Towards a better understanding of magnetic interactions within m-phenylene α-nitronyl nitroxide and imino nitroxide based radicals, part III: Magnetic exchange in a series of triradicals and tetraradicals based on the phenyl acetylene and biphenyl couplin

Article abstract of DOI:10.1002/chem.200400552

The present work completes and extends our previous reports^[1,2] on the determination of the magnetic ground state and on the strength of the through bond exchange coupling within series of biradicals. This knowledge was subsequently exploited

Full text of DOI:10.1002/chem.200400552

Towards a Better Understanding of Magnetic Interactions within
m-Phenylene a-Nitronyl Nitroxide and Imino Nitroxide Based Radicals,
Part III: Magnetic Exchange in a Series of Triradicals and Tetraradicals
Based on the Phenyl Acetylene and Biphenyl Coupling Units
Laure Catala,[b] Jacques Le Moigne,[c] Nathalie Gruber,[a] Juan J. Novoa,[d] Pierre Rabu,[c]
Elie Belorizky,[e] and Philippe Turek*[a]
Abstract: The present work completes
and extends our previous reports[1,2] on
the determination of the magnetic
ground state and on the strength of the
through bond exchange coupling within
series of biradicals. This knowledge
was subsequently exploited for the
analysis of the magnetic interactions in
their crystals. We report here the stud-
ies of series of triradicals incorporating
a-nitronyl nitroxides (NN) or a-imino
nitroxides (IN) as terminal radical frag-
ments connected through a m-phenyl-
ene coupling unit in one case and a
phenyl acetylene unit in other case.
Tetraradical derivatives have also been
studied. The studies of isolated mole-
cules (EPR in solution and DFT calcu-
lations) allow the assessment of the
magnetic interactions through the mag-
netic coupling unit. All triradical deriv-
atives are found to exhibit a quartet
ground state, whereas a singlet ground
state is determined for the tetraradical.
This last result reinforces previous find-
ings that the singlet ground state is fav-
oured in related biradicals involving
similar m-phenylene couplers. More-
over, the through bond magnetic ex-
change coupling for the ortho–meta
connectivity could be demonstrated as
being ferromagnetic, thus ascertaining
our previous hypotheses.[1] The magnet-
ic properties of the triradicals and tet-
raradicals in their solid state have been
rationalized by using a previously pro-
posed methodology,[2] allowing to iden-
tify the most relevant magnetic path-
ways.
Keywords:
theory
density
EPR spectroscopy
functional
·
·
magnetic properties
·
molecular
materials · radicals
[a] Dr. N. Gruber, Prof. P. Turek
Institut Charles Sadron (CNRS-UPR 22)
Universitꢀ Louis Pasteur, 6 rue Boussingault
67083 Strasbourg (France)
Introduction
Molecular-based magnetism is an active research area with
respect to the promising new materials that might be pre-
pared in the near future, whether at the macroscopic scale
or at the so-called nanoscale.[3] Purely organic magnets are
of special interest as they combine properties of organic ma-
terials, such as solubility and transparency, and offer the
possibility of combining magnetic, optical, mechanical prop-
erties by the fine-tuning of organic synthesis.[4] Apart from
the purely organic approach, such paramagnetic molecules
are also good candidates for generating original networks by
coordination with transition metal ions and reinforcing the
magnetic interactions between them.[5] Two different strat-
egies can be distinguished, namely the intramolecular
(“through-bond”) and the intermolecular (“through-space”).
The intermolecular approach consists in using supramolec-
ular engineering tools, for example, hydrogen bonds[6] to
provide a better control of the packing of radical moieties,
aiming at the generation of a 3D magnetic network to get
magnetic order. However, there is still much to do to predict
a priori the magnetic behaviour exhibited by a compound
just looking at the molecular packing.[7] The intramolecular
Fax : (+33)388-414-099
E-mail: turek@ics.u-strasbg.fr
[b] Dr. L. Catala
Institut de Chimie Molꢀculaire et des Matꢀriaux dꢁOrsay
bꢂtiment 420, Universitꢀ Paris-Sud
91120 Orsay (France)
[c] Dr. J. Le Moigne, Dr. P. Rabu
Institut de Physique et Chimie des Matꢀriaux
de Strasbourg (CNRS-UMR 7504), Universitꢀ Louis Pasteur
23 rue du Loess
67034 Strasbourg (France)
[d] Prof. J. J. Novoa
Departament de Quꢃmica Fꢃsica
Facultat de Quꢃmica and CERQT, Park Cientific
Universitat de Barcelona, Av. Diagonal 647
08028 Barcelona (Spain)
Associate Member of the CEPBA-IBM Research Institute
[e] Prof. E. Belorizky
Laboratoire de Spectromꢀtrie Physique (CNRS-UMR 5588)
Universitꢀ Joseph Fourier, B. P. 87
38402 Saint Martin dꢁHꢄres Cedex (France)
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author.
2440
ꢅ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/chem.200400552
Chem. Eur. J. 2005, 11, 2440 – 2454

FULL PAPER
approach is based on the connection of radical fragments
via a conjugated coupler, so as to provide high spin mole-
cules by controlling the nature of the magnetic interactions
of the spins.[8] Exploiting this way may result in the obtain-
ing of magnetic organic polymers.[9] However, understanding
intramolecular interactions in polymers is complicated by
the presence of defects, and requires model compounds to
assess precisely the strength and nature of the magnetic in-
teractions through the conjugated coupler. To do so, it is of
utmost importance to study oligoradicals based on various
coupling units.
Following such guiding principles, some of our previous
work (Part I)[1] considered biradical derivatives consisting of
two imino-nitroxides (IN) radicals connected in a meta con-
figuration through a phenyl acetylene coupler. A singlet
magnetic ground state was found for the meta derivatives,[1]
whereas the widely used “topological rules”[10] predicted a
triplet ground state[11] as long as the torsion angle between
the phenyl and the radicals remains small.[12] Since little was
known for the m-phenylene bridging unit when connected
to nitronyl nitroxides (NN) and imino nitroxides IN, we first
investigated model biradicals and an extensive study has
been previously reported (Part II).[2]
In the present work, we extend the previous results by de-
signing triradicals and tetraradicals (Scheme 1). The work
on triradicals is a direct extension of our previous work,
both for the phenyl acetylene magnetic coupling unit
(Part I) and for the m-phenylene bridging unit (Part II). The
biphenyl bridge[13] has also been studied in order to investi-
gate the effects on the intramolecular exchange coupling of
a more distorted but shorter coupling unit. Furthermore,
some hetero-triradicals were synthesized in order to assess
the influence of the nature of the connected radicals, that is,
NN versus IN. Not only such compounds permitted to study
the variation of the strength of intramolecular magnetic in-
teractions, but they also provided a nice way to understand
the solid state magnetic behaviour within isostructural
series. Following the procedure depicted in Part II, slight
modifications of the geometry at the intermolecular contacts
(structural motives) could be related to the variation of the
magnetic behaviour. Indeed, statistical studies have shown
Abstract in French: Le prꢀsent travail complꢁte et ꢀtend nos
travaux prꢀcꢀdents portant sur la dꢀtermination des couplages
dꢂꢀchange magnꢀtique intramolꢀculaires de sꢀries de biradi-
caux,[1,2] et partant de cette connaissance, de lꢂanalyse des cor-
rꢀlations magnꢀtostructurales dans lꢂꢀtat cristallin. Nous prꢀ-
sentons ici lꢂꢀtude de sꢀries de triradicaux comprenant des ra-
dicaux a-nitronyl nitroxydes (NN) ou a-imino nitroxydes
(IN) substituꢀs en mꢀta dꢂun groupement phꢀnylꢁne dꢂune
part, et en para dꢂun groupement phꢀnylꢁne ꢀthynylꢁne
dꢂautre part. Des dꢀrivꢀs tꢀtraradicalaires ont aussi ꢀtꢀ synthꢀ-
tisꢀs et ꢀtudiꢀs. Les interactions dꢂꢀchange magnꢀtique ꢃ tra-
vers les liaisons sont estimꢀes dꢂaprꢁs lꢂꢀtude de ces molꢀcules
isolꢀes par RPE en solution ainsi que par des calculs effec-
tuꢀs dans lꢂapproximation de la fonctionnelle densitꢀ (DFT).
Alors que lꢂensemble des dꢀrivꢀs triradicalaires prꢀsente un
ꢀtat fondamental magnꢀtique quartet de spin, un ꢀtat singulet
est proposꢀ pour les tꢀtraradicaux. Ce dernier rꢀsultat confir-
me nos travaux prꢀcꢀdents[1] ayant conclu ꢃ un ꢀtat fonda-
mental singulet pour des biradicaux basꢀs sur le mÞme cou-
pleur. De plus, le couplage dꢂꢀchange magnꢀtique au travers
du coupleur phꢀnylꢁne ꢀthynylꢁne pour des biradicaux substi-
tuꢀs dans une topologie ortho–mꢀta est ferromagnꢀtique,
confirmant ainsi les hypothꢁses antꢀrieures.[1] Suivant une mꢀ-
thodologie prꢀcꢀdemment dꢀcrite,[2] les propriꢀtꢀs magnꢀti-
ques des triradicaux ꢃ lꢂꢀtat cristallin ont pu Þtre analysꢀes
dans le dꢀtail et permettent de proposer des chemins dꢂinte-
raction magnꢀtique et des gꢀomꢀtries de contacts intermolꢀcu-
laires prꢀcises pour lꢂꢀtablissement dꢂinteractions magnꢀtiques
bien identifiꢀes.
Abstract in Spanish: El presente trabajo completa y extiende
nuestras anteriores estudios[1,2] sobre la determinacion del
estado fundamental magnꢀtico y la fuerza del intercambio
magnetico a travꢀs del enlace en una serie de biradicales.
Esta informaciꢄn es luego usada en el anꢅlisis de las interac-
ciones magneticas dentro de sus cristales. En este trabajo pre-
sentamos los estudios sobre una serie de triradicales que con-
tienen a-nitroxidos nitronꢆlicos (NN) o a-imino nitrꢄxidos
(IN) como fragmentos radicalarios terminales conectados a
travꢀs de unidades acopladoras del tipo m-phenilꢀnicas, en
un caso, y fenil acetilꢀnicas, en el otro. Se han estudiado tam-
biꢀn tetraradicales. Los estudios con moleculas aisladas
(EPR en disoluciꢄn y DFT) permiten evaluar las interaccio-
nes magneticas a travꢀs de las unidades acopladoras. Todos
los triradicales se ve que presentan un estado fundamental
del tipo cuadruplete, mientras que el cuadruplete se cree que
tienen un estado fundamental singlete. Este fflltimo resultado
refuerza las conclusiones de un estudio previo en el que se
viꢄ que el singlete es el estado fundamental en biradicales si-
milares conectados por una unidad m-fenilꢀnica. Ademꢅs, el
intercambio magnꢀtico a travꢀs del enlace para la conectivi-
dad orto y meta se demuestra que es ferromagnetica, confir-
mando hipꢄtesis anteriores.[1] Las propiedades magnꢀticas de
los triradicales y tetraradicales en su estado sꢄlido se han ra-
cionalizado empleando una matodologꢆa propuesta anterior-
mente,[2] identificando los caminos magnꢀticos relevantes.
Scheme 1. Target radicals and some intermediates.
Chem. Eur. J. 2005, 11, 2440 – 2454
www.chemeurj.org
ꢅ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2441

P. Turek et al.
that considering the shortest contacts in the structure and
applying McConnellꢁs first principle failed to predict the
predominant magnetism, and that comparing similar pat-
terns could ensure reliable magneto-structural correlations.[7]
The overall ideas for the synthesis and for the study of
the tetraradical derivatives was to investigate the possible
effects of the change in the spin configuration in the m-
phenylene based phenyl acetylene magnetic coupling unit
reported in Part I. This is achieved by substitution of the IN
terminal radical fragment studied in Part I with the triplet
ground state m-phenylene based biradicals studied in
Part II. We here report the synthesis of such oligoradicals,
the assessment of their magnetic ground state by EPR study
in dilute solution and by DFT calculations, as well as the
magneto-structural correlations in the solid state.
(selectively through route A) and 54% of diINpIN from re-
duction of diNNpNN by NaNO2 in acidic media. The pre-
cursor of the biphenyl derivative diNNbpNN was obtained
in high yield by using route C by Suzuki coupling[18] of the
5-bromoisophtalaldehyde to the p-formyl boronic acid in the
presence of catalytic amount of tetrakistriphenyl phosphine
palladium catalyst, followed by triscondensation with bishy-
droxylamine. Oxidation of the compound by sodium period-
ate in dichloromethane/water mixture afforded the desired
triradical. Attempts to do the cross-coupling of the NN de-
rivatives instead of the formylated ones did not succeed,
since deboronation of the boronic acid occurred, as reported
in other cases.[19] To obtain the tetraradicals, coupling 1,3-
bis(1,3-hydroxy-4,4,5,5-tetramethylimidazolin-2-yl)-5-ethy-
nylbenzene (Part II) to either dibromobenzene or diioben-
zene in the presence of diisopropylamine and tetrakistri-
phenyl phosphine palladium was performed. The compound
was precipitated and directly oxidised to give the tetraradi-
cal tetraNN with an overall yield of 9% when dibromoben-
zene was used and 17% with diiodobenzene. When the pre-
cipitated compound was dehydrated with 15% SeO2 prior to
MnO2 oxidation,[20] tetraIN could be isolated in a 31%
yield. The other pathway, using 1,3-diethynylbenzene and
the brominated biradical precursor did not lead to the target
tetraradicals, since degradation by homocoupling or poly-
merization side reactions of the 1,3-diethynylbenzene occur-
red. All compounds were characterized by elemental analy-
sis and FAB spectrometry.
Results and Discussion
Synthesis: The target compounds are presented on
Scheme 1. The palladium-based synthesis of the phenyl acet-
ylene based oligoradicals was complicated by side reactions
such as dehydration from NN precursors to IN ones as well
as homocoupling. Various synthetic routes (illustrated in
Scheme 2 for a phenyl acetylene triradical) were explored
for the triradical series to prevent such side reactions and
have been shortly reported elsewhere.[14] Coupling the imi-
dazolidine precursors (NNH) through route A ended up
with side products, even if direct introduction of
[PdCl2(PPh3)2] instead of in situ generation reduced the de-
hydration processes. However, this procedure leads to the
isolation of the hetero-triradicals INNNpNN and diINpNN.
The route consisting in coupling the paramagnetic building
blocks (B), which was nicely used in some related pyridine
compounds,[15] did not give any of the target compounds
even when performed with other amines than triethylamine
to fasten the reaction[16] (diisopropylamine, pyrrolidine), due
to the fast decomposition of the radical fragments. Coupling
the formyl precursors (C) followed by condensation of the
bishydroxylamine with the formylated derivatives and final
oxidation[17] was the best way to obtain compounds bearing
NN fragments. Hence, the four phenyl acetylene based tri-
radicals were obtained with the following yields: 49% for
diNNpNN (overall yield through route C), 15% of
INNNpNN (through via A), 22% for diINpNN triradical
Intramolecular magnetic exchange coupling: The EPR spec-
tra of the oligoradicals were recorded in dilute (ca. 10ꢀ3–
10ꢀ4 mol) dichloromethane/acetone (1:1) solutions.
Triradicals: The NN substituted triradicals diNNpNN
(Figure 1) and diNNbpNN showed the expected thirteen
line spectrum corresponding to the strong exchange limit,
that is, the exchange interaction between radical fragments J
is much larger than the hyperfine coupling constant
(hfcc).[21] The EPR spectra in fluid solution are well repro-
duced[22] with the parameters reported in Table 1. The spec-
trum of diINpNN could not be properly simulated. An
asymmetry is observed, which is attributed to the superposi-
tion of conformations in strong and weak exchange, result-
Figure 1. Room temperature EPR spectrum of triradical diNNbpNN in
solution (CH2Cl2/acetone). The simulated spectrum matches perfectly the
experimental spectrum.
Scheme 2. Synthetic routes A, B, C for obtaining diNNpNN.
2442
ꢅ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 2440 – 2454

Magnetic Interactions
FULL PAPER
Table 1. EPR parameters (14N hfcc values for non-equivalent nuclei, g
factor, line width, Gaussian/Lorentzian line shape ratio) used to simulate
the EPR spectra of the various triradical derivatives in solution at room
temperature.
Compound aN1 [G] aN2 [G] aN3 [G]
g
DB/G Gauss/Lorentz
diNNpNN
diNNbpNN 2.50
INNNpNN 2.50
diINpIN
diINbpIN
2.50
–
–
3.10
1.45
1.51
–
–
1.42
–
–
2.0067 1.8
2.0066 1.4
2.0063 1.1
2.0059 1.4
2.0060 1.4
100
100
100
50
3.05
3.02
80
Figure 2. EPR spectrum of triradical diINpNN recorded in frozen so-
lution (CH2Cl2/acetone) at T=12 K. In the inset is shown the half-field
line (DMS =ꢁ2) recorded with different spectrometer settings.
ing in the triradical signal (strong exchange) superimposed
with the biradical and monoradical species (weak exchange).
These different species have slightly different g values, thus
explaining the asymmetry of the spectrum. The superposi-
tion of contributions from monoradical and biradical is also
required to improve the simulation of the EPR spectrum of
diINpIN. Notably the overall appearance of the spectrum of
for example diINpNN strongly depends on the used solvent,
that is, aromaticity, polarity, bulkiness. This stresses the ef-
fects of molecular conformation on the exchange coupling,
as recently evidenced in trimethylenemethane-based diradi-
cals.[23] Actually, if the rotation barrier of the radical frag-
ments with respect to the magnetic coupler is low enough in
energy, the molecular conformation strongly depends on the
molecular environment. In turn, the exchange coupling de-
pends on the molecular conformation. Therefore, dynamical
effects are expected in fluid solution, and the determination
of the exchange coupling in a frozen solution should better
be considered as a conformational average. It turns out that
in Part II, we could estimate such average values in frozen
solution as being close to the corresponding values in the
crystalline state, that is, for one dominant conformation.
This is also verified in the present work.
In frozen solutions, the characteristic spectrum of a quar-
tet state exhibits five transitions, with a distance of 2D’Q4
(zero field splitting, ZFS parameter) between the two inter-
nal lines and 4D’Q4 between the two external ones.[24] The
value of the D’Q4 can be related to the triplet ZFS parame-
ter D’T by the relation D’Q4 =D’T/3, provided that the dipo-
lar interaction between the biradical part and the monoradi-
cal can be neglected. Since the distance between those is
quite long, this relation can be used for the present triradi-
cals. We could thus compare the conformation of the biradi-
cal part to the biradicals previously studied in frozen solu-
tions (Part II), so as to know whether we could expect the
same magnetic exchange interaction through the m-phenyl-
ene fragment of these triradicals.
Table 2. Experimental ZFS splittings (D’Q4) estimated in Gauss from the
quartet lines of the EPR spectra of the triradical derivatives in frozen so-
lution. The second column refers to the relation: D’Q4 =D’T/3, and the
last column (D*T) is given for the sake of comparison with the previous
results obtained for the isolated m-phenylene based biradical fragments
(Part II).[2]
D’Q4 [G]
D’T [G]
D*T [G][3]
diNNpNN
INNNpNN
diINpNN
diINpIN
diINbpIN
diNNbpNN
32ꢁ4
18ꢁ1
26ꢁ2
26ꢁ2
23ꢁ 2
39ꢁ4
96ꢁ12
54ꢁ3
78ꢁ4
78ꢁ4
69ꢁ6
117ꢁ12
100–115
67
90
90
90
100–115
diNNpNN and diNNbpNN is quite close to the related iso-
lated biradicals; the D’T value has been taken from Part II.[2]
The other derivatives exhibit lower D’T values as compared
with the related isolated biradicals. This modification of the
torsion of the radicals with respect to one another should
thus affect the magnetic interactions through the m-phenyl-
ene magnetic coupling unit.
In the present work, the intramolecular exchange interac-
tion has been assessed within some triradicals from the care-
ful recording of the temperature dependence of the EPR
susceptibility, that is, the doubly integrated EPR signal
[1,2,8e,25]
cEPR
.
The usual Curie plot (cEPR vs 1/T) is linear for
all compounds. It is well known[26] that the observation of
the DMS =ꢁ2 peak does not allow to conclude definitely to
the quartet ground state against a triplet spin species. The
forthcoming assessment of the quartet ground state is sup-
ported by i) the fit of the temperature dependence of the
EPR susceptibility, which relevance is strengthened by the
previous knowledge of the exchange coupling of the m-
phenylene based biradical moiety (Part II), and ii) the re-
sults of computations performed at the DFT level for the
molecular conformation found in the crystalline state. A
consistency check is further given by the present studies of
the magnetic properties of these triradicals in the solid state.
In most cases, the temperature dependence of the peak-to-
peak amplitude, App, of the half-field signal is representative
of the evolution of cEPR versus temperature, since no signal
distortion is observed throughout the studied temperature
range. A similar behaviour has been observed in the study
The frozen solutions were studied in a dichloromethane/
acetone 1:1 mixture at a concentration of 10ꢀ3 m. The central
line present on all spectra is due to the population of the
doublet (Figure 2). A half-field “forbidden” DMS =ꢁ2 tran-
sition is observed for all compounds. However, no DMS =
ꢁ3 transition could be observed. The experimental ZFS pa-
rameters are reported in Table 2 in magnetic field units
(D’,E’=0). As shown in Table 2, the conformation of the m-
phenylene biradical fragment (diNN moiety) within
Chem. Eur. J. 2005, 11, 2440 – 2454
www.chemeurj.org
ꢅ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2443

P. Turek et al.
ꢀ
ꢁ
10 þ expðꢀD1=kTÞ þ expðꢀD2=kTÞ
2 þ expðꢀD1=kTÞ þ expðꢀD2=kTÞ
of the integrated EPR susceptibility for both the DMS =ꢁ2
and the DMS =ꢁ1 absorption lines, as shown on Figure 3.
The cEPRT curves were normalised at high temperature to
1.125 emuKmolꢀ1, which is the value expected for three non
cT ¼ 0:375
ð1Þ
with D1 ¼ 3J1=2 and D2 ¼ ðJ1 þ 2J2Þ=2:
interacting spins S=1= .
The homo-triradical diINbpIN could not be studied be-
cause of the weakness of the half-field transition. The results
of the fits to Equation (1) for the EPR data of all com-
pounds are summarized in Table 3. First of all, the quartet is
found being the ground state for all studied triradicals.
Hence, a ferromagnetic exchange interaction is effective
through the m-phenylene unit and the phenyl acetylene or
biphenyl unit between the present IN/NN radical fragments.
It has to be pointed out that the magnitude found within the
biradical part is in full agreement with those found in the
biradical series (Part II), thus showing that EPR is a power-
ful experimental tool as long as careful studies are per-
formed on dedicated series of compounds. Within the
phenyl acetylene derivatives, the nature of the radical (NN
or IN) on the biradical fragment has a direct influence on
the magnitude of the exchange, as previously demonstrated
in the biradical series. It is worth noting that a similar rule is
found for triradicals, that is, a NN fragment connected para
to the biradical part leads to a better delocalisation of the
spin density on the phenyl acetylene coupler (cf. J1/k) com-
pared with an IN fragment. Moreover, the idea that con-
necting a triplet entity may modify the magnetic interaction
through the phenyl acetylene coupler is demonstrated here,
since the magnetic exchange is greater through the phenyl
acetylene for the diNNpNN triradical than for the reported
biradical NNpNN (Part I).[1] The modification is quite small,
but still corroborates some results found on more robust
triplet entities. Moreover, the ferromagnetic coupling is defi-
nitely proven experimentally for NN or IN radical fragments
substituted in the 1,3- or 1,5-positions, that is, in the para–
meta position, on a phenylene–ethynylene magnetic cou-
pling unit. This could not be assessed experimentally in
Part I.[1]
2
*
Figure 3. Temperature dependence of the cT product of the DMS =1 (
)
!
and DMS =2 ( ) EPR lines together with the product of the peak-to-
peak amplitude by temperature of the DMS =2 (&) for the triradical
diNNpNN. The full line is the fit of the data to Equation (1).
The data are fitted to the model depicted in Scheme 3.
Three interacting S=1= spins are distributed at the apexes
2
of an isosceles triangle (Scheme 4) with two different inter-
actions J1 and J2 assuming that the g factor of each spin is
equal to the free electron value, this model yields the fol-
lowing temperature dependence of the cT product
[Eq. (1)]:[27]
To have an independent indication that the quadruplet is
the ground state in all these triradicals we carried out
B3LYP/3-21G(d) calculations at their crystal geometry de-
picted in the section devoted to the solid-state properties.
Some of the methyl groups are disordered in these crystals.
In this case, we took for our calculations the option in which
the methyl groups are closer to an alternated arrangement,
with the C–C distance equal to the average one for the
other methyl groups and the C–H distance equal to 1.1 ꢆ.
At this geometry, we computed the energy for the quadru-
plet and doublet states, imposing tight accuracy in the com-
putation of the integrals and in the convergence of the self-
consistent process. The results for all radicals show that the
quadruplet is the ground state in all cases. The experimental
and the calculated quadruplet-doublet splitting (D2) are
given in Table 4. The given results are for the doublet spin
being localized in the radical fragment being connected to
the singly substituted phenyl ring. Although two more dou-
blet spin configurations may be considered, this spin distri-
bution yields the lowest energy. A good agreement is found
Scheme 3. Isosceles triangle model with the corresponding Hamiltonian
used to model the magnetic behaviour of the studied triradicals.
Scheme 4. Symmetrical four-spins system modelling the magnetic interac-
tions within the studied tetraradicals.
2444
ꢅ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 2440 – 2454

Magnetic Interactions
FULL PAPER
Table 3. Values of the exchange coupling (expressed in K) through the m-phenylene magnetic coupling unit
(J2/k), and through the phenyl acetylene or the biphenyl magnetic coupling unit (J1/k) deduced from the fits of
the EPR data to Equation (1). The last row gives previous data (Part II) obtained for the isolated m-phenylene
based biradicals, to be compared with J2.
presents sign alternation in all
centers, being the largest
atomic spin population (com-
puted using the Mulliken analy-
sis) 0.06 atomic units (au). The
Exchange coupling
diNNpNN
diNNbpNN
INNNpNN
diINpIN
diINpNN
J1/k [K]
7ꢁ2
7ꢁ2
+7ꢁ2
15ꢁ1
13–25
0.7ꢁ0.1
6.5ꢁ1
5–15
1.0ꢁ0.5
+6.5ꢁ4
5–15
ꢀ
central C C bond connecting
J2/k [K]
23ꢁ4
20–40
10ꢁ7
20–40
the two six-membered rings in
Jbiradical/k [K][2]
the XpX’ radicals preserve the
Table 4. Experimental and calculated quadruplet–doublet splitting (in K)
for the studied triradicals.
Q–D split-
diNNpNN diNNbpNN diINpIN diINpNN INNNpNN
ting; D2/K
experiment
theory
27
67
14
139
7
7
8
19
40
16[a]
[a] the unknown geometry has been chosen similar to the geometry of
diINpIN, as suggested by the EPR results in frozen solution.
Figure 4. Spin density computed for some of the indicated triradicals at
the UB3LYP/3-21G(d) level. The surfaces plotted are those of 0.002
atomic units of electronic density (positive in white, negative in black).
The maximum of spin density is, in all cases, located in the NCNO or
ONCNO five-membered ring group of atoms.
between theory and experiment when the m-phenylene bi-
radical moiety of the triradicals is diIN, as previously ob-
served for the singlet-triplet splitting of the isolated biradi-
cal (Part II). When dealing with diNN as the m-phenylene
biradical moiety of the triradicals, the theoretical values are
quite overestimated as already found in the corresponding
isolated biradical (Part II). It has to be noticed the largest
quadruplet–doublet found theoretically for the biphenyl
magneto-coupler in diNNbpNN, whereas the experimental
results point to a similar efficiency than the phenylene–ethy-
nylene magnetic coupling unit. Indeed, the synthesis and the
study of diNNbpNN were motivated by the idea that the bi-
phenyl magnetic coupling unit may be more efficient than
the phenylene–ethynylene one. However, neither the
“thought experiment” nor the theoretical predictions could
lead to the proper conclusion. Both the diNNpNN and
diNNbpNN radicals present values well above the other tri-
radicals (46.7 and 96.7 cmꢀ1 vs. 5.7, 27.6 and 11.17 cmꢀ1). It
is possible that the density functional methodology comput-
ed doublet–quadruplet energy differences are not complete-
ly reproduced, though previous estimates with other com-
pounds indicate that these differences are within 10% of the
experimental result. We expect that in this family of com-
pound, whose structure is quite similar, the errors to be also
constant. Thus, even if some error could exist, the trends be-
tween compounds are expected to remain. Therefore, these
results suggest new molecular designs to induce stronger fer-
romagnetic interactions (the use of the diNNpNN and
diNNbpNN radicals). We have not analyzed yet the reasons
behind their stronger ferromagnetic interactions, but a first
look indicates that are complex.
sign alternation (one has a spin population of 0.03 au and
the other ꢀ0.03 au). It is also worth noting that the spin al-
ternation is preserved in these radicals despite the lack of
planarity between the two six-membered rings, and also be-
tween the five- and six-membered rings.
The influence of the conformation of the coupling unit
may be considered by comparing the through magnetic cou-
pling unit intramolecular interactions in the diNNpNN and
the diNNbpNN triradicals. The phenyl rings in the phenyl–
acetylene magnetic coupling unit in diNNpNN are twisted
by 138, whereas the biphenyl magnetic coupling unit in
diNNbpNN is quite distorted with a dihedral angle of 368
between the phenyl rings. However, the exchange coupling
J2 is not much weakened through the biphenyl magnetic
coupling unit compared with the phenyl acetylene one. De-
spite the longest through bond exchange pathway, the rigid
and conjugated phenyl–acetylene bridge could be expected
as being more efficient than a shorter but distorted one as is
the biphenyl unit. Related to Part I of our work, this again
points to the role of the sp hybridization through the acety-
lenic bond as a softener of the magnetic transmission.
Tetraradicals: In dilute solution, the EPR spectrum of tet-
raNN is expected to give 17 lines due to the hyperfine inter-
action of the unpaired electron with eight equivalent 14N,
whereas tetraIN should give 25 lines. However, the through-
space dipolar interactions between the biradical fragments
broaden the individual components of the hyperfine struc-
ture to result in a large envelope. Within such tetraradicals,
we assume that the magnetic interactions are symmetrical
with respect to the central phenyl ring (Scheme 4). Having
noticed that the triplet species are not robust enough to con-
sider that the triplet state is fully populated over the studied
According to previous studies on nitronyl nitroxides and
imino nitroxides, the quadruplet presents an electronic
structure where the spin is located essentially over the
ONCNO and ONCN atoms of the five-membered ring,
being negative in the central C(sp2) atoms and negative over
the remaining atoms (see Figure 4). The six-membered ring
Chem. Eur. J. 2005, 11, 2440 – 2454
www.chemeurj.org
ꢅ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2445

P. Turek et al.
temperature range, the following Hamiltonian yields the ei-
genstates of the four-spins system [Eq. (2)] depicted in
Scheme 4:[27]
the 4–100 K range, but the DMS =ꢁ1 signal saturates over
the whole range of temperature at all incident microwave
power. Since the line shape does not change with tempera-
ture within this temperature range, the thermal behaviour of
the EPR spin population involved in the half-field line is
plotted as AppT versus T on Figure 6. The decrease of AppT
^
^
^
^
^
H ¼ ꢀJ1ðS1 ꢂ S2 þ S3 ꢂ S4Þ
ð2Þ
^
^
^
^
^
^
^
^
ꢀ J2ðS1 ꢂ S4 þ S1 ꢂ S3 þ S2 ꢂ S3 þ S2 ꢂ S4Þ
The spin multiplicities are given as one quintet, Q5, three
triplets, Ti, and two singlets, Si. In frozen solution and for
weakly coupled biradical fragments as it is the case here, the
ZFS parameter for the quintet state, D’Q5, can be related to
each radical by the relation D’Q5 =(D’T1 +D’T2)/6.[8a] Both
biradicals are equivalent, so that D’Q5 =D’T/3 allows us to
compare the conformations of the biradical fragments in the
tetraradicals to the isolated biradicals as previously done for
the triradical derivatives. The EPR spectra due to the
DMS =ꢁ1 transition in frozen solutions can be due to the
thermal population of the quintet and the triplets. The exter-
nal lines from these states coincide so that the quintet state
can be followed by its more internal lines. For the tetraradi-
cal tetraNN, the broadening prevents assignation of the
lines, whereas some shoulders appear more distinctly in the
tetraIN spectrum (Figure 5). The internal lines are separated
Figure 6. Temperature dependence of the product of the peak-to-peak
amplitude of the EPR half-field line by the temperature for the tetraradi-
&
cals tetraIN ( ) and tetraNN (&) in frozen solution.
with the temperature means the depopulation of a high spin
state to a lower spin state. The fitting by a four spin model
is not relevant given the required large number of parame-
ters for such a smooth variation. Numerous couples of (J1,
J2) values yield a satisfactory correlation for the least
squares fitting procedure. With respect to the previous re-
sults, modelling with a biradical interaction of +25 K and
an antiferromagnetic interaction of ꢀ5 K through the phenyl
acetylene coupler could reproduce the behaviour of tetraNN.
It can be qualitatively deduced from this study that the
singlet is the ground state for both tetraradicals. This will be
further discussed when considering the magnetic properties
in the solid state. It is worth reminding that the singlet
ground state has been found in related m-phenylene based
phenyl acetylene biradicals (Part I). The present results con-
firm that the phenyl–acetylene magnetic coupling unit with
this connectivity to IN or NN radicals does not obey the
topological rules. It has to be pointed out that for stronger
triplet species such as nitrenes, the same coupler gave ferro-
magnetic coupling.[29]
Figure 5. EPR spectrum of tetraradical tetraIN recorded at 4 K in frozen
solution (CH2Cl2/acetone). In inset is shown the half-field line (DMS =
ꢁ2) recorded with different spectrometer settings.
by 180 G which gives D’Q5 =30ꢁ1 G and D’T =90ꢁ3 G.
This value is found to be the same as for the corresponding
m-phenylene based biradicals (see Table 2). Therefore, the
strength of the exchange coupling is expected to be between
5 and 15 K as in the latter compounds. A peak correspond-
ing to the DMS =ꢁ2 transition is detected for both tetraradi-
cals, whereas the DMS =ꢁ3 and DMS =ꢁ4 lines could not
be detected. As for the triradical derivatives, the knowledge
of the S–T splitting of the biradical fragments together with
the studies in the solid state will allow defining properly the
magnetic ground state. However, two outer extra features
with a lower intensity are observed on the side of the half-
field line (Figure 5). This specific pattern has been observed
for other species and can be considered to be characteristic
of Sꢃ2.[28] The presence of this signal on the whole range of
temperature ascertains the population of the quintet. The
DMS =ꢁ1 and DMS =ꢁ2 transitions have been recorded in
Solid-state properties: A convenient and widely used way to
relate the magnetic properties to the molecular packing is to
consider the signs of the spin densities involved in the short-
est intermolecular contacts. The sign of the intermolecular
exchange coupling is then attributed according to McCon-
nellꢁs first proposal.[30] The same sign of the spin densities at
contact yields an antiferromagnetic coupling, whereas oppo-
site signs would result in a ferromagnetic coupling. Howev-
er, careful analysis have pointed out that the successful ap-
plications result from errors that compensate.[7] Neverthe-
less, we have reported in Part II that the study of a series of
isostructural biradicals leads to a good analysis of the mag-
2446
ꢅ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 2440 – 2454

Magnetic Interactions
FULL PAPER
Table 5. X-ray crystal structure and lattice parameters of the phenyl acet-
ylene based triradical derivatives.
neto-structural properties. Indeed, the substitution of the IN
radical by the NN radical slightly modifies a few contacts
upon suppressing one oxygen atom. A way to simplify the
analysis of a molecular packing within a crystal structure
consists in determining the primary, secondary and tertiary
patterns related to the strength of the hydrogen bonds in
the structure. At intermolecular distances shorter than 5 ꢆ,
the contacts are considered between the molecular frag-
ments known to bear the spin density, for example, between
NO and any other atom. These are reviewed within each
structure, considering the geometry of the contacts by using
the QUEST program of the Cambridge Data Base for crys-
tallographic data.[31] In the present study, such an analysis
suggests a pathway for the intermolecular magnetic ex-
change. It results in a model for the description of the exper-
imentally determined magnetic properties.
diINpNN
INNNpNN
diNNpNN
diNNbpNN
formula
C35H43N6O4· C35H43N6O5· C35H43N6O6· C33H43N6O6
CH2Cl2
696.70
crystal system monoclinic
CH2Cl2
712.70
monoclinic
P21/n
12.297(1)
18.283(2)
16.796(2)
93.88(2)
3767(1)
4
CH2Cl2
728.70
monoclinic
P21/n
12.2541(7)
18.293(2)
16.892(2)
93.877(7)
3777.9(1)
4
Fw
619.75
triclinic
¯
space group
a [ꢆ]
b [ꢆ]
c [ꢆ]
P21/n
P1
12.414(2)
18.2021(9)
16.643(1)
93.712(8)
3752.8(9)
4
9.7410(2)
12.3790(5)
15.3730(7)
104.567(1)
2635.1(3)
2
b [8]
V [ꢆ3]
Z
1 [gcmꢀ3
]
1.23
1.26
1.28
1.27
data collected 5505
5794
6945
9119
data used for
refinement
I>3
2935
2370
2451
4701
s(I)
R(I)
Rw(I)
0.084
0.094
0.058
0.083
0.061
0.083
0.046
0.066
Phenyl–acetylene based triradical: The compounds
diNNpNN, INNNpNN and diINpNN crystallised in the
same space group P21/n with very similar cell parameters
(Table 5), as briefly reported in a previous communica-
tion.[14] A dichloromethane molecule is included in the unit
cell, and the crystals remain solvated at room temperature.
Compound diINpIN crystallises in the same space group but
with different cell parameters, since there is no solvent mol-
ecule. In the case of INNNpNN, disorder is present on the
internal oxygen atoms O4 and O6 (Figure 7). The three ter-
minal radical fragments, that is, the five-membered rings,
are numbered as cycle 1, 2, 3. The oxygen atoms O3 and O5
(external positions) have a probability of presence of 90%,
which is coherent with the external position solely observed
in diINpNN. The internal oxygen atoms have a probability
of presence of 60%. Hence, cycle 1 has an equal probability
to be either a NN or IN fragment. Moreover, the IN cycle
has only 10% probability to have the oxygen in an internal
position. The dihedral angle between the phenyl rings and
the planes of the ONCNO frag-
group 2, which involves short contacts between NO frag-
ments terminated by O1 and O5 and belonging to neighbour-
ing molecules. As there is no occupied O4 position in
diINpNN, only contacts of group 2 are found for this com-
pound. For diNNpNN and INNNpNN, the shortest contact
in group 1 is found at 3.16 ꢆ between the central carbon C8
from cycle 3 (monoradical fragment) and the oxygen O4 of
cycle 1. For INNNpNN, this short contact is present in only
55% of the cases, since O4 site is not always occupied. No
contact at a distance shorter than 4.5 ꢆ is observed between
the described dimers of triradicals, so that they can be con-
sidered as almost isolated with respect to magnetic path-
ways. The interatomic distances measured for the various
contacts within different groups are reported in Table 7. The
magnetic behaviour of the four synthesized derivatives is re-
ported in Figure 8. As expected for such triradicals, the cT
ments is reported for the differ-
ent compounds in Table 6.
In the following, we will
focus on the series of isostruc-
tural compounds, hence discard-
ing the diINpIN compound.
Weak hydrogen bonds (C(sp3)-
H···O) and p–p stacking at
3.6 ꢆ between the aromatic
cycles generates head-to-tail
dimers of triradicals related by
an inversion centre. Two types
of short contacts between the
ONCNO fragments are found
within these dimers (Table 7).
The proximity of oxygen O4
(cycle 1 of molecule 1) and
cycle 3 of molecule 2 generates
contacts gathered in group 1.
Figure 7. The four possible representations of the INNNpNN molecule with respect to the disorder: A, B, C,
Three contacts are gathered in D give the corresponding probability as deduced from the XRD analysis.
Chem. Eur. J. 2005, 11, 2440 – 2454
www.chemeurj.org
ꢅ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2447

P. Turek et al.
Table 6. Dihedral angles formed by the radical cycles (numbering
scheme defined in Figure 7) and the nearest phenyl ring within the four
different triradical derivatives.
nates the magnetic behaviour of these compounds at lower
temperature.
Within the series of isostructural derivatives, the observed
distinct behaviours may have two origins. The first modifica-
tion of the magnetic exchange could be explained by the dif-
ferent contacts within the dimer of triradicals. Indeed, the
presence of the ferromagnetic component in INNNpNN and
diNNpNN can be related to group 2 of contacts, which is
not involved in the structures of diINpNN. It has to be no-
ticed that the short contact between a central carbon (C8)
bearing negative spin density and a nitroxide bearing posi-
tive spin density should result in a ferromagnetic interaction
according to McConnell rules. However, in order to assign a
given magnetic pathway, the detailed geometry of the con-
tacts has to be considered and it should be compared with
other similar compounds as previously discussed[7] and as
demonstrated in Part II. To achieve this, we have chosen the
biradical diNNtmsa (Scheme 5); its structure and magnetic
Compound
Cycle 1[8]
Cycle 2[8]
Cycle 3[8]
diNNpNN
INNNpNN
diINpNN
diINpIN
46
43
37
21
26
25
22
3
18
17
15
17
Table 7. Summary of the shortest interatomic distances (intermolecular
contacts, in ꢆ) found between the ONCNO fragments within dimers of
triradicals. Bold figures point to the shortest contacts.
Contacts
Group
diNNpNN
INNNpNN
diINpNN
O5···O1
O5···N2
O5···N1
N1···O4
N2···O4
O4···O1
O4···C8
2
2
2
1
1
1
1
3.68
3.47
3.80
3.30
3.46
3.77
3.16
3.69
3.53
3.83
3.29
3.42
3.68
3.16
3.84
3.53
3.69
–
–
–
–
Scheme 5. a) Molecular scheme of the diNNtmsa biradical derivative.
b) Definition of the geometrical parameters (distances, angles and tor-
sions) for the description of the intermolecular contacts within dimers of
diNNpNN and diNNbpNN compounds.
Figure 8. Temperature dependence of the static susceptibility (SQUID
magnetometer measurements; applied magnetic field=0.5 T) represented
*
*
^
as the cT product for triradical derivatives: diINpNN; diNNpNN;
~
INNNpNN ; diINpIN. For the sake of clarity only 33% of the experi-
mental data points have been plotted. Full lines represent the fits to the
six-spins model depicted in Scheme 6.
properties have already been reported in Part II.[2] This
compound exhibits a similar pattern, which was found to
generate unambiguously a ferromagnetic intermolecular in-
teraction. The geometries are reported in Table 8 according
to the definitions given in Scheme 5b. It appears that the ge-
ometries at these contacts are very similar, so that we can
infer the existence of a ferromagnetic intermolecular path-
way through these contacts observed within the dimers of
triradicals in compounds diNNpNN and INNNpNN. More-
over, its magnitude could be close to 10 K as found in
diNNtmsa. In a first approach, this may explain why the fer-
romagnetic trend is weaker in INNNpNN than in
diNNpNN, since these contacts are altered by disorder in
the former.
product is close to 1.125 emuKmolꢀ1 at room temperature.
For diINpIN and diINpNN, the cT product remains un-
changed as the temperature is lowered down to 10 K, where-
as for diNNpNN and INNNpNN, a slight increase is ob-
served to reach
a maximum at 14.5 K of 1.39 and
1.28 emuKmolꢀ1 respectively. This behaviour points out the
presence of a ferromagnetic interaction within these molecu-
lar crystals. However, the further decrease at lower tempera-
ture indicates for all compounds that a neat antiferromag-
netic intermolecular interaction dominates the magnetic
properties in the low temperature regime.
The overall magnetic behaviour of diINpIN and diINpNN
underlines weak magnetic interactions, as revealed by the
flat Curie-like behaviour above 10 K. This is well under-
stood when considering the intramolecular ferromagnetic
couplings J1 and J2 for the isolated triradicals. A comparable
or lower intermolecular antiferromagnetic interaction domi-
Not only the possible intermolecular magnetic interac-
tions but also the intramolecular ones are different for the
three compounds, as it has been demonstrated from the
EPR studies of the triradicals in frozen solution (Table 3). It
has been considered that the dihedral angle of the biradical
fragment with respect to the phenylene ring is close to 308.
2448
ꢅ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 2440 – 2454

Magnetic Interactions
FULL PAPER
Dihedral angles of 42 and 218 are found within the biradical
fragment of diNNpNN. Therefore, a similar value of J1 is ex-
pected for diNNpNN in frozen solution and in the solid
state. The moderate torsion of cycle 3 (178) should as well
give a comparable value of J2 in frozen solution and in the
solid state. The simplest way to consider both intramolecular
and intermolecular effects within the dimers of triradicals is
given in Scheme 6 with a six-spins cluster with the parame-
(45/55) of the experimental susceptibility data of diINpNN
and diINpNN with the data of INNNpNN. As concluded in
the analysis of the crystal structure, the group 1 contacts re-
sults in a ferromagnetic interaction depicted by J3 in
diNNpNN and in INNNpNN. Its magnitude is in agreement
with the value found for similar contacts in diNNtmsa
(~10 K). As expected, J4 is antiferromagnetic for all com-
pounds. However, a value of 17 K is probably too high to be
solely ascribed to the small difference of geometry between
diINpNN and diNNpNN. This approach is certainly still
oversimplified, but is interestingly consistent with both pre-
vious magneto-structural correlations and the intramolecular
exchange coupling found through studies in the isolated
state.
Triradicals based on the biphenyl coupling unit: Single crys-
tals were obtained only for diNNbpNN and the crystal struc-
ture could be refined, leading to the following parameters:
¯
triclinic space group P1; with lattice parameters: a=9.741,
b=12.379, c=15.373 ꢆ, a=104.567, b=98.477, g=110.6908,
Z=2. No disorder is observed for the asymmetric unit (see
Supporting Information), and the torsion angles between
the imidazolidine and the phenyl rings range between 10
and 228. Short contacts C(sp3)-H···O and C(sp3)···N lead to a
primary structure, which is made of dimers (molecules A
and A’ in Figures S16 and S17) further connected to build
up chains. Within the dimers, some short contacts are pres-
ent between ONCNO fragments, due to the proximity of
N3O3 groups: two contacts O3···O5 at 3.75 ꢆ and O3···O6 at
4.02 ꢆ, two contacts at 3.30 ꢆ (O3···N5) and 3.50 ꢆ (O3···N6)
and the shortest contact O3···Ca at 3.03 ꢆ. Along the a axis,
the proximity of N4O4 groups of two molecules belonging to
two different dimers leads to a short contact of O4···O4 at
3.85 ꢆ between the dimers: this extends the primary struc-
ture into the chains (see Supporting Information). Further
short contacts are not observed between these chains. As for
the diNNpNN derivative, a comparison of the geometry of
the dimer within the two derivatives diNNtmsa and
diNNbpNN shows that these molecular packings have simi-
lar geometries at contacts except for the O···O contact
(Table 8).
Scheme 6. The six-spins system and its model Hamiltonian used for the
simulation of the behaviour of the head-to-tail stacked dimers of triradi-
cals formed in the solid state of diINpNN and diNNpNN.
ters J1 and J2 as the intramolecular coupling constant, and
J3, J4 as the intermolecular coupling constant related to
group 1 and group 2, respectively, previously defined types
of contacts (see Table 7). Based on the previous results,
some conditions may be fixed so as to reduce the number of
parameters, hence improving the relevance of the fitting pa-
rameters. It is first noted that the previous measurements in
the isolated state yield estimations of J1 and J2. Therefore, J1
has been fixed between 0 and 40 K allowing the experimen-
tal data to be well adjusted for the diINpNN and diNNpNN
(continuous lines in Figure 8) with the following sets of
values in temperature units for diNNpNN (J1,J2)=(+30.5ꢁ
3.1 K,+12.3ꢁ0.6 K); (J3,J4)=(+15.1ꢁ0.9 K,ꢀ17.7ꢁ1.2 K);
and for diINpNN (J1,J2)=(+2.0ꢁ0.1 K,+0.0ꢁ0.02 K);
(J3,J4)=(0 K,ꢀ6.36ꢁ0.03 K). For these simulations, the g
factor has been set to the free electron value. The set of
values refined for the intramolecular exchange coupling is in
good agreement with that found in the isolated state. For
INNNpNN, a rough agreement is found upon setting (J1,J2)
to the values found in the isolated state, that is, (J1,J2)=
(+15 K,+7 K). This procedure yields (J3,J4)=(+12 K,ꢀ14 K),
thus ascertaining the observed behaviour of diNNpNN and
of diINpNN. It is worth noticing that the present set of
(J3,J4) for INNNpNN does represent an average value for
two sets, with and without J3, with respective weight 55 and
45% as deduced from the analysis of the molecular packing.
Along this line, and although being quite fortuitous, it may
be noted the almost perfect overlap of the weighted sum
The cT product reaches 1.11 and 1.03 emuKmolꢀ1 at
room temperature for diNNbpNN and diINbpIN, respective-
ly (Figure 9). For diINbpIN, antiferromagnetic interactions
dominate the overall behaviour down to 2 K. For
Table 8. Comparsion of the detailed geometries as depicted in Scheme 5
for two similar intermolecular contacts within diNNpNN, diNNtmsa[2]
and diNNbpNN. Ca reads for CA or CB in the ONCaNO ring of
Scheme 5.
diNNpNN
diNNtmsa[2]
diNNbpNN
Contact
d[ꢆ]
d[ꢆ]
d[ꢆ]
(A1,A2,T1,T2,T3)[8] (A1,A2,T1,T2,T3)[8] (A1,A2,T1,T2,T3)[8]
3.16 3.03 3.03
(123, 94, 84, 83, 85) (138, 98, 89, 85, 73) (125, 96, 83, 85, 75)
3.77 3.70 3.75
(140, 62, 82, 88, 62) (156, 62, 78, 77, 55) (100, 60, 41, 65, 57)
Ca···O
O···O
Chem. Eur. J. 2005, 11, 2440 – 2454
www.chemeurj.org
ꢅ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2449

P. Turek et al.
Figure 9. Temperature dependence of the static susceptibility (SQUID
magnetometer measurements; applied magnetic field=0.5 T) represented
Figure 10. Temperature dependence of the static susceptibility (SQUID
magnetometer measurements; applied magnetic field=0.5 T) represented
*
*
as the cT product for triradical derivatives: diNNbpNN; diINbpIN.
*
*
as the cT product for tetraradical derivatives: tetraNN; tetraIN. The
full line is the best fit to singlet-triplet equilibrium with a mean-field cor-
rection.
diNNbpNN, a maximum is observed by about 14 K with a
value of 1.37 emuKmolꢀ1. Such a ferromagnetic interaction
is attributed to the intramolecular magnetic exchange as ex-
pected from the results on the isolated triradicals. The de-
crease of cT at lower temperature indicates the competition
with intermolecular antiferromagnetic interactions. The tor-
sion angle between both rings of the biphenyl unit is quite
large in the crystalline state, and we assume a J2 coupling
lower than 7 K as assessed in frozen solution. From an inter-
molecular point of view, the presence of a short contact be-
tween the central carbon and a nitroxide could again be re-
sponsible for some ferromagnetic interactions. The ferro-
magnetic component can thus be attributed to both intramo-
lecular and intermolecular magnetic exchange. However,
the six-spins cluster previously considered for almost isolat-
ed dimers of triradicals within diNNpNN and diINpNN does
not yield a satisfactory fit of the data in the case of
diNNbpNN. The interdimer interactions have to be consid-
ered, since short NO–NO contacts are observed between
the biradical fragments of the triradicals as described
before. Therefore, a proper but quite complicated model
may be a chain of dimers of triradicals, which is not availa-
ble to us. Moreover, the expected close values of the various
acting interactions would strongly limit the relevance of
such a complicated model, whenever available.
crease is observed throughout the whole temperature range
for tetraIN, a plateau with a slight increase is observed by
about 20 K for tetraNN. Let us assume that such bulky mol-
ecules do pack without exhibiting short contacts so that the
tetraradicals are quite isolated in the solid state. Then, the
previously described four-spin model (Scheme 4) may be
considered for the data fitting. Performing such a fit for tet-
raNN shows that neglecting the intermolecular interactions
prevents a proper adjustment to the experimental data. The
results of the fit give complementary indication of the intra-
molecular exchange couplings giving J1 =+12ꢁ1 K, J2 =
ꢀ2.5ꢁ0.1 K. Similar results were obtained with the same
model for the temperature dependence of the EPR signal of
a small amount of polycrystalline powder of tetraNN lead-
ing to J1 =+14.1ꢁ0.3 K, J2 =ꢀ2.0ꢁ0.1 K. The best fit rep-
resented in Figure 10 is obtained with simple singlet–triplet
equilibrium, that is, J2 =0 K, with a mean-field correction
(Curie-Weiss temperature, q) where J1 =+14 K, and q=
ꢀ2 K, but with a quite irrelevant Curie constant, C=
2.1 emuKmolꢀ1. It turns out that, although being not opti-
mized, both of these approaches yield the proper range for
the interaction J1 within each biradical fragment, whereas
the exchange coupling J2 through the phenyl–acetylene mag-
netic coupling unit is suggested either vanishing or antiferro-
magnetic. However, the EPR studies in fluid solution have
shown that the tetraradical does not consist in isolated bi-
radical moieties. Moreover, it may be assumed a negative J2,
when referring to our previous report on J2 being of the
order ꢀ4 to ꢀ6 K (Part I).[1] Therefore, the present estima-
tion of an antiferromagnetic J2 corroborates the antiferro-
magnetic interaction through the meta-substituted phenyl–
acetylene bridge, hence yielding the singlet ground state.
This is unexpected, if one refers to the commonly accepted
topological rules.[10] Furthermore, substituting a biradical
fragment instead of a monoradical does not significantly
change the through-bond exchange coupling. Worst-case,
the qualitative trend is a weakening of J2. The singlet
ground state in these m-phenylene based di- and tetra-radi-
cal derivatives is not yet understood. Among possible ways,
we may consider the non-robustness of the triplet state here
The magnetic behaviour of diINbpIN is not further con-
sidered, due to the lack of knowledge of both, the set of in-
tramolecular exchange couplings (J1,J2) and the crystal struc-
ture determination. Nevertheless, there must be obviously a
strong intermolecular antiferromagnetic interaction.
Tetraradicals: None of the attempts to obtain single crystals
of the synthesized tetraradicals was successful. Therefore,
the previous strategy for the analysis of the magneto-struc-
tural correlations is not applicable. The measurements of
the magnetic properties performed in the solid state
(Figure 10) will be used as a complementary analysis of the
behaviour of the isolated molecule.
For both compounds, the value of cT is close to
1.5 emuKꢀ1 molꢀ1 as expected with four independent S=1=
2
spins. Antiferromagnetic correlations are observed for both
derivatives at low temperature. Whereas a continuous de-
2450
ꢅ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 2440 – 2454

Magnetic Interactions
FULL PAPER
over KOH. All the reagents were used as received and purchased from
ALDRICH. 4-bromoisophtaladehyde,[32] 5-ethynylbenzaldehyde, 1,3-bis-
formyl-5-ethynylbenzene[2] and 1,3-bis(1,3-hydroxy-4,4,5,5-tetramethyl-
imidazolin-2-yl)-5-ethynylbenzene[2] were synthesised as reported else-
where. 2,3-Bis(hydroxyamino)-2,3-dimethylbutane (bishydroxylamine)
was synthesised using the procedure described in the literature.[33] Thin-
layer chromatography (TLC) was performed on aluminium plates coated
with Merck silica gel 60 F254. Microanalyses were performed in the
common facilities of the Institute Charles Sadron. Fast atom bombard-
ment (FAB, positive mode) were recorded on a ZAG-HF-VG-Analytical
considered as compared to for example, nitrenes or car-
benes. Another possible origin may be the spin polarization
combined with the sp hybridization as mentioned when
comparing the biphenyl magnetic coupling unit with the
phenyl–acetylene.
Conclusion
1
apparatus in an m-nitrobenbyl alcohol (mNBA) matrix. H and 13C NMR
spectra were recorded on a Bruker ARX 300 spectrometer. The EPR
spectra have been recorded on X-band Bruker spectrometer (ESP-300-
E) equipped with a rectangular TE 102 cavity. The static field was mea-
sured with an NMR Gaussmeter (Bruker ER035) while the microwave
frequency was simultaneously recorded with a frequency counter (HP-
5350B). Solutions were degassed by bubbling Argon directly in the EPR
tube prior to measurements. The spin concentration has been estimated
for all powder compounds by comparison with a standard sample (Varian
pitch). It is reported hereafter as NS (spin per molecule). Temperature
was measured with a thermocouple (AuFe/Chromel) introduced inside
the tube, at 1.5 cm from the bottom. As described in Part I[1] and
Part II,[1] much care was devoted to the effects of microwave power satu-
ration effects. At each temperature, the signal has been recorded at vari-
ous power settings for the DMS =2 line and for the DMS =1 line. The var-
ious power levels were selected within different temperature range, de-
pending on the limits for the observation of saturation effects. Magnetic
The intramolecular exchange coupling between IN and/or
NN radical substituents within triradical and tetraradical de-
rivatives based on the m-phenylene moiety and linked by
either a phenyl–acetylene or a biphenyl unit, has been prop-
erly assessed in the isolated state with the help of the EPR
technique. The magnetic ground state of the triradicals has
also been determined by DFT calculations. The main con-
clusions are: i) the quartet is the magnetic ground state for
all triradicals and ii) the singlet state is very probably the
ground state for all tetraradicals. Although being not yet
fully understood, this violation of the so-called topological
rules supports our previous findings (Part I)[1] that the
phenyl–acetylene magnetic coupling unit in the meta-con-
nectivity does not yield the high spin ground state. As dis-
cussed in Part II for the related biradical derivatives, the
spin polarisation mechanism is presumably at the origin of
the difference of magnetic coupling between IN or NN radi-
cal substituents. This tends to increase the spin density dis-
tribution over the coupling unit, hence increasing the
through-bond magnetic exchange as exemplified for the
ortho–meta connectivity (J~0 K in NNpNN compared with
J=7 K in diNNpNN). The synthesis of isostructural com-
pounds for series of triradicals afforded the possibility of a
careful study based on a previously developed strategy
(Part II). According to theoretical studies and statistical
analyses, the relation between structure and magnetism re-
quires considering the relative disposition of the whole mol-
ecules in the crystal.[7] This paper shows that the comparison
of peculiar patterns is one objective way to deduce magnetic
interaction pathways. The power and the relevance of this
strategy are corroborated in the present work dealing with
larger molecules bearing more magnetically active sites. In
particular, the magnitude and the sign of most interactions
were found similar in the solid state and in dilute solution,
thus strengthening the validity of the various assessments
performed by using either EPR in frozen solution or
SQUID magnetometry. Comparing the peculiar patterns
found in the contact geometries of basic building blocks,
which individual magnetic properties have been properly as-
sessed, is relevant for the comprehensive use of such com-
plex molecules in a bottom-up approach of molecular mag-
netism.
susceptibility measurements were obtained with
MPMS/XL SQUID magnetometer.
a Quantum Design
CCDC-229617 (diINpNN), -229618 (INNNpNN), -229619 (diNNpNN)
and -229727 (diNNbpNN) contain the supplementary crystallographic
data for this paper. These data can be obtained free of charge from The
Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_
request/cif
Theoretical calculations: All calculations presented in this work were
done with the B3LYP[34] non-local exchange-correlation density function-
al,[35] using the 3-21G(d) basis Gaussian basis set,[36] as implemented in
the Gaussian 98 suite of programs.[37] The tight option was used to obtain
enough accuracy in computing the integrals and energy.
m,m’-Bisformylbenzene 4-ethynyl benzaldehyde (1): 4-Ethynylbenzalde-
hyde (501 mg, 1.1 equiv) and [Pd(PPh3)4] (292 mg, 8%mol) were added
under argon to 5-bromoisophtaldehyde (532 mg, 2.86 mmol). Benzene
(25 mL) and diisopropylamine (8 mL) were added and the mixture was
kept at 808C during 15 h. After the solution was cooled down, the ammo-
nium salt was filtered, the filtrate was evaporated and the residue puri-
fied through chromatography on silica gel (CH2Cl2), to give a light
yellow solid (556 mg, 74%). M.p. 1688C; 1H NMR ([D6]acetone): d=
10.20 (s, 2H, CHO), 10.09 (s, 1H, CHO), 8.47 (t, 1H, Ph), 8.38 (d, 2H,
Ph), 7.99 (d, 2H, Ph), 7.85 ppm (d, 2H, Ph); 13C NMR ([D6]acetone): d=
192.3 (CHO),191.8 (CHO), 137.1, 135.9, 132.2, 129.6, 127.3, 123.5 (C-Ph),
90.6 (-CꢄC-), 90.2 ppm (-CꢄC-); IR (KBr): n˜ =1693 cmꢀ1 (C=O); elemen-
tal analysis calcd (%) for C17H10O3: C 77.86, H 3.84, N 18.30; found C
77.60, H 3.75, N 18.62.
m,m’-Bis(1,3-hydroxy-4,4,5,5-tetramethylimidazolin-2-yl)-4-ethynyl-1-
(1,3-hydroxy-4,4,5,5-tetramethylimidazolin-2-yl)benzene (2): 2,3-Bis(hy-
droxyamino)-2,3-dimethyl butane (186 mg, 3.2 equiv) was added to a so-
lution of trisaldehyde 1 (103 mg, 0.39 mmol) in MeOH (50 mL). A weak
flow of argon was maintained during 3 d until the mixture was nearly dry.
The white paste was filtered, washed with MeOH and dried to give a
white solid (230 mg, 90%). Elemental analysis calcd (%) for C35H52O6N6:
C 64.39, H 8.03, N 12.87; found C 64.71, H 7.98, N 12.50; 1H NMR
([D6]acetone): d=7.67 (m, 1H, Ph), 7.60 (m, 2H, Ph), 7.54 (d, J=9 Hz,
2H, Ph), 7.46 (d, J=9 Hz, 2H, Ph), 7.14 (s, 6H, OH), 4.71 (s, 3H, -HC
imid.), 1.18 (s, 18H, CH3), 1.10 ppm (s, 18H, CH3); 13C NMR
([D6]DMSO): d=142.6, 141.9, 130.9, 130.6, 129.9, 129.6, 128.7 (Ph), 92.1
(HC imid.), 89.9, 89.8 (CꢄC), 66.2 (Cquat imid.), 24.4, 17.1 ppm (CH3).
Experimental Section
Materials and methods: Solvents were distilled under argon before use.
diNNpNN: Compound
2 (100 mg, 0.15 mmol) and MnO2 (198 mg,
In particular, THF was dried over sodium/benzophenone, triethylamine
2.28 mmol, 5 equiv per radical) were added to a solution of CH2Cl2
Chem. Eur. J. 2005, 11, 2440 – 2454
www.chemeurj.org
ꢅ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2451

P. Turek et al.
(30 mL). The mixture turned dark blue and was stirred during 5 h, fil-
tered and the solvent was evaporated. The residue was purified through
chromatography (silica gel, AcOEt/CH2Cl2 1:9) to give diNNpNN
(71 mg, 74%). Dark blue crystals containing one CH2Cl2 molecule per
diNNpNN molecule could be grown by slow evaporation of a CH2Cl2/
pared above and MnO2 (225 mg, ~5 equiv per radical) were introduced.
After 5 h under stirring at room temperature and evaporation of the sol-
vent, the residue was purified by chromatography (silica gel, AcOEt/
CH2Cl2 5:95) and gave diINpNN (60 mg, 22% overall yield). Slow evapo-
ration of
a
CH2Cl2/hexane mixture gave dark green crystals of
hexane mixture. M.p. 2428C (decomp); IR (KBr): n˜ =1363 cmꢀ1 (N O);
ꢀ
ꢀ
diINpNN·CH2Cl2. M.p. 2368C (decomp); IR (KBr): n˜ =1363 (N O),
1544 cmꢀ1 (C=N); UV/Vis: lmax(e)=316 (16370), 385 (62450), 445
(1380), 602 nm (560mꢀ1 cmꢀ1); elemental analysis calcd (%) for
C35H19O4N6·CH2Cl2: C 62.06, H 6.51, N 12.06, O 9.18; found C 62.14, H
6.52, N 11.99, O 9.30; MS FAB+ (mNBA): m/z: 613.2 [M+H]+, 597.2,
580.2, 564.2, 548.2, (successive loss of 4 O), 498.1, 483.1, 384.0, 369.1,
353.1; NS =2.75 spins per molecule.
UV/Vis: lmax(e)=319 (53605), 371 (24680), 592 nm (1130mꢀ1 cmꢀ1); ele-
mental analysis calcd (%) for C35H19O6N6·CH2Cl2: C 59.34, H 6.22, N
11.53; found C 59.11, H 6.20, N 11.28; MS FAB+ (mNBA): m/z: 645.1
[M+H]+, 629.1, 613.1, 597.1, 581.1, 565.1, 549.3, 533.1 (successive loss of
6 O), 514.0, 384.0 (successive loss of 2 fragments of 130 mass units); NS =
3.1 spins per molecule.
diINpIN: Compound diNNpNN (45 mg, 0.069 mmol) and NaNO2
(71 mg) were added to water/CH2Cl2 (1:1, 50 mL) acidified at pH 6.
After 20 min, the bright red solution extracted with CH2Cl2, dried on
Na2SO4 and evaporated. The red solid was purified by chromatography
(silica gel, AcOEt/CH2Cl2 5:95) to give a red solid (25 mg, 59%). Crys-
1,3-Bis(1,3-hydroxy-4,4,5,5-tetramethylimidazolin-2-yl)-5-bromobenzene
(3): 5-Bromoisophtaladehyde (1.51 g, 7.05 mmol) and 2,3-bis(hydroxy-
amino)-2,3-dimethylbutane (2.42 g, 2.3 equiv) were added to MeOH
(30 mL). The mixture was concentrated to almost dryness during 3 d. The
white solid was washed with MeOH and dried, and a white powder was
isolated (3.15 g, 94%). M.p. 1758C (decomp); IR (KBr): n˜ =3455 cmꢀ1
(OH); elemental analysis calcd (%) for C20H33O4N4Br: C 51.40, H 5.82,
tals were grown by slow evaporation of a CH2Cl2/hexane mixture. M.p.
ꢀ1
ꢀ
2058C (decomp); IR (KBr): n˜ =2208 (CꢄC), 1371 (N O), 1543 cm (C=
N); UV/Vis: lmax(e)=273 (34450), 449 nm (1441mꢀ1 cmꢀ1); elemental
analysis calcd (%) for C 70.56, H 7.27, N 14.11; found C 70.32, H 7.26, N
14.04; MS FAB+ (mNBA): m/z: 597.2 [M+H]+, 581.1, 565.2, 549.2, (suc-
cessive loss of 3 O), 483.2, 369.1 459.3, 345.1; NS =2.75 spins per mole-
cule.
1
N 11.99; found C 51.27, H, 5.87, N 11.76; H NMR ([D6]DMSO): d=7.81
(s, 4H, -OH), 7.55 (s, 2H,Ph), 7.51 (s, 1H, Ph), 4.50 (s, 2H, -CH imid.),
1.07 (s, 12H, -CH3), 1.03 ppm (s, 12H, -CH3); 13C NMR ([D6]DMSO):
d=144, 129, 128, 120 (C-Ph), 90 (CH imid.), 66 (Cquat imid.), 24 (CH3),
17 ppm (CH3).
m,m’-Bisformylphenyl-4-benzaldehyde (5): In a three-neck round-bottom
flask containing triethylamine (50 mL), 5-bromoisophtalaldehyde
(191 mg, 0.90 mmol) and 4-formylboronic acid (150 mg, 1.1 equiv) was
1,3-Bis(1-oxyl-3-oxo-4,4,5,5-tetramethylimidazolin-2-yl)-5-bromobenzene
(4): In
a round-bottom flask containing a suspension of 3 (2.1 g,
4.44 mmol) in MeOH (100 mL), SeO2 (74 mg, 6.65 mmol, 15% mol) was
added.[20] After 2 d at room temperature, the yellow solution was evapo-
rated and purified by chromatography (silica gel, CH2Cl2/EtOH 8:2) to
give a yellow powder (1.65 g, 85%). 1H NMR ([D6]DMSO): d=8.90 (s,
1H, Ph), 8.30 (s, 2H, Ph), 1.36 (s, 12H, -CH3), 1.30 ppm (s, 12H, -CH3);
13C NMR ([D6]DMSO): d=141, 131, 129, 127 (Ph), 76 (CH imid.), 61.5
(Cquat imid.), 24 (CH3), 19 (CH3).
added.
A degassed aqueous solution (1.5 mL) containing Na2CO3
(212 mg, 2.2 equiv) was added with a syringe. Dimethoxyethane (5 mL)
freshly distilled under argon were then added and the mixture was
heated at 858C. After 5 h of reaction, the reaction mixture was filtered
and the filtrate was evaporated under vacuum. The residue was purified
by chromatography (silica gel, AcOEt/CH2Cl2 1:99) to give a white solid
(144 mg, 61%). M.p. 1608C (decomp); IR (KBr): n˜ =1691 cmꢀ1 (CHO);
elemental analysis calcd (%) for C15H10O3: C 75.62, H 4.23, N 20.15;
found C 75.35, H 4.15, N 20.40; 1H NMR (CDCl3): d=10.21 (s, 2H,
CHO), 10.12 (s, 1H, CHO), 8.42 (m, 3H, Ph), 8.05 (d, J=8.5 Hz, 2H,
Ph), 7.86 ppm (d, J=8.5 Hz, 2H, Ph); 13C NMR (CDCl3): d=191.5
(CHO), 190.6 (CHO), 144.0, 136.3, 133.9, 131.7, 131.5, 127.9, 124.0 ppm
(Ph).
INNNpNN: In a round-bottom flask under argon, equipped with a
funnel containing triethylamine (12 mL) distilled under argon, 3 (492 mg,
1.06 mmol), [PdCl2(PPh3)2] (37 mg, 5%mol) and CuI (10 mg) were
added. Triethylamine (7 mL) was added dropwise under stirring. 4-Ethyn-
yl-(1,3-hydroxy-4,4,5,5-tetramethylimidazolin-2-yl)benzene
(366 mg,
1.3 equiv) was dissolved in the remaining triethylamine (5 mL). This so-
lution was added in five parts to the reaction mixture during 48 h, and
heated to 808C. After 12 h reaction time, partial dehydratation could be
observed by TLC (silica gel, AcOEt/CH2Cl2 1:9). The mixture was cooled
down, the amine salt filtered and the filtrate precipitated in hexane
(200 mL). The precipitate (560 mg) was dried. Oxidation of this solid
(270 mg) was performed by introducing MnO2 (1.049 g, ~5 equiv per rad-
ical) in CH2Cl2 (25 mL). The suspension was stirred during 6 h at room
temperature and turned to dark blue-green. The solution was evaporated
and the remaining solid was purified by chromatography (silica gel,
AcOEt/CH2Cl2 5:95) to isolate diNNpNN (37 mg) and INNNpNN
(48 mg, 15%). Slow evaporation of a CH2Cl2/hexane mixture gave black
m,m’-Bis(1,3-hydroxy-4,4,5,5-tetramethylimidazolin-2-yl)4-(1,3-hydroxy-
4,4,5,5-tetramethylimidazolin-2-yl)benzene (6): The procedure was simi-
lar to the synthesis of 2, using 5 (132 mg, 0.5 mmol) and 2,3-bis(hydroxya-
mino)-2,3-dimethyl butane (262 mg, 3.3 equiv) in distilled MeOH
(30 mL) and CH2Cl2 (4 mL). After the usual workup, a white powder
(294 mg, 85%) was isolated. M.p. 1758C (decomp); IR (KBr): n=
1
3597 cmꢀ1 (OH); H NMR ([D6]acetone): d=7.72 (s, 6H, OH), 7.63–7.55
(m, 7H, Ph), 4.59 (s, 3H, -HC5), 1.09 (s,18H, CH3), 1.06 ppm (s,18H,
CH3); 13C NMR ([D6]DMSO): d=152.1, 141.8, 140.8, 140.3, 139.2, 131.6,
128.9, 126.0 (Ph), 90.0 (HC imid.), 66.2 (Cquat. imid.), 24.0, 16.9 ppm
(CH3).
needles of INNNpNN·CH2Cl2. M.p. 2408C (decomp); IR (KBr): n˜ =1365
ꢀ1
ꢀ
(N O), 1549 cm (C=N); UV/Vis: lmax(e)=320 (45250), 371 (16570),
diNNbpNN: The procedure used for obtaining diNNpNN was employed
starting from 6 (103 mg, 0.16 mmol) and MnO2 (198 mg, 2.28 mmol,
5 equiv per radical). The residue was purified by chromatography (silica
gel, AcOEt/CH2Cl2 1:99) to give diNNbpNN (65 mg, 66%). M.p. 2288C
450 (712), 595 nm (700mꢀ1 cmꢀ1); elemental analysis calcd (%) for
C35H19O5N6·CH2Cl2: C 60.67, H 6.36, N 11.79; found C 60.55, H 6.39, N
11.65; MS FAB+ (mNBA): m/z: 629.4 [M+H]+, 613.4, 597.4, 581.4,
565.4, 549.3 (successive loss of 5 O), 514.4, 498.2, 384.1, 368.1; NS =2.8
spins per molecule.
(decomp); IR (KBr): n˜ =1360 cmꢀ1 (N O); UV/Vis (CH2Cl2): lmax(e)=
ꢀ
294 (43230), 371 (26420), 589 nm (1380mꢀ1 cmꢀ1); elemental analysis
calcd (%) for C33H19O6N6: C 63.95, H 6.99, N 13.57; found C 63.70, H
7.16, N 13.85; MS FAB+ (mNBA): m/z: 621.2 [M+H]+, 605.2, 589.2,
573.2, 557.2, 541.2, 525.2 (successive loss of 6 O), 490.1, 360.2; NS =3.2
spins per molecule.
diINpNN: Compound 4 (400 mg, 0.91 mmol), [PdCl2(PPh3)2] (30 mg, 5%
mol) and CuI (5 mg, 3%) were dissolved in a solution of triethylamine
(5 mL). A solution of 4-ethynyl-(1,3-hydroxy-4,4,5,5-tetramethylimidazo-
lin-2-yl)benzene (277 mg, 1.2 equiv) in triethylamine (5 mL) was added
to the funnel. This solution was added in portions of 1 mL during 48 h.
The mixture was cooled down, the amine salt filtered and the filtrate pre-
cipitated in hexane. A yellow powder (220 mg) was isolated. 1H NMR
([D6]DMSO): d=8.94 (s, 1H, Ph), 7.93 (s, 2H, Ph), 7.83 (s, 2H, OH),
7.54 (m, 4H, Ph), 4.53 (s, 1H, -HC imid.), 1.11 (s,18H, CH3), 1.04 ppm
(s,18H, CH3).
diINbpIN: Compound diNNbpNN (50 mg, 0.081 mmol) and NaNO2
(168 mg, 10 equiv per radical) were added to a water/CH2Cl2 mixture
(1:1, 50 mL) acidified at pH 6. After 20 min, the bright red solution ex-
tracted with CH2Cl2, dried on Na2SO4 and evaporated. The red solid was
purified by chromatography (silica gel, AcOEt/CH2Cl2 5:95) to give a red
solid (31 mg, 54%). Crystals were grown by slow evaporation of a
CH2Cl2/hexane mixture. M.p. 2458C (decomp); IR (KBr): n˜ =1375 cmꢀ1
In a flask containing MeOH (30 mL), the yellow powder (105 mg) pre-
2452
ꢅ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 2440 – 2454

Magnetic Interactions
FULL PAPER
(N O); UV/Vis (CH2Cl2): lmax(e)=273 (34450), 449 nm (1441mꢀ1 cmꢀ1);
elemental analysis calcd (%) for C 69.32, H 7.58, N 14.70; found C 69.43,
H 7.54, N 14.39; MS FAB+ (mNBA): m/z: 573.3 [M+H]+, 557.3, 541.3,
525.3 (successive loss of 3 O), 459.3, 345.1; NS =3.3 spins per molecule.
[3] For recent general reviews on molecular magnetism, see a) Magnetic
Properties of Organic Materials (Ed.: P. M. Lahti), Marcel Dekker,
New York, 1999; b) Molecular Magnetism (Eds.: K. Itoh, M. Kinosh-
ita), Gordon and Breach, Kodansha, Tokyo, 2000; c) p -Electron
magnetism: From molecules to magnetic materials, Vol. 100 (Ed.: J.
Veciana), Springer, 2001; d) Magnetism: Molecules to Materials,
Vol. I–IV (Eds.: J. S. Miller and M. Drillon), Wiley-VCH, Weinheim,
2001–2003; e) B. Pilawa, Ann. Phys. 1999, 3, 191–254; f) J. A. Cray-
ston, J. N. Devine, J. C. Walton, Tetrahedron 2000, 56, 7829–7857.
[4] a) J. S. Miller, Adv. Mater. 1992, 4, 298–300; b) J. S. Miller, Adv.
Mater. 1992, 4, 435–484; c) A. J. Epstein, MRS Bull. 2003, 28, 492–
498.
ꢀ
tetraNN: In
a
round-bottom flask, 1,3-diiodobenzene (150 mg,
0.45 mmol) and 1,3-bis(1,3-hydroxy-4,4,5,5-tetramethylimidazolin-2-yl)-5-
ethynylbenzene (410 mg, 2.1 equiv), [Pd(PPh3)4] (66 mg, 12 mol%) and
freshly distilled diisopropylamine (10 mL) were added. The reaction mix-
ture was heated at 808C during 6 h. After filtration, evaporation of the
filtrate, the concentrated solution was precipitated with pentane
(100 mL). The beige powder was filtered and dried under vacuum (70%
of estimated coupling).
[5] a) A. Caneschi, D. Gatteschi, R. Sessoli, P. Rey, Acc. Chem. Res.
1989, 22, 392–398; b) H. O. Stumpf, L. Ouahab, Y. Pei, D. Grand-
jean, O. Kahn, Science 1993, 261, 447–449; c) K. Inoue, T. Hayami-
zu, H. Iwamura, D. Hashizume, Y. Ohashi, J. Am. Chem. Soc. 1996,
118, 1803–1804.
[6] a) J. Cirujeda, E. Hernꢇndez-Gasio, C. Rovira, J.-L. Stanger, P.
Turek, J. Veciana, J. Mater. Chem. 1995, 5, 243–252; b) M. Matsuchi-
da, A. Izuoka, T. Sugawara, T. Kobayashi, N. Wada, N. Takeda, M.
Ishikawa, J. Am. Chem. Soc. 1997, 119, 4369–4379; c) F. M.
Romero, R. Ziessel, M. Bonnet, Y. Pontillon, E. Ressouche, J.
Schweizer, B. Delley, A. Grand, C. Paulsen, J. Am. Chem. Soc. 2000,
122, 1298–1309.
The oxidation was performed on this powder dispersed in CH2Cl2
(100 mL). MnO2 (382 mg, ~20 equiv) was added and the suspension was
stirred at room temperature during 3 h. The mixture was worked up as
usual and the residue purified by chromatography on a column and a
second time on a preparative TLC support (silica gel, AcOEt/CH2Cl2/
MeOH 19:80:1) to give a blue powder with an overall yield of 7% using
the dibromobenzene, and 17% using diiodobenzene. IR (KBr): n˜ =
1361 cmꢀ1 (N O); elemental analysis calcd (%) for C 66.80, H 6.50, N
ꢀ
12.46; found C 66.56, H 6.76, N 12.50; UV/Vis (CH2Cl2): lmax(e)=288
(101210), 371 (63350), 585 nm (2725mꢀ1 cmꢀ1); MS FAB+ (mNBA): m/z:
899.3 [M]+, 884.3, 868.3, 852.4, 836.4, 820.4, 804.3, 786.2, 770.2 (successive
loss of 8 O), 769.2, 639.1, 509.3; NS =4.0 spins per molecule.
[7] a) M. Deumal, J. Cirujeda, J. Veciana, M. Kinoshita, Y. Hosokoshi,
J. J. Novoa, Chem. Phys. Lett. 1997, 265, 190–199; b) M. Deumal, J.
Cirujeda, J. Veciana, J. J. Novoa, Adv. Mater. 1998, 10, 1461–1466;
c) M. Deumal, J. Cirujeda, J. Veciana, J. J. Novoa, Chem. Eur. J.
1999, 5, 1631–1642; d) M. Deumal, M. A. Robb, J. J. Novoa, Poly-
hedron 2003, 22, 1935–1944.
[8] a) K. Itoh, Pure Appl. Chem. 1978, 50, 1251–1259; b) H. Iwamura,
Pure Appl. Chem. 1986, 58, 187–196; c) H. Iwamura, Pure Appl.
Chem. 1987, 59, 1595–1604; d) H. Iwamura, Pure Appl. Chem. 1993,
65, 57–64; e) H. Iwamura, Adv. Phys. Org. Chem. 1990, 26, 179–
253; f) D. A. Dougherty, Acc. Chem. Res. 1991, 24, 88–94; g) H.
Iwamura, N. Koga, Acc. Chem. Res. 1993, 26, 346–351; h) A. Rajca,
Chem. Rev. 1994, 94, 871–893; i) K. Itoh, Y. Sawaki, A. Izuoka, T.
Sugawara, F. Miko, H. Iwamura, J. Am. Chem. Soc. 1990, 112, 4074–
4075; j) A. Rajca, J. Wongsriratanakul, S. Rajca, Science 2001, 294,
1503–1505.
[9] a) H. Nishide, Adv. Mater. 1995, 7, 937–941; b) H. Nishide, T.
Ozawa, M. Miyasaka, E. Tsuchida, J. Am. Chem. Soc. 2001, 123,
5942–5946; c) M. Miyasaka, Y. Saito, H. Nishide, Adv. Funct. Mater.
2003, 13, 113–117.
[10] a) H. C. Longuet-Higgins, J. Chem. Phys. 1950, 18, 265–274; b) N.
Mataga, Theor. Chim. Acta 1968, 10, 372–376; c) W. T. Borden,
E. R. Davidson, J. Am. Chem. Soc. 1977, 99, 4587–4594; d) A. A.
Ovchinnikov, Theor. Chim. Acta 1978, 47, 297–304; e) W. T.
Borden, H. Iwamura, J. A. Berson, Acc. Chem. Res. 1994, 27, 109–
116.
tetraIN: The cross-coupling was performed as described for tetraNN, but
after the reaction mixture was evaporated, the solid obtained (450 mg,
0.5 mmol) was treated with SeO2 (10 mg, 15% mol) during 2 d in MeOH
(200 mL). The suspension disappeared and the solution turned to yellow;
the solvent was evaporated and the compound purified by chromatogra-
phy (silica gel, CH2Cl2/EtOH 9:1 then gradually 6:4) to yield a yellow-
brown powder (255 mg, 59%). 1H NMR ([D6]DMSO): d=8.44 (s, 2H,
Ph), 8.22 (s, 4H, Ph), 7.80 (s, 1H, Ph), 7.66–7.63 (m, 3H, Ph), 1.16
(s,24H, CH3), 1.14 ppm (s, 24H, CH3).
To a suspension of this solid (200 mg, 0.22 mmol) in CH2Cl2 (30 mL),
MnO2 (380 mg, 4.4 mmol, 20 equiv) was added and the mixture stirred
during 3 h. After filtration and evaporation of the filtrate the red com-
pound was chromatographed (silica gel, AcOEt/CH2Cl2 15:85) to give
tetraIN (110 mg, 31% overall yield) which gave solvated red needles un-
suitable for structural determination by slow evaporation in a CH2Cl2/
hexane mixture. M.p. 2498C (decomp); IR (KBr): n˜ =2216 (C=C), 1371
ꢀ1
ꢀ
(N O), 1549 cm (C=N); UV/Vis (CH2Cl2): lmax(e)=288 (73390), 302
(62940), 447 nm (1667mꢀ1 cmꢀ1); elemental analysis calcd (%) for tet-
raIN·3CH2Cl2: C 69.12, H 6.78, N 12.77, O 7.29; found C 68.91, H 6.82,
N 12.56, O 7.84; MS FAB+: m/z: 836.3 [M+H]+, 821.3, 805.3, 789.3,
773.3 (successive loss of 4 O), 722.2, 608.1, 494.1; NS =4.1 spins per mole-
cule.
[11] Recent reports: a) D. A. Shultz, A. K. Boal, G. T. Farmer, J. Org.
Chem. 1998, 63, 9462–9469; b) R. M. Fico, M. F. Hay, S. Reese, S.
Hammond, E. Lambert, M. A. Fox, J. Org. Chem. 1999, 64, 9386–
9392; c) W. Adam, M. Baumgarten, W. Maas, J. Am. Chem. Soc.
2000, 122, 6735–6738; d) V. M. Domingo, C. Aleman, E. Brillas, L.
Juliꢈ, J. Org. Chem. 2001, 66, 4058–4061; e) D. A. Shultz, S. H.
Bodnar, H. Lee, J. W. Kampf, C. D. Incarvito, A. L. Rheingold, J.
Am. Chem. Soc. 2002, 124, 10054–10061; f) A. Dei, D. Gatteschi, C.
Sangregorio, L. Sorace, M. G. F. Vaz, Inorg. Chem. 2003, 42, 1701–
1706.
[12] a) M. Dvolaitzky, R. Chiarelli, A. Rassat, Angew. Chem. 1992, 104,
180–181; Angew. Chem. Int. Ed. Engl. 1992, 31, 180–181; b) F.
Kanno, K. Inoue, N. Koga, H. Iwamura J. Am. Chem. Soc. 1993,
115, 847–850; c) S. K. Silverman, D. Dougherty, J. Phys. Chem.
1993, 97, 13273–13283; d) N. Yoshioka, P. M. Lahti, K. Takashi, Y.
Kuzumaki, E. Tsuchida, H. Nishide, J. Org. Chem. 1994, 59, 4272–
4280; e) S. Fang, M. S. Lee, D. A. Hrovat, W. T. Borden, J. Am.
Chem. Soc. 1995, 117, 6727–6731; f) J. Fujita, M. Tanka, H. Sue-
mune, N. Koga, K. Matsuda, H. Iwamura, J. Am. Chem. Soc. 1996,
Acknowledgement
Dr. Marc Drillon (IPCMS, Strasbourg, France) is gratefully acknowl-
edged for discussions of the model Hamiltonians of various spin clusters.
Mr. Maxime Bernard (ICS, Strasbourg, France) is thanked for invaluable
help in the set up of EPR experiments. J.J.N. acknowledges the support
of CESCA/CEBPA in allocating computer times in their machines and
also the financial support of the Spanish “Ministerio de Ciencia y Tecno-
logia” and Catalan Autonomous Government (projects BQU2002-04587-
C02-02 and 2001SGR-0044, respectively). The stay in Barcelona of LC
was supported by the EU through
CEPBA.
a TMR contract with CESCA/
[1] Part I: P. Wautelet, J. Le Moigne, V. Videva, P. Turek, J. Org. Chem.
2003, 68, 8025–8036.
[2] Part II: L. Catala, J. Le Moigne, N. Kyritsakas, P. Rey, J. J. Novoa, P.
Turek, Chem. Eur. J. 2001, 7, 2466–2480.
Chem. Eur. J. 2005, 11, 2440 – 2454
www.chemeurj.org
ꢅ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2453

P. Turek et al.
118, 9347–9351; g) D. Schultz, A. K. Boal, G. T. Farmer, J. Am.
Chem. Soc. 1997, 119, 3846–3847.
[27] a) E. Belorizky, J. Phys. I 1993, 3, 423–445; b) E. Belorizky, P. H.
Fries, J. Chim. Phys. Phys.-Chim. Biol. 1993, 90, 1077–1100; c) E.
Belorizky, M. Jeannin, P. Rey, A. Rassat, Solid State Commun. 1993,
85–88, 735–737.
[28] A. Rajca, S. Rajca, J. Am. Chem. Soc. 1995, 117, 806–816.
[29] a) S. Murata, H. Iwamura, J. Am. Chem. Soc. 1991, 113, 5547–5556;
b) T. Mitsumori, K. Inoue, N. Koga, H. Iwamura, J. Am. Chem. Soc.
1995, 117, 2467–2478.
[13] a) A. Rajca, S. Rajca, J. Am. Chem. Soc. 1996, 118, 8121–8126;
b) A. Rajca, J. Wongsriratanakul, S. Rajca, J. Am. Chem. Soc. 1997,
119, 11674–11686; c) A. Rajca, J. Wongsriratanakul, S. Rajca, R.
Cerny, Angew. Chem. 1998, 110, 1284–1288; Angew. Chem. Int. Ed.
1998, 37, 1229–1232.
[14] L. Catala, P. Turek, J. Le Moigne, A. De Cian, N. Kyritsakas, Tetra-
hedron Lett. 2000, 41, 1015–1018.
[30] H. M. McConnell, J. Chem. Phys. 1963, 39, 1910.
[31] F. H. Allen, Acta Crystallogr. Sect. B 2002, 58, 380–388.
[32] J. A. Wytko, J. Weiss, Tetrahedron Lett. 1991, 32, 7261–7264.
[33] M. Lamchen, T. W. Mittag, J. Chem. Soc. 1966, 2300–2303.
[34] B3LYP is a density functional obtained by taking the three parame-
ter non-local exchange functional of Becke and the non-local corre-
lation functional of Lee–Yang–Parr a) A. D. Becke, J. Chem. Phys.
1993, 98, 5648–5652; b) C. Lee, W. Yang, R. G. Parr, Phys. Rev. B
1988, 37, 785–789.
[35] R. Parr, W. Yang, Density Functional Theory of Atoms and Mole-
cules, Oxford University Press, New York, 1989.
[36] J. S. Binkley, J. A. Pople, W. J. Hehre, J. Am. Chem. Soc. 1980, 102,
939–947.
[37] Gaussian98 (Revision A.7), M.J. Frisch, G. W. Trucks, H. B. Schle-
gel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, V. G. Zakrzewski,
J. A. Montgomery, R. E. Stratmann, J. C. Burant, S. Dapprich, J.M.
Millam, A. D. Daniels, K. N. Kudin, M. C. Strain, O. Farkas, J.
Tomasi, V. Barone, M. Cossi, R. Cammi, B. Mennucci, C. Pomelli,
C. Adamo, S. Clifford, J. Ochterski, G. A. Petersson, P. Y. Ayala, Q.
Cui, K. Morokuma, D. K. Malick, A. D. Rabuck, K. Raghavachari,
J. B. Foresman, J. Cioslowski, J. V. Ortiz, B. B. Stefanov, G. Liu, A.
Liashenko, P. Piskorz, I. Komaromi, R. Gomperts, R. L. Martin,
D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng, A. Nanayakkara,
C. Gonzalez, M. Challacombe, P. M. W. Gill, B. G. Johnson, W.
Chen, M. W. Wong, J.L. Andres, M. Head-Gordon, E. S. Replogle,
J. A. Pople, Gaussian, Inc., Pittsburgh, PA, 1998.
[15] F. M. Romero, R. Ziessel, A. De Cian, J. Fischer, P. Turek, New J.
Chem. 1996, 20, 919–924.
[16] M. Alami, F. Ferri, G. Linstrumelle, Tetrahedron Lett. 1993, 34,
6403–6406.
[17] E. F. Ullman, J. H. Osiecki, D. G. B. Boocock, R. Darcy, J. Am.
Chem. Soc. 1972, 94, 7049–7059.
[18] N. Miyaura, A. Suzuki, Chem. Rev. 1995, 95, 2457–2483.
[19] Z. Z. Zhong, H. N. C. Wong, J. Org. Chem. 1994, 59, 33–41.
[20] G. Ulrich, R. Ziessel, Tetrahedron Lett. 1994, 35, 1215–1218.
[21] J. E. Wertz, J. R. Bolton, Electron Spin Resonance, Elementary
Theory and Practical Applications, Chapman and Hall 1994, p. 112.
[22] WINEPR Simfonia software, Bruker Analytische Messtechnik
GmbH, 1995–1996, Version 1.25.
[23] a) D. A. Shultz, R. M. Fico, Jr., S. H. Bodnar, R. K. Kumar, K. E.
Vostrikova, J. W. Kampf, P. D. Boyle, J. Am. Chem. Soc. 2003, 125,
11761–11771; b) D. A. Shultz, R. M. Fico, Jr., H. Lee, J. W. Kampf,
K. Kirschbaum, A. A. Pinkerton, P. D. Boyle, J. Am. Chem. Soc.
2003, 125, 15426–15432.
[24] a) K. Yoshizawa, A. Chano, A. Ito, K. Tanaka, T. Yamabe, H.
Fujita, J. Yamauchi, M. Shiro, J. Am. Chem. Soc. 1992, 114, 5994–
5998; b) Y. Teki, T. Takui, K. Itoh, H. Iwamura, K. Kobayashi, J.
Am. Chem. Soc. 1986, 108, 2147–2156.
[25] a) J. A. Berson, in The Chemistry of Quinonoid Compounds, Vol. 11
(Eds.: S. Patai and Z. Rappoport), 1988, pp. 481; b) K. Matsuda, H.
Iwamura, J. Chem. Soc. Perkin Trans. 2 1998, 1023–1026.
[26] a) M. Tanaka, K. Matsuda, T. Itoh, H. Iwamura, J. Am. Chem. Soc.
1998, 120, 7168–7173; b) P. M. Lahti, Y. Liao, M. Julier, F. Palacio,
Synth. Met. 2001, 122, 485–493.
Received: June 3, 2004
Revised: September 7, 2004
Published online: January 26, 2005
2454
ꢅ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 2440 – 2454