Role of conformation and electronic structure in the chemistry of ground and excited state o-pyrazolylphenylnitrenes

Article abstract of DOI:10.1021/ja043988w

The chemistry of 2-(1-pyrazolyl)- (2a) and 2-[1-(3,5-dimethylpyrazolyl] phenylnitrene (2b) has been studied in EtOH solution at room temperature, in EtOH glasses at 90 K, and in Ar matrices at 12 K. These nitrenes were chosen as suitable models for clarifying the mechanism of intramolecular reactions because attack at the pyrazole ring can occur according to different modes and the asymmetry of the substituent gives rise to different conformations. Detailed DFT and CASSCF/CASPT2 studies on the conformation and decay paths of both spin states of the nitrenes have been carried out. Ring expansion to dehydroazepines (via benzoazirines) is calculated to be competitive in both nitrenes, but in the dimethyl derivative, 2b, attack onto the N lone pair (which is made more nucleophilic by the methyl groups) is favored. The higher barriers (by 4-8 kcal/mol) in singlet 2a cause 60-70% of this nitrene to decay by intersystem crossing to the triplet. Thus, the seemingly straightforward formation of benzo-fused heterocycles through intramolecular attack of the pyrazoline N lone pair by the singlet phenylnitrene can only overcome ring expansion and intermolecular reactions under favorable circumstances. The comparatively persistent triplet nitrenes are characterized in matrices, and the yields of photocyclization products (mainly pyrazolo[1,5-a]benzimidazole (7) from 2a and 5,6-dihydropyrazolo[1,5-a]quinoxaline (8) from 2b) are shown to depend on the preferred conformation of the starting azide and nitrene.

Full text of DOI:10.1021/ja043988w

Published on Web 03/25/2005
Role of Conformation and Electronic Structure in the
Chemistry of Ground and Excited State
o-Pyrazolylphenylnitrenes
Claudio Carra,†,‡ Thomas Bally,*,† and Angelo Albini*,§
Contribution from the Department of Chemistry, UniVersity of Fribourg,
CH-1700 Fribourg, Switzerland, and Department of Chemistry, UniVersity of PaVia,
I-27100 PaVia, Italy
Received October 2, 2004; E-mail: Thomas.Bally@unifr.ch; Angelo.Albini@unipv.it
Abstract: The chemistry of 2-(1-pyrazolyl)- (2a) and 2-[1-(3,5-dimethylpyrazolyl]phenylnitrene (2b) has been
studied in EtOH solution at room temperature, in EtOH glasses at 90 K, and in Ar matrices at 12 K. These
nitrenes were chosen as suitable models for clarifying the mechanism of intramolecular reactions because
attack at the pyrazole ring can occur according to different modes and the asymmetry of the substituent
gives rise to different conformations. Detailed DFT and CASSCF/CASPT2 studies on the conformation
and decay paths of both spin states of the nitrenes have been carried out. Ring expansion to
dehydroazepines (via benzoazirines) is calculated to be competitive in both nitrenes, but in the dimethyl
derivative, 2b, attack onto the N lone pair (which is made more nucleophilic by the methyl groups) is favored.
The higher barriers (by 4-8 kcal/mol) in singlet 2a cause 60-70% of this nitrene to decay by intersystem
crossing to the triplet. Thus, the seemingly straightforward formation of benzo-fused heterocycles through
intramolecular attack of the pyrazoline N lone pair by the singlet phenylnitrene can only overcome ring
expansion and intermolecular reactions under favorable circumstances. The comparatively persistent triplet
nitrenes are characterized in matrices, and the yields of photocyclization products (mainly pyrazolo[1,5-
a]benzimidazole (7) from 2a and 5,6-dihydropyrazolo[1,5-a]quinoxaline (8) from 2b) are shown to depend
on the preferred conformation of the starting azide and nitrene.
Scheme 1
1. Introduction
Significant advances in the understanding of the chemistry
of arylnitrenes, usually generated by photolysis of aryl azides,
and the role of subsequent intermediates have been recently
achieved by time-resolved1-5 and matrix isolation experiments6-9
supplementing a large body of preceding studies carried out
over several decades.10-17
† University of Fribourg.
‡ Present address: Department of Chemistry, University of Ottawa,
Canada.
§ University of Pavia.
(1) Gritsan, N. P.; Yuzawa, T.; Platz, M. S. J. Am. Chem. Soc. 1997, 119,
5059.
(2) Gritsan, N. P.; Zhu, Z.; Hadad, C.; Platz, M. S. J. Am. Chem. Soc. 1999,
121, 1202.
From the synthetic point of view, arylnitrenes have been
proven disappointing16,18 by not undergoing the useful inter-
molecular insertion and addition reactions of the related
arylcarbenes. This is due to rapid rearrangements that prevent
intermolecular trapping16,19 and are usually followed by po-
lymerization, leading to “tars” (see Scheme 1). Thus, isolated
yields are often quite low, in particular at azide concentrations
(3) Gritsan, N. P.; Gudmundsdottir, A. D.; Tigelaar, D.; Zhu, Z.; Karney, W.
L.; Hadad, C. M.; Platz, M. S. J. Am. Chem. Soc. 2001, 123, 1951.
(4) Gritsan, N. P.; Likhotvorik, I.; Tsao, M. L.; Celebi, N.; Platz, M. S.; Karney,
W. L.; Kemnitz, C. R.; Borden, W. T. J. Am. Chem. Soc. 2001, 123, 1425.
(5) Tsao, M. L.; Gritsan, N.; James, T. R.; Platz, M. S.; Hrovat, D. A.; Borden,
W. T. J. Am. Chem. Soc. 2003, 125, 9343.
(6) Chapman, O. L.; LeRoux, J. P. J. Am. Chem. Soc. 1978, 100, 282.
(7) Hayes, J. C.; Sheridan, R. S. J. Am. Chem. Soc. 1990, 112, 5879.
(8) Tomioka, H.; Ichikawa, N.; Kamatsu, K. J. Am. Chem. Soc. 1993, 115,
8621.
(9) Maltsev, A.; Bally, T.; Tsao, M.-L.; Platz, M. S.; Kuhn, A.; Vosswinkel,
M.; Wentrup, C. J. Am. Chem. Soc. 2003, 126, 237.
(10) Smith, P. A. S. In Azides and Nitrenes: ReactiVity and Utility; Scriven, E.
F. V., Ed.; Academic Press: Orlando, FL, 1984; p 95.
(11) Wentrup, C. In ReactiVe Intermediates; Abramovitch, R. A., Ed.; Plenum
Press: New York, 1980; Vol. 1; p 263.
(12) Wentrup, C. AdV. Heterocycl. Chem. 1981, 28, 279.
(13) Wentrup, C. In ReactiVe Molecules; Wiley-Interscience: New York, 1984;
Chapter 4.
(14) Platz, M. S.; Leyva, E.; Haider, K. Org. Photochem. 1991, 11, 367.
(15) Schuster, G. B.; Platz, M. S. AdV. Photochem. 1992, 17, 69.
(16) Platz, M. S. Acc. Chem. Res. 1995, 28, 487.
(17) Platz, M. S. In ReactiVe Intermediate Chemistry; Moss, R. A., Platz, M.
S., Jones, M., Eds.; Wiley: New York, 2003.
(18) Doering, W. v. E.; Odum, R. A. Tetrahedron 1966, 22, 81.
(19) Borden, W. T.; Gritsan, N. T.; Hadad, C. M.; Karney, W. L.; Kemnitz, C.
R.; Platz, M. S. Acc. Chem. Res. 2000, 33, 765.
9
5552
J. AM. CHEM. SOC. 2005, 127, 5552-5562
10.1021/ja043988w CCC: $30.25 © 2005 American Chemical Society

Ground and Excited State o-Pyrazolylphenylnitrenes
A R T I C L E S
Scheme 2. Products Obtained after 254-nm Photolysis of Azides 1a and 1b in EtOH under Different Conditionsa
a The numbers for 1b are taken from ref 21. Note that in that paper the values for products 4b and 8b were erroneously interchanged.
that are high enough to make reactions preparatively useful.
Intramolecular trapping reactions are better suited both for
obtaining higher yields and for gathering information about the
chemistry of nitrenes.10
The primary products of phenyl azides (PA) are singlet
phenylnitrenes (1PN), which have been identified by flash
photolysis2-5,20 and trapped intermolecularly (in the case of 2,6-
difluoro-1PN)20 or intramoleculary by suitable nucleophiles.8,21
At room temperature 1PN equilibrates rapidly with the isomeric
dehydroazepines (DA) via the metastable benzazirines (BA).
These intermediates have also been identified spectroscopically
and through trapping by nucleophiles.18,22,23
At low temperature (below about 160 K for parent PN and
simple derivatives) intersystem crossing (ISC) to the more stable
triplet phenylnitrenes (3PN) overwhelms the activated decay
processes of the singlet nitrenes and, if diffusion can take place,
3PN dimerize to azobenzenes (AB).24 In solid matrices, photo-
excitation of 3PN yields DA6,7 (BA in the case of 2,5-difluoro-
PN25).
An investigation on the photolyis of pyrazolylphenyl azides
1 (Scheme 2)21,26 contributed to establishing the above picture.
The dimethylpyrazolylnitrene 2b was originally proposed as a
mechanistic probe by Suschitzky,27,28 who expected that in-
tramolecular reactions would allow one to distinguish the
electrophilic attack of the N sp2 lone pair by the singlet vs radical
attack of the methyl group by the triplet nitrene. We found that,
at room temperature, direct irradiation of 1 does indeed lead to
attack of the pyrazole nitrogen by the singlet nitrene, whereas
sensitized irradiation leads mainly to azo compounds, which
arise by dimerization of triplet nitrenes.21 In contrast to the
usually low yields of products from intermolecular trapping
reactions, the material balance was found to be satisfactory
(>70%) and thus provided confidence in assessing competing
pathways by analysis of product distributions. As with all phenyl
azides,14-17 ISC to 32b became the only process below 160 K,
as revealed by UV-vis spectroscopy at 77 K. Upon melting
the glass, the triplet nitrene dimerized, with very little hydrogen
abstraction from the methyl groups. Photolysis of the 160 K
glass resulted in attack at the methyl group, rather than
(20) Gritsan, N. P.; Zhai, H. B.; Yuzawa, T.; D, K.; Brooke, J.; Platz, M. S. J.
Phys. Chem. A 1997, 101, 2833.
3
conversion to dehydroazepine DA as with parent PN (cf. the
discussion in ref 24).
(21) Albini, A.; Bettinetti, G.; Minoli, G. J. Am. Chem. Soc. 1999, 121, 3104.
(22) DeGraff, B. A.; Gillespie, D. W.; Sundberg, R. J. J. Am. Chem. Soc. 1974,
96, 7491.
(23) Carroll, S. E.; Nay, B.; Scriven, E. F. V.; Suschitzky, H.; Thomas, D. R.
Tetrahedron Lett. 1977, 3175.
(26) Albini, A.; Bettinetti, G.; Minoli, G. J. Am. Chem. Soc. 1997, 119, 7308.
(27) McRobbie, I. M.; Meth-Cohn, O.; Suschitzky, H. Tetrahedron Lett. 1976,
825.
(28) Lindley, J. M.; McRobbie, I. M.; Meth-Cohn, O.; Suschitzky, H. J. Chem.
Soc., Perkin Trans. 1 1980, 982.
(24) Leyva, E.; Platz, M. S.; Persy, G.; Wirz, J. J. Am. Chem. Soc. 1986, 198,
3783.
(25) Morawietz, J.; Sander, W. J. Org. Chem. 1996, 61, 4351.
9
J. AM. CHEM. SOC. VOL. 127, NO. 15, 2005 5553

A R T I C L E S
Carra et al.
Figure 2. IR difference spectra for (a) the formation (254-nm photolysis
of azide 1a, red spectrum) and (b) the >515-nm bleaching (blue spectrum)
3
of nitrene 2a in an Ar matrix at 12 K. The bands that increase in a and
decrease in b (dashed lines) are correlated with those of the simulated
spectrum of nitrene 32a at the bottom (B3LYP/6-31G* calculation of most
stable conformer). The bands that increase in spectrum b are correlated
with those of the simulated spectrum of pyrazoloindole 7 shown at the top
(dotted lines).
24%, respectively, at >95% conversion (cf. Scheme 2; prepara-
tive experiments for the isolation of the products were done on
5 × 10-3 M solutions). In the presence of 0.1 M diethylamine
(DEA), the yields of products 3a and 4a were almost unchanged,
but that of 5a was lowered in favor of two products incorporat-
ing the amine in 30% combined yield. These were identified as
the tautomeric 2-diethylamino-7-pyrazolylazepines 6a and 6a′
(see the experimental part).
Figure 1. (a) Stepwise 254-nm photolysis of azide 1a (blue spectrum) in
3
EtOH at 90 K. The visible band of nitrene 2a (red spectrum) is enlarged
3
for better visibility. (b) Photolysis of nitrene 2a (red spectrum) at >450
nm. (c) Difference spectra showing the formation (254 nm photolysis of
azide 1a, red spectrum) and the >515-nm bleaching (blue spectrum) of
3
nitrene 2a in an Ar matrix at 12 K.
As 2b had proven to be a useful model, we decided to extend
our previous study to include the parent pyrazolylnitrene, 2a.
This compound lacks the radicalophilic methyl groups but offers
the possibility of electrophilic attack on the pyrazole π-system,
analogous to the recently reinvestigated case of o-biphenylni-
trene.5 2a also allowed us to examine whether the conformation
of the pyrazole group in the azides or in the incipient nitrenes
would affect the reactions, including the competing modes of
electrophilic attack. To confront this problem, the intermediates
and their chemistry had to be fully characterized in solution
and in matrices and the various reaction paths had to be modeled
by appropriate methods.
Thus, we investigated the chemistry of nitrene 2a at room
temperature and in an ethanol glass at 90 K (by UV detection
and product analysis) and of both 2a and 2b in Ar matrices at
15 K (by UV and IR detection). Furthermore, we examined the
properties and the reactivity of both nitrenes by comprehensive
quantum chemical calculations. As will be shown, the results
revealed some new facets of the chemistry of both states of
phenylnitrenes and indicated the limits for the formation of
heterocycles through intramolecular reactions, i.e., the most
useful synthetic application of phenylnitrenes.10
Irradiation of 1a at 254 nm in an EtOH glass at 90 K until
>90% bleaching led to conspicuous new absorptions at ca. 330
3
and 530 nm (see Figure 1a), assigned to 2a in view of the
similarity to other triplet phenylnitrenes (more stringent evidence
is reported below). On melting the glass, UV and HPLC showed
that 4a was the main product, accompanied by some amine 3a
and a trace of heteropentalene 5a (Scheme 2).
Upon irradiation of the glass containing 32a at >450 nm, the
absorption of this transient was replaced by new bands at ca.
220, 300, and 365 nm (the black line in Figure 1b) which
remained unchanged after melting the glass. HPLC revealed
20% of 3a, a small amount of 5a (absorbing at 365), no 4a,
and, as the main product (55%), a compound not formed at
room temperature, to which belonged the bands at 220 and 300
nm. This was an isomer of 2a (m/e 157, LC/MS), to which the
structure of pyrazolo[1,5-a]benzimidazole 7 (Scheme 2) was
assigned, on the basis of the similarity of its UV spectrum to
that of carbazole. Further evidence to support this assignment
will follow below.
Irradiation (254 nm) of 1a in an Ar matrix at 12 K gave a
similar spectrum, with considerably more fine structure (the red
line in Figure 1c). Bleaching at >515 nm led to the disappear-
ance of the bands above 370 nm, of the system at 310-360
nm, and of a sharp peak at 260 nm (the blue line in Figure 1c),
while the small peak at 367 nm and the sharp bands at 305/
299/293 nm, which all belong to 5a, increased. The correspond-
2. Results and Discussion
2.1. Experimental Studies. Pyrazolylphenylnitrene 2a.
Irradiation of 1 × 10-4 M azide 1a in argon-flushed EtOH at
room temperature gave the amine 3a, the azo derivative 4a,
and pyrazolo[1,2-a]benzopyrazole 5a in yields of 39, 18, and
9
5554 J. AM. CHEM. SOC. VOL. 127, NO. 15, 2005

Ground and Excited State o-Pyrazolylphenylnitrenes
A R T I C L E S
Figure 3. Difference spectra for (a) the formation (254 nm photolysis of
azide 1b, red line) and (b) the >525-nm bleaching of nitrene 2b (blue line)
in an Ar matrix at 12 K. Spectrum c (green line) relates to subsequent
irradiation at 313 nm, which leads to dehydrogenation of dihydroquinoxaline
8 (see the text).
ing IR difference spectra (Figure 2) showed that most of the
3
peaks generated at 254 nm matched those calculated for 2a
Figure 4. IR difference spectra corresponding to spectra b (blue line) and
c (green line) in Figure 3. The bands that decrease in b are correlated with
those of the simulated spectrum of nitrene 32b. (B3LYP/6-31G* calculation
of most stable conformer) at the bottom. Peaks that rise in b and decrease
in c (dashend lines) are due to dihyroquinoxaline 8, while those that rise in
c are correlated with the IR spectrum of quinoxaline 11 at the top.
(bottom of Figure 2a). These features disappeared upon bleach-
ing at >515 nm, while some peaks already present after reaction
of 1a (marked with asterisks in Figure 2a) increased (Figure
2b). Comparison with the calculated IR spectrum of 7 (top of
Figure 2) confirmed the assignment to this species; The IR
spectra showed also that very little 5a was formed.
Subsequent irradiation at 313 nm (traces c in Figures 3 and
4) bleached the UV and IR bands of 8 and gave a band at ca.
350 nm and IR peaks (pointing up in Figure 4c) attributed to
quinoxaline 11 by comparison with an authentic sample (top
of Figure 4). A small band at 1886 cm-1 also rose, indicating
the formation of a dehydroazepine, recognized as 15b rather
than regioisomeric 16b by comparison with the calculated
spectrum (see Supporting Information). On the other hand, we
had previously shown that 365-nm bleaching of 5b gives
triazasemibullvalene 9b, in a reaction that is photoreversible at
313 nm.29
Subsequent photolysis at 365 nm (not shown) converted 5a
into triazasemibullvalene 9a (Scheme 3) in a reaction that can
be reverted at 313 nm,29 while 7 was slowly converted to a
species weakly absorbing at 610 nm, tentatively identified as
radical 10 (Scheme 3) due to the close similarity to the
N-carbazolyl radical.30
Dimethylpyrazolylphenylnitrene 2b. The previously re-
ported photochemistry of the dimethyl derivative 2b in EtOH26,31,32
gave heteropentalene 5b as almost the only product at room
temperature, but in the presence of DEA, two regioisomeric
adducts, 2-diethylamino-3-pyrazolylazepine 6b′′ (22%) and 2-di-
ethylamino-7-pyrazolylazepine (6b, 3%), were formed, along
with low yields of amine 3b and azo compound 4b (cf. Scheme
2). The triplet nitrene (32b) was again cleanly produced at 90
K and melting of the glass gave 4b, though further irradiation
at >450 nm gave dihydropyrazoloquinoxaline 8, with a minor
amount of 5b.
Summing up, at room temperature heteropentalenes 5 are
formed from the singlet nitrenes as the only isolated products
(72% from 1b and 24% from 1a, paths a and e, respectively in
Scheme 4). Triplet nitrene products (after ISC, path b) are
abundant from 1a (3a and 4a, together 54%), while they are
absent from 1b (Scheme 4). In the presence of DEA, amino-
azepines are formed through trapping of dehydroazepine 16 (via
path d, that followed by most o-substituted phenyl-
nitrenes)33-37 in the case of 1b and trapping of 15 (via path c,
as expected e.g. for o-cyanophenylnitrene)4 in the case of 1a.
As in previous cases,21 addition of DEA was found to slightly
enhance the yield of tripled derived products, perhaps because
it increases the overall yield of isolable products.
We now carried out the reaction in an Ar matrix at 12 K,
3
reproducing the UV/vis absorptions at 340 and 530 nm of 2b
(Figure 3a) and strengthening the identification by comparing
the corresponding IR spectrum obtained with that calculated
for 32b (Figure 4). An intense band at ca. 300 nm belonging to
8 (comparison with an authentic sample) and a weak indication
of 5b at 365 nm were also observed. The latter bands increased
markedly upon >515-nm bleaching of 32b, along with the 305-
nm peak of 5b (Figure 3b).
Triplet nitrenes are formed both in EtOH glasses and in Ar
matrices, but subsequent visible light irradiation gives contrast-
3
ing results (Scheme 3). 2a attacks the pyrazole π-system to
give 7 (presumably via 17, path e in Scheme 4), while yielding
only a trace of 5a and no dehydroazepines. With 32b the main
product is 8, which arises by H-abstraction from the methyl
group presumably via diradical 18 (path f), with a significant
(1 to 5) proportion of 5b and a small amount of dehydroazepine
(29) Carra, C.; Bally, T.; Jenny, T. A.; Albini, A. Photochem. Photobiol. 2002,
1, 38.
(30) Levya, E.; Platz, M. S.; Niu, B.; Wirz, J. J. Phys. Chem. 1987, 91, 2293.
(31) Albini, A.; Bettinetti, G.; Minoli, A. J. Am. Chem. Soc. 1991, 113, 6928.
(32) Albini, A.; Bettinetti, G.; Minoli, G. Heterocycles 1995, 40, 597.
9
J. AM. CHEM. SOC. VOL. 127, NO. 15, 2005 5555

A R T I C L E S
Carra et al.
Scheme 3. Products Obtained after 254-nm Photolysis of Azides
1a and 1b in Ar at 12 K
Scheme 4. Summary of the Observed Decay Paths of Nitrenes 2
Thus we analyzed the conformational surfaces of azides 1 and
nitrenes 2 by calculations.
Parent System 1a/2a. Full geometry optimizations (B3LYP/
6-31G*) led to four conformational minima in the case of 1a
(structures in the top row of Figure 5), pairwise related by the
conformation of the azide (1a and 1a′) and the pyrazolyl groups
(denoted NH or NN), respectively, and distinguished by the
dihedral angle between the phenyl and the pyrazole rings. We
calculated that 93% of the parent azide is present as the 1a(NH)
conformer at room temperature,38 suggesting that, unless dea-
zotation involves rotation around the phenyl-pyrazole bond,
the N atom is mainly facing a pyrazole C-H bond in the
incipient nitrene.
15b. In Ar matrices, further photochemistry of the above
products is observed, namely reversible rearrangement of 5a,b
to 9a,b and hydrogen loss from 7 to give 10 and from 8 to give
11.
2.2. Computational Study of the Conformation of Azides
and Nitrenes. The above experiments show that intramolecular
attack of the pyrazole nitrogen by the singlet nitrene is more
efficient (and competes more effectively with ISC to the triplet)
in 2a than in 2b, a finding for which we considered the following
rationalizations:
(a) Cyclization to the heteropentalenes 5 requires that the
pyrazole N lone pair faces the nitrene N atom. The percentage
of this conformation may differ in 2a and 2b.
(b) The nucleophilicity of the pyrazole N lone pair is
increased by the methyl groups.
The corresponding (optimized) structure of triplet nitrene
(32a(NH) in Figure 5) is clearly favored over the other
3
conformation, 2a(NN), where the nitrene N atom faces the
pyrazole N lone pair, presumably due to a stabilizing interaction
between the pyrazole H-atom vicinal to the nitrene N (distance
< 2 Å) in the former and to a repulsive interaction between the
two N atoms (indicated by deviation from planarity) in the latter
conformation. The Arrhenius activation energy [Ea ) ∆H(T)+RT]
for their conversion is 6.66 kcal/mol at room temperature (6.56
kcal/mol at 77 K), i.e., the two conformations cannot intercon-
vert at 77 K, let alone at 12 K, once they are formed.39
(c) Intersystem crossing may be faster in 2a than in 2b, thus
siphoning away singlet nitrene more rapidly than it can cyclize
to 5a.
All of the above factors may be operative simultaneously,
but only the first two are amenable to computational exploration.
(38) The B3LYP/6-31G*-calculated Arrhenius activation energies for the rotation
of the pyrazolyl groups are 2.76 kcal/mol (1a(NH) f 1a(NN)) and 0.52
kcal/mol (1a′(NH) f 1a′(NN)) at 298 K, and those for rotation of the
azide group are 4.31 kcal/mol (1a(NN) f 1a′(NN)) and 3.68 kcal/mol
(1a(NH) f 1a′(NH); full details are given in the Supporting Information).
Thus, the four conformations of azide 1a equilibrate rapidly at room
temperature.
(39) In addition, a CASSCF optimization of the first excited triplet state of 2a
showed that the bond between the phenyl and the pyrazole ring shortens
and that the excited triplet shows no tendency to undergo rotation around
that bond (Carra, C., unpublished results). Hence, no conformational change
is expected to occur on bleaching of 32a.
(33) Berwick, M. A. J. Am. Chem. Soc. 1971, 93, 5780.
(34) Sundberg, R. J.; Sutar, S.; Brenner, M. J. Am. Chem. Soc. 1972, 94, 513.
(35) Lamara, K.; Redhouse, A. D.; Smalley, R. K.; Thompson, J. R. Tetrahedron
1994, 50, 5515.
(36) Levya, E.; Sagredo, R. Tetrahedron 1998, 54, 7367.
(37) Karney, W. L.; Borden, W. T. J. Am. Chem. Soc. 1997, 119, 3347.
9
5556 J. AM. CHEM. SOC. VOL. 127, NO. 15, 2005

Ground and Excited State o-Pyrazolylphenylnitrenes
A R T I C L E S
phenylnitrenes have a comparable electronic structure, but while
in the former the Coulombic repulsion between the unpaired
electrons is minimized “inherently” by the Pauli exclusion
principle, an increased spatial separation is required in the singlet
to achieve a similar effect. As one of the electrons occupies an
in-plane p-AO of the N atom, the other one must be in a
π-orbital that is localized at the far end of the phenyl ring. Thus,
singlet phenylnitrene resembles a cyclohexadienyl radical linked
to an iminyl radical (short CdN bond), whereas the triplet rather
resembles more a benzyl radical (longer C-N bond).
This holds also for 2a (see Table 1), with the additional
consequence that the nonbonded N‚‚‚H distance in the NH
conformer is shorter in the singlet, even though the C-N bond
is longer (and hence the nitrene N atom is closer to the pyrazole
H) in the triplet. This attractive interaction between the nitrene
N and the vicinal pyrazole H atoms makes the singlet-triplet
gap shrink from almost 21 kcal/mol in the NN conformer (or
in the transition state (TS) between the two conformers), to 17.7
kcal/mol in the NH conformer.42 This N‚‚‚H interaction may
also hinder N-C bond formation by attack on the pyrazole π
system (see below).
The two rotamers43 readily equilibrate at room temperature
(although the NN concentration is tiny), and thus, the fate of
12a depends on the relative rates of the competitive decays rather
than on its native conformation. In contrast, at 90 or 10 K, ISC
precludes reactions from 12a and the conformers do not
interconvert; thus the products are dictated by the conformation
of the starting azide and of the incipient nitrene. Indeed, the
main product under these conditions is pyrazolobenzimidazole
7, only accessible from the NH conformer.
Figure 5. Conformers of azide 1a (top line), azide 1b (middle line), and
of the triplet nitrenes 32a and 2b (bottom line). The numbers inside the
phenyl rings denote values of the phenyl-pyrazolyl dihedral angle.
Structures and ∆G values are from B3LYP/6-31G* calculations.
Dimethyl Derivative 1b/2b. To our surprise a systematic
conformational search of azide 1b revealed only two minima,
NN and NMe,44 with the conformers in a ca. 60:40 mixture at
room temperature according to their B3LYP free energy
difference.45 Since the two aromatic rings are nearly perpen-
dicular in both structures, this imposes little bias on the nascent
Table 1. Relative Energiesa of the Two Conformers of Nitrene 2a
in Its Singlet and Triplet State, Activation Energies for Their
Interconversion, S/T Gaps, and Some Critical Geometry
Parameters
3
nitrene. In the triplet nitrene, 2b, the free energy difference
S/T
gapd
ωc,e
deg
/
r(C
−
Åc
N)/
r(N‚‚‚H)/
Åc
3
B3LYP
CASPT2b,c
between the two conformers is much smaller than in 2a, and
the barrier for their interconversion is much lower (3.20 or 1.12
kcal/mol, respectively, at 298 K), since they deviate already by
about 50° from planarity.
32a(NH)
TS
(0)
6.90
4.43
-
(0)
-
0
1.338
1.349
1.348
1.256
1.272
1.272
2.233
-
5.45
3.21
(0)
8.60
6.32
-
93.9
129.0
0
91.8
130.3
32a(NN)
12a(NH)
TS
-
-
17.74
20.88
20.92
2.218
-
-
All attempts to find a CASSCF potential energy minimum
-
1
corresponding to the NN conformer of 2b converged to the
12a(NN)
-
NMe conformer, suggesting that decay to the latter (Table 2)
is barrieless in this case.
a In kcal/mol. b Based on (14,12) CASSCF calculations; for details, see
the Methods section. c At the (14,12) CASSCF geometries. d Singlet/triplet
gap from CASPT2 calculations; E(S) > E(T). e Dihedral angle C-C-N-C
between the phenyl and the pyrazole rings.
(41) Note that the CASSCF dipole moment of 2a(NN) is much higher than that
of 2a(NH) (4.3 vs 1.7 D). Thus, the energy difference between the two
conformers may be strongly attenuated in polar solvents, e.g. ethanol.
(42) The extent of this effect depends quite strongly on the size and the nature
of the active space in the CASSCF calculations. Removing the σ lone pair
on the nitrene N atom from the active space does not affect the energy
difference between the two conformers in the triplet state but reduces that
difference in the singlet state from 6.4 to 4.5 kcal/mol (and increases the
S/T gap in the H-bonded conformer to 20.4 kcal/mol, only 1 kcal/mol less
than in the other conformer).
(43) The transition states were found to be stationary points in a saddle point
search. Unfortunately, we were unable to perform frequency calculations
at the CASSCF(14,12) level to ascertain the nature of these stationary points
and to obtain vibrational data for thermal corrections. However, the
transition state found for the triplet nitrene looks very similar to that found
(and characterized) by B3LYP, so we have reason to be confident that we
have actually located true saddle points also by CASSCF.
DFT methods do not yield meaningful results for singlet 12a,
due to the open-shell nature of this species and the resulting
multideterminantal nature of its wave function,5,19 so we resorted
to the CASSCF/CASPT2 methodology to calculate both spin
states of nitrenes 2a. As shown in Table 1, the NN conformers
are consistently less stable than the (planar40) NH conformers,
presumably due to a repulsive effect of the pyrazole N lone
pair.41 However, the energy difference is nearly twice as high
in the singlet than in the triplet state. As Borden et al. have
explained,19 the triplet and the (open-shell) singlet states of
(44) To convince ourselves that we had not missed a secondary minimum, we
carried out relaxed scans of the dihedral angle, the results of which are
summarized in Figure S1 of the Supporting Information.
(40) CASSCF predicts nonplanar equilibrium conformations for the NH
conformation, both in triplet (dihedral angle 56.7°) and singlet 2a (dihedral
angle 57.5°). In both cases CASPT2 favors, however, a planar conformation
of Cs symmetry, so only this conformation is considered here.
(45) The B3LYP/6-31G* calculated Arrhenius activation energy for the inter-
conversion of the NMe and the NN conformer is 3.71 kcal/mol at 298 K
(full details are given in the Supporting Information). Thus, the two
conformations of 12b equilibrate rapidly at room temperature.
9
J. AM. CHEM. SOC. VOL. 127, NO. 15, 2005 5557

A R T I C L E S
Carra et al.
Table 2. Relative Energiesa of the Two Conformers of Nitrene 2b
in Its Singlet and Triplet State, Activation Energies for Their
Interconversion, S/T Gaps, and Some Critical Geometry
Parameters
1
three intramolecular reactions of 2a are all within ca. 5 kcal/
1
mol at the CASPT2 level, and those for 2b lie even closer,
thus we expect that these processes compete at room temper-
ature.
B3LYP
CASPT2b,c
S/T gapd
ω
e/deg
r(C−N)/Å
The Benzoazirine-Dehydroazepine Path. For the cycliza-
tions to azirines 13a and 14a, separate reaction paths, leading
to separate conformers of the products, were found for the two
32b(NMe)
TS
(0)
(0)
-
2.40
-
-
63.7b
1.347b
1.327f
1.348b
1.272b
3.33
2.13
-
-
95.7f
32b(NN)
12b(NMe)
-
21.25
108.0b
130.3b
1
conformers of singlet 2a. Those originating from the more
stable NH conformers (right-hand side of Figure 6) lead to the
more stable NH conformers of the azirines over lower barriers
compared to the less stable NN conformers (left-hand side).
Thus, we will restrict the discussion to the former intermediates.
Interestingly, formation of the less stable azirines 14a involves
lower barriers than those for formation of their more stable
isomers, 13a. The pyrazole substituent stabilizes azirines 13a
vs isomers 14a, but it has the opposite effect (due to steric
hindrance) on the transition state for cyclization toward itself.
A similar, though less pronounced effect was seen in o-
biphenylnitrene,5 where it did, however, not lead to a curve
crossing. As in that case, the DFT activation energies for
cyclization to azirines are generally higher than those found at
the CASPT2/CASSCF level, but the relative energies of the
transition states are similar.
a In kcal/mol. b Based on (14,12) CASSCF calculations; for details, see
the Methods section. c At the (14,12) CASSCF geometries. d Singlet/triplet
gap from CASPT2 calculations; E(S) > E(T). e Dihedral angle C-C-N-C
between the phenyl and the pyrazole rings. f At the B3LYP geometry.
Indeed, conformational preference appears to play no role at
room temperature, where 5b is virtually the only product,
whereas at 90 K its yield is reduced to 1%. Subsequent
irradiation of the nitrene in matrices does yield 10% 5b, but
most of the nitrene 2b decays by H-abstraction from the
proximate methyl group, reflecting the conformational prefer-
ence of the NMe over the NN conformer, both in the azide and
in the nitrene (cf. Figure 5).
It remains to note that the singlet/triplet gap in 2b as well as
the C-N bond length in 12b are very similar to the values found
in the NN conformer of 2a, demonstrating that the NH
conformer of 2a is the exceptional case, for the reasons discussed
above.
2.3. Competitive Reactions of Singlet Nitrenes: Compu-
tational Studies. The above data on the NN vs NH (NMe)
equilibria offer a rationalization for the results obtained in
matrices, but in general, the competing intramolecular decay
paths of Scheme 4 must also be taken into account. Thus, we
carried out calculations of the activation energies for attack at
the nitrogen or carbon atom of the pyrazole ring (paths a, e),
for insertion into the methyl group (f), and for cyclization to
benzoazirines (c, d) for the conformers of both nitrenes.
Intermolecular reactions of the triplets, such as H abstraction
from the solvent and dimerization to give azo compounds, were
not considered. The results of CASSCF(14,12)/CASPT2 and
B3LYP calculations are presented schematically in Figures 6
and 7 (precise numbers are given in the Supporting Informa-
tion).
As with other phenylnitrenes,3,4,19,37 azirines 13a and 14a
readily open to didehydroazepines 15a and 16a, respectively.
According to B3LYP calculations, the isomers 15a are more
stable than 16a, especially in the NH conformers, where a
hydrogen bond (RN‚‚‚H ) 2.53 Å) contributes probably to the
extra stabilization. There is no steric effect in this step (which
follows the Bell-Evans-Polanyi principle), but the effect on
1
the first step influences the oVerall reaction from 2a to the
didehydroazepines, which leads more rapidly to 16a than to the
more stable 15a. Once formed, 15a must overcome a consider-
1
able barrier to revert to 2a (Ea(B3LYP) ) 21.7 kcal/mol at
298 K),47 whereas equilibration of 16a with 12a is expected to
be rapid at room temperature (Ea(B3LYP) ) 9.9 kcal/mol at
298 K),47 allowing a path back to 15a. Experimentally, only
azepines 6a/6a′ (from the trapping of 15a by DEA) are formed
at the expense of 5a. Apparently, the peculiar structure of 15a,
with its intramolecular H bond, makes it more reactive toward
amines.
We9 and others5 have recently found that, although singlet
nitrenes are not amenable to DFT calculations due to their high
diradicaloid character, the transition states for their “chemical”
decay processes may be modeled fairly well (compared to
CASSCF structures and CASPT2 energies) if R and â-electrons
are not restricted to occupy the same orbitals in the wave
function that models the density.46 We found the same to be
true in the present case, at least as long as spin contamination
(as indicated by the expectation value of the Sˆ2 operator) remains
moderate, which is the case for the transition states leading to
azirines (but less so for those leading to 5a and 17, where Sˆ2
) 0.65 and 0.86, respectively). To be able to compare the results
of the two types of calculations, we “anchored” the B3LYP
numbers on the triplet state of the NH conformer of 2a or the
NMe conformer of 2b, respectively, adding the CASPT2
singlet-triplet gap to arrive at a common origin of the energy
scale ()the most stable conformer of the singlet nitrene). As it
appears from Figures 6 and 7, the activation barriers for the
In the dimethyl derivative (Figure 7), the cyclizations to
azirines are predicted to be more exothermic than in the parent
compound, and the barriers decrease by over 5 kcal/mol for
both modes,48 although the difference in activation energies for
cyclization toward and away from the pyrazole ring remains at
5.5 kcal/mol. Apparently, the methyl groups on the pyrazole
exert a stabilizing electronic effect on the transition states (even
in 14b where there is no steric effect) and attenuate the
difference in stability between the isomers 13 and 14.
The ensuing ring-opening reactions are less affected by the
methyl groups: that of 13b is more exothermic by ca. 2 kcal/
(47) Note that this number is not identical to the energy difference between the
corresponding stationary points in Figures 6 and 7 because the Arrhenius
activation energy, Ea, corresponds to an enthalpy difference (augmented
by RT for a unimolecular reaction).
(48) One may argue that these differences between the parent system and the
dimethyl derivative come about by an error in the calculation of the singlet/
triplet gap. However, to explain a 5 kcal/mol difference in activation
energies would require the relatiVe S/T gap of the parent system and the
dimethyl derivative to be in errror by a similar amount. In the absence of
H-bonds, the S/T gaps of the two compounds are calculated to be about
the same (21 ( 0.2 kcal/mol), which is reasonable. Thus we believe that
the differences in exothermicities and activation barriers for cyclization to
azirines are not artifactual.
(46) Hrovat, D. A.; Duncan, J. A.; Borden, W. T. J. Am. Chem. Soc. 1999, 121,
169.
9
5558 J. AM. CHEM. SOC. VOL. 127, NO. 15, 2005

Ground and Excited State o-Pyrazolylphenylnitrenes
A R T I C L E S
Figure 6. Potential energy surface on which the different rearrangements of nitrene 2a take place. The B3LYP (normal font) and CASPT2 relative energies
3
(italic) are made to coincide for the NH conformer of 2a. Adding the CASPT2 singlet-triplet gap for this species leads to the origin of the energy scale.
Figure 7. Potential energy surface on which the different rearrangements of nitrene 2b take place. The B3LYP (normal font) and CASPT2 relative energies
(italic) are made to coincide for the NMe conformer of 32b. Adding the CASPT2 singlet-triplet gap for this species leads to the origin of the energy scale.
mol with a barrier lowered by ca. 3 kcal/mol and that of 14b is
less exothermic but proceeds over a slightly lowered barrier.
Overall, the less stable dehydroazepine 16b is again predicted
to be formed more rapidly than its more stable isomer 15b,
although the energy difference is smaller than in 12a. The barrier
for reversal of 15b to 12b is twice as high (Ea(B3LYP) ) 20.9
kcal/mol at 298 K)47 than that for 16b (Ea(B3LYP) ) 9.7 kcal/
mol at 298 K).47 Experimentally, trapping of 16b predominates
over that of 15b and this corresponds to what is observed with
most o-substitued phenylnitrenes, which cyclize “away” from
9
J. AM. CHEM. SOC. VOL. 127, NO. 15, 2005 5559

A R T I C L E S
Carra et al.
the substituent,4,10,33-37 just as 2b does, singling out the
abnormally easily trapped 15a (see above).
However, this “laziness” can be overcome by excitation of
32b in a low-temperature matrix, where 8 is formed as the major
product (a little unreacted nitrene abstracts hydrogen from the
solvent and gives 6% 3b).50 This is reasonable, because the
lowest excited state of 32b lies at ca. 30 kcal/mol on the energy
scale in Figure 7, i.e., well above the barrier leading to 8, and
this process is further favored by the predominant NMe
conformation of the triplet nitrene.51 The intramolecular reaction
with the methyl group is diagnostic, because, in the absence of
a specific target for radical attack, excitation of the immobilized
triplet nitrene apparently leads back to the singlet surface, as
indicated in the formation of dehydroazepine from phenylni-
trene19 and carbazole from o-biphenylnitrene.5 Correspondingly,
in the present case a trace of 15b and some 5b are formed from
In the absence of diethylamine, the didehydroazepines form
a reservoir from which singlet nitrenes 12 are replenished.
Indeed, evidence from flash photolysis studies supported that a
part of 5 is formed in a slow reaction,31 thus supporting such
an equilibrium. Other products may be obtained by irreversible
processes if their rates are competitive with that for intersystem
3
crossing to 2 (path b in Scheme 4). Indeed, the activation
barriers for thermal reactions of the singlet nitrenes are generally
lowered by 4-5 kcal/mol on bismethylation of the pyrazole
ring, thus explaining why at room temperature ISC is negligible
1
1
in 2b, whereas it predominates in 2a (70% in the absence of
the amine, 60% in its presence).
3
32b, and 17 is the main product from nonmethylated 2a by
Attack on the Pyrazole Ring. According to the (gas phase)
irradiation in the matrix.
1
CASPT2 calculations, attack of nitrene 2a on the pyrazole
The further reactions in matrices are secondary reactions. The
reversible photorearrangement 5 f 9 has been previously
reported,29 while hydrogen loss from benzimidazole 7 and
dehydrogenation of dihydroquinoxaline 8 are homolytic pro-
cesses caused by the lack of energy dissipation channels in the
Ar matrix.
π-system (yielding 17 and then 7) should be slightly favored
over attack on the N lone pair (yielding 5a). The opposite result
obtained in solution is explained by the much higher dipole
moment of the transition state leading to 5a (3.4 D) compared
to that leading to 17 (1.5 D). This favors formation of 5a in
polar solvents, while the above-mentioned N‚‚‚H interaction
1
with the pyrazole H atom in 2a disfavors formation of 17 in
3. Conclusions
solution. The latter compound is the main product obtained on
photoexcitation of 32a in frozen matrices, but this results from
the fact that the nitrene N-atom faces the CH bond of the
pyrazole in the predominant conformation.
Formation of 5a from 12a is predicted to involve an activation
energy that is lower by ca. 3 kcal/mol than that for cyclization
to azirine 13a (and hence to 6a) at the CASPT2 level. However,
CASPT2 calculations are known to overestimate the latter barrier
by 2-3 kcal/mol.5,49 Accordingly, the two processes compete
in solution, although ISC to the triplet funnels away two-thirds
of the nitrene, leading to a low yield of 5a.
In the case of 12b, the much greater efficiency of ring closure
to yield 5b is explained by the lower calculated barrier. This is
in part due to the greater exothermicity (by 7-8 kcal/mol) in
forming 5b rather than 5a (a product effect, not expressed in
the position of the transition state, however, since the N‚‚‚N
distance at the transition state is 2.07 Å in both cases) and more
importantly due to the enhanced nucleophilicity of the pyrazole
N lone pair caused by the methyl groups. Indeed, B3LYP
calculations on pyrazole show an increase of the energy of the
N lone pair by 0.2 eV upon 3,5-dimethylation.
Nitrene 2b has been proposed almost 30 years ago as a
mechanistic probe designed to distinguish the electrophilic
chemistry of the incipient singlet phenylnitrene from the
expected radical chemistry of its triplet ground state through
intramolecular reactions, and thus to rationalize and direct the
use of these intermediates. In the meantime, work by several
groups has documented through elaborate experiments and
calculations how the cyclization to benzoazirines hampers the
synthetically desirable intermolecular chemistry of singlet
phenylnitrene. A classical intramolecular electrophilic reaction,
that of o-biphenylnitrene, has also been theoretically studied in
detail recently.5
Considering nitrenes 2a and 2b adds a degree of complexity
because two modes of intramolecular electrophilic attack (a and
e in Scheme 4) compete with two cyclizations (paths c and d)
and with intramolecular radical attack (path f). Furthermore,
conformational equilibria must be taken into account. Indeed,
the product distributions shown in Schemes 2 and 3 and the
potential energy surfaces in Figures 6 and 7 are quite intricate,
but it has been possible to explain the chemistry observed as a
complex function of the conformational distribution and the
barriers encountered on the various decay paths, thus evidencing
some key aspects.
Intramolecular Hydrogen Abstraction. The only other
decay process of nitrene 2b, H abstraction from the proximate
pyrazole methyl group, encounters a much higher barrier.
Indeed, 8 is not formed upon direct irradiation of 1b at room
In particular, it is clarified that attack on the pyrazole N lone
pair differs from attack on the π system in that it involves a
more polar transition state. The former process predominates
1
temperature, whereas 2b decays through path a (Scheme 4).
However, some 8 is formed by thermal decomposition at 143
°C of 1b (15%), via the triplet nitrene by photosensitization of
1b in solution (10%),31,32 and from the relaxed triplet nitrene
in an EtOH glass at 90 K (10%), although the calculated thermal
barrier is not negligible (Ea(B3LYP) ) 15.6 kcal/mol at 298
K).47 Thus, our calculations document the well-known “lazy”
character of phenylnitrene triplets toward H-abstraction,16,26-28,32
which is apparently not “cured” by making the reaction
(50) Photoinduced intermolecular H-abstractions of triplet nitrenes are known,
cf.: Levya, E.; Munoz, D.; Platz, M. S. J. Org. Chem. 1989, 54, 5938.
(51) The primary product of hydrogen abstraction from the methyl group is
biradical 18. This turns out to be an interesting species, because it has an
open-shell singlet ground state that is almost degenerate with the corre-
sponding triplet state. The corresponding closed-shell singlet state (which
is what is obtained in a DFT calculation) corresponds to a polyeneimine
and lies vertically about 22 kcal/mol above the open-shell singlet ground
state of 17. However, on geometry optimization that state ends up only ca.
5 kcal/mol above that state (all CASSCF/CASPT2 results). We did not
succeed in finding a transition state for the cyclization of the open-shell
singlet state to dihydroquinoxaline 8 (all attempts to shorten the distance
betweeen the two radical centers resulted in collapse to 8), so we assume
that this process is essentially barrierless.
3
intramolecular as in 2b.
(49) Karney, W. L.; Borden, W. T. J. Am. Chem. Soc. 1997, 119, 1378.
9
5560 J. AM. CHEM. SOC. VOL. 127, NO. 15, 2005

Ground and Excited State o-Pyrazolylphenylnitrenes
A R T I C L E S
Photolysis in Ethanol Glass. A 1 × 10-4 M solution of azides 1a,b
in EtOH (2 mL) in a 1-cm quartz cell was sealed under vacuum after
degassing by four freeze-pump-thaw cycles. The cell was inserted
into an Oxford DN 1704 liquid nitrogen cryostat fitted with a calibrated
ITC4 temperature controller, placed in a Kontron Uvikon UV-vis
spectrometer, and illuminated from the bottom by means of a low-
pressure mercury arc (Helios Italquartz 15W). In a typical experiment,
the solution was frozen to 90 K, irradiated until the azide was fully
consumed (ca. 10 min), warmed to room temperature, and evaporated
in the dark, and the residue was redissolved in 1 mL of MeCN for
HPLC separation with acetonitrile-water mixtures. The product
distribution was determined on the basis of calibration curves. In some
experiments, samples photolyzed at 254 nm were further irradiated from
the side by a focused high-pressure mercury arc (Osram 150 W) through
a cutoff filter (>450 nm) for up to 6 h. Under these conditions, a new
peak from 1a and isomeric with it [base peak 158 (M + 1)+] was
assigned to 7 in view of the close similarity of its UV spectrum with
that of carbazole, and the calibration curve of carbazole was used for
estimating the concentration.
Matrix Isolation and Spectroscopy.52 Crystals of azides 1a or 2a
were placed in a U-shaped tube immersed into a water bath and
connected to the inlet system of a closed-cycle cryostat. While the bath
was kept at 10 °C (1a) or 18 °C (1b), a mixture of high-purity argon
and nitrogen (10:1, to improve the optical quality of the matrices) was
flowing through the tube at a rate of ≈1 mmol/h and swept the
compounds onto a CsI window held at 19 K. A sufficient quantity of
the compound accumulated within 2 h. Electronic absorption (EA)
spectra (200-1200 nm) were taken with a Perkin-Elmer Lambda 19
instrument and IR spectra with a Bomem DA3 interferometer (1 cm-1
resolution) equipped with an MCT detector (500-4000 cm-1). Pho-
tolyses were carried out with low-pressure Hg lamps (254 nm) or with
high-pressure Hg/Xe lamps using cutoff or interference filters for
wavelength selection, as indicated in the text.
in solution, highlighting the electrophilic character of the singlet
nitrene, as seen in the increased efficiency with the more
nucleophilic dimethylpyrazole, a requisite for overcoming ISC
on the triplet. Attack to the pyrazole π system is a more complex
reaction, similar to what has been found with o-biphenylnitrene,5
and is obtained by photoexcitation of the nitrene in a matrix.
As for triplet phenylnitrene, the “lazy” character of this
species has for the first time been documented through a
computational study, indicating also that the excited triplet is a
good hydrogen abstractor. Finally, in matrices, the conforma-
tional equilibration of the o-pyrazole substituent is suppressed
and the phenylnitrene photochemistry is dictated by the con-
formation of the nascent nitrene.
As for the general rationalization of the chemistry of
phenylnitrenes, Platz16 previously pointed out that the rapid
cyclization to benzoazirine (finally leading to tars and/or to ISC)
makes the intermolecular chemistry of the singlet nitrene
synthetically less useful than that of phenylcarbene. The present
study shows that such cyclization competes even with intramo-
lecular attack to a lone pair. Formation of a benzofused
heterocycle such as 5 can compete with other decay processes
only when the nucleophilic site is further activated. The limits
to the use of (hetero)arylphenyl azides for the synthesis of
carbazole analogues are clearly stringent.
Experimental and Theoretical Section
General Information. NMR spectra were run on a Brucker 300
instrument and are reported in CDCl3 with TMS as an internal standard;
attributions are based on the appropriate double irradiation experiments.
IR spectra were collected on a Perkin-Elmer Partaon 1000 spetropho-
tometer and mass spectra on a Finnigan LCQ instrument. Ethanol (95%)
was spectroscopic grade solvent. Column chromatography was per-
formed with silica gel Merck HR 60. The azides were prepared as
previously reported.26
Preparative Irradiation. Preparative irradiations and chromato-
graphic product separations were carried out as in previous reports.26
The characterization of photoproducts 3, 4, 5, 6b, 8, 9, and 11 has
been previously reported.21,26,29 The main analytical data for the newly
isolated photoproducts are reported below (assignment based on DEPT
and NOESY experiments).
3H-2-Diethylamino-7-pyrazolylazepine (6a): slightly yellow oil: 1H
NMR [(CD3)2CO] δ 1.3 (t, 6H), 3.0 (d, 2H, J ) 7 Hz, H2-3), 3.6 (q,
4H), 5.3 (q, 1H, J ) 7 Hz H-4), 6.3 (dd, 1H, J ) 2, 2.5 Hz, pyrazole
H-4), 6.4 (dd, 1H, J ) 7, 9 Hz, H-5), 6.7 (d, 1H, J ) 9 Hz, H-6), 7.55
(d, 1H, J ) 2 Hz, pyrazole H-5), 8.2 (d, 1H, J ) 2.5 Hz, pyrazole
H-3), NOE enhancement between the N-CH2 and the 3-CH2 groups;
13C NMR [(CD3)2CO] δ 13.2 (CH3), 31.1 (CH2, C-3), 43.9 (CH2), 95.4
(CH, C-6), 105.1 (CH, pyrazole C-4), 109.5 (CH, C-4), 126.3 (CH,
pyrazole C-3), 128.4 (CH, C-5), 139.5 (CH, pyrazole C-5); HRMS calcd
for C13H18N4 230.1531, found 230.1528.
Quantum Chemical Calculations. The geometries of all species
except the singlet nitrenes were optimized by the B3LYP density
functional method53,54 as implemented in the Gaussian program
package,55,56 using the 6-31G* basis set. All stationary points were
characterized by second derivative calculations which served also for
the calculation of vibrational spectra. In the calculations of the transition
states for cyclization of singlet phenylnitrenes, the density was modeled
by a spin-unrestricted wave function, to allow for some open-shell
character, a procedure which has proven to give transition state
geometries in good accord with the multideterminantal calculations
described below.5,9 Activation energies computed in this fashion are
useful if the spin contamination does not exceed a certain degree ( Sˆ2
< 0.2). Otherwise, the admixture of high-spin states leads to an
artifactual overstabilization, i.e., to an underestimation of activation
energies, as was found, for example, for the transition states leading to
heteropentalenes 5.
Excitation energies were evaluated by time-dependent response
theory,57 according to which the poles and the residues of the frequency-
7H-2-Diethylamino-7-pyrazolylazepine (6a′): slightly yellow oil; 1H
NMR (CDCl3) δ 1.3 (t, 6H), 3.55(q, 2H), 5.55 (t, 1H, J ) 9 Hz, H-6),
5.7 (t, 1H, Hz, H-4), 6.0 (t, 1H, pyrazole H-4), 6.4 (d, 1H, J ) 9 Hz,
H-7), 6.9 (dd, 1H, J ) 7, 9 Hz, H-5), 7.0 (d, 1H, J ) 7 Hz, H-3), 7.1
(d, 1H, J ) 2, pyrazole H-5), 7.45 (d, 1H, J ) 2 Hz, pyrazole H-3);
HRMS calcd for C13H18N4 230.1531, found 230.1527.
3H-2-Diethylamino-7-(3,5-dimethylpyrazolyl)azepine (6b) was ob-
tained in a chromatographic fraction admixed with 5b: 1H NMR
(CDCl3) δ 1.17 (t, 6H), 2.18 (s, 3H), 2.27 (s, 3H), 2.9 (d, J ) 8 Hz,
1H, H-3), 3.41 (q, 4H), 5.11 (q, J ) 8 Hz, H-4), 5.88 (s, 1H), 6.21 (d,
J ) 6 Hz, H-6), 6.44 (dd, J ) 6, 8 Hz), H-5); 13C NMR (CDCl3) δ
12.7 (CH3), 12.9 (CH3), 13.5 (CH3), 29.6 (CH2), 43.4 (CH2), 103.6
(CH), 106.2 (CH), 110.1 (CH), 128.3 (CH); HRMS calcd for C15H22N4
258.1844, found 258.1843.
(52) Bally, T. In ReactiVe Intemediate Chemistry; Moss, R. A., Platz, M. S.,
Jones, M., Eds.; Wiley: New York, 2003.
(53) Becke, A. D. J. Chem. Phys. 1993, 98, 5648.
(54) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785.
(55) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M.
A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A.; 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.; Pommelli, 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.; 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.; Challacombe, M.; Gill, P. M. W.; Johnson, B. G.;
Chen, W.; Wong, M. W.; Andres, J. L.; Gonzales, C.; Head-Gordon, M.;
Repogle, E. S.; Pople, J. A. Gaussian 98, Rev. A7-A11; Gaussian, Inc.:
Pittsburgh, PA, 1998.
(56) Johnson, B. G.; Gill, P. M. W.; Pople, J. A. J. Chem. Phys. 1993, 98,
5612.
9
J. AM. CHEM. SOC. VOL. 127, NO. 15, 2005 5561

A R T I C L E S
Carra et al.
dependent polarizability are calculated, whereby the former correspond
to vertical excitation energies and the latter to oscillator strengths. We
employed the density-functional-based implementation of this method
(TD-DFT)58 using again the B3LYP functional and the 6-31G* basis
set described above.
parent compound and over 63% in the dimethyl compounds, thus
ensuring that inclusion of dynamic electron correlation by many-body
perturbation theory produces no artifacts. The [C,N]3s2p1d/[H]2s ANO
basis set60 was used in the CASSCF/CASPT2 calculations, which were
carried out with the MOLCAS program.61
The open-shell states of the singlet nitrenes19 require at least two
Slater determinants for a qualitatively correct description. We therefore
resorted to the CASSCF/CASPT2 protocol59 to calculate the energies
of singlet nitrenes, the transition states for their different cyclization
reactions, and the products of these cyclizations (cf. Figures 6 and 7).
To “anchor” the DFT results on a common energy scale, we also
recalculated the triplet phenylnitrenes at this multideterminantal level.
All geometries were fully optimized by the CASSCF procedure, using
an active space comprising 14 electrons in 12 molecular orbitals. In
addition to 12 π-MOs (six on the phenyl ring, six on the pyrazolyl
ring), the active space included two σ-MOs. In the nitrenes these were
the singly occupied in-plane p-AO on the N atom and the in-plane
lone pair on the same atom. In the cyclization transition states the former
MO turned into the incipient C-N σ-bonding MO, whereas in the final
products it was usually replaced by a higher lying σ(C-C) MO.
The above active spaces resulted in zero-order wave functions that
consistently described the CASPT2 wave functions to over 66% in the
Full sets of Cartesian coordinates and absolute energies from all
calculations (including thermal corrections in the case of B3LYP results)
are given in the Supporting Information.
Acknowledgment. This work was supported by the Swiss
National Science Foundation (project no. 200020-105217). We
are very indebted to Krzysztof Piech for mapping the potential
surfaces of the azides and the triplet nitrenes at the DFT level.
Supporting Information Available: Cartesian coordinates and
energies (including enthalpies and free energies where available)
of all stationary points discussed in this study (text file), relaxed
potential energy surface scan of azide 1b, and IR spectrum of
didehydroazepine 15b (pdf). This material is available free of
charge via the Internet at http://pubs.acs.org.
JA043988W
(57) Casida, M. E. In Recent AdVances in Density Functional Methods; Chong,
D. P., Ed.; World Scientific: Singapore, 1995; Part I, p 155.
(58) Stratmann, R. E.; Scuseria, G. E.; Frisch, M. J. J. Chem. Phys. 1998, 109,
8218.
(59) Andersson, K.; Roos, B. O. In Modern Electronic Structure Theory; World
Scientific Publ. Co.: Singapore, 1995; Part 1, Vol. 2, p 55.
(60) Pierloot, K.; Dumez, B.; Widmark, P.-O.; Roos, B. O. Theor. Chim. Acta
1995, 90, 87.
(61) Andersson, K.; Blomberg, M. R. A.; Fu¨lscher, M. P.; Kello¨, V.; Lindh, R.;
Malmqvist, P.-Å.; Noga, J.; Olson, J.; Roos, B. O.; Sadlej, A.; Siegbahn,
P. E. M.; Urban, M.; Widmark, P.-O. MOLCAS, versions 4 and 5; University
of Lund: Lund, Sweden, 1998.
9
5562 J. AM. CHEM. SOC. VOL. 127, NO. 15, 2005