Characterization of a didehydrobenzoxazine intermediate in a novel carbene-to-carbene transformation

Article abstract of DOI:10.1021/ja026300t

Irradiation of N₂ matrix-isolated 3-chloro-3-(2-benzoxazolyl)diazirine gives a mixture of syn- and anti-benzoxazolylchlorocarbenes which could be characterized by IR, UV/vis, and B3LYP modeling. Subsequent irradiation of the carbene caused ring opening to the corresponding quinoimine, which was similarly characterized. In turn, the quinoimine photochemically underwent ring-closure to a novel, highly strained cyclic ketenimine. Unrestricted singlet DFT calculations were required to fit the IR spectrum of the ketenimine, suggesting significant diradical character. Photolysis of the ketenimine led to ring cleavage in yet another fashion, to give an isonitrile-phenoxychlorocarbene, whose spectra were similar to those of the previously characterized phenoxychlorocarbene. Finally, at shorter wavelengths, the phenoxycarbene underwent the expected rearrangement to the corresponding benzoyl chloride. Copyright

Full text of DOI:10.1021/ja026300t

Published on Web 06/11/2002
Characterization of a Didehydrobenzoxazine Intermediate in a Novel
Carbene-to-Carbene Transformation
Asya Nikitina and Robert S. Sheridan*
Department of Chemistry, Mail Stop 216, UniVersity of NeVada, Reno, NeVada 89557
Received March 24, 2002
Scheme 1
Cyclohexa-1,2,4-triene (1), or “isobenzene”, has long fascinated
experimentalists and theoreticians.1 Despite considerable strain (1
lies ca. 75 kcal/mol above benzene1a), 1 and its derivatives have
been postulated as intermediates in thermal electrocyclic ring-
closure of hexa-1,3-dien-5-ynes,1,2 in Diels-Alder reactions of
alkynes with vinyl acetylenes,3 and in a variety of related syntheti-
cally useful transformations in heterocyclic4 and polycyclic5
systems. Our group has recently reported an unusual entry to such
a species, which could be characterized spectroscopically.6 Specif-
ically, we discovered a facile photochemical interconversion be-
tween the benzofurylcarbene 2 and the highly strained didehydro-
benzopyran 4, mediated by the intercession of ring-opened quino-
methide 3.
Even on brief 334 nm irradiations of diazirine 5, the carbene
was accompanied by quinoimine 7, signified by strong 2125 and
1660 cm-1 IR bands. The similarity between the experimental
spectrum and that predicted by DFT calculations9 confirmed this
assignment. Continued irradiation at 334 nm, or more effectively
at 350 nm, converted the initially formed 6 to 7 completely.
Quinoimine 7 also absorbed strongly in the UV at ca. λmax 375
nm, overlapping absorption of 6, but could be distinguished from
the carbene by its long tail extending well into the visible. Ring-
opening of the carbene was reversible; irradiation of 7 at 436 nm
(or more slowly at 578 nm) gave back 6. Carbene 6 and quinoimine
7 could be recycled several times by alternating 350 and 436 nm
irradiations. However, subsequent products could be observed to
grow through this process, at the expense of 6 and 7 (vide infra).
Irradiation of either 6 or 7 at 404 nm caused slow transformation
to a new intermediate, whose IR spectrum was dominated by a
strong band at 1121 cm-1. Simultaneously, a broad and weak visible
absorption centered at 575 nm appeared. The similarity of the IR
and UV/vis spectra to those of 4,6 together with the mode of
generation, suggested that this new photoproduct was cyclic
ketenimine 8. However, the IR predicted by restricted B3LYP/6-
31G** calculations for 8 bore only marginal similarity to the
experimental spectrum.9 Theory suggests that cumulenes constrained
to small rings may have considerable diradical character,1 which
in this case (i.e. 8a) might compromise restricted calculations on
the ketenimine. Indeed, we found that a broken-spin symmetry
unrestricted singlet B3LYP calculation9,10 led to a lower energy
optimized structure for 8. Moreover, the IR spectrum predicted by
the unrestricted singlet calculation fit the experiment quite well
(Figure 1).
We were attracted by the possibility that the corresponding
benzoxazolylcarbene system might afford the corresponding strained
ketenimine. We found, however, that the presence of nitrogen not
only changed the photochemistry of this system significantly, but
also opened an unexpected new fragmentation channel, yielding
yet another carbene. We now report our observations on this unusual
system.
Our experimental observations are summarized in Scheme 1.
Photolysis of matrix-isolated diazirine 57 (ca. 1:800, N2, 10 K) at
366 nm, at the center of its nπ* absorption, produced a complex
IR spectrum that evolved over time and clearly reflected the
presence of several sequential products.8 On the other hand, 334
nm irradiations gave the expected carbene 6 relatively cleanly (but,
together with 7, vide infra). The experimental IR spectra fit
predictions by B3LYP/6-31G** calculations9 nicely for a mixture
of anti-6a and syn-6b conformations of the carbene. A very strong
UV band for the carbene at ca. 360 nm appeared simultaneously.
* Corresponding author. E-mail: rss@chem.unr.edu.
(1) (a) Engels, B.; Scho¨neboom, J. C.; Mu¨nster, A. F.; Groetsch, S.; Christl,
M. J. Am. Chem. Soc. 2002, 124, 287. (b) Johnson, R. P. Chem. ReV. 1989,
89, 1111.
(2) Prall, M.; Kru¨ger, A.; Schreiner, P. R.; Hopf, H. Chem. Eur. J. 2001, 7,
4386.
(3) Burrell, R. C.; Daoust, K. J.; Bradley, A. Z.; DiRico, K. J.; Johnson, R. P.
J. Am. Chem. Soc. 1996, 118, 4218.
(4) (a) Wills, M. S. B.; Danheiser, R. L. J. Am. Chem. Soc. 1998, 120, 9378.
(b) Frey, L. F.; Tillyer, R. D.; Ouellet, S. G.; Reamer, R. A.; Grabowski,
E. J. J.; Reider, P. R. J. Am. Chem. Soc. 2000, 122, 1215. (c) Kimball, D.
B.; Herges, R.; Haley, M. M. J. Am. Chem. Soc. 2002, 124, 1572.
(5) (a) Scott, L. T.; Chen, P.-C.; Hashemi, M. M.; Bratcher, M. S.; Meyer, D.
T.; Warren, H. B. J. Am. Chem. Soc. 1997, 119, 10963. (b) Zimmermann,
G. Eur. J. Org. Chem. 2001, 457.
(6) Khasanova, T.; Sheridan, R. S. J. Am. Chem. Soc. 2000, 122, 8585.
(7) Graham, W. J. J. Am. Chem. Soc. 1965, 87, 4396. See the Supporting
Information for synthetic details
(9) See the Supporting Information for calculational details and additional
experimental IR spectra. See refs 2 and 4c for DFT methodology applied
to similar intermediates. Restricted DFT methods have been shown to be
effective with singlet halo carbenes: Cramer, C. J.; Hillmyer, M. A. J.
Org. Chem. 1999, 64, 4850.
(8) For a description of the matrix-isolation instrumentation and techniques,
see: Rempala, P.; Sheridan, R. S. J. Chem. Soc., Perkin Trans. 2 1999,
2257.
9
7670
J. AM. CHEM. SOC. 2002, 124, 7670-7671
10.1021/ja026300t CCC: $22.00 © 2002 American Chemical Society

C O M M U N I C A T I O N S
characterization we reported a number of years ago.12,13 Carbene
11 showed similar strong absorption in the ca. 1250 cm-1 region,
which calculations attribute to stretching of the C-O partial double
bond in the carbenes. Moreover, both 9 and 11 rearrange to the
corresponding acid chlorides on photolysis.
Theory indicates that cumulenes constrained to small rings can
acquire diradical (8a) or zwitterionic (8b) character as the systems
become increasingly deformed.1 Recent high-level calculations by
Engels et al.1a and McKee et al.14 support the intuitive concept that
the amounts of zwitterionic and open-shell contributions to the
electronic structure of strained allenes are also sensitive to adjacent
hetereoatoms, as well as aromatic stabilization. MR-CI calculations
indicate that the didehydrobenzopyran 4 (without chloro substitu-
tion) should have a twisted allenic ground state, with zwitterionic
and singlet diradical planar excited states 5.4 and 9.6 kcal/mol
higher in energy, respectively.1a Consistent with these theoretical
predictions, our previous work demonstrated that the spectra of
didehydrobenzopyran 4 could be modeled satisfactorily with
restricted DFT calculations;6 all attempts to locate a lower energy
singlet diradical electronic state at the UB3LYP level failed. In
contrast, broken spin symmetry UB3LYP singlet calculations gave
an optimized structure for 8 that was 1.2 kcal/mol lower than that
predicted by restricted DFT, suggesting some diradical character
(i.e. 8a) in the ketenimine.10 Both methods, however, gave rather
similar geometries, somewhat twisted about the cumulenic bonds,
and not unlike that calculated for allene 4. The calculations indicate
that the CdCdN asymmetric stretch in 8 corresponds to the IR
band observed at 1558 cm-1, shifted considerably from that of
Figure 1. (a) Difference IR spectrum showing conversion of ketenimine
8 (“down bands”) to carbene 9 (“up bands”) on 366 nm irradiation. (b)
B3LYP/6-31G** calculated IR for carbene 9. (c) UB3LYP/6-31G**
calculated IR for ketenimine 8. Calculated frequencies are scaled by 0.97.
Although in our earlier work6 4 could be interconverted readily
with the corresponding carbene 2, photolysis of 8 at 366 nm gave
a different product that displayed a puzzling IR with prominent
2135 and ca. 1232 cm-1 bands (Figure 1). This same product could
be generated alternatively from matrices containing 6, 7, and/or 8
with >400 nm broadband irradiation. After considering and
calculationally modeling several possible products, we discovered
that the new bands arose from an unexpected fragmentation of 8;
the calculated IR of ring-opened phenoxycarbene 9 matched the
experimental spectrum very well (Figure 1). Corroborating this
assignment, irradiation of 9 at 313 nm gave the expected re-
arrangement product, acid chloride 10, signified by a strong band
at 1798 cm-1 (confirmed by modeling the IR spectrum with DFT
calculations).
We found previously that in the benzofuryl system, quinomethide
3 was particularly photolabile, and could only be formed in minor
amounts on irradiation of either 2 or 4.6 In contrast, the corre-
sponding quinoimine 7 could be produced cleanly. The dominant
product in the facile 6/7 equilibrium depended on the wavelength
of irradiation. Irradiation on the short-wavelength side of the
overlapping strong ca. 360 nm absorptions favored ring-opened 7.
Conversely, 7 absorbs more strongly at 436 nm, and thus rearranged
selectively to carbene 6 at this wavelength. Interestingly, comparison
with calculated spectra indicated that initial cyclization of 7
produced mainly syn-6b, as might be expected stereoelectronically.11
Longer exposure to 436 nm light slowly converted 6b to the anti
isomer 6a.
The alternate closure of 7 to ketenimine 8 was much less
efficient, but minor amounts of 8 and 9 were also formed during
the photochemical interconversions of carbene 6 and quinoimine
7. Irradiation at 404 nm, however, where both 6 and 7 absorb
strongly, drove the system slowly to 8. The electronic spectrum of
8, with weak broad absorption centered at 575 nm and strong
absorption at 360 nm, was similar to that observed for the
corresponding didehydropyran 4 (545 and 305 nm, respectively).6
The amount of 8 could be maximized with 404 nm light, at a
wavelength between these maxima. However, irradiation of 8 at
366 nm caused rapid ring opening, primarily to carbene/isonitrile
9, but also return to 7 to a minor extent.
unstrained ketenimines normally found at ca. 2000 cm-1 15
.
In summary, we have uncovered a novel transformation of a
benzoxazolyl carbene to a phenoxycarbene, by way of a highly
strained cyclic ketenimine 8. Ring opening of 8 to 9, formally a
double-bond cleavage to two carbenes, appears unlikely at first
glance. However, the pseudopericyclic16 nature of the fragmentation
is more apparent, perhaps, when contributions from 8b to the
electronic structure are considered. The need for unrestricted
calculations10 to model the IR spectrum of 8 may suggest the first
direct spectroscopic support for diradical character in such geo-
metrically constrained cumulenes. Higher level theory1a will be
necessary to confirm this point, however.
Acknowledgment. We thank the National Science Foundation
and the Petroleum Research Fund, administered by the American
Chemical Society, for financial support.
Supporting Information Available: Synthetic details for diazirine
5, calculation results for 6-10, and experimental/calculated IR for
interconversions of 6-10 (PDF). This material is available free of
charge via the Internet at http://pubs.acs.org.
JA026300T
The IR and UV spectra of carbene 9 are similar to those of
phenoxychlorocarbene (11), whose low-temperature generation and
(11) Khasanova, T.; Sheridan, R. S. J. Am. Chem. Soc. 1998, 120, 233.
(12) Kesselmayer, M. A.; Sheridan, R. S. J. Am. Chem. Soc. 1986, 108, 844.
(13) The splitting in the ca. 1230 cm-1 band of 9 may indicate the presence of
several conformations. However, the conformation shown in scheme 9 was
the only realistic energy minimum found calculationally (see the Supporting
Information), and fits the IR well overall.
(10) The similar energies predicted for 8 and 8a indicate that DFT cannot resolve
the exact nature of the observed intermediate (or even whether both 8 and
a planar 8a are present). The broken spin symmetry UB3LYP calculations
on 8 actually result in a mixed singlet-triplet diradicaloid state, with S2 )
0.58. For a discussion of this methodology, see for example: (a) Bally, T.;
Borden, W. T. ReV. Comput. Chem. 1999, 13, 1. (b) Freeman, P. K.; Pugh,
J. K. J. Org. Chem. 2001, 66, 5338.
(14) McKee, M. L.; Shevlin, P. B.; Zottola, M. J. Am. Chem. Soc. 2001, 123,
9418.
(15) Patai, S. The Chemistry of Ketenes, Allenes, and Related Compounds, Part
2; John Wiley & Sons: New York, 1980.
(16) Birney, D. M. J. Am. Chem. Soc. 2000, 122, 10917.
9
J. AM. CHEM. SOC. VOL. 124, NO. 26, 2002 7671