Discrepancy between the spin distribution and the magnetic ground state for a triaminoxyl substituted triphenylphosphine oxide derivative

Article abstract of DOI:10.1002/chem.200400656

The magnetic interaction and spin transfer via phosphorus have been investigated for the tri-tert-butylaminoxyl para-substituted triphenylphosphine oxide. For this radical unit, the conjugation existing between the π* orbital of the NO group and the pheny

Full text of DOI:10.1002/chem.200400656

Discrepancy between the Spin Distribution and the Magnetic Ground State
for a Triaminoxyl Substituted Triphenylphosphine Oxide Derivative
Oscar Benedi Borobia,[a] Philippe Guionneau,[a] Henrike Heise,[b, f] Frank H. Kçhler,*[b]
Laurent Ducasse,[c] Jose Vidal-Gancedo,[d] Jaume Veciana,*[d] StØphane Golhen,[e]
Lahc›ne Ouahab,[e] and Jean-Pascal Sutter*[a]
Abstract: The magnetic interaction and
spin transfer via phosphorus have been
investigated for the tri-tert-butylami-
noxyl para-substituted triphenylphos-
phine oxide. For this radical unit, the
conjugation existing between the p* or-
bital of the NO group and the phenyl p
orbitals leads to an efficient delocaliza-
tion of the spin from the radical to the
neighboring aromatic ring. This has
been confirmed by using fluid solution
high-resolution EPR and solid state
MAS NMR spectroscopy. The spin
densities located on the atoms of the
molecule could be probed since 1H,
13C, 14N, and 31P are nuclei active in
NMR and EPR, and lead to a precise
spin distribution map for the triradical.
The experimental investigations were
completed by a DFT computational
study. These techniques established in
particular that spin density is located at
the phosphorus (1=ꢀ15”10ꢀ3 au),
that its sign is in line with the sign al-
ternation principle and that its magni-
tude is in the order of that found on
the aromatic C atoms of the molecule.
Surprisingly, whereas the spin distribu-
tion scheme supports ferromagnetic in-
teractions among the radical units, the
magnetic behavior found for this mole-
cule revealed a low-spin ground state
characterized by an intramolecular ex-
change parameter of J=ꢀ7.55 cmꢀ1 as
revealed by solid state susceptibility
studies and low temperature EPR. The
X-ray crystal structures solved at 293
and 30 K show the occurrence of a
crystallographic transition resulting in
an ordering of the molecular units at
low temperature.
Keywords: EPR spectroscopy
·
magnetic
spectroscopy
distribution
properties
radicals
·
NMR
· spin
·
Introduction
[a] O. B. Borobia, Dr. P. Guionneau, Dr. J.-P. Sutter
Institut de Chimie de la Mati›re CondensØe de Bordeaux
CNRS, UniversitØ Bordeaux 1, 87 Ave. Dr. Schweitzer
33608 Pessac (France)
It is usually considered that the knowledge of the spin densi-
ty distribution for a paramagnetic compound is the ultimate
information to account for the magnetic behavior exhibited
by a molecule or an assembly of molecules in the crystal.
For a paramagnetic compound the mixing of the SOMO
(single occupied molecular orbital)with several orbitals lo-
calized on the molecular fragments may lead to the delocali-
zation of the unpaired electron.[1] As a consequence, the
spin is not exclusively located on one atom or functionality
but distributed over the molecule. Starting from the core
group bearing the unpaired electron, this spin delocalization
over the entire molecule leads to a pattern which reflects
the polarization of the s- and p-bonding electrons. Thus,
two neighboring atoms will respectively bear a spin (posi-
tive spin)and b spin (negative spin)density, and this sign al-
ternation is spread over the whole molecule. For instance,
for a molecule containing two radical units in ferromagnetic
interaction, the spin polarization pattern developed on the
molecular moiety linking them yields the same sign for both
the radicals. On the contrary, when they are in antiferromag-
netic interaction the sign of their respective spin is opposite.
This spin polarization scheme which can be observed experi-
mentally by spectroscopic techniques,[1–4] is also used as
straightforward predictive tool to determine a priori the sign
Fax : (+33)540-002-649
E-mail: jpsutter@icmcb-bordeaux.cnrs.fr
[b] Dr. H. Heise, Prof. Dr. F. H. Kçhler
Anorganisch-chemisches Institut
Technische Universität München, 85747 Garching (Germany)
E-mail: F.H.Koehler@lrz.tu-muenchen.de
[c] Dr. L. Ducasse
Laboratoire de Physico-Chimie MolØculaire
UniversitØ Bordeaux 1, 33405 Talence (France)
[d] Dr. J. Vidal-Gancedo, Prof. Dr. J. Veciana
Institut de Ci›ncia de Materials de Barcelona
Campus Universitari de Bellaterra, 08913 Cerdanyola (Spain)
E-mail: vecianaj@icmab.es
[e] Dr. S. Golhen, Dr. L. Ouahab
LCSIM UMR 6511 CNRS-UniversitØ de Rennes 1
Institut de Chimie de Rennes
35042 Rennes Cedex (France)
[f] Dr. H. Heise
New address: Max-Planck-Institute for Biophysical Chemistry
Am Fassberg 11, 37077 Gçttingen (Germany)
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author: Field depend-
ence of the magnetization at 2 K, temperature dependence of the
variation of the Dms =2 signal between 5 and 80 K, detailed solid-
state NMR data, and selected geometrical parameters for the
UB3LYP/6-31g* optimized structure for compound 1.
128
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/chem.200400656
Chem. Eur. J. 2005, 11, 128 – 139

FULL PAPER
of the intramolecular exchange interaction in p-conjugated
organic polyradicals.[5–7]
We recently reported on a family of open-shell phosphine
derivatives which were envisaged because of the possibility
offered by these molecules to coordinate to transition-metal
ions by the mean of the phosphine moiety and the aminoxyl
radical groups.[8–13] The information gathered for a series of
nitronyl nitroxide substituted phosphine derivatives estab-
lished the presence of spin density at phosphorus and, more
important, a spin distribution in accordance with high-spin
ground states for the polyradical derivatives.[14] Unfortunate-
ly, too weak intramolecular exchange interactions due to the
poor spin delocalization for nitronyl nitroxide radicals ham-
pered the observation of the actual magnetic behavior ex-
hibited by the molecules.
In order to gain further and conclusive information on the
capability of a P atom to mediate the exchange interaction
between the paramagnetic centers it bridges, we designed a
molecule with an increased strength of the interaction be-
tween the spin carriers. This was achieved by enhancing the
spin density located on the triphenylphosphine core and
choosing a spin distribution scheme such as the C atoms
bound to P bear positive spin. The molecule investigated in
the present study is the tri-tert-butylaminoxyl substituted tri-
phenylphosphine oxide. For this radical unit, the conjugation
that exists between the p* orbital of the NO group and the
phenyl p orbitals leads to an efficient delocalization of the
spin from the radical on the aromatic ring. This has been
confirmed by the spin distribution mapping undertaken by
fluid solution high-resolution EPR[4] and NMR[1] spectrosco-
py which have proven to be very efficient tools in assessing
the magnitude and sign of the spin density borne by the
atoms of organic open-shell molecules. The spin densities lo-
cated on all atoms of the molecule could be probed since all
Scheme 1. Synthesis routes to triradical 1. Route A: a) nBuLi; b)PCl 3,
c) nBuLi, d)( tBuNO)2, e)NaIO 4
; route B: a) nBuLi, b)( tBuNO)2,
c)ClSiMe 2tBu, d) nBuLi, e)PCl 3, f) Bu4NF, g)NaIO 4.
the phosphine derivative to the corresponding phosphine
oxide.
1
are active nuclei in NMR and EPR, that is, H, 13C, 14N, and
31P, and lead to a precise spin distribution map. These exper-
imental investigations were completed by a DFT computa-
tional study. Surprisingly, whereas the spin distribution
scheme supports ferromagnetic interactions among the radi-
cal units, the magnetic behavior found for this molecule re-
vealed a low-spin ground state.
By route B, the organic fragment containing the radical
precursor unit, 4a, b, was first prepared from para-dibromo-
benzene and 2-methyl-2-nitrosopropane, and protected as a
silane derivative. Its lithiation with nBuLi and subsequent
reaction with PCl3 yielded tris[p-{(O-tert-butyldimethylsilyl)-
N-tert-butyl-oxylamino}phenyl]phosphine (2a). The conver-
sion to the corresponding tris(hydroxylamine) 2b was effi-
ciently achieved with Bu4NF in THF. Previous studies on
such systems showed that it is possible to selectively obtain
the radical without oxidizing the PIII center when Ag2O is
used as oxidant.[9,12,15] This proved not to be the case here,
only a partial oxidation was observed independent of the
stoichiometry considered. The oxidation was therefore con-
ducted as in route A to obtain the triradical 1. The overall
yield with respect to PCl3 is about 15% either by route A or
B. Route B has the advantage that the products of each step
can be purified.
Results
Synthesis
The triradical 1 was prepared according to the two synthesis
pathways outlined in Scheme 1. Following route A, the core
of the molecule was first formed from para-dibromobenzene
and trichlorophosphine leading to tri-(4-bromophenyl)phos-
phine (3). The latter was lithiated with BuLi and treated
with 2-methyl-2-nitrosopropane to give the correspond-
ing tris(hydroxylamine). Based on 31P NMR spectrum the
crude reaction product appeared to be a mixture of phos-
phine 2b and phosphine oxide 2c. Triradical 1 was then
obtained by oxidation with NaIO4 which also converted
X-ray crystal structure
Single crystals suitable for X-ray crystallographic analysis
were obtained by slow diffusion of Et2O to a solution of
Chem. Eur. J. 2005, 11, 128 – 139
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
129

F. H. Kçhler, J. Veciana, J.-P. Sutter et al.
compound 1 in CHCl3. The structure was investigated at 293
and 30 K, the crystallographic parameters are given in
Table 1. A view of the molecular structure at 293 K is de-
picted in Figure 1 with selected geometrical features. The
for other radical substituted phosphine oxide derivatives.[8,16]
and the O-P-C angles close to 1128 do not reflect any steric
constraint in the molecule. The phosphorus atom is located
at 0.670(2)Š out of the plane defined by C1-C11-C21, the
corresponding phenyl rings are tilted from this plan by
Table 1. Crystal and experimental data for triradical 1 at 293 and 30 K.
ꢀ
73.5(1), 72.6(2) and 72.5(2)8, respectively. The C C bonds
Triradical 1
T=293 K
T=30 K
lengths for the phenyl groups are in the range [1.381(4)–
1.373(5)Š] and those connecting the aminoxyl N atoms to
the aromatic rings are found between 1.409(4)and
1.431(4)Š excluding a conceivable quinonoid structure for
1. The dihedral angles of the aminoxyl units with the phenyl
groups they are linked to are 31.1(2), 16.8(3) and 34.5(2)8.
Finally, the intramolecular separations between the aminox-
yl groups taken as the O···O distances are 9.863(6), 9.896(6),
and 9.975(6)Š. These features reveal that each Ph-NO( tBu)
moiety has slightly different geometric characteristics. How-
ever, the later appear in good agreement with those found
usually.[11,12,17,18]
formula
Mw
crystal system
space group
l [Š]
a [Š]
b [Š]
c [Š]
a [8]
C30H39N3O4P
536.61
C30H39N3O4P
536.27
triclinic
P1
0.71073
trigonal
P3
0.71069
6.0984(3)20.799(5)
12.3080(7)20.799(5)
12.3112(7)6.005(5)
117.363(3)90.0
101.582(3)90.0
96.494(3)120.0
781.19(7)2250(2)
1
b [8]
g [8]
V [Š3]
Z
3
1.188
23.4
1 [gcm3]
2qmax [8]
1.141
26.0
0.124
Figure 2 shows the crystalline organization in the bc and
ac planes. The molecular stacking can be described as a su-
perposition of layers along the a axes. In these layers, all the
m [mmꢀ1
]
0.129
reflections
measured
independent
used in refinement
parameters
R1
4866
4866
4866
343
0.047
9713
3876
3876
344
0.077
wR2
0.1193
0.19/ꢀ0.18
0.1738
0.45/ꢀ0.38
ꢀ3
residue [eꢀ·Š
]
Figure 2. Projection of the crystal packing for triradical 1 at 293 K along
a (top)and along b axis (bottom).
Figure 1. ORTEP drawing of compound 1 at 293 K with thermal ellipsoid
plots at 30% probability level. The hydrogen atoms are omitted for clari-
ꢀ
ꢀ
ty. Selected distances [Š] and angles [8]: O1 N1 1.271(3), O2 N2
O atoms of the aminoxyl groups are pointing in the same di-
ꢀ
ꢀ
ꢀ
ꢀ
1.260(4), O3 N3 1.272(4), N1 C4 1.420(4), N1 C7 1.500(5), N2 C14
ꢀ
rection but in an opposite direction to that of the P O
ꢀ
ꢀ
ꢀ
ꢀ
1.409(4), N2 C17 1.475(5), N3 C24 1.431(4), N3 C27 1.474(5), P O4
bond, and are almost parallel to each other. The shortest in-
terlayer distance involves the PO unit and a phenyl group,
O4···H22, with 2.490(2)Š, and the shortest intermolecular
separation within a layer is found for H9b(tBu)and
H13(Ph)with 2.54 Š. As far as the aminoxyl units are con-
cerned, the shortest intermolecular separations are found
for O1···H5(tBu), 2.678(4) Š, O2···H15(Ph), 2.652(5) Š, and
O3···H25(Ph), 2.644(3) Š, and the minimum intermolecular
separation between two NO groups is 5.82 Š.
ꢀ
ꢀ
ꢀ
ꢀ
1.490(2), P C21 1.802(3), P C11 1.805(3), P C1 1.805(3), O1 O2
ꢀ
ꢀ
9.864(6), O2 O3 9.896(5), O1 O3 9.975(5); O1-N1-C4 115.9(2), O2-N2-
C14 116.3(3), O3-N3-C24 115.7(3), O4-P-C21 111.63(12), O4-P-C11
111.94(12), O4-P-C1 111.82(12), O1-N1-C4-C3 27.3(4), O2-N2-C14-C13
16.8(6), O3-N3-C24-C23 33.0(5).
crystal structure confirms the formulation of compound 1 as
a phosphine oxide derivative with the three tBuNO units in
para position of the phenyl rings. At room temperature, the
ꢀ
P O bond length (1.490(2)Š)is in the order of that found
130
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 128 – 139

Magnetic Properties
FULL PAPER
A view of the molecular structure of the molecule at 30 K
is depicted in Figure 3a. Compared with the structure at
293 K, a higher symmetry is observed at low temperature,
the space group is now P3. It can be noticed that both high
and low temperature crystal packings are chiral. The unit
cell contains three independent molecules, each located on a
C3 symmetry axis. As a result, at 30 K the three Ph-
NO(tBu)moieties of a molecule are strictly equivalent,
which is not the case within the 293 K structure (Figure 3b).
The three independent molecules of the unit cell display a
dihedral angle between the aminoxyl units and the phenyl
groups of 5.9, 25.0 and 37.58, respectively. The other intera-
tomic distances and angles are close to those observed at
293 K. The organization of the molecules in layers is also
maintained at 30 K, the shortest separations involving the
aminoxyl units being in the order of those found at 293 K
with O1···H5(tBu), 2.529(5) Š, O4···H15(Ph), 2.671(5) Š,
and O6···H27(tBu), 2.644(6) Š. The crystallographic transi-
tion is then mainly associated with an ordering of the molec-
ular units which, at 30 K, have a three-fold axis passing
through the central phosphorus atom. As a general remark,
it can be noted that structural rearrangements of the crystal
lattices when cooling from room to cryogenic temperatures
are often observed for organic radicals and may have signifi-
cant incidence on the magnetic behavior exhibited by these
species.[19–22]
Spin distribution mapping
The spin distribution mapping for triradical 1 was undertak-
en by fluid solution high-resolution EPR and solid state
MAS NMR spectroscopies as well as by DFT calculations.
High-resolution isotropic EPR: When performed in dilute
fluid solutions under high-resolution conditions, EPR meas-
urements provide precise isotropic hyperfine coupling con-
stants for nuclei with both large and small spin densities.[4]
The value of the hyperfine coupling observed by EPR is di-
rectly related to the absolute spin density on the considered
nuclei (see below). The X-band EPR spectrum of compound
1 was recorded in degassed CH2Cl2 solutions (<10ꢀ4 m). The
spectrum obtained at 295 K is centered at g=2.0056 and ap-
parently consists of eight lines equally spaced by 3.75 G.
Figure 4 shows the spectrum together with the simulated
signal. This simulation required to consider a linewidth of
0.78 G and isotropic hyperfine coupling constants (hfcc)
with three equivalent N nuclei of the aminoxyl units along
with the hfcc values with the central phosphorus atom and
with six equivalent H atoms in positions 2/6 and six in posi-
tions 3/5 of the phenyl groups. The value found for the hfcc
with the N atom, aN =3.80(1)G, is, as expected, one third of
the related constant found for the mono-radical derivative
(aN =11.6 G).[15] The same result was found for the hfcc
values with the aromatic H atoms in 2/6 and 3/5 positions
with 0.7(1)and 0.29(5G) , respectively, while the corre-
sponding values for the mono-radical derivative are 2.15 and
0.86 G.[15] By contrast, the hfcc with the P atom for com-
pound 1, aP =3.30(1)G, remains the same as for the mono-
radical derivative (aP =3.284 G). This result can be ascribed
to the particular position of the P atom at the center of this
trigonal molecule which shares the three singly-occupied
molecular orbital (SOMO)of this triradical. The very close
values for aN and aP lead to an overlapping of the central
lines and indeed yield a spectrum apparently composed of
only eight lines (see stick diagram in Figure 4). The further
splitting of these main lines by the coupling with the H
Figure 3. a)ORTEP drawing of triradical 1 at 30 K with thermal ellipsoid
plots at 30% probability level. The hydrogen atoms are omitted for clari-
ꢀ
ty. Selected distances [Š] and angles [8]: Molecule 1: O1 N1 1.281(8),
ꢀ
ꢀ
ꢀ
ꢀ
N1 C4 1.451(8), N1 C7 1.514(8), P1 O2 1.497(8), P1 C1 1.823(8),
O1···O1 9.957(8), O1-N1-C4 115.7(1), C4-N1-C7 124.3(1), O2-P1-C1
ꢀ
111.7(1), C1-P1-C1 107.1(1), C3-C4-N1-O1 25.0(1); molecule 2: O4 N2
ꢀ
ꢀ
ꢀ
ꢀ
1.293(8), N2 C10 1.425(8), N2 C16 1.523(8), P2 O3 1.489(8), P2 C13
1.812(8), O4···O4 10.190(9), O4-N2-C10 117.0(1), C10-N2-C16 124.6(1),
O3-P2-C13 111.4(1), C13-P2-C13 107.5(1), C11-C10-N2-O4 37.5(5); mol-
ꢀ
ꢀ
ꢀ
ꢀ
ecule 3: O6 N3 1.275(8), N3 C23 1.405(8), N3 C26 1.513(8), P3 O5
ꢀ
1.479(8), P3 C20 1.792(8); O6···O6 9.810(9), O6-N3-C23 116.8(1), C23-
N3-C26 126.4(1), O5-P3-C20 111.8(1), C20-P3-C20 107.0(5), C24-C23-N3-
O6 5.9(1). b) Comparison of the crystal packing and the unit cells at 293
and 30 K. At a given temperature, the geometry of the molecular parts
drawn with the same color are strictly identical for symmetry reasons.
Chem. Eur. J. 2005, 11, 128 – 139
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
131

F. H. Kçhler, J. Veciana, J.-P. Sutter et al.
vides information on both the value and the sign of the spin
density of the magnetically active nuclei. Moreover, it ap-
peared ideally suited for the compound envisaged in the
present study since all atoms of the molecular fragment link-
1
ing the radical units are nucleus active in NMR, that is, H,
13C, and 31P.
The MAS NMR spectra of the solid compound 1 are de-
1
picted in Figure 5. The H spectrum shows three isotropic
signals that are assigned to the triradical. The methyl pro-
tons of the tert-butyl groups appear at ꢀ0.13 ppm, and the
protons in position 3/5 and 2/6 of the phenyl groups are
seen at 70 and ꢀ138 ppm, respectively. An additional peak
near ꢀ244 ppm is due to nickelocene which has been used
for internal calibration of the temperature in much the same
way as described previously.[26] The 13C NMR signals
(middle spectrum)for the triradical with negative shifts are
assigned to the tert-butyl quaternary Ca atoms and to the
phenyl C in position 3/5 at ꢀ1211 and ꢀ535 ppm, respective-
ly. The sharp signal at 1066 ppm is assigned to the methyl Cb
atoms while the signal for the carbon atoms in position 4
and 2/6 appear as shoulders of the methyl signal. The latter
signal shifts had to be estimated and are considered to have
Figure 4. Top: Experimental and simulated EPR spectra of triradical 1 in
fluid solution. Bottom: Stick diagram showing the eight overlapping lines
along with the numbering scheme for 1.
atoms is not resolved, instead the observed signals are
1
rather broad with an apparent linewidth of DH= = 2 G.
2
The fact that the coupling with the protons is not resolved is
probably due to the small hfcc values of H atoms which lead
to an overlap of the signals. In addition, electron–electron
dipolar interactions, which may result from the not so large
inter-radical separation within the molecule broaden the
lines and hence contribute to an additional overlap. Such a
line broadening due to dipolar interactions, when going
from a mono-radical to a triradical derivative, was observed
for related series of polyradical compounds.[14]
The high-resolution spectrum obtained for triradical 1 in
fluid solution suggests that the spin density is spread over
the core of the molecule. Moreover, the observation of a
coupling with the P clearly indicates that spin density is lo-
cated on this atom. To gather a more precise information on
the spin distribution for this compound we applied solid
state MAS NMR spectroscopy.
Spin-density distribution by 1H, 13C and 31P MAS NMR
spectroscopy: Solid state magic angle spinning NMR spec-
troscopy has recently been shown to be a powerful and
simple alternative to polarized neutron diffraction for map-
ping the distribution of the spin density in paramagnetic mo-
lecular and supramolecular solids.[23–25] This technique pro-
Figure 5. 1H, 13C, and 31P MAS NMR spectra from top to bottom for tri-
radical 1 (T=310.8 K, spinning rate: 15 kHz). Spinning side-bands are
marked by asterisks. The signal of Cp2Ni was used as internal tempera-
ture standard; X = impurity (probably the diradical derivative).
132
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 128 – 139

Magnetic Properties
FULL PAPER
an error limit of ꢁ30 ppm. We were unable to detect the
signal of C1 probably because of its proximity to the spin
source, which entails unfavorable relaxation.[1] Finally, in the
31P MAS NMR spectrum a signal was found at ꢀ1781 ppm.
Conversion of the experimental signal shifts into the con-
tact shifts, dcon, (see Supporting Information)and further
into the hyperfine coupling constants, aX, of the nucleus X
by using Equation (1)gave the data collected in Table 2. In
Equation (1), gX is the nuclear gyromagnetic ratio, h and k
are the Planck and Boltzmann constant, T is the absolute
temperature, gX is the isotropic g factor of nucleus X, bE is
the Bohr magneton, and S is the spin quantum number. The
contact shifts were also converted into the spin densities, 1X
at the nucleus X by means of Equation (2)where m0 is the
vacuum permeability and a0 the Bohr radius, while the other
symbols have been mentioned with Equation (1). The abso-
lute value of spin density at a nucleus X can also be deduced
from the experimental hyperfine coupling constant obtained
by EPR, the conversion is done with the Fermi equation
given Equation (3), where bX is the nuclear magnetic
moment. The spin densities derived from MAS NMR data
as well as those obtained from EPR data are given in
Table 2.
Figure 6. a)Delocalization of the positive spin density as a result of the
mixing of the p* orbital of a NO unit with the aromatic p-orbitals; b)
Spin polarization map for triradical 1 (positive spin=› and negative
spin=fl).
Magnetic properties
Solid-state behavior: The magnetic properties of triradical 1
were investigated on a polycrystalline sample in the temper-
ature domain 2 to 300 K. The temperature dependence of
the molar magnetic susceptibility, cM, is given in Figure 7 as
a plot of cMT versus T. At 300 K the value for cMT is
1.10 cm3 Kmolꢀ1, close to the 1.125 cm3 Kmolꢀ1 expected for
3gXhkT
aX
¼
dTconðXÞ
dTconðXÞ
ð1Þ
ð2Þ
ð3Þ
gx2bE2SðSþ1Þ
three non-interacting S=1= spins, and steadily decreases as
2
the temperature is lowered to reach 0.28 cm3 Kmolꢀ1 at 2 K.
Such a behavior is indicative for overall antiferromagnetic
interactions among the spin carriers. In order to reveal pos-
sible contributions arising from intermolecular exchange in-
teractions, a diluted sample of 1 (6% mass)in polyvinyl
chloride (PVC)was measured in the same conditions. The
temperature dependence of cMT (Figure 7)shows however
only slight differences with respect to the crystalline sample
in the low temperature domain. This suggests that the main
contribution to the magnetic behavior arises from intramo-
lecular exchange and that intermolecular interactions are
9kT a03
m0gx2bE2ðSþ1Þ
1X
1X
¼
¼
3 S
aX
m0gXbX
The qualitative and quantitative data concerning the spin
born by the atoms of the molecule permit to draw a pattern
of the spin distribution where the sign of the spin alternates
from one atom to the other, as shown schematically in
Figure 6.
Table 2. Contact shifts (dcon in ppm), hyperfine coupling constants (a in G), and spin densities (1 in au”10ꢀ3)obtained from high-resolution EPR, solid-
state NMR and from DFT calculations.
EPR[a]
NMR[b]
DFT
exptl
311
com
298
[e]
[f]
jaj
j1j
d
d
a [G]
1
1HS
1HS
H2/6
H3/5
Hg
0.7
1.34[d]
0.55[d]
ꢀ138
ꢀ152
ꢀ2.03
ꢀ1.28
0.55
ꢀ0.01
ꢀ6
ꢀ6
0.29
70
65
0.87
2
3
[c]
ꢀ0.13
ꢀ1.4
ꢀ0.019
15/0/16[g]
ꢀ100
124
17/0/17]g]
ꢀ136
149
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
C1
C2/6
C3/5
C4
Ca
Cb
1055
ꢀ535
945
967
ꢀ696
846
3.26
ꢀ2.35
2.85
8.1
ꢀ5.86
7.1
ꢀ67
ꢀ59
126
143
ꢀ1211
ꢀ1325
ꢀ4.47
ꢀ11.15
ꢀ29
ꢀ27
1066
1087
3.66
9.15
P
O2
N1
O1
3.30
15.31[d]
98.9[d]
ꢀ1781
ꢀ1889
ꢀ10.26
ꢀ15.89
4
4
339
519
25
0
491
353
[c]
[c]
[c]
[c]
[c]
[c]
[c]
3.80
[c]
[a] Absolute values of the hfcc values obtained in CH2Cl2 at 293 K. [b] For details see Experimental Section and Supporting Information. [c] Not ob-
served. [d] Obtained from Equation (3)by using three times the observed hfcc values. [e] By using the UB3LYP/6-31g* approach. [f] By using the
UB3LYP/6-31+g* approach. [g] Spin densities on the H of the Me groups.
Chem. Eur. J. 2005, 11, 128 – 139
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
133

F. H. Kçhler, J. Veciana, J.-P. Sutter et al.
radical 1. The spectrum recorded at 90 K consists in a broad
symmetrical signal with non-symmetrical weak broad satel-
lite lines centered at g=2.00, as depicted in Figure 8. This
Figure 7. Experimental (&)and calculated ( c) cMT versus T behavior
for compound 1, best fit was obtained for J = ꢀ7.55 cmꢀ1 (see text). cMT
^
behavior ( )for diluted solid solution of triradical in PVC.
weak. Actually, there are intermolecular contacts as short as
about 2.6 Š, but in all cases the spin density of at least one
engaged nucleus is very small (Table 2).
3J
Ng 2b 2T
4kTFꢀzJ0
1þ1 expð2kT
Þ
Þ
Figure 8. X-band EPR spectrum of triradical 1 in a frozen CH2Cl2/toluene
matrix. Top: at 90 K in the g=2 and g=4 (insert)regions. Middle: Ex-
perimental spectrum in the Dms =1 region at 5 K. Bottom: Simulated
spectrum of the Dms =1 signal of the doublet ground state of 1.
cMT ¼
where F ¼
ð4Þ
3J
1þ5 expð2kT
The low temperature X-ray structure revealed the molecules
to be highly symmetrical, therefore the magnetic behavior
for triradical 1 was analyzed with a model considering three
set of signals corresponds to the Dms =1 transitions among
the sublevels of the electronic states of the triradical that
are populated at this temperature, that is, assuming the
same geometry as in solid state the sublevels should corre-
spond to both the doubly degenerate ground doublet state
and the first excited quartet state. A weaker signal at g=4
(see insert of Figure 8)which corresponds to the Dms =2
transitions and is characteristic of a quartet species, is also
visible. However, the signal at g=6 corresponding to the
Dms =3 transition which is characteristic for the quartet
state of triradical species was not detected, but this forbid-
den transition is rarely observed.[27] The temperature de-
pendence of the signal centered at g=4 was followed be-
tween 5 and 80 K, the variation of its intensity is plotted
versus 1/T (see Figure SI2 in the Supporting Information).
For the higher temperatures the signal intensity is found to
vary linearly in good agreement with a Curie law while for
the lower temperature domain a downward deviation is ob-
served confirming the occurrence of antiferromagnetic inter-
actions among the spin carriers of the molecule. Analysis of
this behavior by the model described above [Eq. (4)] yielded
an intramolecular exchange parameter of J=ꢀ3.9ꢁ
0.3 cmꢀ1, a value in good agreement with the exchange inter-
action deduced for solid state behavior (Figure SI2). A
worth noticing point of the EPR spectrum of triradical 1 is
the evolution of the signal at the g=2 region that becomes
progressively asymmetric when the temperature is lowered.
Thus, at 5 K it appears as three broad overlapped lines that
equivalent S=1= spin units located at the corners of an
2
equilateral triangle, that is, with equivalent exchange inter-
actions among them. The spin Hamiltonian appropriate to
ˆ
ˆ
ˆ
ˆ
ˆ ˆ
this situation is given by H = ꢀJ(S1·S2+S2·S3+S3·S1)where J
is the exchange parameter. A rigorous description of the
system would require to consider three of such interactions
because the asymmetric crystal unit contains three slightly
different molecules. But their respective intramolecular in-
teraction are anticipated to be very similar and, therefore, in
order to avoid over-parameterization the system was consid-
ered formed by a single type of molecule. The expression of
cMT used to analyze the magnetic behavior of 1 is given in
Equation (4)where N denotes Avogadroꢁs number, the
other symbols have been defined above. The weak intermo-
lecular interactions have been considered in the mean-field
approximation as zJ’. Least-squares fitting to the experimen-
tal data lead to J=ꢀ7.55ꢁ0.06 cmꢀ1 and zJ’ = ꢀ1.40ꢁ
0.06 cmꢀ1, by using g=2.00 as a fixed constant. The antifer-
romagnetic nature of the intramolecular exchange interac-
tion reveals that the ground doublet state, which is doubly
degenerate, is more stable than the quartet state by ꢀ3 J/2 =
11.3 cmꢀ1. In order to confirm definitively this intramolecu-
lar antiferromagnetic interaction a frozen matrix EPR study
of a diluted solution of triradical 1 in a solvent was under-
taken.
Frozen matrix EPR spectroscopy: The frozen matrix was
formed from a 10ꢀ3 m CH2Cl2/toluene (1:2)solution of tri-
can be simulated as a randomly oriented S=1= species
2
134
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 128 – 139

Magnetic Properties
FULL PAPER
showing a coupling with one 14N by using the following ani-
sotropic parameters: gxx =2.0080, gyy =2.0055, gzz =2.0018,
Ax =1.90, Ay =3.60 and Az =25.0 G. This result agrees with
the fact that at 5 K the doublet state is the most populated
state of triradical 1 and this species shows the anisotropic
components of g factor and 14N hyperfine coupling constant
similar to those exhibited by another reported substituted
tert-butylaminoxyl radicals.[27] This observation suggests that
in frozen solution at low temperatures triradical 1 is distort-
ed, probably due to solvent interactions, adopting a geome-
try in which its magnetic behavior is better described by
ground state is the high spin state but the calculated high
spin–low spin energy difference is about 30 kcalmolꢀ1. It is
thus much larger than the experimental estimate and it dem-
onstrates that the RODFT approach does not perform effi-
ciently. In a second step, we used the unrestricted DFT ap-
proach and we first optimized the geometry of the quartet
state without any symmetry constraint and by using the RT
crystal structure as a guess. The resulting energy of the opti-
mized structure is ꢀ1619.201939 au. This geometry exhibits
a C3 symmetry (see Supporting Information for selected
bond lengths, bond angles and dihedral, Table SI2). The cal-
culated parameters are consistent with the experimental
values. For example, the calculated P–O distance of 1.504 Š
is quite similar to the crystal RT distance of 1.490 Š while
the calculated dihedral angle O4-P1-C4-C3 of 1548 agrees
with the crystal value of 1638. For the doublet state, the
same experimental guess induces a rather large departure
from the C3 symmetry. Instead, we used the quartet opti-
mized geometry as a starting point and the optimized dou-
blet state is found to be the ground state with an energy of
ꢀ1619.201984 au, and this state is almost degenerate with
the quartet state. The HS–LS difference is less than
0.03 kcalmolꢀ1, and it may be considered as non significant.
The optimized geometries of both states are quite similar.
The eigenvalue of the total spin operator S2 gives a good
indication of the intrinsic quality of the unrestricted type
calculations which in some cases may lead to spin contami-
nation effects. For the high spin state, the S2 eigenvalue is
3.785 (reducing to 3.751 after annihilation of the first spin
contaminant)which is very close to the expected value of
3.750. For the low spin state, there exists a spin contamina-
tion because the calculated value is 1.784 (which reduces to
0.871 after annihilation of the first spin contaminant)for a
theoretical value of 0.750.
three S=1= spins arranged in an isosceles triangle with two
2
different antiferromagnetic J values for which a non degen-
erated doublet ground state is expected. Unfortunately, the
experimental uncertainties of the intensity of g=4 signal
does not permit to discriminate if its temperature depen-
dence is better described by either the equilateral or the
isosceles triangle models. It can be mentioned that the anal-
yses of the experimental data by both models yielded very
close exchange parameters.
Molecular modelisation of magnetic states and spin
populations
The sign and magnitude of the effective magnetic coupling
constant is intimately related to the balance between the
direct exchange and the inter-site electronic delocalization.
These constants indeed correspond to quite small differen-
ces between otherwise large total energies whose determina-
tion requires spin-dependent optimized geometries and ac-
curate treatment of the effects of electron correlation. From
comparative studies on molecular systems of reduced size, it
is generally stated that the magnetic parameters extracted
from Hartree–Fock (HF)theory underestimate the observed
values while a density functional theory analysis leads to an
overestimation of such terms, because in DFT the magnetic
orbitals are excessively delocalized between the centers.[28]
However, in DFT, these shortcomings can be corrected by
an appropriate choice of the (hybrid)functional and thus
makes such an approach quite attractive for these stud-
ies.[29,30] Mitani and co-workers compared the effective ex-
change integrals in m-phenylene coupling units. Their results
show that DFT integrals are rather close to CASSCF and
CASPT2 values while MPn gives a wrong sign even at
fourth order.[31]
The HS Mulliken spin densities are reported in Table 2
for the P and O4 atoms and for one of the three radicals.
For the quartet state, each radical carries a total spin density
of about 1.0. For the doublet state, two radicals exhibit a
+1.0 density (and 1i(LS) = 1i(HS), where i labels an atom
of these two radicals, and the corresponding values are not
reported in Table 2)and the third radical is antiferromagnet-
ically coupled with a spin density of ꢀ1.0 (and 1i(LS) =
ꢀ1i(HS), where i labels an atom of this third radical).
The electronic energies, the optimized geometries and the
spin densities of the doublet state and of the quartet state of
triradical 1 were computed by means of the density func-
tional theory method using the B3LYP functional and the
standard 6-31g* basis set. All the calculations were per-
formed with the Gaussian03 package.[32] It is necessary to in-
clude polarisation functions in the basis set in order to ac-
count for the hypervalency of the P atom. In order to save
computer time, the tert-butyl end groups were substituted by
methyl groups.
Discussion
The magnetic behavior of open-shell molecule-based sys-
tems is usually understood on the basis of the spin polariza-
tion Scheme displayed by the molecules. As far as conjugat-
ed hydrocarbon radicals are concerned, the spin distribution
is rather well established and follows the sign alternation
principle. But when a heteroatom is involved in the ex-
change pathway of a polyradical, the situation appears much
less straightforward and very little is known on the effective
implication of the heteroatom in the spin distribution. The
experimental results gathered in this study unambiguously
In a first step, we checked the performance of the restrict-
ed-open shell density functional theory (ROB3LYP). The
Chem. Eur. J. 2005, 11, 128 – 139
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
135

F. H. Kçhler, J. Veciana, J.-P. Sutter et al.
establish the presence of spin density at the phosphorus in
bridging position and show that its sign is in line with the al-
ternation principle. More surprising however, is the discrep-
ancy observed between the spin distribution pattern which
suggests a high-spin ground state for triradical 1, that is, an
intramolecular ferromagnetic interaction, and the actual in-
tramolecular antiferromagnetic interaction between the
three radical units.
ly for related phosphine oxide derivatives whereas it does
not apply for the corresponding phosphine compounds.[14] It
is useful to refer to the recent work of Schatzschneider and
Rentschler who have analyzed the spin density distributions
in substituted tert-butyl phenyl nitroxides by using the
2ꢀ
B3YP/6-31g* approach.[33] In PO3 nitroxide, the (non-nor-
malized)absolute values of the spin densities are 0.6937 on
2ꢀ
NO, 0.5214 on the phenyl ring and 0.2000 on the PO3 sub-
The spin density distribution found for compound 1
(Table 2)shows that the spin delocalization from the ami-
noxyl units onto the core of the molecule leads to the ex-
pected alternation of the sign of the spin densities among
the atoms (Figure 6). The conjugation between the p*–
SOMO orbital of the aminoxyl units and the p orbitals of
the phenyl groups is anticipated to yield substantial positive
spin on the aromatic carbon atoms in positions 2, 4 and 6,
and weaker negative spin for the other positions.[1] This is
indeed observed for 1 and confirmed by the opposite sign
for the spin densities found on the atoms bound to these
carbon centers. Thus, the H atoms 3/5 in meta with respect
to the aminoxyl units bear positive spin whereas negative
spin is found at H in position 2/6 and at the phosphorus
atom. It can be noticed that the coupling constant aP found
by EPR for compound 1 is an order of magnitude larger
than the one determined for the related nitronyl nitroxide
triradical derivative.[14] The same trend is observed by
31P NMR for the molecule which exhibits a resonance at
ꢀ1781 ppm, confirming that more spin is located on P for 1
than for the corresponding nitronyl nitroxide substituted
triradical. This increased spin density can be attributed to
the aminoxyl radical involved here but also to the fact that
for compound 1 the phosphorus is linked to phenyl C atoms
bearing positive spin, that is, atoms with a higher degree of
spin while for the related nitronyl nitroxide derivative the
phosphorus is linked to C atoms with smaller negative spin
densities. The amount of spin found at P (ꢀ15.9.10ꢀ3 au)for
1 is in the order of the spin density found on the aromatic C
atoms (j5”10ꢀ3 j to j10”10ꢀ3 j au)suggesting that the spin
transfer by polarization is as efficient with the heteroatom
as it is for the carbon atoms of the phenyl rings. The larger
spin density found at P might be related to the fact that this
atom is polarized by three C atoms bearing positive spin (C
in position 4)whereas the C atoms in position 3/5 are sur-
rounded by only two of such atoms, that is, C in positions 4
and either 2 or 6.
stituent. On the last group, the spin densities are ꢀ0.0367 on
P, and 0.1161 on the O atom which is perpendicular to the
plane of the phenyl ring while the two other O atoms carry
a spin density of 0.0373 and 0.0199. Thus, we might expect
that the negative spin density should be accounted for by
using a more extended 6-31+g* basis set. We thus calculated
the UB3LYP/6-31+g* energies and spin densities of triradi-
cal 1 by using the optimized UB3LYP/6-31g* geometries.
The 6-31+g* HS–LS splitting does not differ a lot from its
6-31g* counterpart: the doublet is now the ground state and
D is ꢀ0.02 kcalmolꢀ1. The corresponding spin densities of
the triradical 1 in its quartet state are reported in the last
column of Table 2. We found that 1P(6-31+g*)is even larger
than 1P(6-31g*). To check further the dependence of the
spin density on the basis set, we built a simplified model of
triradical 1, by replacing each phenyl group by a C=C bond,
while keeping the same geometrical parameters. This small-
er system allowed us to perform UB3LYP calculations using
6-31+g* (357 basis functions)and cc-pVTZ (724 basis func-
tions)basis sets. In each case, the doublet is the ground
state and D is ꢀ0.3 kcalmolꢀ1. No definite conclusion may
be drawn about the spin population on the P atom because
1P(6-31+g*) = ꢀ0.015 while 1P(cc-pVTZ) = +0.014. Thus,
it appears rather difficult to assess from theoretical calcula-
tions the spin density on the P atom. The size of this radical
precludes any post-HF calculations (CASSCF, CASPT2,
etc.)which are required to give quantitative results.
The rather efficient delocalization of the spin from the
aminoxyl units onto the core of the molecule is also ac-
counted for by the exchange interactions which exist be-
tween the spin carriers. But this magnetic behavior is in ap-
parent contradiction with the behavior anticipated for the
spin distribution scheme established for the molecule. The
magnetic behaviors exhibited by both the crystalline sample
and the diluted frozen sample reveal that the intramolecular
radical–radical interactions are antiferromagnetic with J=
ꢀ7.55 cmꢀ1. Based on topological rules, for a XY3 system
with spin density at Y, the simple spin polarization scheme
would be expected to lead to ferromagnetic interaction.
Precedents for intramolecular antiferromagnetic interaction
It is gratifying that good agreement is found between the
EPR, NMR, and DFT spin density distributions. In particu-
lar that the densities on the phenyl ring are alternate with
more (positive)spin at C2/6 and C4 than (negative)spin at
C3/5. The calculated spin densities on these carbon atoms
are, however, larger than the experimental ones. This dis-
crepancy may be related to the use of unrestricted approach
which is known to generally overestimate the difference be-
tween a-spin and b-spin populations.
in related poly-aminoxyl derivatives with an heteroatom (N,
[18,34–36]
B or Si)in bridging position have been reported.
For
these compounds, the antiferromagnetic interactions ob-
served between the radical units were attributed either to a
super-exchange mechanism involving a lone-pair, or to geo-
metric considerations, that is, dihedral angles close to 908
which hamper conjugation, or to the absence of spin at the
bridging atom. But these arguments do not apply for com-
pound 1. For the triradical the phosphorus has oxidation
The density on P deserves a particular comment. The cal-
culated density is positive while the experimental one is neg-
ative. This sign opposition on P has been observed previous-
136
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 128 – 139

Magnetic Properties
FULL PAPER
state (v)and thus no lone-pair which could take part in the
exchange mechanism.[14] In addition, the spin distribution
mapping establishes unambiguously that spin is located at
the phosphorus center. Structural features might be invoked
because the molecule is not planar, the P center induces a
pyramidal geometry which does not favor a good conjuga-
tion between the p system of the aromatic rings. As a conse-
quence, spin transfer from one radical unit to another
through the P is certainly hindered but does take place as
clearly shown by EPR and by the degree of spin localized at
P. This could lead to a pseudo-disjoint system with almost
degenerated doublet and quartet states favoring the low-
spin ground state.[37,38] While these considerations would ex-
plain the sign of the exchange interaction among the spin
carriers they do not account for the striking contradiction
that exists between the spin distribution found and the effec-
tive magnetic interaction between the radicals units. Conse-
quently, what appears as a discrepancy may simply reveal
that the observed exchange interaction is not governed by a
spin polarization mechanism.
of spin borne by the P suggests that the spin transfer by po-
larization is as efficient with the heteroatom as it is for the
aromatic carbon atoms of the phenyl rings.
Quite surprisingly, whereas the spin distribution scheme
for the triradical supports ferromagnetic interactions among
the radical units, the magnetic behavior found for this mole-
cule revealed a low-spin ground state characterized by an in-
tramolecular exchange parameter of J=ꢀ7.55 cmꢀ1. This ap-
parent discrepancy between the found spin distribution and
the effective intramolecular exchange interaction among the
aminoxyl units suggests that the latter is not governed by a
spin polarization mechanism.
Finally, it may be stressed that triradical 1 represents a
further example of open-shell derivatives for which the
structural features at low temperature are modified with re-
spect to those found at room temperature. It underlines the
importance that when a magneto-structural correlation is
envisaged, the structural properties should also be studied in
the temperature domain relevant for the magnetic proper-
ties.
The spin system realized by triradical 1 deserves a short
comment. This compound can be schematically described as
consisting of three spins located at the corners of an equilat-
eral triangle, each of which interacting equally with its two
neighbors. When governed by antiferromagnetic interactions
as it is the case here, there is no possibility to organize the
three spins in an antiparallel orientation, a situation known
as spin frustration.[39–41] In such a system, different organiza-
tion for the spins of the system are possible which all lead to
a magnetization at saturation lower than MS = 1mB expected
Experimental Section
Commercially available reagents were used without further purification,
except for those mentioned below which were purified as described. All
solvents used in the reactions were purified by conventional methods and
under inert atmosphere. Dry THF was obtained by distillation from Na/
benzophenone, CH2Cl2 was distilled from CaH2, MeCN was dried over
P2O5, DMF was dried over MgSO4 and distilled under vacuum, MeOH
was dried over Mg(OMe)2 prior to distillation. All reactions were carried
out under an atmosphere of dry N2 unless otherwise specified. 2-Methyl-
2-nitrosopropane was prepared by a procedure reported in the litera-
ture.[43] Infrared spectra were recorded in the range 4000–400 cmꢀ1 by
using a Perkin Elmer FT-IR Paragon 1000. Elementary analysis were per-
formed by the “Service Central dꢁAnalyse du CNRS” at Vernaison,
for a S=1= spin ground state.[42] This was also observed for
2
the field dependence of the magnetization recorded for
compound 1 at 2 K, for any field the experimental value is
below the value calculated by the Brillouin function for a
S=1= spin (g = 2.00)at this temperature (see Supporting
2
France. H, 13C{1H} et 31P{1H} NMR spectra in solution were obtained on
1
Information). The magnetic behavior for compound 1 but
also the structural features at low temperature are in agree-
ment with a frustrated spin system. Such a characteristic
makes the triradical a valuable building block for the prepa-
ration in association with paramagnetic metal ions of supra-
molecular magnetic coordination frameworks exhibiting
properties in relation with spin frustration.
a
Bruker ARX 300 spectrometer operating at 299.5, 75.42 and
121.42 MHz, respectively; chemical shifts (d)are given in ppm. EPR
spectra were recorded on a Bruker EMX X-band (9.4 GHz)equipped
with an Oxford ESR-900 helium cryostat, the spectra simulations were
done with the program WINEPR and EPRFTSM programs.[44] Magnetic
measurements down to 2 K were carried out in a Quantum Design
MPMS-5S SQUID susceptometer. All magnetic investigations were per-
formed on a polycrystalline sample. The molar susceptibility was correct-
ed for the sample holder and for the diamagnetic contribution of all
atoms by using Pascalꢁs tables.[45,46]
Concluding Remarks
Synthesis of triradical 1 (route A)
Tri(4-bromophenyl)phosphine (3):[47] A solution of p-dibromobenzene
(20 g, 84.7 mmol)in freshly distilled THF (100 mL)was added dropwise
to a solution of nBuLi (58.28 mL, 93.1 mmol)at ꢀ788C. When the addi-
tion was complete, PCl3 (2.43 mL, 84 mmol)in THF (45 mL)was intro-
duced to the mixture at ꢀ788C over a period of 2 h. The reaction mixture
was allowed to warm to room temperature, then filtered and the solvent
was removed under reduced pressure. The yellow residue was dissolved
in Et2O (50 mL)and MeOH (30 mL)was added. The solution was stirred
in order to slowly evaporated Et2O and 3 was obtained as a white micro-
crystalline solid (21.79 g, 52%). 1H NMR (CDCl3, 293 K): d=7.46 (d,
J(H,H)=8.17 Hz, 6H, Ph), 7.13 (dd, J(H,H)=8.17, J(H,P)=7.27 Hz, 6H,
Ph); 13C{1H} NMR (CDCl3, 293 K): d=134.97 (d, J(C1,P)=12.20 Hz,
Ph), 134.82 (d, J(C2,P)=20.74 Hz, Ph), 131.68 (d, J(C3,P)=7.31 Hz, Ph),
123.69 (s, Ph); 31P{1H,13C} NMR (CDCl3, 293 K): d=ꢀ7.95 (s); IR (KBr):
The results gathered for triradical 1 unambiguously establish
that the bridging P atom is involved in the spin distribution
exhibited by the molecules. This has been confirmed by
solid state MAS NMR and fluid EPR spectroscopies which
proved to be powerful tools for gaining access to the spin
distribution for an organic radical. The spin densities located
on all atoms of the molecule could be probed, and led to a
precise spin distribution map for the triradical. These tech-
niques permitted to establish that spin density is located at
the phosphorus (1=ꢀ15.10ꢀ3 au)and that its sign is in line
with the sign-alternation principle. Moreover, the magnitude
Chem. Eur. J. 2005, 11, 128 – 139
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
137

F. H. Kçhler, J. Veciana, J.-P. Sutter et al.
n˜ = 2986 (w), 2932 (w), 2871(w), 1568 (m), 1475 (s), 1381 (s), 1067 (s),
Ph), 7.08 (dd, J(H,H)=7.84, J(H,P)=7.08 Hz, 6H, Ph), 1.10 (s, 27H,
tBu), 0.88 (s, 27H, tBu-Si), ꢀ0.12 (s, 18H, (CH3)2-Si); 13C{1H} NMR
(CDCl3, 293 K): d=154.69 (s, Ph), 132.55 (d, J(C2,P)=20.14 Hz, Ph),
131.37 (d, J(C1,P)=10.99 Hz, Ph), 125.15 (d, J(C3,P)=7.32 Hz, Ph),
61.36 (s, Me3C-Si), 61.26 (s, Me3C-N), 26.87 (s, (CH3)3C-N), 17.95 (s,
(CH3)3C-Si), ꢀ4.69 (s, (CH3)2-Si); 31P{1H,13C} NMR ([D6]DMSO, 293 K):
d = ꢀ8.38 (s).
Tris[p-(tert-butyl-N-hydroxylamino)phenyl]phosphine (2b): A solution of
2a (2.29 g, 2.61 mmol)in THF (10 mL)was reacted under nitrogen with
a 0.75n THF solution of Bu4NF (15 mL, 11.5 mmol)for 36 h. Afterwards,
a saturated aqueous solution of NH4Cl (50 mL)was added and the organ-
ic fraction extracted with Et2O. The organic phase was dried over MgSO4
and Et2O partly evaporated under reduced pressure. The addition of
814 (s), 725 (s), 514 cmꢀ1 (m).
Tri[p-(tert-butyl-N-hydroxylamino)phenyl]phosphineoxide (2c):
A so-
lution of n-butyllithium (11 mL, 18 mmol, 1.6m)was added at ꢀ408C to
a solution of tri(4-bromophenyl)phosphine (3; 3 g, 5.97 mmol)in Et 2O
(50 mL). After the reaction mixture was allowed to warm to 08C, the
mixture was cooled to ꢀ788C and a solution of 2-methyl-2-nitrosopro-
pane (1.56 g, 8.95 mmol)in THF (10 mL)was added dropwise. Stirring
was continued for 2 h at room temperature. Afterwards, a saturated
aqueous solution of NH4Cl in H2O and Et2O were added. The organic
phase was collected and dried over MgSO4 before the solvent was evapo-
rated in vacuo. The treatment of the residue with CH3CN gave 2 as a
white powder (1.21 g, 38%). 31P{1H,13C} NMR ([D6]DMSO, 293 K): d=
27.95 (s, PV), ꢀ9.07 (s, PIII); IR (KBr): n˜ = 3296 (s), 2982 (s), 2926 (s),
2876 (s), 1593 (s), 1458 (m), 1388 (m), 1189 (m), 1119 (s), 838 (m), 744
(m), 575 cmꢀ1 (s).
CH3CN under strong agitation then resulted in the formation of
2
(1.28 g, 95%)as white powder. IR (KBr): n˜ = 3315 (F), 2981 (m), 2932
(m), 1591 (m), 1486 (s), 1200 (m), 1173 (m), 837 (s), 747 (w), 573 cmꢀ1
(m); 1H NMR ([D6]DMSO, 293 K): d=8.33 (s, 3H, N-OH), 7.24 (dd,
J(H,H)=7.63, J(H,P)=1.21 Hz, 6H, Ph), 7.09 (dd, J(H,H)=7.63,
J(H,P)=6.48 Hz, 6H, Ph), 1.06 (s, 27H, tBu); 13C{1H} NMR ([D6]DMSO,
293 K): d=151.63 (s, Ph), 132.52 (d, J(C2,P)=19.53 Hz, Ph); 132.05 (d,
J(C1-P)=9.77 Hz, Ph); 124.47 (d, J(C3,P)=6.71 Hz, Ph), 59.79 (s, Me3C-
N), 26.29 (s, (CH3)3C-N); 31P{1H,13C} NMR: d = ꢀ8.99 (s).
Tri[p-(tert-butyl-N-oxylamino)phenyl]phosphineoxide (1): A solution of
NaIO4 (400 mg, 1.8 mmol)in H 2O (30 mL)was added to a solution of
tri[p-(tert-butyl-N-hydroxylamino)phenyl]phosphineoxide (2c; 0.150 g,
0.288 mmol)in CH 2Cl2 (30 mL)at 0 8C. The mixture was stirred for
15 min and the organic phase was collected, dried over MgSO4 and the
solvent removed to obtain a red powder. The latter was purified by chro-
matography over silica gel by using CH2Cl2/MeOH 9:1 as eluent and re-
crystallized from CHCl3/Et2O to obtain 1 (0.118 g, 78%)as red crystals.
IR (KBr): n˜ = 2977 (m), 2930 (m), 2871 (w), 15781 (m), 1482 (m), 1189
(s), 1116 (F), 745 (s), 564 cmꢀ1 (m); EPR (CH2Cl2, 293 K): g=2.0051; 8
lines, aN =3.86, aP =3.28 G; elemental analysis calcd (%)for
C30H39N3O4P (536.6): C 67.14, H 7.32, N 7.83; found: C 67.32, H 7.36, N
7.86.
Tri[p-(tert-butyl-N-oxylamino)phenyl]phosphineoxide (1): The oxidation
of 2b to triradical 1 was performed as described above for 2c.
31P, 13C, and 1H MAS NMR spectra: The 31P, 13C, and 1H MAS NMR
spectra were recorded with a Bruker MSL 300 spectrometer by using mi-
crocrystalline samples of 1. A small amount of nickelocene has been
added to 1 and used as internal temperature standard.[26,51] The powder
was packed into 4 mm ZrO2 rotors and sealed with Kel-F caps. The free
induction decays were sampled after applying single 908 pulses of 2–4 ms
duration with a delay time of 100–500 ms between successive scans. Data
handling included exponential multiplication up to the matched filter and
base line correction. The experimental signal shifts, dexptl, were deter-
mined relative to external adamantane (d(1H)=2.0)and ammoniumhy-
drogenphosphate ((d(31P)=1.1 ppm). Contact shifts were obtained after
Synthesis of triradical 1 (route B)
Bromo-4-[tert-butyl-N-hydroxylamino]aniline (4a) (adapted from Inoue
et al.[48]): n-Butyl lithium (39.7 mL, 63.5 mmol, 1.6m in hexane)was
added dropwise at ꢀ788C to a solution of p-dibromobenzene (15 g,
63.5 mmol)in THF (100 mL.) The reaction mixture was stirred for
15 min at low temperature then gradually warmed to room temperature.
The solution was again cooled to ꢀ788C and a solution of 2-methyl-2-ni-
trosopropane (6.08 g, 63 mmol)in THF (10 mL)was added slowly. The
mixture was then stirred for 2 h at room temperature before a saturated
aqueous solution of NH4Cl was added followed by Et2O. The organic
layer was separated, dried over MgSO4 and the solvents removed in
vacuo. The resulting residue was purified by chromatography over silica
gel by using hexane as eluent to obtain 4a (10.86 g, 71%)as a colorless
subtracting from the experimental signal shifts, dexptl, the signal shifts, ddia
,
of corresponding nuclei of diamagnetic reference compounds, that is, tri-
phenylphosphine oxide with d(31P)=28.5 ppm, d(13C2/6)=128 ppm,
d(13C3/5)=132 ppm, d(13C4)=134 ppm, and d(1H)=7.5 ppm for all aro-
matic protons, and compound 2b was the reference compound for the
proton signals of the radical moiety with d(13CCH3)=58 ppm, d(13CH3)=
25 ppm, and d(C,1H3)=1.2 ppm. Note that the dipolar shift contribution
can be neglected, because the g factor anisotropy of organic radicals is
generally very small.[52] Further details are given in Table S1 of the Sup-
porting Information.
1
oil. H NMR (CDCl3, 293 K): d=7.35 (d, J(H,H)=8.68 Hz, 2H, Ph), 7.09
(d, J(H,H)=8.68 Hz, 2H, Ph), 1.09 (s, 9H, tBu).
Bromo-4-[N-tert-butyl-N-(tert-butyldimethylsilyloxy)]aniline (4b): The
protection of the hydroxylamino group of 4a was performed according to
a method described by Corey et al.[49] To a degassed solution of 4a (4 g,
16.3 mmol)in DMF (15 mL)were added imidazole (3.34 g, 49.1 mmol)
and tert-butylchlorodimethylsilane (4.8 g, 32 mmol), and the reaction mix-
ture was stirred at 50–608C for 24 h. After cooling and addition of H2O
Crystallographic studies: Small orange pyramidal shaped single crystals
of approximate dimensions 0.12”0.12”0.08 mm3 were mounted on a
Bruker-Nonius four circle diffractometer equipped with a CCD camera
and a graphite monochromated MoKa radiation source. Effective absorp-
tion correction was performed (SCALEPACK).[53] The structures were
solved and the atomic parameters were refined by full-matrix least-
squares method on F2 by using the SHELX-97 package.[54] All H atoms
were calculated and treated according to the riding model during refine-
ment with isotropic displacement corresponding to the atom they are
linked to.
(20 mL)the organic layer was extracted with hexane, dried over MgSO
4
and the solvent removed in vacuo. The residue was purified by chroma-
tography over silica gel by using hexane as eluent to obtain 4b as a color-
less oil (4.85 g, 83%). 1H NMR spectra in agreement with the one previ-
ously reported.[50]
CCDC-242664 and -242665 contain the supplementary crystallographic
data for this paper. These data can be obtained free of charge via
www.ccdc.cam.ac.uk/conts/retrieving.html, or from the Cambridge Crys-
tallographic Data Centre, 12 Union Road, Cambridge CB2 1EZ, UK;
fax: (+44)1223-336033; or email: deposit@ccdc.cam.ac.uk.
Tris[p-{(O-tert-butyldimethylsilyl)-tert-butyl-N-oxylamino}phenyl]phos-
phine (2a): n-Butyl lithium (7.2 mL, 12 mmol, 1.6m in hexane)was
added to a solution of 4b (4.25 g, 11.9 mmol)in Et 2O (50 mL)at ꢀ788C.
The reaction mixture was stirred for 45 min at low temperature, then a
solution the PCl3 (0.35 mL, 3.95 mL)in THF (10 mL)was added drop-
wise over a period of 2 h. The stirring was maintained 2 h at low temper-
ature followed by 6 h at room temperature. After evaporation of solvents,
the residue was taken off in CH2Cl2 (40 mL)and filtered under nitrogen
to obtain a clear yellow/green solution. After removing the CH2Cl2 in
vacuum triturating of the residue with CH3CN yielded 2a (1.48 g, 43%)
as a white powder. IR (KBr): n˜ = 2989 (m), 2932 (w), 2871(w), 1578 (m),
1483 (m), 1252 (s), 1173 (w), 857 (s), 743 (s), 586 cmꢀ1 (m); 1H NMR
([D6]DMSO, 293 K): d=7.19 (dd, J(H,H)=7.84, J(H,P)=1.02 Hz, 6H,
Acknowledgement
This work was supported by the European Union TMR Research Net-
work ERB-FMRX-CT-98-0181 (1998–2003)entitled “ Molecular Magnet-
138
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 128 – 139

Magnetic Properties
FULL PAPER
ism: From Materials toward Devices”. J.V. and J.V.-G. thank the Direc-
ciꢂn General de Investigaciꢂn, Spain (Project MAT2003–04699)and the
Generalitat de Catalunya (Grant 2001SGR00362).
[30] T. Soda, Y. Kitagawa, T. Onishi, Y. Tanako, Y. Shigeta, H. Nagao, Y.
Yoshioka, K. Yamaguchi, Chem. Phys. Lett. 2000, 319, 223–230.
[31] M. Mitani, D. Yamaki, Y. Takano, Y. Yoshioka, K. Yamaguchi, J.
Chem. Phys. 2000, 113, 10486–10504.
[32] Gaussian 03, Revision Bed.04, M. J. Frisch, G. W. Trucks, H. B.
Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, J. A. J.
Montgomery, T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam,
S. S. Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scal-
mani, N. Rega, G. A. Petersson, H. Nakatsuji, M. Hada, M. Ehara,
K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y.
Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P.
Hratchian, J. B. Cross, C. Adamo, J. Jaramillo, R. Gomperts, R. E.
Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W.
Ochterski, P. Y. Ayala, K. Morokuma, G. A. Voth, P. Salvador, J. J.
Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C.
Strain, O. Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari,
J. B. Foresman, J. V. Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cio-
slowski, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaro-
mi, R. L. Martin, D. J. Fox, T. Keith, Al-M. A. Laham, C. Y. Peng,
A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W.
Chen, M. W. Wong, C. Gonzalez, J. A. Pople, Gaussian Inc., Pitts-
burgh PA, 2003.
[1] F. H. Kçhler, in Magnetism: Molecules to materials, Vol. 1 (Eds.: J. S.
Miller, M. Drillon), Wiley-VCH, Weinheim, 2001, pp. 379–430.
[2] J. Schweizer, B. Gillon, in Magnetic propereties of organic materials
(Ed.: P. M. Lahti), Marcel Dekker, New York, 1999, pp. 449–473.
[3] J. Schweizer, E. Ressouche, in Magnetism: Molecules to materials
Vol. 1 (Eds.: J. S. Miller, M. Drillon), 2001, pp. 325–355.
[4] J. Cirujeda, J. Vidal-Gancedo, O. Jürgens, F. Mota, J. J. Novoa, C.
Rovira, J. Veciana, J. Am. Chem. Soc. 2000, 122, 11393–11405.
[5] H. McConnell, J. Chem. Phys. 1963, 39, 1910.
[6] P. M. Lahti, Magnetic Properties of Organic Materials, Marcel
Dekker, New York, 1999.
[7] K. Itoh, M. Kinoshita, Molecular magnetism: new magnetic materi-
als, Gordon and Breach, Amsterdam, 2000.
[8] C. Rancurel, J.-P. Sutter, O. Kahn, P. Guionneau, G. Bravic, D. Chas-
seau, New J. Chem. 1997, 21, 275–277.
[9] C. Rancurel, J.-P. Sutter, T. Le Hoerff, L. Ouahab, O. Kahn, New J.
Chem. 1998, 22, 1333–1335.
[33] U. Schatzschneider, E. Rentschler, J. Mol. Struct. 2003, 63, 163–168.
[34] T. Itoh, K. Matsuda, H. Iwamura, Angew. Chem. 1999, 111, 1886–
1888; Angew. Chem. Int. Ed. 1999, 38, 1791–1793; .
[35] E. C. Brown, W. T. Borden, J. Phys. Chem. A 2002, 106, 2963–2969.
[36] Y. Liao, M. Baskett, P. M. Lahti, F. Palacio, Chem. Commun. 2002,
252–253.
[10] C. Rancurel, D. B. Leznoff, J.-P. Sutter, S. Golhen, L. Ouahab, J.
Kliava, O. Kahn, Inorg. Chem. 1999, 38, 4753–4758.
[11] D. B. Leznoff, C. Rancurel, J.-P. Sutter, S. J. Rettig, M. Pink, C. Paul-
sen, O. Kahn, J. Chem. Soc. Dalton Trans. 1999, 3593–3599.
[12] D. B. Leznoff, C. Rancurel, J.-P. Sutter, S. J. Rettig, M. Pink, O.
Kahn, Organometallics 1999, 18, 5097–5102.
[37] T. B. Borden, E. R. Davidson, J. Am. Chem. Soc. 1977, 99, 4587–
4594.
[38] T. Matsumoto, T. Ishida, N. Koga, H. Iwamura, J. Am. Chem. Soc.
1992, 114, 9952–9959.
[39] G. Toulouse, Commun. Meth. Phys. 1977, 2, 115–119.
[40] H. Y. Woo, H. So, M. T. Pope, J. Am. Chem. Soc. 1996, 118, 621–
626.
[13] C. Rancurel, D. B. Leznoff, J.-P. Sutter, P. Guionneau, D. Chasseau,
J. Kliava, O. Kahn, Inorg. Chem. 2000, 39, 1602–1605.
[14] C. Rancurel, H. Heise, F. Kçhler, U. Schatzschneider, E. Rentschler,
J. Vidal-Gancedo, J. Veciana, J.-P. Sutter, J. Phys. Chem. A 2004,
108, 5903–5914.
[15] K. Torssell, Tetrahedron 1977, 33, 2287–2291.
[16] C. Rancurel, N. Daro, O. Benedi Borobia, E. Herdtweck, J.-P.
Sutter, Eur. J. Org. Chem. 2003, 167–171.
[41] O. Kahn, Chem. Phys. Lett. 1997, 265, 109–114.
[42] M. F. Collins, O. A. Petrenko, Can. J. Phys. 1997, 75, 605–655.
[43] J. C. Stowel, J. Org. Chem. 1971, 36, 3055–3056.
[44] Simulation of EPR spectrum in isotropic conditions was made with
Simfonia (V1.0)program (Bruker Instruments, Germany)and of
EPR in frozen solution with EPRFTSM program (B. Kriste, Freie
Universität Berlin, Germany).
[17] A. W. Hanson, Acta Crystallogr. 1953, 6, 32–34.
[18] T. Itoh, K. Matsuda, H. Iwamura, K. Hori, J. Am. Chem. Soc. 2000,
122, 2567–2576.
[19] W. Fujita, K. Awaga, Science 1999, 286, 261–262.
[20] D. A. Shultz, R. M. Fico, P. D. Boyle, J. W. Kampf, J. Am. Chem.
Soc. 2001, 123, 10403–10404.
[45] G. A. Baker, Jr., G. S. Rushbrooke, H. E. Gilbert, Phys. Rev. A
1964, 135, 1272–1277.
[46] O. Kahn, Molecular Magnetism, VCH, New York, 1993.
[47] H. Ravindar, H. Hemling, H. Schumann, J. Blum, Synth. Commun.
1992, 22, 841–851.
[21] M. Fettouhi, E. B. Ali, E.-A. M. Ghanam, S. Golhen, L. Ouahab, N.
Daro, J.-P. Sutter, Inorg. Chem. 2002, 41, 3705–3712.
[22] M. Fettouhi, E. B. Ali, M. Morsy, S. Golhen, L. Ouahab, L. B. Guen-
nic, J.-Y. Saillard, N. Daro, J.-P. Sutter, E. Amouyal, Inorg. Chem.
2003, 42, 1316–1321.
[48] K. Inoue, H. Iwamura, Angew. Chem. 1995, 107, 973–974; Angew.
Chem. Int. Ed. Engl. 1995, 34, 927–928.
[23] G. Maruta, S. Takeda, R. Imachi, T. Ishida, T. Nogami, K. Yamagu-
chi, J. Am. Chem. Soc. 1999, 121, 424–431.
[49] E. J. Corey, A. Venkateswarlu, J. Am. Chem. Soc. 1972, 94, 6190–
6191.
[24] H. Heise, F. H. Kçhler, F. Mota, J. J. Novoa, J. Veciana, J. Am.
Chem. Soc. 1999, 121, 9659–9667.
[50] P. M. Lahti, Y. Liao, M. Julier, F. Palacio, Synth. Met. 2001, 122,
485–493.
[25] H. Heise, F. H. Kçhler, M. Herker, W. Hiller, J. Am. Chem. Soc.
2002, 124, 10823–10832.
[51] F. H. Kçhler, X. Xie, Magn. Reson. Chem. 1997, 35, 487–492.
[52] R. J. Kurland, R. B. McGarvey, J. Magn. Reson. 1970, 2, 286–301.
[53] Z. M. Otwinowski, W. Minor, Processing of X-ray Diffraction Data
Collected in Oscillation Mode, Vol. 276, Academic Press, 1997.
[54] G. M. Sheldrick, Program for the refinement of crystal structures,
University of Gçttingen (Germany), 1997.
[26] H. Heise, F. H. Kçhler, X. Xie, J. Magn. Reson. 2001, 150, 198–206.
[27] N. M. Atherton, Principle of Electron Spin Resonance, Ellis Hor-
wood Prentice Hall, New York, 1993.
[28] J. Cabrero, C. J. Calzado, D. Maynau, R. Caballol, J. P. Malrieu, J.
Phys. Chem. A 2002, 106, 8146–8155.
[29] V. Baron, B. Gillon, O. Plantevin, A. Cousson, C. Mathoniere, O.
Kahn, A. Grand, L. ꢃhrstrçm, B. Delley, J. Am. Chem. Soc. 1996,
118, 11822–11830.
Received: June 28, 2004
Published online: November 11, 2004
Chem. Eur. J. 2005, 11, 128 – 139
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
139