2,3,5,6-Tetrafluorophenylnitren-4-yl: Electron paramagnetic resonance spectroscopic characterization of a quartet-ground-state nitreno radical

Article abstract of DOI:10.1021/ja078171s

2,3,5,6-Tetrafluorophenylnitren-4-yl (5) was synthesized in argon at 4 K via the photolysis of 2,3,5,6-tetrafluoro-4-iodo-phenyl azide (6). Electron paramagnetic resonance (EPR) spectroscopy allows us to observe triradical 5 in its quartet state with the

Full text of DOI:10.1021/ja078171s

Published on Web 03/08/2008
2,3,5,6-Tetrafluorophenylnitren-4-yl: Electron Paramagnetic
Resonance Spectroscopic Characterization of a
Quartet-Ground-State Nitreno Radical
Wolfram Sander,*,† Dirk Grote,† Simone Kossmann,‡ and Frank Neese*,‡
Lehrstuhl fu¨r Organische Chemie II, Ruhr-UniVersita¨t Bochum, D-44780 Bochum, Germany, and
Lehrstuhl fu¨r Theoretische Chemie, UniVersita¨t Bonn, D-53115 Bonn, Germany
Received October 25, 2007; E-mail: wolfram.sander@rub.de
Abstract: 2,3,5,6-Tetrafluorophenylnitren-4-yl (5) was synthesized in argon at 4 K via the photolysis of
2,3,5,6-tetrafluoro-4-iodo-phenyl azide (6). Electron paramagnetic resonance (EPR) spectroscopy allows
us to observe triradical 5 in its quartet state with the zero-field splitting (ZFS) parameters |D/hc| ) 0.285
and |E/hc| ) 0.043 cm-1. The quartet ground state of 5 is in accordance with our previous infrared (IR)
spectroscopic investigation, in which the high-spin quartet state, but no low-spin doublet state, of 5 was
observed in solid argon at 4 K [Wenk, H. H.; Sander, W. Angew. Chem., Int. Ed. 2002, 41, 2742-2745].
Because annealing of the matrix at temperatures of >10 K results in the rapid recombination of the highly
reactive species 5 with I atoms produced during the photolysis of 6, the Curie-Weiss behavior could not
be investigated. However, the absence of low-spin states in the IR investigations, as well as the results of
ab initio and density functional theory (DFT) calculations, strongly suggest that 5 has a robust quartet
ground state that is best-described as an unprecedented σ,σ,π-triradical. The ZFS of 5 has been successfully
reproduced by DFT calculations, which furthermore provide qualitative insight into the origin of the observed
EPR parameters.
Scheme 1. Mechanism of the Rearrangement of Phenylnitrene 1
Introduction
Arylnitrenes (1; see Scheme 1) are important reactive
intermediates with triplet ground states and large singlet triplet
splittings. Convenient precursors for spectroscopic studies are
the corresponding aryl azides 2, which photochemically or
thermally split off N2 to produce the nitrenes 1. The kinetics,
spectroscopy, and computational chemistry of these intermedi-
ates has recently been reviewed.1 Under the conditions of matrix
isolation a major photochemical process that diminishes the yield
of nitrenes 1 is the ring expansion to 1,2-didehydroazepines 4,
via benzazirines 3 as intermediates. After many years of intense
experimental and theoretical work in this field, the mechanistic
details are now well established (see Scheme 1).
The introduction of a radical center in para position of the
phenyl ring leads to phenylnitreno radicals, which represent a
new class of reactive intermediates.2-4 The first example of these
intermediates that could be isolated in low-temperature matrices
is 2,3,5,6-tetrafluorophenylnitren-4-yl 5.2 Ultraviolet (UV) pho-
tolysis (254 nm) of aryl azide 6, matrix-isolated in argon at 4
K, produced nitrene 7 that could be easily identified by
comparison of the experimental infrared (IR) spectrum with
density functional theory (DFT) calculations (UB3LYP/6-311G-
(d,p)). Further UV irradiation resulted in the formation of two
new products: ketenimine (9) (presumably formed via azirine
(8) that could not be detected) and nitreno radical 5 (see Scheme
2). The radical 5 was only observed if the matrix was kept at a
temperature of 3-4 K during irradiation, whereas at 10 K, only
traces of 5 were formed. The formation of 5 is reversible, and
annealing the matrix resulted in recombination with the I atom
to yield nitrene 7. This explains why the nitreno radical 5 was
only observed at temperatures of <10 K.
The IR spectra calculated for the lowest-lying doublet state
D-5 and quartet state Q-5 exhibit characteristic differences
that allowed these two states to be differentiated by IR
spectroscopy. The IR spectrum of matrix-isolated 5 was in good
agreement with that calculated for Q-5, whereas D-5 was not
† Ruhr-Universita¨t Bochum.
‡ Universita¨t Bonn.
(1) Gritsan, N. P.; Platz, M. S. Chem. ReV. 2006, 106, 3844-3867.
(2) Wenk, H. H.; Sander, W. Angew. Chem., Int. Ed. 2002, 41, 2742-2745.
(3) Bettinger, H. F.; Sander, W. J. Am. Chem. Soc. 2003, 125, 9726-9733.
(4) Sander, W.; Winkler, M.; Cakir, B.; Grote, D.; Bettinger, H. F. J. Org.
Chem. 2007, 72, 715-724.
9
4396
J. AM. CHEM. SOC. 2008, 130, 4396-4403
10.1021/ja078171s CCC: $40.75 © 2008 American Chemical Society

EPR Characterization of a Quartet Nitreno Radical
A R T I C L E S
Scheme 2. Photochemistry of Aryl Azide (6)
Figure 1. Electron paramagnetic resonance (EPR) spectra (9.5877 GHz)
produced by 308-nm irradiation of arylazide (6), matrix-isolated in argon
at 4 K: (a) spectrum obtained after short-time irradiation is dominated by
unspecified radicals and triplet nitrene (T-7) with |D/hc| ) 1.108 cm-1
and |E/hc| ) 0.012 cm-1 (g ) 2.003); (b) same matrix after prolonged
irradiation. Several new signals are assigned to quartet nitreno radical
(Q-5).
observed.2 This suggests a quartet ground state of 5, in
accordance with the results from a theoretical study, which
predict quartet ground states for aryl nitreno radicals if the
radical center is in the ortho or para position of the phenyl ring,
whereas a radical center in the meta position should result in a
doublet ground state.3 Attempts to synthesize meta-aryl nitreno
radicals to verify the latter assumption failed, because these
systems easily ring-open to thermally preferred nitriles.4
a significant contribution of spin-orbit coupling (SOC) to the
ZFS parameters of T-7 is very likely. This notion is also
consistent with our preliminary calculations.
Results and Discussion
To obtain more insight into the effect of the halide substituent
in the para position on the ZFS, calculations were performed
EPR Spectrum of T-7. An argon matrix doped with small
amounts of aryl azide 6 was deposited on top of a copper rod
that was maintained at 4 K. The copper rod was then positioned
inside a quartz tube (within the high-vacuum system), which
allowed the matrix-isolated azide 6 to be irradiated with a XeCl
excimer laser (308 nm) and EPR spectra (X-band spectrometer)
of the photoproducts to be recorded. After irradiation with a
few laser pulses, three strong signals appear in the area between
7000 and 8500 G, which are assigned to the x2, y2, and z1
transitions of triplet nitrene T-7. A simulated EPR spectrum
with the parameters g ) 2.003, |D/hc| ) 1.108 cm-1, and |E/
hc| ) 0.012 cm-1, assuming randomly oriented molecules,
nicely reproduces the experimental spectrum. The large zero-
field splitting (ZFS) parameter D is characteristic of triplet
arylnitrenes, which generally show values in the range of 0.8-
1.1 cm-1, depending on the substituents.5 The splitting of the
x- and y-signals in the EPR spectrum of T-7 of ∼600 G
corresponds to a ZFS parameter |E/hc| of 0.012 cm-1, which is
larger than that in most arylnitrenes. Thus, phenylnitrene (|D/
3
for models 7-X with X ) H, F, Cl, Br, I. Using B3LYP/
3
TZVPP, the D value for 7-H is calculated to be 0.978 cm-1
with E/D ) 0.0074. This value is well within the range typically
observed for aromatic nitrenes. Changing the para-substituent
from H to F leads to a moderate increase of the value of D (to
0.993) while the value of E/D decreases by a factor of ∼2 (to
0.003). This is attributed to the higher symmetry of the system,
compared to that of 37-H. On the other hand, replacement of F
by Cl yields D ) 1.048 and an increased E/D value of 0.056.
Unfortunately, for the heavier halides (Br and I), we apparently
face a breakdown of perturbation theory and the calculations
provide unrealistic values for the ZFS. Hence, we have focused
our attention on 37-Cl, where no problems are apparent. In this
compound, the spin-spin interaction accounts for 0.882 cm-1
of D while the spin-orbit contribution already amounts to 0.227
cm-1, which is a factor of ∼2 larger than the SOC contribution
in 37-H. In fact, practically all of the increase in the E/D values
between 37-H and 37-Cl results from the spin-orbit contribution.
This is readily rationalized by the mechanisms that govern the
ZFS. For the spin-spin contribution, it will be explained in
detail below that the principal axis points along the C-N bond.
However, for the SOC contribution, the easy axis is perpen-
dicular to the C-N bond and the plane of the molecule. Hence,
after adding the spin-spin and spin-orbit contributions together,
increased rhombicity will result.
hc| ) 0.9978 cm-1, |E/hc| < 0.002 cm-1 5
) and 2,3,5,6-
tetrafluorphenylnitrene (|D/hc| ) 1.051 cm-1, |E/hc| ) 0.0008
cm-1) show E values that are close to zero. This indicates a
significant influence of the iodine substituent on the ZFS
parameters of T-7. Hall et al. investigated several 4-substituted
phenyl nitrenes in methylcyclohexane at 77 K and observed,
compared to the remainder of the investigated derivatives, a
very broad EPR spectrum for 4-iodophenylnitrene.6 This was
attributed to hyperfine coupling (HFC) with the iodine sub-
stituent. In our experiments in argon matrices, no HFC was
observed, but a very well-resolved splitting between transitions
x2 and y2 was observed. It is well-known that not only spin-
spin dipolar interactions contribute to the ZFS parameters, but
also spin-orbit effects.7,8 Because iodine (I) is a heavy atom,
EPR Spectrum of Q-5. Prolonged irradiation with the XeCl
excimer laser (308 nm) at 4 K leads to new EPR signals at
1170, 1470, 2395, 4740, 4830, 5940, 7660, and 9490 G which
continuously grow during the photolysis (Figure 1). When
annealed at 30 K, these signals disappear but can be recovered
when subjected to 308 nm irradiation at 4 K. This is charac-
teristic behavior of radical pairs produced by the photolysis of
(5) Wasserman, E. Prog. Phys. Org. Chem. 1971, 8, 319-336.
(6) Hall, J. H.; Fargher, J. M.; Gisler, M. R. J. Am. Chem. Soc. 1978, 100,
2029-2034.
(7) Glarum, S. H. J. Chem. Phys. 1963, 39, 3141-3144.
(8) Weltner, W., Jr.; Van Zee, R. J.; Li, S. J. Phys. Chem. 1995, 99, 6277-
6285.
9
J. AM. CHEM. SOC. VOL. 130, NO. 13, 2008 4397

A R T I C L E S
Sander et al.
assignment
Table 1. Assignment of the EPR Transitions of Q-5
Signal
simulation
calculation
experiment
transition
1165
1470
1490
2395
1165
1465
1525
2390
4125
1170
1465
1480
2395
z, |1 T |2
y, |3 T |4
z, |2 T |4
x, |3 T |4
z, |2 T |4
z1
y1
z4
x1
z4′
A
x3
A
y3
z3
4725
4825
5935
7655
9495
4740
4830
5940
7660
9490
4810
x, |1 T |2
7620
9450
y, |1 T |2
z, |3 T |4
spectrum. The field dependence of the magnetic energy levels
was calculated for the principal axes, using the solution of the
Hamilton matrix of a quartet state11,12 with the parameters
derived from the simulation. The axial transitions for the external
magnetic field parallel to the x-, y-, and z-direction, and their
positions in the spectrum (depending on the ZFS parameters)
are shown in Figures 4 and 5, respectively. Transition z4 appears
in the spectrum if D is greater than ∼0.16 cm-1 and moves
with increasing D value to higher field values (see Figure 5).
Figure 2. (a) EPR spectrum (9.5875 GHz) of Q-5 in solid argon at 4 K
produced by 308 nm irradiation of 6. (b) Simulation of the spectrum of
Q-5 using the following parameters: g ) 2.003, |D/hc| ) 0.285 cm-1 and
|E/hc| ) 0.043 cm-1
.
Figure 3. Low field area showing details of the EPR spectrum (9.5752
GHz) of Q-5: (a) experimental spectrum and (b) simulated spectrum. The
shape of the signals was adapted to match the experimental spectrum.
matrix-isolated precursors. The radical pair is formed in close
proximity in the matrix, and annealing leads to rapid recombina-
tion. This is exactly the same behavior previously observed for
Q-5 and I atoms, using IR detection;2 thus, the labile signals
are assigned to the nitreno radical in its quartet ground state.
The I atoms formed during photolysis could not be observed in
the electron spin resonance (ESR) spectra. This has been also
found in other experiments and is attributed to the orbital
degeneracy and strong spin-orbit coupling in the I atoms,
resulting in very broad signals. EPR spectra of matrix-isolated
I atoms were only observed in solid xenon and other matrices
with stronger interactions between I atoms and the matrix.9,10
Simulation of the EPR Spectrum. The simulation of the
EPR spectrum of Q-5 with ZFS parameters |D/hc| ) 0.285 and
|E/hc| ) 0.043 cm-1 resulted in excellent agreement with the
experimental spectrum (see Figures 2 and 3). The assignment
of the transitions (given in Table 1) is based on the simulated
(9) Iwasaki, M.; Toriyama, K.; Muto, H. J. Chem. Phys. 1979, 71, 2853-
2859.
(10) Rai, U. S.; Symons, M. C. R.; Wyatt, J. L.; Bowman, W. R. J. Chem. Soc.,
Faraday Trans. 1993, 89, 1199-1201.
(11) Atherton, N. M. Principles of Electron Spin Resonance; Prentice Hall: New
York, 1993.
(12) Fleischhauer, P.; Gehring, S.; Saal, C.; Haase, W.; Tomkowicz, Z.; Zanchini,
C.; Gatteschi, D.; Davidov, D.; Barra, A. L. J. Magn. Magn. Mater. 1996,
159, 166-174.
Figure 4. Energy of the four quartet sublevels, as a function of the external
field H parallel to the z, x, and y-axis, respectively, calculated for |D/hc| )
0.285 cm-1 and |E/hc| ) 0.043 cm-1 (giso ) 2.003).
9
4398 J. AM. CHEM. SOC. VOL. 130, NO. 13, 2008

EPR Characterization of a Quartet Nitreno Radical
A R T I C L E S
Figure 7. Spin-density distribution in Q-5. Positive values are contoured
in red at the level of 0.003 e-/(Bohr3); negative values are contoured in
yellow at a level of -0.002 e-/(Bohr3).
transition z4, also split for E > 0; however, these signals do
1
not merge with corresponding transitions at E/D ) /3. This
Figure 5. Simulation of the field dependence of the axial transitions of a
1
quartet system with D ) 0.05-0.285 cm-1 and E/D values up to /3. The
behavior is expected for off-axis transitions. Therefore, although
transition z4 can be assigned to an axial transition, it also exhibits
some behavior that is characteristic of an off-axis transition.
This z4 is a transition between levels |2 and |4 . According to
the literature, such transitions are forbidden for H || z, but
become allowed if the z-axis is rotated away from the direction
of the magnetic field.12 This quasi off-axis nature of z4 is also
supported by the calculation of its angular dependence, and this
might be the reason for the significant error of 35 G between
the calculation and the simulation of this transition. Transitions
y1 and z4 are predicted by the simulation to be very close at
∼1470 G, and, indeed, a closer inspection of the experimental
spectrum reveals a splitting of this peak into two components
at 1465 and 1480 G.
signal positions of 5 with |D/hc| ) 0.285 and |E/hc| ) 0.043 cm-1 are
marked. Note that only the most intense signals were traced.
Electronic Structure Calculations. The quartet ground state
of Q-5, calculated with DFT, shows a fairly complex spin-
density distribution, which involves π- as well as σ-components
(Figure 7). More insight is obtained by identifying the singly
occupied molecular orbitals (SOMOs) of the system. This is
most conveniently done by examining the exactly singly
occupied spin unrestricted natural orbitals (UNOs) transformed
to a localized representation (Figure 8). The analysis suggests
the following interpretation: The electronic structure of Q-5 is
best described as having two parallel-spin unpaired electrons
localized on the nitrogen and one on the unsaturated carbon
atom. The unpaired electron sitting in the in-plane nitrogen “lone
pair” is fairly well-localized and only slightly polarizes the
sigma-system of the ring. However, the unpaired electron sitting
in the nitrogen out-of-plane lone pair is heavily delocalized into
the π-system of the ring. Similarly, the carbon σ-lone pair is
fairly delocalized into the σ-system of the ring. Schematically,
the electronic structure may therefore be best-represented by
the following leading resonance structure:
Figure 6. Angular dependence of the EPR signals of Q-5 when varying
the field orientation from parallel x to parallel z and parallel y. The signals
marked as “A” can be assigned to off-axis transitions. The plot also shows
the off-axis behavior of transition z4.
As for all z signals, the position of z4 in the spectrum does not
change significantly by changing the value of E, if D is constant.
By contrast, the position of the x and y signals are very sensitive
to changes in the value of E. The splitting between the
corresponding x and y signals increases as the value of E
1
increases, and, with an E/D ratio of /3, the corresponding z
and y signals, as well as the x1 and x3 signals, merge into one
signal (see Figure 5). This behavior is known for triplet systems
and confirms that the assignment of the transition (see Table
1) is consistent. In addition, there are two transitions A that are
not assignable to axial transitions. (See Figure 6.) These
transitions correspond to higher-order solutions of the spin
Hamiltonian and are characteristic for high-spin systems (S >
1) with large ZFS parameters.13,14 Transition A, as well as
This is consistent with the notion that Q-5 is best-described as
a σ,σ,π-triradical.
9
J. AM. CHEM. SOC. VOL. 130, NO. 13, 2008 4399

A R T I C L E S
Sander et al.
Figure 8. Three singly occupied molecular orbitals of Q-5: spin-unrestricted natural orbitals (UNOs) in a localized representation; Pipek-Mezey60 localization;
and B3LYP/TZVPP.
Table 2. Analysis of the Calculated D-tensor (B3LYP/TZVPP) of
Ab initio calculations were performed to evaluate the extent
to which Q-5 indeed has an isolated quartet ground state.
Because not all of the doublet states are properly represented
by a single determinant, we resorted to correlated multireference
ab initio methods in form of the spectroscopy-oriented config-
uration interaction (SORCI) approach,15 as implemented in
the ORCA package.16 Calculations were performed on top of a
state-averaged complete active space self-consistent field
(SA-CASSCF) wavefunction for one quartet and two doublet
states with three electrons in three orbitals. This constitutes the
proper model space for answering the question raised previously.
Starting orbitals were taken from the quasi-restricted orbitals
(QROs)17 of a calculation with the BP18,19 function and the
TZVPP20 basis set (BP86/TZVPP). The active orbitals transform
in the C2v point group under the b2 (in-plane nitrogen lone-
pair), b1 (out-of-plane nitrogen lone pair), and a1 (carbon σ-lone
pair) irreducible representations; hence, the lowest quartet state
of the system is designated as 1-4A2.
Q-5 (All Numbers Are Given in Units of cm-1
)
D
E
total (calc)
experiment
0.261
0.285
0.017
0.043
spin-spin
spin-orbit
0.226
0.035
0.024
-0.007
spin-spin
1-center
2-center
3-center
4-center
coulomb
exchange
0.273
-0.049
0.002
0.000
0.182
0.034
-0.010
0.000
0.000
-0.018
0.042
0.044
spin-orbit
M ) 0 (RfR)
M ) 0 (âfâ)
M ) +1 (âfR)
M ) -1 (Rfâ)
-0.002
-0.001
-0.001
0.040
0.006
0.006
-0.003
-0.003
The SORCI calculation was then performed on top of the
SA-CASSCF(3,3) solution with Tsel ) 10-6Eh, Tpre ) 10-4, and
Tnat ) 10-4 which provides essentially converged results for
state energy differences.21 The calculations indeed predict 1-4A2
to be the lowest state, followed by the first doublet (1-2A2) at
∼0.26 eV (6 kcal/mol) and the second doublet (2-2A2) at ∼0.95
eV (22 kcal/mol). Both low-lying doublet states are dominated
by the same (b1)1(b2)1(a1)1 configuration that also applies to the
1-4A2 ground state and correspond to the two linearly indepen-
dent spin-doublet couplings that can be formed for three
unpaired electrons in three orbitals. This result has two
implications. First, it calls the spin-unrestricted Kohn-Sham
solutions for the lowest doublet state (but not for the quartet
state) into question, because it is not described well by a single
determinant. Indeed, if no special care is taken, B3LYP18,22,23
DFT calculations predict the first doublet state to be ∼0.7 eV
above the ground state, which corresponds to an error of ∼0.5
eV (∼11 kcal/mol), relative to the more-reliable multireference
ab initio calculation. Second, the quartet nature of the ground
state becomes intelligible: The three SOMOs form an (acci-
dentally) quasi-degenerate set (upon taking the ROHF solution
of the 1-4A2 state as a reference, the orbital energies are within
<2 eV of each other). Because the SOMOs are not well-
separated spatially, the exchange interactions between the
unpaired electrons must be large. Hence, the quartet state, which
features the largest exchange stabilization, must be the ground
state. These results shows that, in agreement with the available
experimental data, Q-5 has a clear spin-quartet ground state and
a strong preference for three unpaired electrons, even in the
low-lying doublet states.
Calculations of the EPR Parameters. The EPR properties
of Q-5 (1-4A2) were calculated with the B3LYP hybrid
functional and a reasonably large TZVPP basis set, as described
in the Experimental Section. The agreement between theory and
experiment is very good for D, with respect to sign and
magnitude, and is still reasonable for E (see Table 2). The near-
perfect agreement observed for the value of D is, to some extent,
certainly fortuitous; however, good agreement with the experi-
ment is indeed anticipated from the results of Sinnecker and
Neese.24 The decomposition of D into individual contributions
in Table 2 is interesting. Approximately 87% of D results from
the direct spin-spin coupling (DSS), and still ∼13% comes from
the SOC contribution. Interestingly, the calculated SS contribu-
tion is essentially local, with the largest contributions resulting
from one-center integrals. Of the remaining contributions, the
two-center terms reduce the local contributions by ∼18%, which
(13) Teki, Y.; Takui, T.; Yagi, H.; Itoh, K.; Iwamura, H. J. Chem. Phys. 1985,
83, 539-547.
(14) Matsushita, M.; Momose, T.; Shida, T.; Teki, Y.; Takui, T.; Itoh, K. J.
Am. Chem. Soc. 1990, 112, 4700-4702.
(15) Neese, F. J. Chem. Phys. 2003, 119, 9428-9443.
(16) Neese, F. Version 2.6.04 ed.; University of Bonn, Germany, 2007.
(17) Neese, F. J. Am. Chem. Soc. 2006, 128, 10213-10222.
(18) Becke, A. D. Phys. ReV. A 1988, 38, 3098.
(19) Perdew, J. P. Phys. ReV. B 1986, 33, 8822-8824.
(20) Scha¨fer, A.; Huber, C.; Ahlrichs, R. J. Chem. Phys. 1994, 100, 5829.
(21) Wanko, M.; Hoffmann, M.; Strodel, P.; Thiel, W.; Neese, F.; Frauenheim,
T.; Elstner, M. J. Phys. Chem. B 2005, 109, 3606-3615.
(22) Becke, A. D. J. Chem. Phys. 1993, 98, 1372-1377.
(23) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B: Condens. Matter Mater.
Phys. 1988, 37, 785-789.
(24) Sinnecker, S.; Neese, F. J. Phys. Chem. A 2006, 110, 12267-12275.
9
4400 J. AM. CHEM. SOC. VOL. 130, NO. 13, 2008

EPR Characterization of a Quartet Nitreno Radical
A R T I C L E S
Figure 9. Orientation of the D-tensor of Q-5, based on the calculations
Figure 10. Orientation of the g-tensor of Q-5, based on the calculations
(B3LYP/TZVPP).
(B3LYP/TZVPP).
emphasizes the importance of multicenter two-electron spin-
spin integrals if quantitative agreement with the experiment is
desired. The alternative breakdown of the SS term into Coulomb
and exchange contributions shows that the latter are not
negligible and possess the same sign as the Coulomb terms that
are usually solely held responsible for the SS-term within a point
dipole model. Indeed, the exchange terms, which, to the best
of our knowledge, have rarely been discussed in the experi-
mental EPR literature, account for ∼20% of the SS term. This
means that point dipole models are not realistic in the present
systems; this is, of course, also consistent with the fact that the
SS terms result from local one-center contributions, rather than
from two-center Coulomb contributions as anticipated for a point
dipole model.
The limited SOC contributions to D are dominated by the
spin-lowering spin-flip contributions, which result from the low-
lying spin-doublet states. This is consistent with previous results
for other systems25-31 for which the SOC term represents the
largest contribution to D. The SS contributions are calculated
without any spin polarization. Unfortunately, it has been found
previously that inclusion of spin polarization into the calculation
of the SS term (which is technically feasible, of course)
significantly degrades the agreement with the experimental
values.24 Thus, currently, there does not seem to be a satisfactory
solution to this problem in a DFT framework.
Interestingly, the easy axis of the D-tensor does not point
out of the plane but rather along the C-N bond (Figure 9).
This result will be interpreted later in the Conclusions and
Discussion section. The calculated g-shifts are -86.5, 588, and
1012 ppm. The small shifts are typical of radicals that are
composed of light atoms. The orientation of the g-tensor is
displayed graphically in Figure 10. Interestingly, the largest
g-valueswhich also corresponds to the largest g-shiftsis
oriented perpendicular to the molecular plane and, therefore,
also occurs perpendicular to the easy axis of the D-tensor.
Nevertheless, the anisotropy in the g-tensor is sufficiently small,
such that it is hardly possible to observe it under the present
experimental conditions.
Conclusions and Discussion
The EPR measurements confirm our previous assignments
of the products of the matrix photolysis of 6, based on IR
spectroscopy.2 As expected, the primary photoproduct is nitrene
T-7 formed in very high yield. The splitting of the C-I bond
in 7 is much less efficient and competes with the rearrange-
ment to 9. Nevertheless, under conditions that allow the
nitreno radical 5 to be detected via IR spectroscopy, a clear
EPR signal of the quartet Q-5 can be observed. It is pleasing
that the previous assignment of the quartet state, based solely
on IR spectroscopy, is now unequivocally confirmed by EPR
spectroscopy. As observed in the IR experiments,2 the yield of
5 is strongly dependent on the matrix temperature. At 10-15
K, only traces of 5 are formed, whereas the highest yield is
found if 6 is irradiated at 4 K. Annealing at temperatures of
>10 K results in the recombination of 5 with the I atom;
therefore, it was not possible to measure the Curie-Weiss
behavior of 5.
The electronic structure of Q-5 can be best-described as a
σ,σ,π-triradical with one unpaired electron located in an in-
plane orbital at the N atom (σ) and one at the para-C atom of
the phenyl ring (σ), while the third unpaired electron is in an
out-of-plane orbital of the N atom and is heavily delocalized
into the π-system. Because of the significant delocalization of
the out-of-plane SOMO, Q-5 is expected to show both properties
of a nitrene that is linked to a phenyl radical and of a carbene
(cyclohexadienylidene) that is linked to a nitrogen-centered
radical (iminyl radical), as shown in the resonance structures
below.
In accordance with this qualitative picture, the natural spin
population32 is calculated to 1.55 unpaired electrons at the N
atom and to 1.21 at the para-C atom (UB3LYP/TZVPP).
The closest related molecules to Q-5 are a series of quartet-
ground-state nitreno radicals 10-13 synthesized by Lahti et
al.33-35
(25) Zein, S.; Duboc, C.; Lubitz, W.; Neese, F. Inorg. Chem. 2007, 47 (1), 134-
142.
(26) Neese, F. J. Chem. Phys. 2007, 127, 164112.
(27) Duboc, C.; Phoeung, T.; Zein, S.; Pecaut, J.; Collomb, M. N.; Neese, F.
Inorg. Chem. 2007, 46, 4905-4916.
(28) Carmieli, R.; Larsen, T. M.; Reed, G. H.; Zein, S.; Neese, F.; Goldfarb, D.
J. Am. Chem. Soc. 2007, 129, 4240.
(32) Reed, A. E.; Weinstock, R. B.; Weinhold, F. J. Chem. Phys. 1985, 83,
1736.
(29) Neese, F. J. Inorg. Biochem. 2006, 100, 716-726.
(30) Neese, F. J. Am. Chem. Soc. 2006, 128, 10213-10222.
(31) Schoneboom, J. C.; Neese, F.; Thiel, W. J. Am. Chem. Soc. 2005, 127,
5840-5853.
(33) Lahti, P. M.; Esat, B.; Liao, Y.; Serwinski, P.; Lan, J.; Walton, R.
Polyhedron 2001, 20, 1647-1652.
(34) Taylor, P.; Serwinski, P. R.; Lahti, P. M. Org. Lett. 2005, 7, 3693-
3696.
9
J. AM. CHEM. SOC. VOL. 130, NO. 13, 2008 4401

A R T I C L E S
Sander et al.
simultaneously exhibits nitrene and carbene character, then leads
to a simple and appealing geometric interpretation of the large
value of E, in contrast to 10-13, which have E values of zero.
We assume, for Q-5, a coordinate system where the z-axis points
along the C-N bond, the x-axis is perpendicular to the molecular
plane, and the y-axis, consequently, lies in the molecular plane.
In the nitrene part of the electronic structure, the two SOMOs
are comprised of px and py orbitals and their dipolar field is
consequently directed along the z-direction. For the carbene part
of the 1-4A2 ground state, the SOMOs are comprised of pz and
px contributions and, consequently, the resulting dipolar field
points along the y-direction. Because the nitrene character
prevails over the carbene character of Q-5, the main magnetic
axis is still pointing along the C-N bond in the z-direction.
However, the carbene character is reflected by an enhanced
dipolar field in the y-direction, which is strong enough to give
significantly magnetically inequivalent y- and x-directions. We,
therefore, come to the conclusion that the nonzero E-value in
Q-5 mainly reflects the carbene character of the 1-4A2 ground
state. This carbene character, in turn, is controlled by the
properties of the π-system, because the out-of-plane spin-density
arises from the conjugation of the nitrene out-of-plane SOMO
and the π-system of the ring.
Finally, the fact that the D value for Q-5 is a factor of ∼3
smaller than those of typical phenylnitrenes5 is readily explained
by the prefactor 1/[S(2S - 1)] of the DSS term (vide infra), which
amounts to 1 for a ground state spin of 1 and 1/3 for a ground-
state spin of 3/2. Assuming an axial D-tensor, the level splitting
of 2D (in S ) 3/2) versus D (in S ) 1) is more similar for both
multiplicities but is still slightly smaller for the quartet system.
This is explained by the partial cancellation of the carbene and
nitrene contributions to the D value, which oppose each other
along the magnetic z-direction, as explained previously in some
detail.
The major difference from 5 is that, in 10-13, a π-radical is
attached to the phenylnitrene unit and, thus, the electronic
structure is best-described as a σ,π,π-triradical. Consequently,
a carbene-type resonance structure cannot be formulated for
these molecules, which, in turn, has consequences for the ZFS
parameter E (see below). The values for |D/hc| of 10, 11, and
12 were determined to be 0.277, 0.256, and 0.288 cm-1
,
respectively, which is similar to that of 5, whereas |E/hc| was
found to be zero, contrary to that observed for 5, which has a
very large value of |E/hc| (0.043). The quartet ground state of
the para-connected triradicals 10-12 was rationalized in a spin
fragment analysis by their nondisjointed nature favoring high-
spin ground states.33,35 By contrast, the corresponding meta-
connected systems prefer low-spin ground states. Only 13 has
a doublet ground state with a reasonably small doublet-quartet
splitting of 12 cal/mol, as determined by the analysis of the
temperature dependence of the signal intensity (the Curie-Weiss
plot).34 The corresponding meta-connected system shows a
larger preference for the doublet state, in accordance with the
spin fragment analysis. In 13, the ZFS parameters were
determined to be |D/hc| ) 0.300 and |E/hc| ) 0.0 cm-1, which
indicates a higher spin density at the nitrene center and less
delocalization of the nitrene π-electron. The value of D in
nitreno radical 5 (|D/hc| ) 0.285) is similar to that in 10 and
12, which indicates a similar spin density at the nitrene center.
However, the large E value in 5 deserves a closer interpretation.
In agreement with our calculations, Wasserman assumed that
the direction of the principal magnetic axis z of phenyl nitrenes
lies parallel to the N-C bond.5,36,37 It is readily verified (e.g.,
In summary, nitreno radical Q-5 is an interesting σ,σ,π-
triradical with unusual electronic and magnetic properties. The
EPR spectrum could be completely analyzed and understood
based on the qualitative and quantitative arguments. This makes
it a benchmark system for understanding the organic quartet
states. In addition, the robust quartet ground state makes this
type of molecule interesting as models for new building blocks
for the design of organic magnetic materials.
3
by a calculation on linear CH2) that the dipolar field created
by two unpaired electrons in px and py orbitals indeed points
along the z-direction and that the axial symmetry of the spin
distribution in such an arrangement remains untouched and leads
to |E/hc| ) 0. Indeed, for carbenes, it has been reported that
the z-axis lies parallel to a hypothetical 180° bond angle,38 and
that the E value consequently increases with decreasing bond
angle at the carbene center. The addition of a symmetrical
π-radical in the para-position, as in 10-12,33 does not change
the pseudo-cylindrical symmetry of these triradicals; the z-axis
remains in the same position, and, hence, the E value of these
quartet systems should remain zero.
Experimental Section
Electron Paramagnetic Resonance (EPR) Measurements. X-band
EPR spectra were recorded with a Bruker Model Elexsys E500 EPR
spectrometer with a Model ER077R magnet (with a 75-mm gap between
pole faces), a Model ER047 XG-T microwave bridge, and a Model
ER4102ST resonator with a TE102 cavity. The matrices were deposited
on an oxygen-free high-conductivity copper rod (75 mm in length, 3
mm in diameter) that was cooled by a Sumitomo SHI-4-5 closed-cycle
4.2 K cryostat. Because saturation of the spectra might be a problem,
especially at 4 K, most spectra were recorded with a relatively high
microwave power of 20.1 mW at 4 K and, in addition, with 0.64 mW
at 15 K. The spectra were identical, only the signal-to-noise (s/n) ratio
was much worse at higher temperature and lower power. Therefore,
we assume that saturation is not significant under the experimental
conditions used in our experiments.
However, this latter assumption cannot be verified. Our
calculations demonstrate that the bulk of the E value results
from one-center contributions of the direct spin-spin coupling
interaction. This information, together with the fact that Q-5
(35) Serwinski, P. R.; Esat, B.; Lahti, P. M.; Liao, Y.; Walton, R.; Lan, J. J.
Org. Chem. 2004, 69, 5247-5260.
The vacuum system consisted of a vacuum shroud that was equipped
with a sample inlet valve and a half-closed quartz tube (75 mm in length,
10 mm in diameter) at the bottom and a vacuum pump system with a
Pfeiffer Vacuum TMU071P turbo pump backed by a Leybold two-
(36) Smolinsky, G.; Wasserman, E.; Yager, W. A. J. Am. Chem. Soc. 1962, 84,
3220-3221.
(37) Smolinsky, G.; Snyder, L. C.; Wasserman, E. ReV. Mod. Phys. 1963, 35,
576-577.
(38) Wasserman, E.; Hutton, R. S. Acc. Chem. Res. 1977, 10, 27-32.
9
4402 J. AM. CHEM. SOC. VOL. 130, NO. 13, 2008

EPR Characterization of a Quartet Nitreno Radical
A R T I C L E S
1
stage, rotary-vane pump. To avoid contamination of the high-vacuum
segment by pump oil from the backing pump, a catalytic oxidation
filter was placed between the rotary-vane pump and the turbo pump.
During deposition, the inlet port was positioned at the same height as
the tip of the copper rod. For irradiation, the copper rod was lowered
into the quartz tube at the bottom of the shroud, and for the measurement
of the EPR spectra, the entire apparatus was moved downward, so
that the quartz tube and copper rod were positioned inside the EPR
cavity.
Here, R is the fine structure constant (∼ /137 in atomic units); ge is the
free-electron g-value (2.002319...); the indices µ,ν,κ,τ refer to basis
R-â
µν
functions; and P
is an element of the spin-density matrix. No
approximations have been made to the two-electron spin-spin dipole
integrals in eq 1. As discussed previously,24 it is advantageous to apply
an open-shell spin-restricted formalism for this quantity, in conjunction
with present-day DFT methods. We have used the spin densities from
the spin-unrestricted natural orbital (UNO) determinant for this purpose,
as discussed in detail in the work of Sinnecker and Neese.24
Second-order SOC contributions also have been treated throughout
the study. Following the general formulation, in terms of infinite sums
over states,41 a linear response treatment was recently proposed in which
the SOC contribution can be written as26
Azide 6 was evaporated for 1 h at 0 °C and co-deposited with a
large excess of argon (Messer-Griesheim, 99.9999%) on the tip of
the copper rod at 4 K. The matrix-isolated sample was subsequently
irradiated with a Lambda Physik Model Lextra 200 Excimer Laser
(XeCl, 308 nm), and spectra were recorded at various irradiation
times.
D(kSl OC;M) ) fM(S) hSkOC;hSl OC
(2)
In eq 2, M denotes the contributions to the SOC term (M ) 0, (1)
from excited states with S′ ) S ( 1 (where S is the total spin quantum
The computer simulation of the EPR spectrum was performed using
the Xsophe computer simulation software suite (version 1.0.4),39 which
was developed by the Centre for Magnetic Resonance and Department
of Mathematics, University of Queensland, Brisbane (Australia) and
Bruker Analytik GmbH, Rheinstetten (Germany). The angular depen-
dence of the quartet transitions was calculated using EasySpin.40
number of the electronic state for which the ZFS tensor is computed
1
(S > /2)), fM(S) is a spin-dependent prefactor (f0 ) -1/(4S2), f-1
)
1/[2S(2S - 1)], f+1 ) 1/[2(S + 1)(2S + 1)]), and hSkOC;hlSOC is a
shortcut notation for a spin-orbit linear response function. In a DFT
framework, it is related to the derivatives of generalized spin densities,
as explained in detail in ref 26. This treatment supersedes the earlier
proposals for the SOC contribution to the ZFS tensor in the reports of
Neese and co-workers54,55 and, in our opinion, also the work of Pederson
and Khanna.56 For alternative approaches, see the work of Reviakine
et al.57
EPR Calculations. The EPR properties were calculated according
to previously published methods that are implemented in the ORCA
package.24,26,30,41-47 Geometries were optimized using the BP86
functional,18,19 the TZVP basis set48 (featuring a single set of polariza-
tion functions on each atom), and the resolution-of-the-identity ap-
proximation with matching auxiliary basis sets.49-51 Structures were
verified to represent local minima through numeric frequency calcula-
tions.
The spin-orbit operator used in eq 2 was assumed to be of the spin-
orbit mean-field (SOMF) type58 in the multicenter implementation of
Neese46 that is equivalent to Berning et al.59 It is believed to provide
an accurate representation of the full Breit-Pauli two-electron SOC
operator. The g-tensor has been calculated according to established
procedures.43,46
Property calculations were done with the B3LYP hybrid functional,
because, on average, it has been proven to be superior to nonhybrid
functionals in EPR property calculations. In these calculations, a more
extensively polarized triple-ú basis set (TZVPP, amounting to 2d1f
polarization shells all atoms) was used. The spin-spin (SS) contribution
to the zero-field splitting tensor was treated in the mean-field ap-
proximation based on the Kohn-Sham determinant. The SS contribu-
tion is given by24,52,53
Acknowledgment. This work was financially supported by
the Deutsche Forschungsgemeinschaft and the Fonds der
Chemischen Industrie. F.N. and S.K. gratefully acknowledge
the SFB 624 (“Template-vom Design chemischer Schablonen
zur Reaktionssteuerung”, Universita¨t Bonn) as well as the SFB
663 (“Molekulare Antwort nach elektronischer Anregung”,
Universita¨t Du¨sseldorf) for financial support.
2
R2
ge
D(kSl S) ) -
×
S(2S - 1) ∑∑
16
µν κτ
JA078171S
{PR-âPRκτ-â - PR-âPνRτ-â} µν|r12-5{3r12,kr12,l - δklr122}|κτ (1)
µν
µκ
(50) Eichkorn, K.; Treutler, O.; O¨ hm, H.; Ha¨ser, M.; Ahlrichs, R. Chem. Phys.
Lett. 1995, 240, 283-290.
(51) Eichkorn, K.; Weigend, F.; Treutler, O.; Ahlrichs, R. Theor. Chem. Acc.
1997, 97, 119-124.
(39) Griffin, M.; Muys, A.; Noble, C.; Wang, D.; Eldershaw, C.; Gates, K. E.;
Burrage, K.; Hanson, G. R. Mol. Phys. Rep. 1999, 26, 60-84.
(40) Stoll, S.; Schweiger, A. J. Magn. Reson. 2006, 178, 42-55.
(41) Neese, F.; Solomon, E. I. Inorg. Chem. 1998, 37, 6568-6582.
(42) Neese, F. J. Phys. Chem. A 2001, 105, 4290-4299.
(43) Neese, F. J. Chem. Phys. 2001, 115, 11080-11096.
(44) Neese, F. Curr. Opin. Chem. Biol. 2003, 7, 125-135.
(45) Neese, F. J. Chem. Phys. 2003, 118, 3939.
(46) Neese, F. J. Chem. Phys. 2005, 122, 034107/034101-034113.
(47) Neese, F. J. Biol. Inorg. Chem. 2006, 11, 702-711.
(48) Scha¨fer, A.; Huber, C.; Ahlrichs, R. J. Chem. Phys. 1994, 100, 5829-
5835.
(52) McWeeny, R.; Mizuno, Y. Proc. R. Soc. (London), A 1961, A259, 554.
(53) Petrenko, T. T.; Petrenko, T. L.; Bratus, V. Y. J. Phys: Condens. Matter
2002, 14, 12433-12440.
(54) Ray, K.; Weyhermu¨ller, T.; Neese, F.; Wieghardt, K. Inorg. Chem. 2005,
44, 5345-5360.
(55) Pederson, M. R.; Khanna, S. N. Phys. ReV. B 1999, 60, 9566-9572.
(56) Reviakine, R.; Arbuznikov, A. V.; Tremblay, J.-C.; Remenyi, C.; Malkina,
O. L.; Malkin, V. G.; Kaupp, M. J. Chem. Phys. 2006, 125, 054110.
(57) Hess, B. A.; Marian, C. M.; Wahlgren, U.; Gropen, O. Chem. Phys. Lett.
1996, 251, 365.
(58) Berning, A.; Schweizer, M.; Werner, H. J.; Knowles, P. J.; Palmieri, P.
Mol. Phys. 2000, 98, 1823-1833.
(59) Pipek, J.; Mezey, P. G. J. Chem. Phys. 1989, 90, 4916-4926.
(49) Vahtras, O.; Almlof, J.; Feyereisen, M. W. Chem. Phys. Lett. 1993, 213,
514-518.
9
J. AM. CHEM. SOC. VOL. 130, NO. 13, 2008 4403