals which suggested
Ϫ1
6
3h
an assignment in the 1700–1800 cm region. Specifically,
they proposed two absorptions observed at 1695 and
essentially a distorted benzene ring in which alternate atoms
of the hexagon have moved radially in or out, was first pre-
dicted to be the most stable form by Raghavachari and
Ϫ1
15,20
1
715 cm in Ar matrices,
as leading contenders and
13
18
provided a complete set of C isotopic shifts for compari-
1
0,11
co-workers.
They optimized geometries at the Hartree–
son with future experimental studies.
Fock ͑HF͒ level using a 6-31G* basis set and found that
beyond second-order many-body perturbation theory the D3h
structure was energetically favored. The frequencies of the
In the present paper, with the support of detailed 13C
Ϫ1
isotopic measurements, we assign the 1694.9 cm band to
␯4
(eЈ), the most intense mode of cyclic C with D sym-
6 3h
infrared-active carbon stretching modes were calculated as
13
metry. Extensive C isotopic data are in very good agree-
ment with present B3LYP/cc-pVTZ DFT calculations as
well as with Martin and Taylor’s predictions at the B3LYP/
Ϫ1
1
726 and 1365 cm by the same group. These results were
12
later challenged, by Parasuk and Alml o¨ f who predicted on
the basis of multireference configuration interaction ͑MRCI͒
calculations with complete active space self-consistent field
18
cc-pVDZ level.
͑
CASSCF͒ optimized geometries that the linear form was the
II. THEORETICAL CALCULATIONS
most stable and that the cyclic isomer lay 37.2 kcal/mol
higher in energy. These results were in turn challenged by
Hutter and co-workers’ 13,14 higher level density functional
theory ͑DFT͒ calculations at the BP86/4s3p1d level and
Recently, in conjunction with the presentation of their
new coupled cluster calculations, Martin and Taylor sum-
marized the history of theoretical calculations of the C clus-
18
6
6032
J. Chem. Phys. 107 (16), 22 October 1997
0021-9606/97/107(16)/6032/6/$10.00
© 1997 American Institute of Physics
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
138.38.151.151 On: Sun, 23 Nov 2014 16:01:57

6033
Wang, Rittby, and Graham: Detection of cyclic C6
ter. In the pioneering work of Raghavachari et al.10,11 a best
estimate of 13.9 kcal/mol was obtained for the energy sepa-
ration between the linear and cyclic isomers. This value was
derived from a single-point coupled cluster doubles ͑CCD͒
calculation with partial inclusion of fourth-order effects
͓
CCDϩST͑CCD͒/6-31G*͔ at HF/6-31G* geometries. This
energy ordering has subsequently been confirmed by most
higher level calculations. Martin and Taylor’s18 range of val-
ues of 11–15 kcal/mol for the energy separation was based
on their extensive CCSD͑T͒/cc-pVTZ calculations at opti-
mized geometries. These coupled cluster calculations include
single and double excitations as well as a quasiperturbative
inclusion of connected triple excitations. Given that ten years
has passed since the work of Raghavachari and
FIG. 1. Optimized geometry parameter for cyclic C obtained at the DFT-
B3LYP/cc-pVTZ level.
6
1
0,11
co-workers
it may seem that little progress has been
made. This is, however, not true since the evolution of hard-
ware, software, and new theoretical techniques has led to the
possibility of including electron correlation effects to high
orders when optimizing geometries as well as when calculat-
ing vibrational frequencies. With regards to the identification
of new carbon clusters in the laboratory this development has
been crucial.
ab initio calculation and the first-order singly substituted iso-
topic shift ͑for nondegenerate modes͒ of the mode with fre-
quency ␻ is obtained from
␻
mi
⌬1
ϭ
ͩ
1Ϫ
ͪ
͉u͉
2,
2
mf
Based on the theoretical energy separation between the
where m and m are the initial and final isotopic masses,
21
i
f
cyclic and linear isomers Slanina et al. have shown that
thermodynamical considerations make the linear isomer
dominate strongly in the high-temperature production of C6.
respectively, and u is the mass weighted normal-mode vector
for the isotopically substituted nucleus. Since first-order ef-
fects should dominate for small values of ͉m Ϫmf͉ and in-
i
This explains the experimental situation where linear C can
6
teractions with the remaining vibrational modes occur at
higher order, this index should provide us with a feel for the
importance of describing the full vibrational spectrum accu-
rately.
be readily produced by various means whereas the cyclic
structure has been more elusive.
In our matrix Fourier transform infrared ͑FTIR͒ experi-
ments identifications of new molecular species rely heavily
on the method of isotopic substitution. In the case of shorter
carbon chains a majority of the isotopomers can be identi-
fied, and by comparison with theoretical predictions a par-
ticular absorption band can be assigned with a high level of
confidence. As longer carbon chains are considered the large
number of isotopomers gives rise to an increasing number of
overlapping bands and the identification of a subset of isoto-
pomers ͑primarily the singly substituted ones͒ can give suf-
ficient support for a new identification. An additional com-
plication originates from the increasing density of high-
frequency stretching modes which can lead to strong
interactions between modes as the symmetry of the molecule
is broken upon isotopic substitution. This provides a further
challenge to the theorist in that it may require a delicate
To investigate the behavior of calculated isotopic shifts
of cyclic C as well as to aid in the experimental identifica-
6
tion of this cluster we performed a series of calculations at
various levels of theory. All calculations were made on an
25
IBM RS/6000-370 workstation using the GAUSSIAN 94 pro-
gram package using standard basis sets. For comparison with
our experimental data ͑see Sec. IV͒ we initially performed a
DFT calculation with the standard B3LYP functionals and
25
cc-pVTZ basis set available in GAUSSIAN 94. Figure 1 gives
the DFT-B3LYP/cc-pVTZ optimized geometry parameters
and Table I gives the harmonic frequencies and infrared in-
tensities obtained at this geometry. These results are in line
with previous calculations ͑see Refs. 18 and 19͒.
‘
‘fine-tuning’’ of the full force constant matrix in order to
TABLE I. Harmonic frequencies calculated at the DFT-B3LYP/cc-pVTZ
level of cyclic C6.
obtain isotopic shifts accurate enough to be useful to the
experimentalist. Recently we have investigated this situation
in studies of linear C and C ͑Refs. 22 and 23͒ as well as for
7
9
Frequency
Infrared intensity
24
Ϫ1
the linear SiC H cluster.
Vibrational mode
(cm
)
͑km/mol͒
2
In an attempt to quantify the behavior described above
we defined an ‘‘isotopic sensitivity index’’ as the ratio of
higher order to first-order effects:
␯1
␯2
␯3
␯4
(a1Ј)
(a1Ј)
(a2Ј)
(eЈ)
(eЈ)
␯6(eЈ)
(a2Љ)
(eЉ)
1220.9
650.4
1408.5
1758.5
1228.1
644.9
0
0
0
435ϫ2
0.7ϫ2
22ϫ2
6.2
⌬2
ϩ⌬ ϩ•••
⌬␻ϱ
⌬1
␯5
3
␫s
ϭ
ͯ
ͯ
ϭ
ͯ
1Ϫ
ͯ
.
⌬1
␯1
␯8
438.1
538.7
0
Here ⌬␻ denotes the infinite order calculated shift from an
ϱ
J. Chem. Phys., Vol. 107, No. 16, 22 October 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
138.38.151.151 On: Sun, 23 Nov 2014 16:01:57

6034
Wang, Rittby, and Graham: Detection of cyclic C6
TABLE II. Isotopic sensitivity index calculated at the DFT-B3LYP/cc-
pVTZ level for the ␯ (eЈ) mode of cyclic C .
4
6
Isotopomer
C -C -C -C -C -C
␤
⌬1
⌬␻ϱ
␫s
ϳ0
0.025
0.045
0.027
␣
␤
␣
␤
␣
1
3-12-12-12-12-12
2-13-12-12-12-12
0.03
0.55
5.09
0.04
29.82
4.87
3
1
9.55
9.29
Subsequently, we used the normal-mode coordinates
from the B3LYP/cc-pVTZ calculation to calculate the isoto-
13
pic sensitivity index for the ␯ (eЈ) mode for single C sub-
4
stitutions. Since this is a degenerate mode, the first-order
shift must be obtained from the diagonalization of the first-
order matrix ⌬ with matrix elements given by
1
␻
mi
mf
͓
⌬ ͔ ϭ
ͩ
1Ϫ
ͪ
u •u .
1
rs
r
s
2
The results are presented in Table II. Since there are two sets
of three equivalent carbon atoms ͑labeled ␣ and ␤ in Fig. 1͒,
there are only two distinct singly substituted isotopomers.
The lifting of the degeneracy of the eЈ mode gives rise to
four singly substituted isotopic shifts. From our previous ex-
perience with linear carbon clusters2 a sensitivity index of
less than 0.1, i.e., a contribution from higher orders of less
than 10% of that from first order, usually indicates that the
calculated shift will be relatively stable with respect to a
change of method or basis set, provided post-HF methods are
2,23
considered. For cases where ␫ Ͼ0.1, care has to be taken
s
when trying to compare theoretical shifts with those from
experiment. A modest change in the calculation may give
rise to a dramatic change in the shift. To illustrate the situa-
FIG. 2. Simulated spectra for the ␯ (eЈ) mode of cyclic C at three different
levels of theory using a 6-31G* basis set.
4
6
tion for cyclic C , we simulated isotopic spectra for the
6
␯4
(eЈ) mode at three levels of theory, HF, second-order
Furthermore, we note that the simulated spectra are relatively
stable as we go beyond the HF level as anticipated from the
behavior of the sensitivity index in Table II and the discus-
many-body perturbation theory ͓MBPT͑2͔͒, and DFT-
B3LYP, using the same 6-31G* basis set. These spectra
assume a ratio of concentrations between C and C iso-
13
12
1
3
12
sion above. The isotopic spectrum of C is thus expected to
6
topes C: Cϵcϭ1/9, and have been scaled so that the main
band corresponds to the experimental value obtained in this
work. From Table II we find that we should expect three
strong bands corresponding to the singly substituted species,
with the fourth overlapping the main band. Each singly sub-
stituted band should have an intensity of 3ϫcϫI where 2
ϫI is the intensity of the ͑degenerate͒ main band. The rela-
tive intensity of a singly substituted species and the main
band is thus
be relatively well described by theory with a distinct position
and intensity pattern for the singly substituted isomers.
III. EXPERIMENT
Matrix samples were prepared by vaporizing the surface
of carbon rods using a slightly focused Nd:YAG ͑DCR-11,
Spectra-Physics͒ pulsed laser operating at 532 nm. The
evaporated species were condensed in solid Ar ͑99.999%
pure͒ on a gold plated surface cooled to ϳ10 K by a closed
cycle refrigerator ͑APD Cryogenics, Displex͒ in a vacuum of
3
ϫcϫI
ϭ1/͓͑2/3c͒ϩ1͔.
2
ϫIϫ3ϫcϫI
Ϫ7
As can be seen in Fig. 2, one of the singly substituted
bands is persistently stronger than the others in all three cal-
culations. This is due to an overlap with one of the compo-
nents of the doubly substituted isotopomer containing dia-
10 Torr or better during the experiment.
The carbon rods were fabricated by mixing carbon-12
͑Alfa, 99.9999% purity͒ and carbon-13 ͑Isotec, 99.3% pu-
rity͒ powder in various ratios and then pressing the mixtures
into rods. A degassing procedure was used to remove any
gas trapped in the rods by heating them to ϳ200 °C under
vacuum for 24 h.
metrically positioned 13C atoms, i.e., combining the 9 cm
Ϫ1
Ϫ1
␤
-site shift with the ϳ0 cm ␣-site shift. This should be an
13
observable effect for large enough concentrations of C.
J. Chem. Phys., Vol. 107, No. 16, 22 October 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
138.38.151.151 On: Sun, 23 Nov 2014 16:01:57

6035
Wang, Rittby, and Graham: Detection of cyclic C6
During evaporation, the carbon rod was continuously ro-
tated and translated to provide a fresh target surface for the
laser beam. A filter was used to prevent unvaporized large
particles sputtered off the rod surface from reaching the
deposition site. After the evaporation, controlled, thermally
induced diffusion in the matrix was used to enable the
growth of large clusters by warming the matrix to a specific
temperature and maintaining it there for ϳ10 min before
cooling back to ϳ10 K.
FTIR absorption spectra of the matrix samples were re-
Ϫ1
corded in the region of 550–3900 cm at a resolution of
Ϫ1
0
.2 cm using a Bomem DA 3.16 Fourier transform spec-
trometer equipped with a liquid-nitrogen-cooled MCT ͑Hg–
Cd–Te͒ detector and KBr beamsplitter. Details of the optical
26
system have been reported previously. All frequencies re-
Ϫ1
ported were measured to Ϯ0.1 cm
.
IV. RESULTS AND DISCUSSION
The most intense infrared band for the cyclic C struc-
6
ture with D3h symmetry is predicted to be the doubly degen-
erate ␯ (eЈ) mode with frequency in the region of
4
Ϫ1 18,19
1
700 cm
.
Four isotopomer bands are possible on
1
3
single C-substitution with one of them predicted to have
zero shift and therefore, to underlie the absorption for the
unsubstituted isotopomer. As discussed earlier ͑see Sec. II͒,
the intensity of each of the three observable singly substi-
FIG. 3. ͑a͒ FTIR spectrum showing singly substituted isotopomer absorp-
1
3
tions of cyclic C produced by the laser evaporation of a C-enriched rod
6
with measured 13C: Cϭ7.5% in the products. ͑b͒ A simulated spectrum
based on the predictions of the B3LYP/cc-pVTZ DFT calculations from the
present study.
12
1
3
tuted C isotopomer absorptions relative to the main absorp-
13
12
tion should then be 1/͓(2/3c)ϩ1͔, where c is the C: C
isotopic ratio for the sample used. This result differs from the
situation for linear carbon clusters where the relative inten-
sity of the singly substituted 13C isotopomers is 2c for clus-
ters with an even number of atoms, and 2c or c for the ‘‘odd
crepancy is believed to originate with the different laser
13
12
evaporation characteristics for the C and C powders ow-
ing to differences in average particle size.
13
number’’ clusters. For example, if the C originates with the
The only absorption found in our experiments which is
1
% natural abundance of the 13C isotope, the relative inten-
consistent with the predicted behavior of the ␯ (eЈ) mode is
4
Ϫ1
sity of singly substituted bands for cyclic C will be
the absorption located at 1694.9 cm and shown in Fig.
6
ϳ1.5%, and increasing the 13C abundance to ϳ7% through
enrichment of the carbon rod will lead to ϳ10% relative
intensity, a readily observed difference from the intensities
expected for a linear chain.
3͑a͒. This band appears strongly in spectra recorded for
13
samples prepared using rods with only ϳ1% natural
C
C
abundance, although the intensities of the three single 13
isotopomer bands are too weak for accurate measurements.
The situation was drastically improved by the addition of 7%
In our experiments, all absorptions observed in the
Ϫ1
13
1
600–1900 cm region have been considered as potential
to 8% C in the rods during fabrication. The assignment of
Ϫ1
candidates for the ␯ (eЈ) vibration. Each band has been thor-
the 1694.9 cm band to the ␯ (eЈ) mode is based on very
4
4
oughly investigated by comparing its observed frequency,
and the frequencies and relative intensities of its isotopic
shift bands with the predictions of the available DFT calcu-
lations.
good agreement between the experimental values and the
13
DFT calculated values for the ␯ mode’s frequency, C iso-
4
topic shift frequencies, and relative intensities. A second eЈ
Ϫ1
mode is predicted at 645 cm ; however, its low calculated
Instead of relying on the nominal 13C: C ratio in the
12
intensity and the decline in the MCT detector’s sensitivity in
this region made a definite identification impossible.
mixture used to fabricate the rod, the relative intensities for
1
3
Ϫ1
the singly substituted C isotopomer bands of cyclic C are
Figures 3͑a͒ and 4 show the absorption at 1694.9 cm
6
based on the actual 13C: C ratio in the product species as
12
and other bands in the immediate vicinity. The spectra were
produced by the evaporation of a 12C/ C rod for 15 and 20
13
determined from the intensities of the well-known isoto-
5
13 12
pomer bands of linear C and C . The reason for this pro-
min, respectively, with measured C: C ratio in the prod-
3
6
1
3
12
cedure is that the actual C: C ratio in the molecular clus-
ters formed in the matrix can differ significantly from the
nominal ratio in the rod, especially when a relatively high
C: C ratio is used. For C: C ratios of less than 15%,
however, the effect is found not to be significant. The dis-
ucts of 7.5%. The difference between the two spectra can be
attributed in part to different evaporation times but perhaps
more importantly, to the greater compactness or rigidity of
the rod used to produce the spectrum in Fig. 4, owing to
improvements in the rod fabrication technique.
1
3
12
13 12
J. Chem. Phys., Vol. 107, No. 16, 22 October 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
138.38.151.151 On: Sun, 23 Nov 2014 16:01:57

6036
Wang, Rittby, and Graham: Detection of cyclic C6
length of time spent at each temperature. Indeed, the main
Ϫ1
absorption at 1694.9 cm in Fig. 3͑a͒, which was originally
Ϫ1
weak and broad (ϳ10 cm ) before annealing, increased in
Ϫ1
intensity by a factor у5 and narrowed to less than 3 cm
after multiple annealing cycles. The band in Fig. 4, although
originally strong, was very broad, but grew in intensity and
sharpened after only one cycle at the relatively low, 20 K
annealing temperature.
In Figs. 3͑a͒ and 4, additional bands to the low-
Ϫ1
frequency side of the main band at 1694.9 cm can also be
clearly seen. While some of these bands are related to the
main band and are enhanced by 13C enrichment, others are
not. The most prominent of the secondary bands, those lo-
Ϫ1
cated at 1663.9, 1685.9, and 1690.5 cm ͓See Fig. 3͑a͔͒,
which show the same behavior during annealing and for ex-
periments with different 13C: C ratios, are clearly related.
12
Ϫ1
Their shifts of 31.0, 9.0, and 4.4 cm , respectively, enable
FIG. 4. FTIR spectrum showing both single and double substituted isoto-
them to be easily assigned to the three, singly substituted,
13
pomer absorptions of cyclic C produced by the laser evaporation of a
6
C isotopomers ͑B͒, ͑D͒, and ͑E͒, which as shown in Table
13
C-enriched rod with measured 7.5% enrichment in the products.
Ϫ1
III, have predicted shifts of 28.8, 9.0, and 4.7 cm , respec-
tively. ͑The labeling scheme identifying the isotopomers of
cyclic C is explained in Table III.͒ The intensities of the
three bands relative to the 1694.9 cm absorption are also
6
Prior to recording the spectra shown in Figs. 3͑a͒ and 4
Ϫ1
the matrix samples were subjected to successive annealing
cycles of warming and quenching to ϳ10 K. The spectrum
in Fig. 3͑a͒, was observed after the sample had been an-
nealed in cycles to 20, 25, 30, 33, and 37 K with 10 min
spent at each of the successively higher temperatures; how-
ever, the spectrum in Fig. 4 resulted from only one annealing
cycle to 20 K. Usually, annealing increases the intensity and
sharpness of bands for large carbon clusters. Furthermore,
these effects are enhanced by increasing the number of an-
nealing cycles at temperatures lower than ϳ40 K and the
consistent with this assignment, averaging 9.5% compared to
1
0% predicted by the formula 1/͓(2/3c)ϩ1͔, discussed ear-
lier. Moreover, as predicted earlier ͑see Sec. II͒, the ͑D͒ iso-
topomer band is consistently observed to be slightly stronger.
The very good agreement between the observed spectrum
and the results of the DFT calculations is further illustrated
by Fig. 3͑b͒ which shows a simulated spectrum generated
13
12
from the B3LYP/cc-pVTZ calculation assuming C: C
ϭ1:9. Comparison of the observed spectrum in Fig. 3͑a͒
TABLE III. Comparison between isotopic shifts (cmϪ1) observed for the ␯4 mode of (D3h) cyclic C and those
6
predicted by DFT calculations.
B3LYP/
B3LYP/
Isotopomer
Observed
cc-pVDZa
cc-pVTZb
C -C -C -C -C -C
␤
Freq.
Shift
Scaled shift
Scaled shift
␣
␤
␣
␤
␣
1
1
2-12-12-12-12-12
3-12-12-12-12-12
͑A͒
͑B͒
1694.9
1663.9
--
--
31.0
--
9.0
4.4
--
28.7
0.0
9.1
4.5
--
28.8
0.0
9.0
4.7
͑
C͒
͑D͒
E͒
͑F͒
G͒
͑H͒
I͒
͑J͒
12-13-12-12-12-12
13-13-12-12-12-12
13-12-13-12-12-12
12-13-12-13-12-12
13-12-12-13-12-12
1685.9c
1690.5
1657.4
overlappedd
1652.4
1681.9
1679.4
1682.7
1658.7
overlappedc
͑
37.5
overlapped
42.5
36.6
5.5
36.6
5.7
͑
44.8
13.9
16.2
11.4
34.0
9.1
44.7
14.0
16.2
11.7
35.4
9.0
͑
13.0
15.5
12.2
36.2
͑K͒
͑L͒
͑M͒
overlapped
aValues scaled by a factorϭ0.9583 obtained from the ratio of the measured and predicted frequencies at the
B3LYP/cc-pVDZ level by Martin and Taylor ͑Ref. 18͒.
b
Values scaled by a factorϭ0.9638 obtained from the ratio of the measured and predicted frequencies at the
B3LYP/cc-pVTZ level in the present work.
Absorptions of the ͑D͒ and ͑M͒ isotopomers overlap.
Absorption ͑G͒ is overlapped by the 1687.6 cm absorption.
c
d
Ϫ1
J. Chem. Phys., Vol. 107, No. 16, 22 October 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
138.38.151.151 On: Sun, 23 Nov 2014 16:01:57

6037
Wang, Rittby, and Graham: Detection of cyclic C6
with the simulated spectrum shows a good match for the
dominant singly C-substituted bands.
sured for the first time based on the excellent agreement of
the frequency, isotopic shifts, and their relative intensities
1
3
Additional, weaker features in the simulated spectrum in
Fig. 3͑b͒ belong to the doubly substituted isotopomers. Cor-
responding absorptions observed in the FTIR spectrum are
clearly observable in the spectrum of Fig. 4 at 1652.4 ͑H͒,
with the predictions from DFT calculations. Cyclic C is the
first cyclic carbon cluster so far discovered spectroscopically.
6
ACKNOWLEDGMENTS
1
1
657.4 ͑F͒, 1658.7 ͑L͒, 1679.4 ͑J͒, 1681.9 ͑I͒, and
682.7 cm ͑K͒. Under a variety of conditions they also
Grants from the Welch Foundation ͑Graham, No.
P-0786; Rittby, No. P-1259͒ and the TCU Research Fund in
support of this research, and from the W. M. Keck Founda-
tion for the purchase of the Bomem spectrometer, are grate-
fully acknowledged. The authors are grateful to Mike Mur-
dock for his skill and advice in the construction of the laser
evaporation and rod fabrication systems.
Ϫ1
exhibit intensity behavior which is similar to that for the
three singly substituted isotopomer bands. As shown in
Table III, these weaker absorptions are readily assigned to
isotopomers with double 13C substitutions. The experimental
frequency values are in reasonably good agreement with the
DFT predicted values despite the fact that some of the iso-
topomer bands overlap. For example, a band was not as-
signed to the ͑M͒ isotopomer because it is overlapped by the
singly substituted 13C isotopomer ͑D͒ band. Similarly, an
1
W. Weltner, Jr. and R. J. Van Zee, Chem. Rev. 89, 1713 ͑1989͒.
J. M. L. Martin, J. P. Franc c¸ ois, and R. J. Gijbels, Mol. Struct. 294, 21
2
͑1993͒.
3
K. R. Thompson, R. L. DeKock, and W. Weltner, Jr., J. Am. Chem. Soc.
assignment for ͑G͒ was not made because it is overlapped by
a band at 1687.6 cm
Ϫ1
93, 4688 ͑1971͒.
.
4
M. Vala, T. M. Chandrasekhar, J. Szcepanski, and R. Pellow, High Temp.
Additional bands appearing at 1667.4, 1671.1, 1675.9,
Sci. 27, 19 ͑1990͒.
R. H. Kranze and W. R. M. Graham, J. Chem. Phys. 98, 71 ͑1993͒.
D. W. Arnold, S. E. Bradforth, T. N. Kitsopoulos, and D. M. Neumark, J.
Ϫ1
5
and 1687.6 cm in Fig. 4 are not, however, related to the
1
the evidence of their different behavior during annealing and
on the correlation of their intensities with other bands to the
high-frequency side of the 1694.9 cm absorption. The in-
6
Ϫ1
694.9 cm absorption. This conclusion is based both on
Chem. Phys. 86, 5212 ͑1986͒.
J. Kurtz and D. P. Huffman, J. Chem. Phys. 92, 30 ͑1990͒.
H. J. Hwang, A. Van Orden, K. Tanaka, E. W. Kuo, J. R. Heath, and R. J.
7
8
Ϫ1
Saykally, Mol. Phys. 79, 769 ͑1993͒.
T. F. Giesen, A. Van Orden, H. J. Hwang, R. S. Fellers, R. A. Proven c¸ al,
9
Ϫ1
tensity of the band at 1687.6 cm decreases slightly on an-
nealing while the intensities of the 1694.9 cm band and its
and R. J. Saykally, Science 265, 756 ͑1994͒.
K. Raghavachari, R. A. Whiteside, and J. A. Pople, J. Chem. Phys. 85,
Ϫ1
10
associated isotopic shift bands all increase. Although appar-
ently not a member of the 1694.9 cm band group, the
6623 ͑1986͒.
Ϫ1
11
K. Raghavachari and J. S. Binkley, J. Chem. Phys. 87, 2191 ͑1987͒.
1
1
2
3
Ϫ1
V. Parasuk and J. Alml o¨ f, J. Chem. Phys. 91, 1137 ͑1989͒.
J. Hutter, and H. P. L u¨ thi, J. Am. Chem. Soc. 101, 2213 ͑1994͒.
J. Hutter, H. P. L u¨ thi, and F. Diederich, J. Am. Chem. Soc. 116, 750
͑1994͒.
J. Szcepanski, and M. J. Vala, J. Phys. Chem. 95, 2792 ͑1991͒.
C. M. L. Rittby ͑unpublished͒.
R. H. Kranze ͑unpublished͒.
J. M. L. Martin and P. R. Taylor, J. Phys. Chem. 100, 6047 ͑1996͒; J. M.
L. Martin ͑private communication͒.
1
687.6 cm band remains unidentified. The intensities of
Ϫ1
the bands at 1667.4, 1671.1, and 1675.9 cm seem corre-
lated to other moderately intense bands located to the high-
frequency side of the 1694.9 cm absorption ͑e.g., 1705.7,
14
Ϫ1
15
1
1
6
7
Ϫ1
1
710.6, and 1714.2 cm ͒, but not shown in both Figs. 3͑a͒
and 4, and may belong to unidentified carbon clusters
trapped in the matrix.
18
19
J. M. L. Martin, J. El-Yazal, and J. P. Fran c¸ ois, Chem. Phys. Lett. 242,
Table III, comparing the observed frequencies and cor-
responding shifts with the values calculated at two different
levels of DFT calculations, summarizes the proposed assign-
5
70–579 ͑1995͒.
20
W. Kr a¨ tschmer and K. Nachtigall, in Polycyclic Aromatic Hydrocarbons
and Astrophysics, edited by A. Leger et al. ͑Reidel, Dordrecht, 1987͒, p.
´
13
ments. It can be seen that all of the single C shifts and
75.
2
2
1
2
¯
almost all of the doubly 13C substituted isotopic shifts, ex-
Z. Slanina, J. M. Rudzi n´ ski, and E. Osawa, Z. Phys. D 19, 431 ͑1991͒.
R. H. Kranze, P. A. Withey, C. M. L. Rittby, and W. R. M. Graham, J.
Chem. Phys. 103, 6841 ͑1995͒.
R. H. Kranze, C. M. L. Rittby, and W. R. M. Graham, J. Chem. Phys. 105,
5313 ͑1996͒.
cept for one singly and two doubly substituted shifts which
are overlapped by other bands, can be unambiguously as-
signed to isotopomers of cyclic C (D ).
2
2
3
4
6
3h
D. Han, C. M. L. Rittby, and W. R. M. Graham, J. Chem. Phys. 106, 6222
͑
1997͒.
V. CONCLUSION
25
M. J. Frisch, G. W. Trucks, H. B. Schlegel, P. M. W. Gill, B. G. Johnson,
M. A. Robb, J. R. Cheeseman, T. Keith, G. A. Petersson, J. A. Montgom-
ery, K. Raghavachari, M. A. Al-Laham, V. G. Zakrzewski, J. V. Ortiz, J.
B. Foresman, J. Cioslowski, B. B. Stefanov, A. Nanayakkara, M. Challa-
combe, C. Y. Peng, P. Y. Ayala, W. Chen, M. W. Wong, J. L. Andres, E.
S. Replogle, R. Gomperts, R. L. Martin, D. J. Fox, J. S. Binkley, D. J.
Defrees, J. Baker, J. P. Stewart, M. Head-Gordon, C. Gonzalez, and J. A.
Pople, GAUSSIAN 94, Revision D.2, Gaussian, Inc., Pittsburgh, PA, 1995.
L. N. Shen, T. J. Doyle, and W. R. M. Graham, J. Phys. Chem. 93, 1597
͑1990͒.
FTIR measurements in an Ar matrix combined with pre-
dictions from a new B3LYP/cc-pVTZ DFT calculation per-
formed in the present work, and previous B3LYP/cc-pVDZ
calculations by Martin and Taylor,18 have resulted in the first
spectroscopic identification of cyclic C . The most intense
6
Ϫ1
infrared active mode, ␯ (eЈ)ϭ1694.9 cm of the cyclic C6
26
4
structure with D3h symmetry, has been assigned and mea-
J. Chem. Phys., Vol. 107, No. 16, 22 October 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
138.38.151.151 On: Sun, 23 Nov 2014 16:01:57