Characterizing triplet states of quinonoidal dinitrenes as a function of conjugation length

Article abstract of DOI:10.1021/ja963648d

Photolysis of para,para'-diazide precursors in 2-methyltetrahydrofuran gives biradical ESR spectra assignable to quinonoidal dinitrenes with excited triplet biradical states. The zero field splitting (zfs) parameters for the biradicals at 77 K are as foll

Full text of DOI:10.1021/ja963648d

J. Am. Chem. Soc. 1997, 119, 2187-2195
2187
Characterizing Triplet States of Quinonoidal Dinitrenes as a
Function of Conjugation Length
Masaki Minato and Paul M. Lahti*
Contribution from the Department of Chemistry, UniVersity of Massachusetts,
Amherst, Massachusetts 01003
ReceiVed October 18, 1996X
Abstract: Photolysis of para,para′-diazide precursors in 2-methyltetrahydrofuran gives biradical ESR spectra assignable
to quinonoidal dinitrenes with excited triplet biradical states. The zero field splitting (zfs) parameters for the biradicals
at 77 K are as follows: 1,4-phenylenedinitrene (|D/hc| ) 0.169 cm-1, |E/hc| ) 0.004 cm-1); 4,4′-biphenyldinitrene
(|D/hc| ) 0.188 cm-1, |E/hc| e 0.002 cm-1); 4,4′-stilbenedinitrene (|D/hc| ) 0.122 cm-1, |E/hc| < 0.002 cm-1);
1,4-bis(p-nitrenophenyl)buta-1,3-diene (|D/hc| ) 0.0865 cm-1, |E/hc| < 0.002 cm-1); 1,8-bis(p-nitrenophenyl)octa-
1,3,5,7-tetraene (|D/hc| ) 0.0442 cm-1, |E/hc| < 0.001 cm-1). These zfs parameters are inconsistent with simple
dipole-dipole analysis of the ESR spectra in localized biradicals, but are reconcilable with a spin-polarized model
involving the π-electrons of the biradicals.
Introduction
calized, unpaired electrons in σ-type orbitals with cojugated
π-clouds that allow the possibility of π-spin polarization.
Through such a spin-polarized mechanism, indirect communica-
tion between the conjugatively isolated unpaired electrons is
enabled. We were interested in probing how quickly such an
indirect exchange interaction would decrease as a function of
conjugation length in the quinonoid structure. In principle,
dilution of spin delocalization would lead to isolation of the
unpaired spins on the nitrogen termini of these molecules,
resulting in a pair of non-interacting radicals. Since spin
polarization is an important exchange mechanism, these mol-
ecules are worthy of investigation, even though they are too
unstable to be of practical use. Herein we report details on
generation and variable temperature ESR spectroscopy of the
biradicals 1,4-phenylenedinitrene, 4,4′-biphenyldinitrene, 4,4′-
stilbenedinitrene, 1,4-bis(p-nitrenophenyl)-1,3-butadiene, and
1,8-bis(p-nitrenophenyl)-1,3,5,7-octatetraene, 1-5 respectively.
Synthetic Methods. Figure 1 details the syntheses of diazide
precursors 6-10. Standard chemical transformations were used
to reduce appropriate dinitroarenes to the corresponding di-
amines. The diamines were then transformed into diazides via
diazotization/azidification by the Sandmeyer route. All diazides
were thermally stable colored solids that rapidly discolored in
room light. They can be stored for many months in the dark at
10 °C or less.
As part of our studies of interelectronic exchange interactions
in open-shell molecules, we have investigated intramolecular
exchange in dinitrene model systems.1-11 By studying these
and other organic diradicals and polyradicals, we aim to
understand better the structure-property relationships control-
ling ground state spin multiplicity. Such information is a critical
part of efforts to design and synthesize new, molecule-based
magnetic materials that incorporate organic building blocks as
part of their structures.12-15
One type of system of considerable interest is the class of
quinonoidal dinitrenes. These are expected to have two lo-
X Abstract published in AdVance ACS Abstracts, February 15, 1997.
(1) Lahti, P. M. In Research Trends; Menon, J.; Ed.; Council of Scientific
Research Integration: Trivandrum, India, 1992; Vol. 3, pp 179-192.
(2) Ling, C.; Minato, M.; Lahti, P. M.; van Willigen, H. J. Am. Chem.
Soc. 1992, 114, 9959.
(3) Minato, M.; Lahti, P. M. J. Am. Chem. Soc. 1993, 115, 4532.
(4) Minato, M.; Lahti, P. M. J. Phys. Org. Chem. 1993, 6, 483.
(5) Ling, C.; Lahti, P. M. J. Am. Chem. Soc. 1994, 116, 8784.
(6) Minato, M.; Lahti, P. M. J. Phys. Org. Chem. 1994, 5, 495.
(7) Ling, C.; Lahti, P. M. Chem. Lett. 1994, 1357,2447.
(8) Lahti, P. M.; Minato, M.; Ling, C. Mol. Cryst. Liq. Cryst. 1995, 271,
147.
Computational Methods. All computations were carried out
on a Silicon Graphics Indigo R4000 computer. Geometry
optimizations were performed using either the GAUSSIAN9416a
or GAMESS16b programs using the standard criteria within these
programs for convergence. Spin density calculations on dini-
trene 1 were carried out at a fixed geometry using the MELDF
suite of programs developed by Davidson and co-workers.17 Spin
density plots were visualized using SPARTAN (version 4.1.1)
by Wavefunction Inc.
(9) Ichimura, A. S.; Sato, K.; Kinoshita, T.; Takui, T.; Itoh, K.; Lahti,
P. M. Mol. Cryst. Liq. Cryst. 1995, 272, 57.
(10) Minato, M.; Lahti, P. M. J. Phys. Org. Chem. 1991, 4, 459.
(11) Ling, C.; Lahti, P. M. Chem. Lett. 1993, 769-772.
(12) Miller, J. S.; Dougherty, D. A., Eds. Mol. Cryst. Liq. Cryst. 1989,
176, 1.
(13) Gatteschi, D.; Kahn, O.; Miller, J. S.; Palacio, F. Magnetic Molecular
Materials; Kluwer Academic Publishers: Dordrecht, The Netherlands, 1991;
Vol. 198E.
(14) Iwamura, H.; Miller, J. S., Eds. Mol. Cryst. Liq. Cryst. 1993, 232-
233, 1.
(15) Miller, J. S.; Epstein, A. J., Eds. Mol. Cryst. Liq. Cryst. 1995, 271-
274, 1.
(16) (a) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.;
Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.; Keith, T.; Petersson, G.
A.; Montgomery, J. A.; Raghavachari, K.; Al-Laham, M. A.; Zakrzewski,
V. G.; Ortiz, J. V.; Foresman, J. B.; Peng, C. Y.; Ayala, P. Y.; Chen, W.;
Wong, M. W.; Andres, J. L.; Replogle, E. S.; Gomperts, R.; Martin, R. L.;
Fox, D. J.; Binkley, J. S.; Defrees, D. J.; Baker, J.; Stewart, J. J. P.; Head-
Gordon, M.; Gonzalez, C.; Pople, J. A. Program Gaussian 94, Revision
B.3; Gaussian, Inc.: Pittsburgh, PA, 1995. (b) Schmidt, M. M.; Baldridge,
K. K.; Boatz, J. A.; Jensen, J. H.; Koseki, S.; Gordon, M. S.; Nguyen, K.
A.; Windus, T. L.; Elbert, S. T. QCPE Bull. 1990, 10, 52 (Program
GAMESS).
S0002-7863(96)03648-7 CCC: $14.00 © 1997 American Chemical Society

2188 J. Am. Chem. Soc., Vol. 119, No. 9, 1997
Minato and Lahti
Figure 1. Synthesis of diazides 6-10 and generation of dinitrenes 1-5: (a) MTHF, 77 K, hν > 300 nm, 5 min; (b) NaOH/acetone/EtOH, heat;
(c) I2/hexane or acetone, heat; (d) SnCl2‚H2O/EtOH, heat or (for 10) Fe/dimethylformamide/HCl, heat; (e) NaNO2, aq H+, then NaN3; (f) (p-
nitrobenzyl)triphenylphosphonium bromide/EtOH/LiOEt; (g) p-nitrobenzaldehyde/EtOH/NaOEt; (h) p-azidobenzaldehyde/EtOH/NaOEt.
Results
X-band ESR spectra in the 100-9900 G range produced by
Figures 4a-e show temperature vs intensity plots for the
biradical spectra of 1-5. The decreasing intensity of the spectra
at <30 K can make them difficult to observe, and clearly shows
the spectra to be due to thermally populated excited states. Due
to the instability of the biradical spectra in MTHF at >85 K,
we were limited in our ability to investigate the Curie behavior
of the dinitrenes. If one assumes thermal equilibrium between
ground state singlet and excited state triplet states, the spectral
intensity I of the ESR active triplet state is given by eq 1:19
5 min each of photolysis at 77 K of 6-10 in 2-methyltetrahy-
drofuran (MTHF) are shown in Figures 2a-e. Similar spectra
could be generated in ether/pentane/alcohol (EPA) glasses, but
all work described below was carried out in the simpler solvent
system. All spectra can be divided into three major regions:
(1) radical spectral peaks in the g ) 2 region, (2) mononitrene
peaks in various positions over the 5800-6800 G region, and
(3) biradical spectral features attributable to ∆Ms ) 1, 2
transitions. The spectral features show considerable thermal
stability at 77 K, but fade rapidly and irreversibly when the
matrix is warmed above 85 K. The radical features in the g )
2 region often dominate the spectra, and may be attributed to
various species, including the photolysate of MTHF and the
products of reaction by the nitrenes or dinitrenes with MTHF.
We did not study the g ) 2 spectral region, save to note that it
increases with photolysis time, and is substantially decreased
by photolysis at wavelengths >300 nm. As part of the process
of evaluating the biradical portions of the spectra, we carried
out triplet ESR spectral simulations using the eigenfield
method,18 the results of which are shown in Figure 3. The zfs
values from the simulations for |D/hc| and |E/hc| of 1-5 are
summarized in Table 1.
3 exp(-2J/RT)
1 + 3 exp(-2J/RT)
IT )
(1)
J is the exchange coupling between the unpaired electrons, T is
absolute temperature, and R is 1.987 cal/(mol‚K). The singlet-
triplet gap ∆ET-S ) 2J for a biradical. Nonlinear least-squares
curve fitting of the data20 in Figure 4 to eq 1 yields the J values
for 1-5 that are listed in Table 1; the fitted curves are shown
in the figure.
The spectral intensities by this model should exhibit maxima,
as is observed only for 5 at temperatures below the MTHF
thawing point. This limitation renders our estimates of J
somewhat imprecise, despite the apparently similar shapes of
the curves for 1-4. Studies in solid state matrices may
eventually allow a wider temperature range for these Curie
studies. For example, in a p-dinitrobenzene matrix dinitrene 1
is stable for many months at room temperature.9 Under these
conditions, ∆ET-S ) 820 cal/mol for 1, reasonably similar to
(17) Davidson, E. R.; Feller, D.; Daasch, R.; Langhoff, S.; Elbert, S. T.;
Martin, R.; Rawlings, D.; McMurche, L.; Cave, R.; Phillips, P.; Day, S.;
Iberle, K.; Nitzsche, L.; Stenkamp, L.; Frey, G.; Jackels, C. Program
MELDF, Quantum Chemistry Group, Indiana University, Bloomington, IN,
1988 (QCPE 580).
(18) Teki, Y.; Fujita, I.; Takui, T.; Kinoshita T.; Itoh, K. J. Am. Chem.
Soc. 1994, 116, 11499. Cf. also: Sato, K. Ph.D. Thesis, Osaka City
University, 1992.
(19) Bleaney, B.; Bowers, K. D. Proc. R. Soc. London, A 1952, 214.
(20) All nonlinear least-squares analyses were carried out using the
program PSI-Plot v. 3.0 from Polysoft Corp in Salt Lake City, UT.

Characterization of Triplet States of Quinonoidal Dinitrenes
J. Am. Chem. Soc., Vol. 119, No. 9, 1997 2189
As an additional complication, other work has suggested that
additional mononitrenes (with zfs slightly smaller than those
of the monodeazetation products) may be formed by a slow
dark reaction of biradicaloid dinitrenes with MTHF.21 We found
the temperature versus intensity behavior for the ESR spectra
of 1-5 to be reproducible over the time scale of our experi-
ments, hence this latter reaction should cause only limited
imprecision in our results.
Discussion
Qualitative Zero Field Splitting Trends in 1-5. Biradical
1 was first generated in powdered matrix in trailblazing work
by Trozzolo et al.22 and was further analyzed in frozen solution
matrix by Singh and Brinen,23 who reported the zfs splitting as
|D/hc| ) 0.171 cm-1 (no E-value was described, although the
literature spectrum showed |E/hc| > 0). Singh and Brinen also
identified a dimerization mechanism for 1 under high concentra-
tion conditions. Subsequently, we reported4 the |D/hc| and |E/
hc| zfs parameters for 1 in frozen MTHF solution, as well as
the qualitative singlet ground state nature of 1. The present
work (Table 1) found ∆ET-S for 1 based on Figure 4a.
Biradical 1 has been known for a considerable time,22 but its
identification as a singlet biradical was an important theoretical
test of exchange models. In a simple molecular orbital (MO)
approximation, one may apply a Hund’s rule type argument to
predict that localized biradicals such as 1-5 would have triplet
ground states, since all have nearly degenerate SOMOs and two
unpaired electrons. But, by valence bond (VB) or spin-
polarization (SP) arguments, all of 1-5 are expected to have
singlet ground states by a small margin. The experimental
finding of singlet ground states for all of 1-5 is a powerful
argument for the VB and SP arguments that have been
eloquently set forth by Itoh and co-workers in their discussions
of biradical dicarbenes such as 11 and 12,24 which are related
by connectivity to 1 and 2 and which also have singlet ground
states with low-lying triplet excited states. Any perturbations
by the higher electronegativity of nitrogen in 1-5 are insuf-
ficient to reverse the simple qualitative effects of the SP model.
Moreover, biradicals 1 and 2 are structurally simpler tests of
the VB and SP models than are 11 and 12, which have
conformational flexibility about their divalent carbons that is
impossible for the corresponding dinitrenes.
Figure 2. Full-range ESR spectra from photolyses of 6-10 at 77 K
in MTHF (spectral traces a-e, respectively). The range in all spectra
is 100-9900 G. All spectra were obtained at ν0 ) 9.5-9.6 GHz. The
symbol “N” denotes a mononitrene peak.
the value of 580 cal/mol found in glassy MTHF. Although solid
state studies have complications associated with possible
microaggregation of photoprecursors and with guest-host
interactions in the crystal lattice (leading to different average
conformations in solid versus frozen solution), they hold
promise for considerable future work. For the remainder of
this contribution, however, we will confine ourselves to describ-
ing frozen MTHF matrix work.
The mononitrene region of the spectrum exhibits some
complexity as a function of time of photolysis and of precur-
sor. Weak mononitrene peaks may be observed with consider-
ably larger zfs than the major peaks associated with monodea-
zetation, especially in the longer conjugated systems. We
tentatively assign these to mononitrenes in which considerable
twisting of the π-network decreases conjugation of the π-elec-
tron at the nitrene site, leading to an increased zfs. We are
uncertain whether such twisting is due to a conformational mix
of the diazide precursors in the frozen matrix before photolysis
or whether twisting occurs during photolysis.
Preliminary ab initio SCF-MO-CI computations on biradical
3
1
1 showed nearly degenerate B1u and Ag states, with a slight
preference for the triplet in violation of the experimental result.25
More recent post-Hartree-Fock computations reverse this
preference, and find the 1Ag state to lie below 3B1u by 890 cal/
mol when the two states are both optimized at a 6-31G*
CASSCF(8,8) level of theory,26 in very good agreement with
the experimental gap of 500-900 cal/mol. The post-Hartree-
Fock optimizations are necessary to obtain the slight differences
in geometry between the 3B1u and 1Ag states which lead to the
(21) Harder, T.; Bendig, J.; Scholz, G.; Sto¨sser, R. J. Am. Chem. Soc.
1996, 118, 2497.
(22) Trozzolo, A. M.; Murray, R. W.; Smolinsky, G.; Yager, W. A.;
Wasserman, E. J. Am. Chem. Soc. 1963, 85, 2526.
(23) Singh, B.; Brinen, J. S. J. Am. Chem. Soc. 1971, 93, 540.
(24) Itoh, K. In Magnetic Molecular Materials, NATO ASI Series;
Gatteschi; D., Kahn, O., Miller, J. S., Palacio, F., Eds.; Kluwer: Dordrecht,
The Netherlands, 1991; p 67.
(25) Ichimura, A. S.; Lahti, P. M. Mol. Cryst. Liq. Cryst. 1993, 233, 33.
(26) Ichimura, A. S. Unpublished results.

2190 J. Am. Chem. Soc., Vol. 119, No. 9, 1997
Minato and Lahti
Figure 3. Biradical spectral regions (1000-5000 G) from photolysis of 6-10 at 77 K in MTHF (spectral traces a-e, respectively). All spectra
were obtained at ν0 ) 9.5-9.6 GHz. Dotted curves show spectral simulations using zfs parameters described in Table 1. All simulations used giso
) 2.003 amd Gaussian line shapes with varying line widths.
form of 2.31 The same group recently looked at the zfs and
Table 1. ESR Spectral Observations and Predictions for Species
Derived from Photolysis of 6-10
dinitrene ∆E(SfT)a |D/hc| (|E/hc|)b |Ddipolar/hc| |D/hc| nitrened
singlet-triplet gaps of 3 and the related 4,4′-azobenzene
dinitrene by ESR.32 Finally, based on recent work with planar-
c
constrained versions of 2 and with unconstrained 3, Harder et
1
2
3
4
5
574 (200) 0.169 (0.004)
580(202) 0.188 (e0.002)
480(168) 0.122 (<0.002)
677(237) 0.0865 (<0.002)
0.0156
0.0028
0.0015
0.0009
0.0004
0.90
0.83
0.74 (0.96)
0.66 (0.92)
0.63 (0.94)
al.21 suggested that a bisected form of 2 need not be invoked to
explain the extra mononitrene peaks observed by Yabe, but that
the new peak may be caused by products of dark reactions of
planar 2 with the matrix solvent over 2 h or more. In addition,
the studies of Harder et al. strengthened the argument that
formation of the quinonoidal dinitrenes occurs via sequential
deazetation reactions, rather than by double deazetation. The
various studies of biradicals 2 and 3 and their corresponding
mononitrenes have shown that not all the minutiae of these
photoreactions are understood. Fortunately, our primary interest
lies in comparison of trends among the zfs and ground state
spin multiplicities of the biradicals 1-5 and related mononi-
trenes, which are readily obtained by ESR spectroscopy alone.
Table 1 shows that the zfs values of both the biradicals and
the peaks in the corresponding mononitrene regions decrease
monotonically along the series 2-5. Biradical 1 has a smaller
zfs than one might expect from this series. However, 1 is the
only biradical of this set that does not have two phenyl rings,
and therefore may be inappropriate for zfs comparison to 2-5,
save as being a member of the general family of these
quinonoidal biradicals. The decease in mononitrene zfs as
conjugation is extended in 6-10 is consistent with increasing
delocalization of the mononitrene π-electron.
145(51)
0.0442 (<0.001)
a Singlet-triplet energy gaps in cal/mol (cm-1) from Figure 3 plots.
bExperimental zero field splitting (zfs) parameters for dinitrenes in cm-1
.
cPredicted zfs parameters for dinitrenes in cm-1 by the dipolar
approximation of eq 2. d Experimental zfs parameters for mononitrene
peaks associated with dinitrenes 1-5, in cm-1. Numbers in parentheses
are assigned to less conjugated dinitrene geometries, as described in
the text.
singlet ground state. Such difficulty in finding a qualitative
match between theory and experiment has been observed in
other weakly exchange coupled systems, notably tetramethyl-
eneethane (TME).27,28
Biradicals 2 and 3 were reported in early UV-vis studies by
Reiser and co-workers.29,30 We later described some ESR
spectroscopic studies of biradicals 2 and 3, and found them to
have singlet biradical ground states with apparently quinonoidal
nature.3,4 Yabe and co-workers have studied 2 by UV-vis and
IR spectroscopy, and used ESR spectroscopic evidence as a
function of irradiation time to suggest the existence of a bisected
(27) Nachtigall, P.; Jordan, K. D. J. Am. Chem. Soc. 1992, 114, 4743.
(28) Nachtigall, P.; Jordan, K. D. J. Am. Chem. Soc. 1993, 115, 270.
(29) Reiser, A.; Wagner, H. M.; Marley, R.; Bowes, G. Trans. Faraday
Soc. 1967, 63, 3162.
(30) Reiser, A.; Wagner, H. M.; Marley, R.; Bowes, G. Trans. Faraday
Soc. 1967, 63, 2403.
(31) Ohana, T.; Kaise, M.; Yabe, A. Chem. Lett. 1992, 1397. Ohana,
T.; Kaise, M.; Nimura, S.; Kikuchi, O.; Yabe, A. Chem. Lett. 1993, 765.
(32) Nimura, S.; Kikuchi, O.; Ohana, T.; Yabe, A.; Kaise, M. Chem.
Lett. 1996, 125. Cf. also the discussion of the comparative ESR spectra for
4,4′-stilbenedinitrene (3) and 4,4′-azobenzenedinitrene in ref 10.

Characterization of Triplet States of Quinonoidal Dinitrenes
J. Am. Chem. Soc., Vol. 119, No. 9, 1997 2191
Figure 4. Curie law plots of ESR spectral intensity (I) versus reciprocal absolute temperature (1/T) for dinitrenes 1-5. Fitted curves were obtained
by nonlinear least-squares analysis based on eq 1; in some cases an ordinate offset parameter was used to allow for a non-zero intercept.
Conformations of Biradicals 1-5. The narrowness of the
ESR spectral lines for 3-5 suggests that the triplet biradicals
are not formed in a distribution of greatly different conformers.
Further support for this notion was provided by photolysis of a
sample of diazide precursor 10 that consisted of a mixture of
geometric isomers, rather than being the all-trans material shown
in Figure 1. When all-trans 10 is photolyzed, the fairly clean
biradical spectrum of Figure 3e is observed. Figure 5 shows
the ESR spectrum from photolysis of the geometrically mixed
10. One can identify two sets of biradical peaks, one from the
all-trans biradical with |D/hc| ) 0.0442 cm-1, and an inner
spectrum with |D/hc| ≈ 0.018 cm-1, which we assign to a
biradical with one or more cis or otherwise twisted ethylenic
units. Using simulations (Figure 5, dotted curve) generated by
summing various ratios of the two spectra, an approximately
1:5 mix of all-trans to highly twisted biradicals appears to be
present in this sample. But, based on the simplicity of the
biradical spectra when all-trans polyenes are used, we feel the
evidence points to one major conformation for each triplet ESR
spectral carrier, presumably an extended or nearly extended
conformation.
The mononitrene regions of the spectra in Figures 2 and 5
also suggest that some complexity of geometry is possible. In
addition to the mononitrenes whose zfs parameters are descrbed
in Table 1 as corresponding to partial photolysis of diazides
6-10, it is also possible to discern other peaks corresponding
to mononitrenes with large zfs. Such a peak is not observed
from diazide 6, which has no possible conformational uncer-
tainty. Biphenyl-based diazide 7 is known to show complex
behavior of its mononitrene region, for reasons that have been
assigned both to conformational torsion31,32 and to low-
temperature reactions21 in the matrix. However, the variation
in zfs of the mononitrene peaks derived from 7 is fairly small.
Photolysis of diazides 8-10 could in principle lead to
conformational mixtures, if bond rotations or cis-trans isomer-
ization were to occur in the matrix due to local heating effects.
Table 1 shows in parentheses the zfs values of low-intensity
mononitrene peaks which appear to be due to such effects. The
high zfs in these species probably is due to production of
deconjugating torsion during photolysis (or possibly due to
photolysis of highly twisted conformers that are frozen in place
by the low-temperature matrix). Torsional deconjugation
reduces delocalization of the mononitrene π-electron, causing
the higher zfs. A variety of geometric torsions is possible as
the polyene chain length extends. Despite the presence of these
“extra” dinitrene peaks, it appears that fairly specific geometries
are required to produce the biradical spectra of 1-5, based on
the narrow lines in these spectra. This is consistent with the
notion that torsion which decreases conjugation would both
increase zfs in the mononitrenes derived from 8-10 and tend
to truncate the exchange interactions which lead to the biradical
states. This analysis, of course, evades the possibility of nitrenes
derived from arylnitrene “ring-walk” rearrangement chemistry,
which might well occur upon extensive photolysis.33 In general,
our assignment of the aryl mononitrene ESR peaks is perhaps
best supported for putative structures with low zfs caused by
increasing conjugation of the π-electrons, with the assignments
for the low-intensity, higher zfs peaks being more tentative.
Ground State Spin Multiplicity. One may consider the
expected ground state spin multiplicities of 1-5 from either a
molecular orbital (MO) or valence bond (VB) point of view. In
a MO model, Hund’s multiplicity rules have frequently been
(33) Review of the rearrangement chemistry of arylnitrenes may be found,
for example, in Wentrup, C. ReactiVe Molecules. The Neutral ReactiVe
Intermediates in Organic Chemistry; Wiley: New York, 1984; pp 212-
219 and references cited therein.

2192 J. Am. Chem. Soc., Vol. 119, No. 9, 1997
Minato and Lahti
Figure 6. Spin polarization diagram for dinitrene 1, and a spin-spin
density map for 1 (6-31G* UHF-SCF level, standard SPARTAN
settings). In the spin polarization diagram, the bottom π-orbital lobes
of distant carbons were omitted for clarity in presentation.
Figure 5. ESR spectrum at 9.58 GHz from photolysis at 77 K in frozen
MTHF of a mixture of geometric isomers of diazide 10 (trace a, 0-10,-
000 G). Trace b shows a 1000-5000 G blowup of curve a (solid line),
compared to the simulation (dotted line) of a spectrum that is a 1:5
of the spin density in the triplet states of 1-5 is localized in
two singly-occupied σ orbitals (SOMOs) on the terminal
nitrogen atoms. These large spin populations polarize the
π-electrons that are nearby, particularly the nitrogen π-elec-
trons. It is important to remember that the total π spin den-
sity sums to zero oVer all sites, but that indiVidual sites may
be polarized to haVe positiVe or negatiVe non-zero densities.
Each nitrogen SOMO has a large R spin density, which by SP
causes a small, positive R spin density in the nitrogen π-space.
Throughout the rest of the π-orbital subspace, one expects the
usual SP rules to apply, such that each orbital of R spin den-
sity prefers to be attached to sites with â spin density, and vice
versa. This is illustrated for biradical 1 in Figure 6. The same
sum of triplet spectra with |D/hc| ) 0.0442 cm-1 (E ) 0.000 cm-1
)
and |D/hc| ) 0.0180 cm-1 (E ) 0.000 cm-1); both simulations use giso
) 2.003 and Gaussian line shapes. The symbol “N” denotes a
mononitrene peak.
applied to molecules having degenerate or near-degenerate
biradical orbitals, with the result that a triplet ground state is
predicted. However, in a variety of conjugated systems the
Hund criterion is experimentally not followed, but rather a low-
spin singlet state is found.34 TME and its derivatives are simple
hydrocarbon examples of such systems in the disjoint class of
connectivity.35
The low-spin natures of TME and related disjoint systems
have been interpreted as supporting VB-type models of elec-
tronic exchange. Various derivations have been given by
Ovchinikov,36,37 Klein,38-40 Borden and Davidson,35,41 and
Sinanoglu,42 inter alia, in order to model intramolecular
exchange in connectivity-conjugated systems that are not
properly treated by the simplistic MO plus Hund’s rule model.
A more general theme involves use of spin polarization (SP)
arguments to understand the distribution of spin densities in an
open-shell system.
3
figure also shows a spin density map of the B1u state of 1,
using a 6-31G* UHF wave function and the SPARTAN
graphical interface.
For triplet states of 1-5 a spin mismatch occurs in the
π-space (Figure 6), with destructive interference occurring
between adjacent sites having the same sign of spin density,
destabilizing the triplet relative to the singlet state. Thus, the
singlet should be the ground state by a small amount, as is
experimentally observed. Any effects of heteroatom substitution
appear insufficient to reverse SP effects favoring a singlet state.
In a similar manner, the connectivity analogous dicarbenes 11-
1243 are found to be low-spin ground states with low-lying triplet
biradical states. The spin density distribution in 12 is found
by ENDOR to be qualitatively just as shown in Figure 6, further
supporting the SP model for 1-5.
A Model for Zero Field Splitting Magnitudes in 1-5. The
same SP model used to explain the qualitative ground state
multiplicities for 1-5 can explain the semiquantitative trends
in zero field splitting (zfs). In the most simplistic picture of
these biradicals, the triplet state has two electrons localized into
different σ SOMOs on the terminal nitrogens. A simple dipole-
dipole approximation to the zfs may be made using eq 2,44 where
SP arguments are particularly useful for interpreting the
behavior of localized biradicals 1-5, since these are not readily
treated by the connectivity-based models described above. Most
(34) Borden, W. T.; Iwamura, H.; Berson, J. A. Acc. Chem. Res. 1994,
27, 109.
(35) Borden, W. T.; Davidson, E. R. J. Am. Chem. Soc. 1977, 99, 4587.
(36) Misurkin, I. A.; Ovchinnikov, A. A. Russ. Chem. Res. (Engl). 1977,
46, 967.
(37) Ovchinnikov, A. A. Theor. Chim. Acta 1978, 47, 297.
(38) Klein, D. J.; Nelin, C. J.; Alexander, S.; Matsen, F. E. J. Chem.
Phys. 1982, 77, 3101.
(39) Klein, D. J. Pure Appl. Chem. 1983, 55, 299.
(40) Klein, D. J.; Alexander, S. A. In Graph Theory and Topology in
Chemistry; King, R. B., Rouvray, D. H., Eds.; Elsevier: Amsterdam, The
Netherlands, 1987; Vol. 51, pp 404.
(41) Diradicals; Borden, W. T., Ed.; John Wiley and Sons, Inc.: New
York, 1982.
(42) Shen, M.; Sinanoglu, O. In Graph Theory and Topology in
Chemistry; King, R. B., Rouvay, D. H., Eds.; Elsevier: Amsterdam, The
Netherlands, 1987; Vol. 51, pp 373-403.
(43) Teki, Y.; Sato, K.; Okamoto, M.; Yamashita, A.; Yamaguchi, Y.;
Takui, T.; Kinoshita, T.; Itoh, K. Bull. Magn. Reson. 1992, 14, 24.
(44) A variant of eq 2 with different units is found in: Eaton, S. S.;
More, K. M.; Sawant, B. M.; Eaton, G. R. J. Am. Chem. Soc. 1983, 105,
6560.

Characterization of Triplet States of Quinonoidal Dinitrenes
J. Am. Chem. Soc., Vol. 119, No. 9, 1997 2193
g is the free electron g value and rij is the distance in angstroms
between interacting dipoles.
ogy, but should suffice for order of magnitude estimates to
determine the major contributions to the overall zfs of 1. Details
of the geometry and spin population numbers used in the
calculation are given in the Supporting Information. While
variation in the exact spin populations used in this calculation
will effect the value estimated for 1, we shall see below that
the simple extension of the dipolar model to include one-center
interactions, as done in the earlier work of Serre and Schneider,
makes a critical difference in the order of magnitude estimate
of the zfs.
Table 1 gives the zfs expected from eq 3 when π-spin
polarization is ignored, and all spin density is assumed to be in
the localized orbitals. All such estimated zfs are found to be
far too small to fit the experimental facts. But, when the 6-31G*
SCF-MO-SDTQ-CI spin density distribution is used for 1sthus
incorporating one-center interactionssthe estimated zfs for 1
is ∼0.2 cm-1, reasonably close to the observed value of 0.169
cm-1. The one-center s-p interactions on the nitrogen sites
contribute most of the zfs in the estimate based upon eq 3. As
a result, we feel that the SP model for the triplet state of 1 is
both appropriate and sufficient to explain its large zfs.48
At present, we have not extended this semiquantitative model
to the larger systems 2-5, due to the difficulty involved in ab
initio spin density computations for the larger systems. Single-
determinant UHF calculations tend to be badly spin contami-
nated, even with 6-31G* and higher basis sets. In addition,
without reasonably good geometries, we feel that spin density
distribution estimates for 2-5 will not be quantitatively useful.
Multiconfigurational CASSCF calculations appear to be promis-
ing for spin density distributions in these molecules but are
computationally very demanding. At present, we are still
working on a practical theoretical approach to this problem. Still,
the qualitative SP-based arguments appear to be consistent with
experimental facts observed for zfs trends in 2-5.
Summary. The singlet state biradicals 1-5 are the products
of fascinatingly complex photochemistry. Although this class
of biradicals have been known for about 30 years, it has
remained for recent work to show clearly that the triplet ESR
spectra observed for these molecules are due to excited states,
and that the unusually large zfs observed for the triplets is
commensurate with the spin polarization models that are have
become most useful to describe open shell molecules. Although
these molecules are not obvious candidates for practical use in
the development of new organic-based magnetic materials, they
are of considerable use in allowing the investigation of the
exchange interaction between unpaired electrons, a key facet
in molecular magnetism. We anticipate that these biradicals
will be of further interest in solid state matrix investigations9
that are continuing at present.
3
D (cm-1) ) 1.299g/rij
(2)
This equation shows that zfs should decrease rapidly as rij
increases. Table 1 compares experimental zfs for 1-5 to zfs
predicted using eq 2, where rij is taken as the distance between
the nitrogen atoms in a triplet UHF AM1 optimized geometry
for each biradical. Use of slightly more sophisticated modelss
such as splitting the unpaired spin population on each nitrogen
into the two halves of a localized Slater type p-orbitalsdoes
not change these numbers appreciably. As the distance between
nitrogen atoms increases in 1-5, the predicted zfs rapidly
decreases to a negligible amount. This is not what is experi-
mentally observed. Conformational effects cannot be invoked
to explain the whole discrepancy, since planar 1-3 already have
zfs far larger than expected by the dipole-dipole model. In
addition, none of 1-3 have sufficient conformational flexibility
to allow the terminal nitrogens to come close enough to explain
the observed zfs by a through space effect.
But, if unpaired spin density is distributed throughout the
molecules by the SP model, there are interactions between the
large, localized spin population in the nitrogen σ space and the
small, polarized spin sites that are distributed throughout the
conjugated π-space. All of the π-spin populations are quite
small, but due to the rij3 dependence of the dipolar interaction,
the one-center terms involving the large localized σ spin density
and the small π-spin density on each nitrogen dominate. This
one-center interaction looks similar to the interaction which
makes the zfs in a mononitrene so large,45 but is different
because the nitrogen π-spin population in 1-5 is very small.
In fact, the R π-spin density on each nitrogen in the biradicals
is explicable only when spin polarization effects are considered,
and hence is not a simple zeroth-order effect. As the length
grows of the π-conjugated network separating the localized
nitrogen electrons, the polarized π-spin density on each nitrogen
grows smaller through delocalization, while the localized spin
density remains about the same. As a result, the zfs should
decrease throughout the series 2-5, as observed.
An early model of the zfs for 1 was given by Serre and
Schneider,46 using a simple MO approximation to the spin
density distribution that is qualitatively similar to the depiction
of Figure 5. Details of the calculation were not given, but the
predicted result of (-)0.181 cm-1 compared favorably to the
known zfs of 0.17 cm-1. In an effort to make a similar estimate,
we used the Lowdin spin density distribution at a 6-31G* SCF-
MO-SDTQ-CI level of theory.25 We then computed the sum
of purely dipolar zfs contributions between all pairs of spin
density sites Fi and Fj, as shown in eq 3. We simulated each
Experimental Section
3
D (cm-1) ) 1.299g( F F /r )
(3)
∑
i
j
ij
General. All chemicals were obtained from Aldrich Chemical Co.
unless otherwise noted. 2-Methyltetrahydrofuran was distilled from
lithium aluminum hydride and tetrahydrofuran from potassium/ben-
zophenone, both under argon.
i>j
π-type spin density site as a point spin split equally onto the
two lobes of a p-orbital, with each half-density placed at a
distance from the associated atom equal to the radius of the
equivalent Slater type orbital. This method follows the approach
used by Berson and co-workers for zfs predictions, details of
which are given elsewhere.47 This is an approximate methodol-
Ultraviolet-visible spectra were obtained on a Shimadzu UV-260
double-beam spectrometer, and tetrahydrofuran was used as a solvent
in all measurements. Infrared spectra were obtained on a Perkin-Elmer
1420 spectrophotometer, and were referenced against polystyrene at
1601 cm-1
.
1H-NMR spectra were obtained on an IBM Instruments
NR-80A Fourier transform spectrometer, and were referenced against
(45) Wasserman, E. Prog. Phys. Org. Chem. 1971, 8, 319.
(46) Serre, J.; Schneider, F. J. Phys. Chim. 1964, 61, 1655.
(47) Rule, M.; Matlin, A. R.; Seeger, D. E.; Hilinshi, E. F.; Dougherty,
D. A.; Berson, J. A. Tetrahedron 1982, 38, 787. Cf. also: Hilinski, E. F.
Ph.D. Thesis, Yale University, New Haven, CT, 1982. The latter reference
contains a listing of the program used for zfs prediction by the methods
listed this paper.
(48) It is also plausible that spin-orbit coupling contributes to the overall
zfs of 1-5. However, the spin-orbit coupling constant of atomic nitrogen
is small, only 76 cm-1 (Gordy, W. In Techniques of Chemistry; West, W.,
Ed.; Wiley, New York, 1980; Vol. XV, p 604). While a spin-orbit
contribution is thus possible, the analysis in the main text suggests that
this contribution is not required to explain the size of the zfs.

2194 J. Am. Chem. Soc., Vol. 119, No. 9, 1997
Minato and Lahti
internal tetramethylsilane on the delta (δ) scale in parts per million.
Chemical analyses were performed by the University of Massachusetts
Microanalysis Laboratory.
which decomposed and turned black upon heating at 120-130 °C.
Samples of the solid darkened within 3 min of exposure to fluorescent
lighting. Anal. Calcd: C, 66.7; H, 4.2; N, 29.1. Found: C, 66.43;
H, 4.01; N, 28.83. 1H-NMR (CDCl3, 80 MHz; ppm): δ 6.6-7.0 (m,
4 H, -CHd), 7.0-7.4 (para AA′BB′ q, 8H, J ) 8.7 Hz, Ar-H). UV-
vis (λmax [ꢀ]; nm [M-1 cm-1]): 361 [47 500], 301 [33 700], 348
[39 100]. IR (KBr; cm-1): 2130 (s, -N3 str), 990 (s, trans -CHdCH-
oop bend).
1,8-Bis(p-azidophenyl)-1,3,5,7-octatetraene (10). Method A: In
a 100-mL round-bottom flask were placed 37 mL of distilled water,
2.7 mL of concentrated sulfuric acid, and 2.25 g (7.80 mmol) of 1,8-
bis(p-aminophenyl)-1,3,5,7-octatetraene (Supporting Information). The
mixture was heated to 60 °C by an oil bath and stirred for 2 h, then
quickly cooled with an ice bath and treated with dropwise addition of
a solution of 1.19 g (17.2 mmol) of sodium nitrite in 3.7 mL of distilled
water over 40 min. The mixture was stirred for an additional 20 min
in the ice bath and then diluted with 148 mL of cold distilled water
and treated with a solution of 1.01 g (15.5 mmol) of sodium azide in
5.0 mL of distilled water. The mixture foamed and a precipitate
appeared, as the reaction color changed from deep purple to colorless.
The precipitate was collected by filtration in subdued light and dried
under vacuum to give 0.5 g (19%) of ruddy solid product. Samples of
the solid turned dark within 3 min of exposure to fluorescent lighting.
Upon heating, the compound blackened without liquefaction at
temperatures above 180 °C. 1H-NMR (DMSO-d6, 80 MHz): δ 6.4-
6.9 (br m, 8 H, -CHd), 6.9-7.4 (para AA′BB′ q, 8 H, J ) 8 Hz,
Ar-H). IR (KBr; cm-1): 2110 (s, -N3 str).
Method B: All glassware was flame dried before use. In a 250-
mL three-necked round-bottom flask were placed 140 mL of absolute
ethanol, 4.5 g (7.0 mmol) of 4-azidobenzaldehyde (Supporting Informa-
tion), and 2.6 g (18 mmol) of hexa-2,4-diene-1,6-diylbis(tributylphos-
phonium) dibromide (Supporting Information). The flask was purged
with nitrogen. A sodium ethoxide solution was prepared by dissolving
0.48 g (21 mmol) of sodium metal in 21 mL of absolute ethanol. The
flask was placed in an ice bath, then treated dropwise with a sodium
ethoxide solution while stirring. The flask was warmed to room
temperature and stirred under nitrogen overnight. This compound is a
mixture of geometric isomers which may be isomerized by boiling in
acetone with a crystal of iodine for up to 10 h under inert atmosphere
in the dark. The final product gives the same spectoscopic data as the
product produced by method A. Upon heating, the compound did not
melt, but blackened without liquefaction at temperatures above 180
°C. No further purification was done due to the photolytical instability
of the product.
Diazide Photolysis and Variable Temperature Analyses. Azides
were irradiated for 5 min using a 1000-W Xenon arc lamp with Pyrex
filter at 77 K. The samples were prepared by dissolving the appropriate
azide compound in dry 2-methyltetrahydrofuran, approximate concen-
tration of 1 mg/1 mL. The sample was placed in a Wilmad 707SQ
Suprasil 4 mm o. d. quartz ESR tube, subjected to three freeze-pump-
thaw vacuum degassing cycles, and sealed under vacuum. Next, the
sample was frozen at 77 K in a Wilmad liquid nitrogen finger dewar
(WG-816-B-Q Suprasil), subjected to photolysis, and directly placed
into a precooled ESR cavity. For fixed 77 K experiments, the cavity
cooling was provided by a second finger dewar charged with liquid
nitrogen. For lower temperature work, an Oxford Instruments ESR-
900 liquid helium cryostat precooled to 50 K was used, and the sample
transferred from the finger dewar into the crystat with sufficient
swiftness that sample thawing did not occur. Typically, upon thawing
the irradiated sample solutions turned deep red within 60 s, then faded
within another 60 s to a straw colored solution containing a red solid.
For all Curie plots, intensities were judged by double integration of
a reasonably strong, free-standing peak in each spectrum. Integration
limits were kept constant for any spectrum as the temperature was
varied. Intensity measurements were obtained both with decreasing
and increasing temperature, to assure that there were no appreciable
effects of sample decomposition or annealing over the temperature
ranges used. Temperatures were measured in the Oxford ESR-900
cryostat by an AuFe/chromel thermocouple placed just below the sample
position; the temperatures were calibrated against carbon electrodes.
Saturation effects were avoided by running all spectra at microwave
powers of e1 mW (100 kHz modulation amplitude).
Electron spin resonance (ESR) spectra were obtained on a Bruker
ESP-300 X-band spectrometer attached to a standard Bruker data system
allowing multiscan spectral digitization and analysis. ESR peak
positions are reported in units of gauss referenced against neat
diphenylpicrylhydrazyl (g ) 2.0037), and peak intensities were obtained
by computer-aided double integration using standard Bruker software
routines. Further details of ESR spectral acquisition are described in
the section on Nitrene Generation. All ESR spectral simulations were
carried out on a Silicon Graphics R4000 computer using a program
that uses the eigenfield method described elsewhere.18
1,4-Diazidobenzene. In a 200-mL round-bottom flask was placed
a solution of 1.0 g (15 mmol) of sodium azide in 77 mL of distilled
water. The flask was placed in an ice bath, treated with 1.9 g (6.3
mmol) of 1,4-phenylenebis(diazonium tetrafluoroborate) (see Supporting
Information), and stirred for 30 min. The resultant precipitate was
collected by filtration, washed with water, then dissolved in ethyl ether.
The ether solution was dried over magnesium sulfate and treated with
decolorizing charcoal. The solvent was evaporated to give 0.7 g (70%)
of light yellow/orange powder with mp 80-81 °C (lit.30 mp 80 °C)
Anal. Calcd: C, 45.0, H, 2.5; N, 52.5. Found: C, 44.50; H, 2.66; N,
51.90%. 1H-NMR (80 MHz, CDCl3): δ 7.6 (s, Ar-H). IR (KBr; cm-1):
2080 and 2110 (s, -N3 str). UV-vis (λmax [ꢀ]; nm [M-1 cm-1]):
272 [12 400].
4,4′-Diazidobiphenyl (7). This compound was synthesized as
described previously by us.3 The product was a buff-colored solid with
mp 120-122 °C (lit.30 mp 127-128 °C). Anal. Calcd: C, 61.0; H,
3.4; N, 35.6. Found: C, 61.06; H, 3.78; N, 33.4. 1H-NMR (CDCl3,
80 MHz): δ 6.8-7.2 (para AA′BB′ q, 4 H, J ) 8.0 Hz, Ar-H). IR
(KBr; cm-1): 2130 (s, -N3 str). UV-vis (λmax [ꢀ]; nm [M-1 cm-1]):
297 [36 700].
(E)-4,4′-Diazidostilbene (8). In a 30-mL round-bottom flask were
placed a solution of 2.1 mL of concentrated hydrochloric acid in 3.8
mL of distilled water, followed by the addition of 1.0 g (4.8 mmol) of
(E)-4,4′-diaminostilbene (Supporting Information). The flask was
placed in an ice bath. The mixture was treated dropwise with a solution
of 0.69 g (10 mmol) of sodium nitrite in 2.4 mL of distilled water.
The mixture was stirred for an additional 1 h in the ice bath, then filtered
to remove any undissolved impurities. The solution was placed in the
ice bath and treated with a solution of 0.62 g (9.5 mmol) of sodium
azide in 2.4 mL of distilled water. Gas evolution was observed
immediately after the addition of sodium azide solution. The mixture
was stirred in the ice bath for 15 min. The resultant precipitate was
collected by filtration, dried under vacuum, dissolved in chloroform,
treated with decolorizing charcoal, and dried over magnesium sulfate.
The solvent was evaporated and the resultant solid was dried under
vacuum. The product was 0.71 g (57%) of light brown solid with mp
154-155 °C (lit.30 mp 164 °C). Anal. Calcd: C, 64.1; H, 3.8; N,
32.9. Found: C, 63.19; H, 3.73%; N, 31.42. 1H-NMR (CDCl3, 80
MHz; ppm): δ 7.0 (s, 4 H, -CHd), 6.95-7.75 (para AA′BB′ q, 8 H,
J ) 8.8 Hz, Ar-H). IR (KBr; cm-1): 2130 (s, -N3 str), 960 and 975
(s, trans -CHdCH- oop bend). UV-vis (λmax [ꢀ]; nm[M-1 cm-1]):
341 [34 400].
(E,E)-Bis(p-azidophenyl)-1,3-butadiene (9). In a 25-mL round-
bottom flask were placed 0.3 g (1.3 mmol) of (E,E)-bis(p-aminophenyl)-
1,3-butadiene (Supporting Information) and 1.6 mL of a 5.6/10 (v/v)
solution of dilute hydrochloric acid in water. The flask was placed
into an ice bath. The mixture was then treated with a solution of 0.18
g (2.6 mmol) of sodium nitrite in 0.6 mL of distilled water over 30-
40 min, the solution foamed during the addition of sodium nitrite
solution. The mixture was further stirred for 1 h in the ice bath. The
mixture was treated dropwise with a solution of 0.17 g (2.6 mmol) of
sodium azide in 0.6 mL of distilled water; the mixture foamed during
the addition of sodium azide solution. The mixture was next diluted
with distilled water and stirred for 20 min. The resultant precipitate
was collected by filtration and dried under vacuum, redissolved in
methylene chloride, treated with decolorizing charcoal, and dried over
magnesium sulfate. The solvent was evaporated, and the residue was
passed through a silica gel column with methylene chloride as the
eluent. The solvent evaporation gave 0.07 g (18%) of orange solid

Characterization of Triplet States of Quinonoidal Dinitrenes
J. Am. Chem. Soc., Vol. 119, No. 9, 1997 2195
Acknowledgment. This work was supported in part by the
National Science Foundation (CHE-9204695 and CHE-9521594).
Acknowledgment is made to the donors of the Petroleum
Research Fund, administered by the American Chemical Society,
for support of this work (PRF 25968-AC4 and 29379-AC4).
We thank Profs. Takeji Takui and Kazunobu Sato for a copy
of the general eigenfield ESR line shape fitting computer
program18 used by us.
Supporting Information Available: Synthetic and analytical
data for a number of intermediates in the syntheses of diazides
8-10 and for 4-azidobenzaldehyde, a tabular summary of peak
positions for the ESR spectra in Figure 2, and details of data
used in the zfs estimate for 1 (8 pages). See any current mast-
head page for ordering and Internet access instructions.
JA963648D