Benzdiynes (1,2,4,5-tetradehydrobenzenes): Direct observation by wavelength-selective photolyses of benzenetetracarboxylic dianhydrides in low-temperature nitrogen matrixes

Article abstract of DOI:10.1021/ja017566n

Toward direct observation of benzdiynes, we investigated the wavelength-selective photolyses of five kinds of benzenetetracarboxylic dianhydride derivatives in nitrogen matrixes at 13 K. In the first step of the photolyses, all dianhydrides were converted into benzynedicarboxylic anhydrides with loss of CO and CO₂ upon irradiation at 308 nm. In the second step, the benzyne intermediates were photolyzed at 266 nm. In these photolyses, the generation of two kinds of benzdiynes, 3,6-difluoro-1,4-benzdiyne and 3,6-bis(trifluoromethyl)-1,4-benzdiyne, was confirmed by good correspondence between observed and calculated IR spectra. These benzdiynes were converted into the corresponding hexatriynes upon further irradiation at 266 nm. Benzdiynes were not observed in the photolyses of the other three dianhydrides: only hexatriynes were observed as major photoproducts. These results suggested that benzdiynes were generated first and then converted into hexatriynes and that the efficiency of the decomposition of benzdiynes depended on the substituents. The dynamics of the generation and decomposition of benzdiynes in the matrixes was analyzed by using a successive reaction scheme.

Full text of DOI:10.1021/ja017566n

Published on Web 03/30/2002
Benzdiynes (1,2,4,5-Tetradehydrobenzenes): Direct
Observation by Wavelength-Selective Photolyses of
Benzenetetracarboxylic Dianhydrides in Low-Temperature
Nitrogen Matrixes
Tadatake Sato,* Sundaram Arulmozhiraja, Hiroyuki Niino, Shigekuni Sasaki,†
Tohru Matsuura,† and Akira Yabe*
Contribution from the Photoreaction Control Research Center, National Institute of AdVanced
Industrial Science and Technology (AIST), 1-1-1 Higashi, Tsukuba, Ibaraki 305-8565, Japan
Received November 19, 2001
Abstract: Toward direct observation of benzdiynes, we investigated the wavelength-selective photolyses
of five kinds of benzenetetracarboxylic dianhydride derivatives in nitrogen matrixes at 13 K. In the first step
of the photolyses, all dianhydrides were converted into benzynedicarboxylic anhydrides with loss of CO
and CO2 upon irradiation at 308 nm. In the second step, the benzyne intermediates were photolyzed at
266 nm. In these photolyses, the generation of two kinds of benzdiynes, 3,6-difluoro-1,4-benzdiyne and
3,6-bis(trifluoromethyl)-1,4-benzdiyne, was confirmed by good correspondence between observed and
calculated IR spectra. These benzdiynes were converted into the corresponding hexatriynes upon further
irradiation at 266 nm. Benzdiynes were not observed in the photolyses of the other three dianhydrides:
only hexatriynes were observed as major photoproducts. These results suggested that benzdiynes were
generated first and then converted into hexatriynes and that the efficiency of the decomposition of benzdiynes
depended on the substituents. The dynamics of the generation and decomposition of benzdiynes in the
matrixes was analyzed by using a successive reaction scheme.
the pyrolyses of dianhydrides,5-7 direct spectroscopic proof of
their existence was obtained only recently. During the interven-
ing years, studies on benzdiynes were limited to theoretical
studies.8-10
1. Introduction
The highly reactive intermediate o-benzyne (didehydroben-
zene, C6H4) participates in a variety of organic chemical
reactions and is among the most widely studied reactive
intermediates in organic chemistry.1,2 Its high reactivity origi-
nates from the strained triple bond in the benzene ring.
Benzdiynes (tetradehydrobenzenes), on the other hand, are
further unsaturated C6H2 chemical species having two triple
bonds in the benzene ring. In contrast to the number of studies
on o-benzyne, the number of studies on benzdiynes is small,
probably because of their extraordinary lability owing to the
high ring strain that arises from the two triple bonds. Because
of their inherent instability, the benzdiynes are among the most
challenging reactive intermediates to study. Although the
existence of the benzdiynes was inferred as early as 1966 from
the results of trapping experiments3,4 and product analyses of
In 1997, we reported that the photolysis of 3,6-bis(trifluo-
romethyl)-1,2;4,5-benzenetetracarboxylic dianhydride in an
argon matrix gave 3,6-bis(trifluoromethyl)-1,4-benzdiyne (1a).11
This was the first direct experimental proof of the existence of
a free benzdiyne structure as an intermediate in organic
reactions. However, we could not detect the generation of
unsubstituted benzdiyne (1b) under similar experimental condi-
tions.12,13 Why were we unable to observe the unsubstituted
benzdiynes?
(4) Hart, H.; Raju, N.; Meador, M. A.; Ward, D. L. J. Org. Chem. 1983, 48,
4357-4360.
(5) Fields, E. K.; Meyerson, S. J. Org. Chem. 1966, 31, 3307-3309.
(6) Fields, E. K.; Meyerson, S. AdV. Phys. Org. Chem., 1968, 6, 1-61.
(7) Fields, E. K. In Organic ReactiVe Intermediates; McManus, S. P., Ed.;
Academic Press: New York, 1973; p 449.
* To whom correspondence should be addressed. Fax: +81-298-61-
4560. E-mail: sato-tadatake@aist.go.jp, akira.yabe@aist.go.jp.
† NTT Advanced Technology Corporation, Nihon-Seimei Musashino
Building, 1-16-10, Naka-cho, Musashino-shi, Tokyo 180-0006, Japan.
(1) For instance: (a) Hoffmann, R. W. Dehydrobenzene and Cycloalkynes;
Academic Press: New York, 1967. (b) Levin, R. H. In ReactiVe Intermedi-
ates; Jones, M., Moss, R. A., Eds.; John Wiley & Sons: New York, 1978;
Vol. 1, Chapter 1.
(2) Radziszewski, J. G.; Hess, B. A., Jr.; Zahradnik, R. J. Am. Chem. Soc.
1992, 114, 52-57 and references therein.
(3) Hart, H.; Lai, C.-Y.; Nwokogu, G.; Shamouilian, S.; Teuerstein, A.;
Zlotogorski, C. J. Am. Chem. Soc. 1980, 102, 6649-6651.
(8) Adam, W.; Grimison, A.; Hoffmann, R. J. Am. Chem. Soc. 1969, 91, 2590-
2599.
(9) Radom, L.; Nobes, R. H.; Underwood, D. J.; Li, W.-K. Pure Appl. Chem.
1986, 58, 75-88.
(10) Zahradnˆık, R.; Hobza, P.; Burcl, R.; Hess, J. B. A.; Radziszewski, J. G. J.
Mol. Struct. (THEOCHEM) 1994, 313, 335-349.
(11) Moriyama, M.; Ohana, T.; Yabe, A. J. Am. Chem. Soc. 1997, 119, 10229-
10230.
(12) Moriyama, M.; Ohana, T.; Yabe, A. Chem. Lett. 1995, 557-558.
(13) Moriyama, M.; Sato, T.; Uchimaru, T.; Yabe, A. Phys. Chem. Chem. Phys.
1999, 1, 2267-2274.
9
4512
J. AM. CHEM. SOC. 2002, 124, 4512-4521
10.1021/ja017566n CCC: $22.00 © 2002 American Chemical Society

Direct Observation of Benzdiynes
A R T I C L E S
intermediates. Although we previously tried to detect 1a and
1b by the wavelength-selective photolyses of dianhydrides, we
could detect only 1a. The concentrations of these benzdiynes
during the course of the wavelength-selective photolyses may
be different owing to substituent effects. Substituent effects
induced by kinetic factors or thermodynamic factors are possible.
In the former case, the decomposition of benzdiynes by
rearrangement may be suppressed, since a CF3 group is more
reluctant to migrate than a light H atom.20 In the latter case,
the electron-withdrawing effect of the CF3 group may thermo-
dynamically stabilize the benzdiyne with high electron density
in an aromatic ring.
Against this backdrop, we investigated in detail the wavelength-
selective photolyses of five kinds of dianhydrides with different
substituents. In addition to the dianhydrides previously studied
(2a, 2b, and 2b-d), two kinds of anhydrides (2c and 2d) were
newly examined. The F group21 (2d) and CF3 group (2c) have
often been used as substituents reluctant to migration.22,23 The
aims of the present study is two-fold: (1) to observe directly
new benzdiyne derivatives and (2) to elucidate the factors
affecting the possible direct observation of benzdiynes.
In 1999, Bettinger, Schleyer, and Schaefer reported high-level
ab initio and density functional theory (DFT) computational
results for benzdiynes including 1a and 1b.14 They argued that
the IR band due to the asymmetrically coupled CtC stretching
mode in 1b should be observable since this band had a predicted
intensity of 8 km‚mol-1 at the CCSD(T)/TZ2P level of
computation. In addition to this band, three IR bands with
intensities > 30 km‚mol-1 (619 cm-1, 157 km‚mol-1; 826 cm-1
,
33 km‚mol-1; 956 cm-1, 35 km‚mol-1) were also predicted by
this computation.14 While the detection of a CtC stretching
band is an important topics in matrix-isolation studies on
arynes,2 the existence of intense IR bands is rather essential for
the direct spectroscopic observation of 1b.15 Indeed, IR bands
whose intensities were predicted to be >0.1 km‚mol-1 were
observed for o-benzyne in a 6 K neon matrix by Radziszewski
et al.2 We confirmed the generation of 1-naphthyne in an 11 K
argon matrix by detecting most of the IR bands with intensities
>5 km‚mol-1 with our experimental setup.16 Given these
experimental results, direct observation of 1b, that is, detection
of IR bands with intensities >30 km‚mol-1, should be possible.
However, these bands could not be detected in our experiments.
Thus, there were likely other factors affecting the direct
observation.
2. Experimental Section
2.1. Materials. Dianhydrides 2b and 2b-d (97.8 atom % D) were
purchased from Tokyo Kasei Co. and C/D/N Isotopes, respectively.
The other dianhydrides were synthesized at NTT. The synthetic
procedures for 2a and 2c are reported elsewhere.24 Precursor 2d was
synthesized from 3,6-difluoro-1,2,4,5-tetracyanobenzene (SDS Biotech
Co. Ltd.) as follows. The difluorotetracyanobenzene was hydrolyzed
to the tetracarboxylic acid by an aqueous H2SO4 solution at 150 °C for
5 h and then converted into the dianhydride by treatment with acetic
anhydride at 130 °C for 2 h. Dianhydride 2d25 was obtained with a
yield of 27%.
All five dianhydrides were purified by sublimation before matrix
isolation experiments.
2.2. Matrix Isolation Experiments. Matrix isolation experiments
were performed with a closed-cycle helium cryostat (Air Products
Displex CS-202). The pressure in the sample chamber was kept at 10-4
It should be noted that direct observation of benzdiyne
depends on the concentration of benzdiynes in the matrix as
well as the intensities of the IR bands. When benzdiynes are
formed by stepwise decarboxylation and decarbonylation of
benzenetetracarboxylic dianhydrides, several kinds of reactive
intermediates that partially lose CO, CO2, or both must be
formed in the matrix.12,13,17 Wavelength-selective irradiation is
useful for controlling such consecutive photochemical
reactions.12,13,16-19 By careful choice of wavelength for the
photolyses, photochemical reactions that would proceed simul-
taneously under broadband irradiation can be promoted con-
secutively. Consequently, the concentration of specific inter-
mediates can be increased. In addition, wavelength-selective
irradiation with high-intensity lasers is effective in driving
photochemical reactions to completion, which facilitates analy-
ses of IR spectra containing IR bands of several kinds of
(20) Evans, R. A.; Wentrup, C. J. Chem. Soc., Chem. Commun. 1992, 1062-
1063.
(21) In determining the substituent effect of an F group, we have to consider
both inductive and resonance effects. Basically, on one hand the F group
(14) Bettinger, H. F.; Schleyer, P. v. R.; Schaefer, H. F. J. Am. Chem. Soc.
1999, 121, 2829-2835.
works as a σ-electron-withdrawing substituent because of the high
electronegativity of the F atom (Langenaeker, W.; Proft, F. D.; Geerlings,
P. J. Phys. Chem. A. 1998, 102, 5944-5950; Jones, G. B.; Warner, P. M.
J. Am. Chem. Soc. 2001, 123, 2134-3145). On the other hand, the F group
often works as an electron-donating group owing to the resonance effect
(Bibas, H.; Wong, M. W.; Wentrup, C. Chem. Eur. J. 1997, 3, 237-248).
(22) Breidung, J.; Bu¨rger, H.; Ko¨tting, C.; Kopitzky, R.; Sander, W.; Senzlober,
M.; Thiel, W.; Willner, H. Angew. Chem., Int. Ed. Engl. 1997, 36, 1983-
1985.
(15) Currently, we can identify the reactive intermediates generated in the matrix
on the basis of the correspondence between the observed and calculated
IR spectra, in particular, those calculated by the DFT method. The DFT
method is known to be one of the most useful computational methods for
predicting the vibrational spectra of various molecules.(Scott, A. P.; Radom,
L. J. Phys. Chem. 1996, 100, 16502-16513., Wong, M. W. Chem. Phys.
Lett. 1996, 256, 391-399., Kudo, S.; Takayanagi, M.; Nakata, M. Chem.
Phys. Lett. 2000, 322, 363-370.). In this case, the identification is done
by the comparing the whole spectral pattern. Therefore, detection of the
IR bands ascribed to a specific functional group such as the CtC bond
stretching mode in o-benzyne is no longer essential for the identification.
Instead, the existence of several groups of intense IR bands is indispensable
for identifying the intermediates.
(23) O’Gara, J. E.; Dailey, W. P. J. Am. Chem. Soc. 1992, 114, 3581-3590.
(24) Matsuura, T.; Ishizawa, M.; Hasuda, Y.; Nishi, S. Macromolecules 1992,
25, 3540-3545.
(25) Spectroscopic data of 4d are as follows: 13C NMR (DMSO) δ 163, 2 (Cd
O), 153.0, 150.4 (C-F, 1JCF ) 261 Hz) 124.7, 124.8, 125.0 (2JCF ) 8-15
Hz); FTIR (v/cm-1
) observed in an nitrogen matrix and predicted
(16) Sato, T.; Moriyama, M.; Niino, H.; Yabe, A. Chem. Commun. 1999, 1089-
frequencies (cm-1) and intensities (km‚mol-1): - (595, 7), 614w (607,
5), 753w (725, 7), 763m (736, 51), - (768, 4), 918s (897, 386), 940m
(921, 119), 1186s, 1208w (1163, 724), 1312m (1287, 47), 1380w (1369,
13), 1508m (1479, 220), 1797vs, 1813m (1808, 1041), 1828m, 1834m,
1861m, 1873m, 1888m (1852, 413); MS m/z Observed 253.972 (M+),
Calculated 253.966.
1090.
(17) Moriyama, M.; Yabe, A. Chem. Lett. 1998, 337-338.
(18) Simon, J. G. G.; Mu¨nzel, N.; Schweig, A. Chem. Phys. Lett. 1990, 170,
187-192.
(19) Sato, T.; Niino, H.; Yabe, A. J. Phys. Chem. A 2001, 105, 7790-7798.
9
J. AM. CHEM. SOC. VOL. 124, NO. 16, 2002 4513

A R T I C L E S
Sato et al.
to 10-5 Pa during the experiments. A CsI plate cooled to 13 K was
used as a substrate, on which vaporized precursors were codeposited
with nitrogen (99.9999%) at 11 K. The sample chamber of the cryostat
had two pairs of windows: a pair of KBr windows for FTIR
measurements and a pair of quartz windows for UV-vis measurements.
With this sample chamber, the photoreaction in the matrix could be
followed by FTIR and UV-vis spectroscopies simultaneously. The
FTIR measurements were carried out on a Perkin-Elmer Spectrum-
GXI spectrometer with a resolution of 1 cm-1. UV-vis absorption
spectra were measured with a Shimadzu UV-3100 spectrometer.
Wavelength-selective photoirradiation was carried out with the
following nanosecond-pulsed lasers: (1) a XeCl excimer laser (308
nm: Lambda Physik Compex-102), whose repetition rate and laser
fluence were 5 Hz and 3-4 mJ‚cm-2‚pulse-1, respectively; and (2)
fourth harmonic generated (FHG) pulses of a Nd:YAG laser (266 nm:
Lotis LS-2125 with a YHG-34 accessory), whose repetition rate and
laser fluence were 10 Hz and ca. 3 mJ‚cm-2‚pulse-1, respectively. The
fluence of the irradiation was kept constant in each batch.
2.3. Computational Methods. All DFT calculations were performed
with the Gaussian 98 program package.26 The geometries of the
compounds were optimized by using the B3LYP method27,28 in
combination with the 6-31G* basis set. The nature of the stationary
points was assessed by means of vibrational frequency analysis.
Theoretical IR spectra were obtained by vibrational frequency analysis.
For the benzdiynes, additional calculations using the cc-pVTZ basis
set were executed for both the geometry optimization and the frequency
analysis. A symmetry-broken spin-unrestricted method was adopted
for the calculation of the benzdiynes since unstable solutions were
obtained by the restricted method.29,30 Vibrational frequencies predicted
at the B3LYP/6-31G* level were scaled by 0.9614 on the basis of
literature,31 whereas vibrational frequencies predicted at the B3LYP/
cc-pVTZ level were used without scaling. All calculations were done
on the IBM RS/6000-SP system at the Tsukuba Advanced Computing
Center (TACC).
Figure 1. UV-vis absorption spectra of 2a upon irradiation with a XeCl
excimer laser (λ ) 308 nm): (a) 0, (b) 600, (c) 3000, (d) 6000, (e) 12 000,
(f) 24 000, and (g) 60 000 pulses.
anhydride is converted into the corresponding aryne by stepwise
decarboxylation and decarbonylation. For instance, in the
wavelength-selective photolysis of naphthalenedicarboxylic
anhydride,19 a cyclopropanone intermediate was initially formed
by decarboxylation of the anhydride, and then decarbonylation
of the cyclopropanone intermediate produced naphthyne. In the
photolysis of phthalic anhydride to o-benzyne, Radziszewski
et al. confirmed temporal formation of a cyclopropanone
intermediate by the characteristic IR band at 1852 cm-1
,
although decarboxylation and decarbonylation proceeded si-
multaneously.2 In our previous report on the photolysis of 2a,11
we tentatively assigned the IR band at 1858 cm-1 to cyclopro-
panone intermediate 3a.
3. Results and Discussion
3.1. Photolyses of Dianhydrides at 308 nm. Five kinds of
dianhydrides were employed as precursors of benzdiynes. As
the first step of our wavelength-selective photolyses, these
precursors were photolyzed with a XeCl excimer laser (λ )
308 nm). Results for the photolysis of 2a are described first.
Precursor 2a was sublimed at 85-90 °C, and codeposited with
nitrogen on the cold substrate at 13 K, and then photolyzed.
The photoreactions induced by irradiation were followed by
FTIR and UV-vis spectroscopies. Upon irradiation, the IR
bands of 2a decreased and were gradually replaced by a set of
new IR bands. The appearance of IR bands due to CO2 (662
and 2347 cm-1) and CO (2140 cm-1) indicated decarboxylation
and decarbonylation of 2a. In general, an arenedicarboxylic
Participation of 3a was suggested in the UV-vis absorption
spectra (Figure 1). At the beginning of the irradiation, distinct
absorption bands appeared at 303 and 292 nm (Figure 1b). These
bands decreased gradually with irradiation, and finally disap-
peared. After prolonged irradiation, the UV-vis absorption
spectra reached a photostationary state, where only structureless
absorption was observed (Figure 1g). These results revealed that
3a was temporally formed and then converted into 4a.
The dynamics of the photochemical reactions from 2a to 4a
could be studied by analyzing the intensities of the IR bands
due to CO and CO2. The concentration of CO2 generated in the
matrix must correspond to the total concentration of the
photoproducts, in this case, 3a and 4a. On the other hand, the
concentration of CO must be equal to that of 4a. Therefore, a
(26) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M.
A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.; Stratmann,
R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin,
K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi,
R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.;
Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.;
Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz,
J. V.; Baboul, A. G.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.;
Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham,
M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.;
Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Andres, J. L.;
Gonzalez, C.; Head-Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian
98, Revision A.9; Gaussian, Inc.: Pittsburgh, PA, 1998.
(27) Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5662.
(28) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785-789.
(29) Arulmozhiraja, S.; Sato, T.; Yabe, A. J. Comput. Chem. 2001, 22, 923-
930.
(30) Arulmozhiraja, S.; Sato, T.; Yabe, A. Submitted for publication.
(31) Scott, A. P.; Radom, L. J. Phys. Chem. 1996, 100, 16502-16513.
9
4514 J. AM. CHEM. SOC. VOL. 124, NO. 16, 2002

Direct Observation of Benzdiynes
A R T I C L E S
Figure 2. Change in the scaled integrated intensities of the IR bands due
to CO2 (circles) and CO (triangles) upon irradiation with a XeCl excimer
laser (λ ) 308 nm). Intensities are relative to the photostationary state
(intensity at the photostationary state ) 1), which was reached after
irradiation with 60 000 pulses.
difference between the concentrations of CO2 and CO indicates
the existence of 3a. Since 2a was completely converted into 4a
at the photostationary state, the concentrations of CO and CO2
at this state must be equal to the concentration of 4a. Accord-
ingly, the molar ratios of CO and CO2 in the matrix could be
estimated by scaling their IR intensities to those at the
photostationary state (intensities at the photostationary state )
1). Figure 2 shows the molar ratios of CO2 and CO during
irradiation at 308 nm. At the beginning of the irradiation, the
molar ratio of CO2 was larger than that of CO, indicating
the existence of 3a.32 Continued irradiation resulted in disap-
pearance of the difference, indicating that 3a had been com-
pletely converted into 4a. The FTIR spectrum observed at the
photostationary state is shown in Figure 3. The observed
spectrum is in good agreement with the calculated IR spectrum
of 4a. No IR bands ascribed to other chemical species ap-
peared in this spectrum, indicating that there was no side
reactions. Thus, it was confirmed that more than 99% of the 2a
present initially was converted into 4a upon irradiation at 308
nm.
Figure 3. (a) FTIR absorption spectrum observed at the photostationary
state upon irradiation of 2a at 308 nm and (b) theoretical IR spectrum of
4a calculated at the B3LYP/6-31G* level.
3.2. Photolyses of Benzyne Intermediates at 266 nm. 3.2.1.
Photolysis of Bis(trifluoromethyl)benzynedicarboxylic An-
hydride. To avoid inducing multiple reactions during irradiation,
photolysis should be carried out under conditions as mild as
possible, that is, at a wavelength as long as possible. We chose
to use the FHG pulses of a Nd:YAG laser (266 nm) for the
successive photolyses of the benzyne intermediates. All of the
benzyne intermediates could be photolyzed at 266 nm, whereas
they were stable upon irradiation at 308 nm. We first describe
our experimental results for the photolysis of 4a at 266 nm.
Upon irradiation, we observed further increases in the intensities
of the IR bands due to CO and CO2, indicating decarboxylation
and decarbonylation of 4a. This photolysis is assumed to proceed
as follows:
The other dianhydrides (2b, 2b-d, 2c, 2d) behaved similarly
to 2a upon irradiation at 308 nm; that is, they were converted
into the corresponding benzyne intermediates. The FTIR spectra
observed at the photostationary state were depicted with
computational results in the Supporting Information (Figures
S5, S10, S15). These benzyne intermediates were identified on
the basis of good agreement between the observed and calculated
IR spectra. None of these intermediates showed further reaction
upon continued irradiation at 308 nm. Except for difluorinated
benzyne intermediate 4d, they showed structureless UV-vis
absorption spectra (Figure 1g). Figure 4 shows the UV-vis
absorption spectrum of 4d: prominent absorption bands were
observed at 294, 282, and 272 nm. To clarify the origin of these
UV-vis absorption bands, we examined the time-dependent
(TD) DFT calculations of 2d and 4d (Figure 4b). These
computational results showed that the prominent absorption
bands of 4d originated from the resonance effect of the fluorine
group (see Supporting Information).
The dynamic behavior of the intensities of the IR bands due
to CO and CO2 is shown in Figure 5. As in the case of 308 nm
photolysis, the molar ratios of CO and CO2 generated by 266
nm photolysis could be estimated by scaling their IR intensities
to those at the photostationary state for 308 nm irradiation. The
dynamics of the CO and CO2 bands was almost the same (Figure
5), indicating that an anhydride group was removed from 4a
upon irradiation at 266 nm and that the concentration of the
cyclopropanone intermediate 5a was negligible.
(32) The temporal formation and decomposition of 3a were also confirmed
by FTIR spectra. The FTIR spectrum of 3a being in good agreement with
the calculated result was depicted in the Supporting Information (Figure
S3).
9
J. AM. CHEM. SOC. VOL. 124, NO. 16, 2002 4515

A R T I C L E S
Sato et al.
behaviors: (1) IR bands that were observed upon initial
irradiation and then gradually decreased and (2) IR bands that
were not observed upon initial irradiation but increased over
the course of irradiation. The IR band at 1465 cm-1, which is
the characteristic IR band of 1a,11 belonged to class 1. In
addition, IR bands at 797, 1125, 1152, 1166, 1182, and 1239
cm-1 were class 1 bands, whereas IR bands at 779, 1141, 1167,
1173, 1206, 1227, 1358, 1381, and 2234 cm-1 were class 2
bands. These results revealed that one chemical species was
generated in the initial irradiation and then gradually converted
into another chemical species. The IR spectra corresponding to
these two chemical species can be extracted by applying the
weighted subtraction method to the IR spectra observed in the
initial and the last stages of the photolysis. Thus, the IR spectrum
for class 1 was in good agreement with the theoretical spectrum
of 1a, while the IR spectrum for class 2 was in good agreement
with the theoretical spectrum of 6a (Figure 7). At the UB3LYP/
cc-pVTZ level of computation,30 the asymmetrically coupled
CtC stretching mode in 1a was predicted to be at 1867 cm-1
with an intensity of 5 km‚mol-1. However, we were not able
to observe this weak IR band, probably because it is obscured
by CdO stretching bands of the remaining 4a. The IR band of
1a corresponding to the band predicted at 606 cm-1 overlapped
with the band of CO2 at 662 cm-1, which was confirmed from
the temporal increase and decrease of the line width of this IR
band during irradiation.
Figure 4. (a) UV-vis absorption spectrum of 4d observed after complete
photolysis of 2d at 308 nm and (b) TDDFT computational results for
4d.
These results indicated that 1a was initially formed by the
photolysis of 4a and then converted into 6a. Changes in
intensities of the IR bands ascribable to 4a, 1a, and 6a are
plotted in Figure 8. The dynamic behaviors of these species
revealed that 1a and 6a were formed from 4a following a
successive reaction scheme. A detailed analysis of the dynamic
behavior of these species is discussed later.
Figure 5. Dynamic behavior of the scaled integrated intensities of the IR
bands due to CO2 (circles, 0-72000 pulses) and CO (triangles), and
decomposition ratio of 4a (crosses) upon irradiation with FHG pulses of a
Nd:YAG laser (λ ) 266 nm). The integrated intensities of the CO and CO2
bands were scaled on the basis of those at the photostationary state upon
irradiation at 308 nm. The results for CO2 are shown only between 0 and
72 000 pulses: because the intensities were too high, the correct intensities
of the CO2 bands could not be estimated after 72 000 pulses.
Figure 6 shows the FTIR spectra of the photoproducts formed
upon irradiation at 266 nm. These spectra were obtained by
eliminating the IR bands of 4a from the FTIR spectra observed
over the course of the photolysis at 266 nm. This elimination
was done by a weighted subtraction using the FTIR spectrum
of 4a obtained at the photostationary state for 308 nm irradiation.
The scaling factor (SF) applied in the weighted subtraction
corresponds to the molar fraction of the surviving 4a at the
stage.33 Accordingly, a (1 - SF) value denotes the decomposi-
tion ratio of 4a (see Figure 5). The dynamics of the decomposi-
tion ratio almost corresponds to that of both CO and CO2,
indicating that the generated photoproducts were C6(CF3)2
species. The observed IR bands of the photoproducts could be
classified into two groups on the basis of their dynamic
In our previous study, the assignment of the IR bands in the
C-F stretching region was unclear because several chemical
species contributed to the large number of IR bands in this
region.11 In the present study, a more reliable assignment became
possible by the separation of the IR spectra of the coexisting
chemical species by the weighted subtraction method. The IR
bands at 1381, 1358, 1173, 1167, and 1141 cm-1 in Figure 7
correspond to those at 1377, 1354, 1176, 1169, and 1139 cm-1
in the argon matrix.11 Although these bands were tentatively
assigned to an intermediate species between 1a and 6a, they
were reassigned to 6a with the aid of DFT calculations.
(33) When the FTIR spectrum of the photoproducts was obtained by weighted
subtraction with a scaling factor (SF) of 0.30, 30% of 4a survived with the
remaining 70% being converted into photoproducts. Thus, SF corresponds
to the molar fraction of surviving 4a, while (1-SF) corresponds to the molar
fraction of decomposed 4a, or, equivalently, the total molar ratio of
photoproducts generated from 4a.
3.2.2. Photolysis of (Trifluoromethyl)benzynedicarboxylic
Anhydride. In contrast to 1a, the existence of (trifluoromethyl)-
benzdiyne 1c in the photolysis of benzyne intermediate 4c was
unclear, although the existence of intense IR bands (>200
9
4516 J. AM. CHEM. SOC. VOL. 124, NO. 16, 2002

Direct Observation of Benzdiynes
A R T I C L E S
Figure 6. FTIR absorption spectra of photoproducts formed by irradiation
of 4a with FHG pulses of a Nd:YAG laser (λ ) 266 nm): (a) 6000, (b)
27 000, (c) 108 000, and (d) 396 000 pulses. The decomposition ratios of
4a were estimated as (a) 5, (b) 18, (c) 51, and (d) 85%. IR bands of
remaining 4a were subtracted.
km‚mol-1) was predicted for 1c. As in the case of 1a, generation
of H-C6-CF3 species was suggested by the similar dynamic
behavior of the IR bands due to CO and CO2 and the
decomposition ratio of 4c in the photolysis at 266 nm (Figure
S6 in the Supporting Information). Figure 9 shows the IR spectra
of the photoproducts formed upon irradiation of 4c at 266 nm
together with the calculated IR spectrum of hexatriyne derivative
6c. As seen in Figure 9, the IR bands of the photoproducts were
mainly ascribed to 6c on the basis of the agreement between
the observed and calculated IR spectra. The dynamic behavior
of the IR bands ascribed to 6c was similar to that of the
decomposition ratio of 4c. Therefore, the major photoproduct
in the photolysis of 4c was identified as 6c.
Figure 7. (a) FTIR absorption spectrum of the species for class 1 and
theoretical IR spectrum of 1a calculated at the UB3LYP/cc-pVTZ level.
(b) FTIR absorption spectrum of the species for class 2 and theoretical
spectrum of 6a at the B3LYP/6-31G* level.
3.2.3. Photolysis of Benzynedicarboxylic Anhydride and
Its Deuterated Isotopomer. In a previous study,13 only two
IR bands due to photoproducts were observed in the 248 nm
photolysis of unsubstituted 2b or dideuterated dianhydride 2b-
d. These bands were assigned to the C-H(D) bending and
stretching modes of the -CtC-H(D) moiety. In the present
study, benzyne intermediates 4b and 4b-d were photolyzed at
266 nm. Upon irradiation, the IR bands of the -CtC-H(D)
moiety appeared at 642 and 3319 cm-1 4b and at 504 and 2589
cm-1 4b-d. No IR bands ascribable to other photoproducts were
observed (Figure S11 in the Supporting Information). Consider-
ing that hexatriynes 6a, 6c, and 6d were the final products in
the 266 nm photolyses of other systems, these IR bands should
be ascribed to hexatriynes. From DFT calculation, these bands
are predicted at 555 and 3358 cm-1 6b and at 428 and 2609
cm-1 6b-d. There are no other intense IR bands predicted for
6b and 6b-d. Thus, the obtained photoproducts were ascribed
to 6b and 6b-d, respectively on the basis of the good cor-
respondence between the observed and calculated IR spectra.
The dynamic behavior of the C-H(D) stretching band corre-
sponds well to the dynamic behavior of the IR band due to CO
Figure 8. Dynamic behavior of the scaled intensities of the IR bands of
4a (circles), 1a (triangles), and 6a (squares). The IR intensities were scaled
on the basis of the parameters estimated in the curve fitting. Curves are
fitted results to eqs 1, 2, and 3.
and that of the decomposition ratio of 4b (4b-d) (Figure S12 in
the Supporting Information). These results indicated that the
hexatriynes were the major photoproducts in the photolyses.
3.2.4. Photolysis of Difluorobenzynedicarboxylic Anhy-
dride: Direct Observation of Difluorobenzdiyne. In contrast
to the other benzyne intermediates, 4d showed prominent UV-
9
J. AM. CHEM. SOC. VOL. 124, NO. 16, 2002 4517

A R T I C L E S
Sato et al.
the case for 1a, ring opening of 1d upon irradiation at 266 nm
led to the formation of 6d. In addition to the IR bands ascribed
to 1d and 6d, broad IR bands were observed around 2100 cm-1
(Figure 11). Corresponding IR bands were not observed or were
very weak in the photolyses of other benzyne intermediates.
Bands in this wavenumber region are typical for ketene species.
Careful analysis of these bands revealed that there were two IR
bands with different dynamic behaviors: a band at 2109 cm-1
was replaced by a band at 2129 cm-1 during the photolysis,
indicating that two ketene species were involved in the
photolysis. The dynamic behaviors of these two IR bands
indicated that the first ketene species (k-A) was formed directly
from 4d, and that the second species (k-B) was formed from
1d or k-A. The dynamic behaviors of these IR bands were
compared with those of IR bands due to 1d, 6d, CO and CO2
and with the decomposition ratio of 4d. From the dynamics,
we concluded that the amount of k-A in the photolysis was
not negligible and that k-B was a minor product (Figure S19
in the Supporting Information). The reaction scheme for the
photolysis of 4d at 266 nm can be depicted as follows. The
present experimental results suggested that neither ketene is
analogous to cyclopentadienylideneketene, which is formed from
benzyne and CO. Studies directed at identifying these ketenes
are in progress, and the details of these studies will be reported
elsewhere.
Figure 9. (a) Theoretical IR spectrum of 6c and observed FTIR absorption
spectra of photoproducts formed by irradiation of 4c with FHG pulses of a
Nd:YAG laser (λ ) 266 nm): (b) 18000, (c) 36000, (d) 78000, (e) 144000,
and (f) 432000 pulses. The decomposition ratios of 4c were estimated as
(b) 22%, (c) 36%, (d) 62%, (e) 74%, and (f) 93%. IR bands of remaining
4c were subtracted.
3.3. Calculated Energies for Benzynes, Benzdiynes, and
Hexatriynes. Although the reactions from benzynes to hexatriynes
are photoinduced processes, the calculated energies of the
species involved in the reactions may be helpful for interpreting
the different photochemical behaviors observed for the five kinds
of derivatives. Figure 12 compares the energies of 2a, 4a, and
6a with the energy of 1a. These energies were calculated at the
B3LYP/6-31G* level and were corrected with scaled zero-point
vibrational energies.31 The energy difference (∆1) between
vis absorption bands. Hence, 4d was more efficiently photolyzed
at 266 nm than the other benzyne intermediates. The FTIR
spectra of the photoproducts obtained upon irradiation at 266
nm are shown in Figure 10. In the initial photolysis, major IR
bands were observed at 579, 929, and 1439 cm-1 (Figure 10a).
The predicted IR spectrum of 1d was simple compared to that
of 1a because of the simple chemical structure of 1d. Therefore,
the observed IR bands were ascribed to 1d with certainty on
the basis of good agreement between observed and calculated
spectra (Figure 11). In addition to these bands, a weak absorption
band at 1174 cm-1 was also assigned to 1d on the basis of the
dynamic behavior of the IR bands; this band was also predicted
in the theoretical spectrum. Thus the generation of 1d was
confirmed. This is the second benzdiyne derivative that has been
directly observed. Although the asymmetrically coupled CtC
stretching mode in 1d was predicted to be at 1529 cm-1 with
an intensity of 0.5 km‚mol-1 (UB3LYP/cc-pVTZ calculation),
this IR band could not be observed experimentally, probably
because of very low intensity. The IR bands ascribed to 1d
decreased upon continued irradiation at 266 nm, while the IR
bands at 830, 875, 1424, and 2322 cm-1 increased. These IR
bands are in good agreement with those in the calculated IR
spectrum of 6d (Figure 11). Thus, we confirmed that, as was
benzyne 4a and benzdiyne 1a was estimated as 85 kcal‚mol-1
:
this is a rather large value compared to the energy difference
between 2a and 4a (55 kcalmol-1). This large energy difference
(destabilization) and the ring strain in the benzene ring make
benzdiyne 1a highly unstable and susceptible to decomposition
by ring opening. The corresponding energy differences for the
other derivatives are summarized in Table 1.
In all cases, ∆1 was estimated as 85-95 kcal‚mol-1
,
indicating that all of the benzdiynes are equally unstable. No
remarkable difference was observed between the derivatives in
which benzdiynes were detected (1a and 1d) and those in which
benzdiynes were not observed (1b and 1c). On the other hand,
the energy difference between 1d and 6d (∆2) was rather small
(26 kcal‚mol-1) compared to the energy differences for the other
three cases. This small difference would come from the
9
4518 J. AM. CHEM. SOC. VOL. 124, NO. 16, 2002

Direct Observation of Benzdiynes
A R T I C L E S
Figure 10. FTIR absorption spectra of photoproducts formed by irradiation
of 4d with FHG pulses of a Nd:YAG laser (λ ) 266 nm): (a) 1200, (b)
4800, (c) 9000, (d) 15,000, and (e) 36,000 pulses. The decomposition ratios
of 4d were estimated as (a) 33%, (b) 69%, (c) 84%, (d) 93%, and (e) 99%.
IR bands of remaining 4d were subtracted. Peaks labeled by (k) are ascribed
to ketene species.
resonance effect of the F atoms. In the other three cases, energy
differences are similar, indicating that ∆2 would not affect direct
observation of benzdiynes. In a separate report, we will discuss
the geometries of benzdiyne derivatives and their relative
energies on the basis of a more detailed computational study.30
3.4. Analysis of Reaction Dynamics. In the present study,
five kinds of benzyne intermediates (4a, 4b, 4b-d, 4c, and 4d)
were obtained by complete photolysis at 308 nm followed by
photolysis at 266 nm. Photoreactions at 308 nm (conversion of
dianhydrides to benzyne intermediates) were similar, whereas
photoreactions at 266 nm were different. In the 266 nm
photolyses, only two benzdiynes, 1a and 1d, were directly
observed. These benzdiynes decomposed upon prolonged ir-
radiation, and hexatriynes were generated as the final products.
These results indicated that benzdiyne could be decomposed
into hexatriyne by photoinduced ring-opening upon irradiation
at 266 nm. Hexatriynes were also the final products in the other
three cases. These results indicated that, in all five cases,
benzdiyne was generated first and then decomposed into
hexatriyne, and that the efficiencies of the decompositions of
the benzdiynes depended on the nature of the substituents. From
this viewpoint, we analyzed the reaction dynamics of photo-
conversion of benzyne to hexatriyne.
Figure 11. (a) Observed and calculated (UB3LYP/cc-pVTZ) IR spectra
of 1d and (b) observed and calculated (B3LYP/6-31G*) IR spectra of 6d.
Peaks labeled by (k) are ascribed to ketene species.
photolyses could be analyzed as a successive reaction sequence
under constant photoirradiation. We analyzed the reaction
dynamics of benzyne, benzdiyne, and hexatriyne on the basis
of the successive reaction scheme depicted below.
[A] ) [A]0 exp(-k1t)
(1)
(2)
In the photoconversion of 4a to 6a, benzyne intermediate 4a
was first converted into the benzdiyne 1a and then into
hexatriyne 6a. The dynamics of their IR band intensities (Figure
8) was typical of the dynamics in a successive reaction. The
reaction of 4a to 1a and the reaction of 1a to 6a are not
elementary reactions; rather each of these reactions involves
several elementary reactions. Nonetheless, the observed dynam-
ics suggested that the reaction dynamics of the 266 nm
k1
[B] )
{exp(-k1t) - exp(-k2t)}[A]0
k2 - k1
k2
k1
[C] ) 1 +
exp( -k1t) -
exp(-k2t) [A]0
{
}
k1 - k2
k1 - k2
(3)
[A], [B], and [C] denote the concentrations of benzyne,
9
J. AM. CHEM. SOC. VOL. 124, NO. 16, 2002 4519

A R T I C L E S
Sato et al.
apparent successive reaction, in which the contributions of other
chemical species are negligible. These results indicate that the
intermediates in the matrix were in a near-homogeneous
environment and that each reaction was unimolecular in nature.
In this context, the estimated k1 and k2 values represent the total
efficiencies of the reaction of 4a to 1a and the reaction of 1a to
6a. From the curve for 1a, the maximum yield of 1a in the
matrix was estimated to be 47%. Because the contribution of
ketene species was not negligible, the reaction dynamics for
4d could not be analyzed with eqs 1-3.
In the photolyses of 4b, 4b-d, and 4c, we found no evidence
for chemical species other than those shown in the above
successive reaction scheme, although we did not confirm
formation of benzdiynes. For these cases, we attempted to use
eqs 1 and 3 to analyze the reaction dynamics of the photocon-
version of benzynes to hexatriynes (Figures S7 and S13 in the
Supporting Information). For 4b, 4b-d, and 4c, the k1 values
were estimated to be 4.7 × 10-6, 9.0 × 10-6, and 9.3 × 10-6
pulse-1, respectively. Next, k2 values were estimated from the
dynamic behavior of the IR band of hexatriyne and eq 3. For
4b-d and 4c, the k2 values were estimated to be 1.0 × 10-2 and
7.7 × 10-5 pulse-1, respectively. The obtained dynamics showed
that the maximum yields of benzdiynes 1b-d and 1c were 0.2
and 8.8%, respectively. On the other hand, because k2 . k1,
the least-squares fitting did not converge in the case of 4b.
Therefore, the maximum yield of 1b would be lower than 0.2%.
These results show that it is essentially impossible to detect 1b
and 1b-d in the photolysis of 2b and 2b-d, respectively, even if
they have intense IR bands. Meanwhile, 1c should be formed
with the maximum yield of 8.8%. When the yield of 1c was
near the maximum, a very weak IR band was observed at 1445
cm-1. This band was ascribed to the characteristic ring deforma-
tion mode of the benzdiyne structure (see Supporting Informa-
tion (Figure S8)).
Figure 12. Energy diagram of some of the species involved in the formation
of benzyne, benzdiyne, and hexatriyne.
Table 1. Energy Differences between Benzynes and Benzdiynes
(∆1) and between Benzdiynes and Hexatriynes (∆2) Estimated at
the B3LYP/6-31G* Level.
compounds
∆1 (kcal‚mol-1
)
∆2 (kcal‚mol-1
)
benzdiyne
4a - 1a - 6a
4b - 1b - 6b
4c - 1c - 6c
4d - 1d - 6d
85
95
91
91
56
53
51
26
observed
not observed
not observed
observed
benzdiyne, and hexatriyne within the matrix, respectively. [A]0
denotes the initial concentration of the benzyne intermediate.
The apparent rate constants k1 and k2 could be estimated by
using eqs 1-3. k1 and k2 are not true rate constants: their values
depend on the concentration of intermediates, the wavelength
and fluence of irradiation, and other experimental conditions.
Moreover, it must be noted that all the reactions involved in
the generation and decomposition of the benzdiynes are
photochemical reactions. Therefore, they should be significantly
affected by the absorption coefficients of the benzynes and
benzdiynes at the wavelength of laser irradiation. However, none
of the benzynes and benzdiynes, except for 4d,34 showed
prominent absorption bands similar to those observed for 4a
(Figure 1g). In addition, the changes in absorption spectra during
the 266 nm photolyses of 4a, 4b, and 4c were slight. Thus, the
differences in the absorption coefficients of the benzdiynes were
not a major cause of the different photochemical behaviors of
these systems. This conclusion was supported by the results of
TDDFT calculations (see Supporting Information). IR bands of
4a, 1a, and 6a were analyzed with eqs 1, 2, and 3, respectively,
by nonlinear least-squares fitting. First, with eq 1, the k1 value
was estimated to be 8.6 × 10-6 pulse-1 from the decay curve
of 4a. Then, with eq 2 the k2 value was estimated to be 5.0 ×
10-6 pulse-1 from the dynamics of 1a. The dynamics of 6a
fitted with eq 3 also gave the same value for k2. The curves
thus obtained fit the experimental values well (Figure 8).
Therefore, the reaction from 4a to 6a can be expressed as an
As described above, the k1 and k2 values reflect the total
efficiencies of the reaction of 4a to 1a and the reaction of 1a to
6a. The maximum yield of benzdiyne is high, when the
efficiency of benzdiyne decomposition (k2) is low compared to
that of benzdiyne generation (k1). Therefore, the k2/k1 ratio may
be used to judge whether benzdiynes could be directly observed.
The k2/k1 ratios for 1a, 1b-d, and 1c were estimated to be 0.59,
1.1 × 103, and 8.2, respectively. The corresponding ratio for
1b would be larger than 1.1 × 103. The k2/k1 ratio for 1d could
not be obtained because of a complicated reaction scheme.
However, considering that the IR spectrum of 1d was clearly
observed, we can deduce that the k2/k1 ratio for 1d must be
low. Thus, a small k2/k1 ratio was estimated for disubstituted
species compared to the species containing H(D) atoms,
implying that the migration of an H(D) atom plays an important
role in the increase of the apparent rate constant for the ring-
opening reaction. That is to say, the substitution by CF3 or F
groups results in the higher concentration of benzdiynes because
the rearrangement and ring-opening reactions are suppressed.
4. Conclusions
We carried out wavelength-selective photolyses of five kinds
of benzenetetracarboxylic dianhydride derivatives in the nitrogen
matrixes at 13 K. In these photolyses, all dianhydrides were
converted into benzyne intermediates upon irradiation at 308
nm. These benzyne intermediates could be further photolyzed
(34) In the photolysis of 4d, the generated ketenes showed prominent absorption
bands that obstructed the absorption spectrum of benzdiyne.
9
4520 J. AM. CHEM. SOC. VOL. 124, NO. 16, 2002

Direct Observation of Benzdiynes
A R T I C L E S
at 266 nm. In the 266 nm photolyses, the generation of only
two kinds of benzdiynes, 3,6-bis(trifluoromethyl)-1,4-benzdiyne
(1a) and 3,6-difluoro-1,4-benzdiyne (1d), was confirmed on the
basis of good correspondence between the observed and
calculated IR spectra. Benzdiyne 1d is the second trapped
benzdiyne to be directly observed after the first, 1a. Both 1a
and 1d were converted into hexatriynes upon prolonged
irradiation. In the other three cases, only hexatriynes were
observed as major photoproducts. These results indicated that
the generated benzdiynes could be photolyzed into hexatriynes
upon irradiation at 266 nm. The photoconversion of benzyne
to hexatriyne could be analyzed by assuming a successive
reaction scheme. The ratio of the efficiency for benzdiyne
decomposition to that for benzdiyne generation was shown to
be useful for judging whether benzdiynes could be directly
observed spectroscopically. This ratio was affected by the nature
of the substituents. For 1a, this ratio was estimated to be 0.59,
which corresponds to a maximum yield of 47%. The migration
of a light H(D) atom plays an important role in increasing the
of efficiency of the ring-opening reaction. Suppressing the ring-
opening reaction by replacing the H(D) atom with the F or CF3
group made the direct observation of benzdiynes possible.
Acknowledgment. We are grateful to Dr. Masaya Moriyama
(The University of Tokyo) and Dr. Hans P. Reisenauer (Justus-
Liebig University) for helpful discussions.
Supporting Information Available: Details of the photolyses
of all precursors and computational results (Geometry, calculated
and observed frequencies, energy diagram for intermediates, and
vibrational modes of benzdiynes) (PDF). This material is
available free of charge via the Internet at http://pubs.acs.org.
JA017566N
9
J. AM. CHEM. SOC. VOL. 124, NO. 16, 2002 4521