Light-driven coordination-induced spin-state switching: Rational design of photodissociable ligands

Article abstract of DOI:10.1002/chem.201201698

The bistability of spin states (e.g., spin crossover) in bulk materials is well investigated and understood. We recently extended spin-state switching to isolated molecules at room temperature (light-driven coordination-induced spin-state switching, or LD-CISSS). Whereas bistability and hysteresis in conventional spin-crossover materials are caused by cooperative effects in the crystal lattice, spin switching in LD-CISSS is achieved by reversibly changing the coordination number of a metal complex by means of a photochromic ligand that binds in one configuration but dissociates in the other form. We present mathematical proof that the maximum efficiency in property switching by such a photodissociable ligand (PDL) is only dependent on the ratio of the association constants of both configurations. Rational design by using DFT calculations was applied to develop a photoswitchable ligand with a high switching efficiency. The starting point was a nickel-porphyrin as the transition-metal complex and 3-phenylazopyridine as the photodissociable ligand. Calculations and experiments were performed in two iterative steps to find a substitution pattern at the phenylazopyridine ligand that provided optimum performance. Following this strategy, we synthesized an improved photodissociable ligand that binds to the Ni-porphyrin with an association constant that is 5.36times higher in its trans form than in the cis form. The switching efficiency between the diamagnetic and paramagnetic state is efficient as well (72 % paramagnetic Ni-porphyrin after irradiation at 365nm, 32 % paramagnetic species after irradiation at 440nm). Potential applications arise from the fact that the LD-CISSS approach for the first time allows reversible switching of the magnetic susceptibility of a homogeneous solution. Photoswitchable contrast agents for magnetic resonance imaging and light-controlled magnetic levitation are conceivable applications. Turn the spin: Nickel-porphyrins with appropriately designed axial photochromic ligands change their coordination number and consequently their spin state reversibly upon irradiation. Rational design led to a substituted 3-phenylazopyridine as a photodissociable ligand with a switching efficiency of 40 % (see figure). Thus, the magnetic susceptibility of a homogeneous solution was switched by a factor of more than two at room temperature.

Full text of DOI:10.1002/chem.201201698

DOI: 10.1002/chem.201201698
Light-Driven Coordination-Induced Spin-State Switching: Rational Design of
Photodissociable Ligands
Steffen Thies,[a] Hanno Sell,[a] Claudia Bornholdt,[a] Christian Schꢀtt,[a] Felix Kçhler,[a]
Felix Tuczek,[b] and Rainer Herges*[a]
Dedicated to G. A. Olah on the occasion of his 85th birthday
Abstract: The bistability of spin states
(e.g., spin crossover) in bulk materials
is well investigated and understood.
We recently extended spin-state switch-
ing to isolated molecules at room tem-
perature (light-driven coordination-in-
duced spin-state switching, or LD-
CISSS). Whereas bistability and hyste-
resis in conventional spin-crossover
materials are caused by cooperative ef-
fects in the crystal lattice, spin switch-
ing in LD-CISSS is achieved by reversi-
bly changing the coordination number
of a metal complex by means of a pho-
tochromic ligand that binds in one con-
figuration but dissociates in the other
form. We present mathematical proof
that the maximum efficiency in proper-
ty switching by such a photodissociable
ligand (PDL) is only dependent on the
ratio of the association constants of
both configurations. Rational design by
using DFT calculations was applied to
develop a photoswitchable ligand with
a high switching efficiency. The starting
point was a nickel–porphyrin as the
transition-metal complex and 3-phenyl-
azopyridine as the photodissociable
ligand. Calculations and experiments
were performed in two iterative steps
to find a substitution pattern at the
phenylazopyridine ligand that provided
optimum performance. Following this
strategy, we synthesized an improved
photodissociable ligand that binds to
the Ni–porphyrin with an association
constant that is 5.36 times higher in its
trans form than in the cis form. The
switching efficiency between the dia-
magnetic and paramagnetic state is ef-
ficient as well (72% paramagnetic Ni–
porphyrin after irradiation at 365 nm,
32% paramagnetic species after irradi-
ation at 440 nm). Potential applications
arise from the fact that the LD-CISSS
approach for the first time allows re-
versible switching of the magnetic sus-
ceptibility of a homogeneous solution.
Photoswitchable contrast agents for
magnetic resonance imaging and light-
controlled magnetic levitation are con-
ceivable applications.
Keywords: coordination number ·
nickel · photochromism · porphyri-
noids · spin crossover
Introduction
metastable spin states in isolated molecules at room temper-
ature, however, would open a number of new applications in
the field of switchable contrast agents for magnetic reso-
nance imaging (MRI),[36] light-controlled diamagnetic levita-
tion,[7,8] data storage, or spintronics.[9] Probably the first ap-
proach to spin-state switching that is not based on coopera-
tive effects of neighboring spin centers was proposed by
Boillot and Zarembowitch et al.[10–13] They chose FeII as the
transition-metal ion in conjunction with ligands that isomer-
ize upon irradiation and concomitantly change their ligand
field strength, which leads to a change in magnetic proper-
ties (light-driven ligand-induced spin change, or LD-LISC).
Another approach is based on changing the spin state by
changing the oxidation state of the transition-metal ion.
Again the switchable redox behavior of photochromic li-
gands is used to change the oxidation state and concomi-
tantly the spin state of the transition-metal ion.[14,15] Howev-
er, the effects in both systems are weak and reversibility was
restricted to only a few cycles.[16]
Magnetic bistability due to the orientation of magnetization
in ferromagnetic materials or spin-state bistability based on
spin crossover in transition-metal complexes are typical
solid-state phenomena. The reason for the hysteresis is the
cooperative interaction between a large number of spin cen-
ters.[1] In isolated molecules, bistability is therefore observed
only at low temperatures.[2] Switching between two stable/
[a] Dr. S. Thies, H. Sell, Dr. C. Bornholdt, C. Schꢀtt, Dr. F. Kçhler,
Prof. Dr. R. Herges
Institut fꢀr Organische Chemie, Universitꢁt Kiel
Otto-Hahn-Platz 4, 24118 Kiel (Germany)
Fax : (+49)431-880-1558
E-mail: rherges@oc.uni-kiel.de
[b] Prof. Dr. F. Tuczek
Institut fꢀr Anorganische Chemie, Universitꢁt Kiel
Max-Eyth-Strasse 2, 24118 Kiel (Germany)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201201698. It includes UV/Vis
spectra, extinction coefficients, half-lives of thermal isomerization, a
complete list of association constants, the numbering of the structures
for the assignment of the NMR spectroscopic signals, and the compu-
tational details of the DFT calculations.
We recently developed the light-driven coordination-in-
duced spin-state switching (LD-CISSS) approach.[17–19] In
contrast to conventional spin-crossover complexes, LD-
CISSS systems are bistable over a very large temperature
16358
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 16358 – 16368

FULL PAPER
range (from the melting point of the solvent to well above
room temperature), they are air stable, completely fatigue
resistant, and most importantly they are bistable in homoge-
neous solution. The idea is based on the fact that a number
of transition metals in some oxidation states (e.g., MnIII,
FeII, FeIII, CoII, CoIII, and NiII) can exist in two or three dif-
ferent spin states depending on their coordination number.
We chose NiII because it is stable in its oxidation state in air
(which facilitates synthesis and handling), and the spin
states are reliably predicted by quantum chemical calcula-
tions (which makes the molecules easier to design). Both as-
pects are distinct advantages over FeII, which is most fre-
quently used in spin-crossover experiments. The benefits
more than outweigh the disadvantage that the attainable
spin change is smaller in NiII (DS=1) than in FeII (DS=2).
Square-planar NiII complexes (coordination number n=4)
are always diamagnetic, and square-pyramidal (n=5) as
well as octahedral (n=6) complexes are paramagnetic (S=
1).[20] The change in spin state of NiII is due to a change in
occupancy of the dx2ꢀy2 and dz2 orbitals. In square-planar NiII
complexes, the dz2 orbital is doubly occupied, and dx2ꢀy2 is
empty (Figure 1, left, Ni porphyrin). Axial donor ligands in-
crease the energy of the dz2 MO close to the level of the
empty dx2ꢀy2 orbital (Figure 1, Ni porphyrin with one axial
pyridine, singlet). In electron-poor square-planar complexes,
the approach of a single, strong donor ligand is sufficient to
induce spin crossover, thus leading to a triplet high-spin
complex with one electron in dz2 and one in dx2ꢀy2 (Figure 1,
Ni porphyrin with one pyridine ligand triplet). A second
axial ligand further stabilizes the triplet high-spin state
(Figure 1, Ni porphyrin with two axial pyridine ligands).
A change in coordination number from n=4 to n=5 (or
to n=6), therefore, gives rise to a change in spin state from
S=0 to S=1. To put this concept into operation, we chose
porphyrins as platforms (n=4) and photochromic molecules
as axial ligands to change the coordination number. Por-
phyrins have a distinct advantage over other square-planar
complexes coordinated by, for example, cyclam or salen li-
gands, in that they are more rigid, and therefore would not
(or only slightly) deform upon axial coordination or decoor-
dination.
Two approaches were pursued to put the light-driven
axial coordination/decoordination into practice: 1) the
“record player” design with the photoswitchable ligand co-
valently tethered to the porphyrin,[19,21] and 2) the photodis-
sociable ligand (PDL) approach, which is based on switch-
AHCTUNGTRENNUNG
able steric hindrance (Figure 2).[18] Both strategies have pros
Figure 2. Two approaches to the light-driven, coordination-induced spin-
state switching (LD-CISSS): a) the “record player” design and b) photo-
dissociable ligand (PDL) strategy.
Figure 1. Molecular orbital scheme depicting the electronic structure of
4-coordinate, square-planar NiII in the singlet state (low spin); 5-coordi-
nate, square-pyramidal NiII in the singlet state (low spin) and in the trip-
let state (high spin); and 6-coordinate square-bipyramidal NiII in the trip-
let state (high spin). Orbital energies (Eabs) are given in eV and are calcu-
lated from Ni–tetrakis(pentafluorophenyl)porphyrin 1 and pyridine at
the B3LYP/6-31G* level of DFT. For further details, see the Supporting
Information.
and cons. Record-player molecules, if properly designed,
have higher switching efficiencies; however, they undergo
intermolecular coordination at high concentrations. PDL-
based systems do not suffer from the latter restriction; how-
ever, a large excess amount of the photoswitchable ligand is
needed at low concentrations. In a preceding study, we used
Chem. Eur. J. 2012, 18, 16358 – 16368
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
16359

R. Herges et al.
symmetric azopyridines as PDLs. Since azopyridines contain
two coordinating pyridine rings, they form coordination
polyACHTUNGTRENNUNGmers with porphyrins that precipitate at high concentra-
In practice, it is particularly interesting to know which
ligand concentration would lead to a maximum switching ef-
ficiency SEmax. The optimal ligand concentration [L]opt is ob-
tained by partial differentiation of SE with respect to [L]tot
[Eq. (5)]:
tions.[18] Therefore, we set out to design phenylazopyridines
with only one coordination site that would still operate at
high concentrations. Consequently, large changes in the
magnetic susceptibility of homogeneous solutions could be
obtained, which are important for applications such as light-
controlled magnetic levitation. So far, no method other than
LD-CISSS has the potential to achieve that.
1
pffiffiffiffiffiffiffiffiffiffiffi
½Lꢁopt
¼
ð5Þ
KaKb
Surprisingly, [L]opt is independent of the porphyrin concen-
tration [P] as long as there is a large molar excess amount
of the ligand in solution ([L]tot ꢂ[L]). The maximum switch-
ing efficiency (SEmax) [%] is obtained by substitution of
[L]tot for [L]opt in Equation (4) [Eq. (6)]:
Results and Discussion
General considerations: The photodissociable ligand (PDL)
approach can be used to control any property that depends
on the coordination[22–25] or preferentially on the coordina-
tion number of a metal complex. The maximum change in a
property that can be obtained in such a system is achieved if
one were able to switch the coordination of the photodisso-
ciable ligand between complete binding and no binding at
all. In the present case, this would correspond to changing
all Ni ions in solution between diamagnetic and paramag-
netic. If we assume the following coordination equilibria
(only 1:1 complexes are formed) [Eq. (1)]:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 ꢀ Kb=Ka
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
SEmax ¼ 100
ð6Þ
1 þ Kb=Ka
SEmax only depends on the ratio of the association constants
of the photodissociable ligand (PDL) in both switching
states. For the design of an efficient PDL, it is important to
note that the switching efficiency is 100% if Kb is zero, inde-
pendent of the association constant Ka. Therefore, it is prob-
ably a reasonable strategy to put more emphasis within
ligand design to prevent the ligand in its conformation b
(Lb) from binding, rather than to optimize the binding prop-
erties of conformation a (La).
Ka
P þ L
P þ L
PL
aG
bG
H
a
ð1Þ
Kb
PL
H
b
The optimal ligand concentration ([L]opt) and the maxi-
mum switching efficiency (SEmax) are system parameters
that characterize the performance of a photodissociable
ligand with a given transition-metal complex. Equally impor-
tant in affecting the switching efficiency are the photosta-
tionary states (PSS1 and PSS2) of the PDL upon irradiation
with light of wavelengths hn1 and hn2 [Eq. (7)]:
in which P is the porphyrin (or other metal-base complex),
La is the ligand in configuration a, and Lb is the ligand in
configuration b. The switching efficiency (SE) [%] can be
defined as [Eq. (2)]:
ꢀ
ꢁ
½PLaꢁ ꢀ ½PLbꢁ
½Pꢁtot
SE ¼ 100
ð2Þ
hv1
L
L
b
aG
H
hv2
½Lꢁb
hv1 : PSS1 ¼
hv2 : PSS2 ¼
in which [PLa] is the concentration of complex PLa, [PLb] is
the concentration of complex PLb, and [P]tot is the total con-
centration of P.
½Lꢁaþ½Lꢁb
ð7Þ
½Lꢁb
½Lꢁaþ½Lꢁb
In practice, at low concentrations of P and low association
constants, a large excess amount of the ligand is needed to
enforce complexation. We therefore set [L]tot =[L] and
[P]tot =[P]+Ka[L][P], and determined the concentration of
free P and of the complex PL as [Eq. (3)]:
Incomplete conversion of the ligand from one configuration
to the other can be accounted for by setting up pseudo asso-
ciation constants K’a and K’b as linear combinations of Ka
and Kb weighted by the photostationary states PSS1 and
PSS2 [Eq. (8)]:
½Pꢁtot
K½Pꢁtot½Lꢁtot
1 þ K½Lꢁtot
ð3Þ
½Pꢁ ¼
and½PLꢁ ¼
Ka0 ¼ ð1 ꢀ PSS2ÞKa þ PSS2Kb
1 þ K½Lꢁtot
ð8Þ
Kb0 ¼ ð1 ꢀ PSS1ÞKa þ PSS1Kb
The switching efficiency can now be expressed as a function
of the total ligand concentration ([L]tot) and the association
constants (Ka, Kb), since [P]tot cancels out [Eq. (4)]:
From substitution of Ka and Kb by K’a and K’b in Equa-
tions (4) and (6), we obtain [L]opt and SEmax for systems with
incomplete photochemical conversion. Again we draw the
conclusion that an optimal PDL should exhibit a complete
conversion to the nonbinding configuration. An efficient
½LꢁtotðKa ꢀ KbÞ
ð4Þ
SE ¼ 100
ð1 þ Ka½LꢁtotÞð1 þ Kb½Lꢁtot
Þ
16360
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 16358 – 16368

Rational Design of Photodissociable Ligands
FULL PAPER
back-isomerization to the binding isomer is less important
as long as there is a large excess amount of the ligand.
other 2-substituted pyridines) is a very weak ligand.[23] The
trans configuration does not bind to Ni–porphyrins above
the detection limit (UV), and association is extremely weak
in the cis form. This is clearly due to the steric hindrance of
the substituent in the 2-position.
At higher concentrations of the porphyrin (lower ligand
to porphyrin ratios) we expect deviations from Equa-
tions (4) and (6). In these cases, nonlinear methods should
be applied to determine [L]opt and SEmax. For systems that
include more than two complex species, that is 2:1 com-
plexes, such as the Ni–porphyrin/phenylazopyridine system
presented here, an analytical solution of the underlying
mathematical problem is not possible. However, iterative
solutions based on up to six association constants that de-
scribe the formation of 1:1 and 2:1 complexes demonstrate
that [L]opt and SEmax are still independent of the porphyrin
concentration if a large excess amount of the ligand is pro-
vided (see the Experimental Section). Therefore, we consid-
er [L]opt and SEmax as defined in Equations (4) and (6) to be
important parameters for ligand design and for setting up
the optimal conditions for property switching with PDLs.
3-Phenylazopyridine is probably the most suitable starting
point for the development of photodissociable ligands. The
parent compound (similar to azobenzene) isomerizes from
trans to cis upon irradiation with UV light of 365 nm. In the
photostationary state (PSS-365), a mixture of 37% trans and
63% cis isomer is attained. Back-isomerization to the cis
isomer is achieved at 440 nm (PSS-440: 80% trans, 29%
cis). Heating of the photostationary equilibrium to 708 for
several hours yields the pure trans isomer. However, the
parent 3-phenlyazopyridine is not suitable as a photodissoci-
able ligand, because it binds to Ni–porphyrin with similar as-
sociation constants in the trans and cis configurations. This
is due to the fact that the trans and the cis isomer each can
have two conformations (trans-a, trans-b, cis-a, and cis-b;
Scheme 1, bottom). Whereas both trans conformations
(trans-a, trans-b) are able to bind, only the cis-a conforma-
tion experiences some steric repulsion with the porphyrin
framework. The cis-b conformation still binds.
In our first step towards the development of a photodisso-
ciable ligand, we concentrated on disfavoring the cis-b con-
formation in such a way that only nonbinding cis-a would
be left in the conformational equilibrium of the cis isomer.
The most straightforward way to achieve this is to increase
the steric hindrance of the pyridine and the phenyl ring by
substitution with large substituents in the 4-position of the
pyridine ring and/or in the 2,5-positions of the phenyl ring.
Ortho substitution of the rings with respect to the azo
group, moreover, has the advantage that light-induced
trans–cis as well as cis–trans isomerization usually are more
efficient than in the parent systems.[27] Seven different 4-sub-
stituted 3-phenylazopyridines (2a–g; Scheme 2) were syn-
thesized and calculated at the PBE/SVP level of density
functional theory.[28,29]
Ligand design: Phenylazopyridines are an appealing class of
compounds to develop photodissociable ligands. They com-
bine the switching capabilities of azobenzene with the coor-
dination properties of pyridine. The three regioisomers, 2-,
3-, and 4-phenylazopyridine (Scheme 1), have been investi-
Scheme 1. Three regiosiomers of phenylazopyridine in cis and trans con-
figurations and simplified binding modes to a square-planar platform
complex (e.g., Ni–porphyrin). For 3-phenylazopyridine, the two confor-
mations of both configurations are given (trans-a, trans-b, cis-a, and cis-
b).
Scheme 2. NiII–tetrakis(pentafluorophenyl)porphyrin (1) and the 4-substi-
tuted 3-phenylazopyridines 2a–g that were used as axial ligands. The
PDLs of the second generation are substituted on the phenyl ring by tert-
butyl substituents (3a and 3b).
gated by Otsuki et al. as ligands for the light-triggered lumi-
nescence modulation of Zn–porphyrins.[22,24] Preliminary
studies in combination with Ni–porphyrins[26] revealed that
4-phenylazopyridine exhibits limited photochromic proper-
ties and a fast thermal back-isomerization, which is even fur-
ther accelerated by coordination to transition-metal ions.
Moreover, extremely large substituents are needed to imple-
ment sterical hindrance.[25] 2-Phenylazopyridine (alongside
The calculated relative energies (Scheme 3) clearly show
that large substituents in the 4-position at the pyridine ring
indeed disfavor the trans-a and cis-b conformations (steric
collisions are indicated in curved bold lines in Scheme 3,
bottom). As a consequence, the binding trans-b and the non-
Chem. Eur. J. 2012, 18, 16358 – 16368
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
16361

R. Herges et al.
2d, however, decomposes after prolonged irradiation with
UV light. The corresponding reisomerization can be ach-
ieved by irradiation with visible light (440 nm) to a percent-
age of more than 80% trans in all but one case. No signifi-
cant conversion could be obtained for 2g (Table 1), proba-
bly because the n–p* and the p–p* bands have similar ab-
sorption maxima and intensity in both configurations.[23] The
thermal half-lives of all cis compounds range from 30 to
900 h in toluene at 258C. Addition of the Ni–porphyrin 1
(106 mm) reduces the half-life of cis-2c (105 mm) from 316 to
249 h in [D8]toluene at 258C. Thus, all 3-phenylazopyridines
except 2d and 2g are suitable as photoswitchable com-
pounds. However, their applicability as photodissociable li-
gands still has to be proven by the determination of their as-
sociation constants to Ni–porphyrin 1 in trans (binding) and
cis configuration (nonbinding).
Scheme 3. DFT (PBE/SVP)-calculated energies of 4-substituted 3-phenyl-
azopyridines 2a–g in their trans and cis configurations, and their a and b
conformations (trans-a, trans-b, cis-a, and cis-b). The values are in kcal
molꢀ1 relative to the a conformation of each configuration. Regions of
steric hindrance induced by the substituents are indicated with black
curves.
Determination of association constants: Previously, we have
demonstrated that the chemical shift of the pyrrol protons
of the Ni–porphyrin is an excellent probe to accurately de-
termine association constants of axial ligands.[17,18] Upon co-
ordination of one or two axial ligands, the Ni ion changes
the spin state from diamagnetic (S=0) to paramagnetic (S=
1), thus leading to a very large downfield shift of the pyrrole
protons from d=8.54 to 52.4 ppm. Ligand exchange is fast
on the NMR spectroscopic timescale. Therefore the ob-
served average chemical shift is an accurate measure of the
ratio between dia- and paramagnetic species in solution,
which was determined as reported previously.[31]
binding cis-a forms should be the predominant species in
solution, thus leading to a large discrimination of the bind-
ing affinity.
Photochemical properties of 3-azopyridines in solution: To
test our hypothesis and evaluate the photochemical proper-
ties, we synthesized the 3-phenylazopyridines 2a–g and de-
termined the photostationary states for 2a–g and 3b upon
irradiation at 365 and 440 nm (PSS-365 and PSS-440) in the
absence of Ni–porphyrin 1. Additionally, the thermal half-
life of the cis isomer was determined by NMR spectrosco-
py[30] (see Table 1; for the UV spectra, see the Supporting
Information).
We performed 1H NMR spectroscopic titration experi-
ments with 3-phenylazopyridines 2a–g and 3b. Different
amounts (10, 50, 100, 175, 350, 500, 650, 800, 1000, 1500,
2000 equiv) of the trans-azopyridines were added to a
106 mm solution of 1 in [D8]toluene. The cis to trans ratio
and the association constants were determined as reported
previously.[18] To obtain a more accurate extrapolation to the
association constants of the pure cis isomers we performed
the NMR spectroscopic titration with three different trans/
cis ratios: 1) pure trans isomer, 2) equilibrium at the PSS-
365, and 3) approximate 1:1 ratio of trans and cis isomer
(obtained from PSS-365 by partial thermal reisomerization).
For the determination of the association constants it is im-
portant to note that the pure Ni–porphyrin is the only dia-
magnetic species in solution, whereas there are five Ni–por-
phyrin complexes that are paramagnetic: the porphyrin with
one and two trans ligands, with one and two cis ligands, and
the mixed complex with one cis and one trans ligand. Hence
there are six association constants that have to be deter-
mined for each ligand (i.e., 56 association constants in
total). For each ligand, 10 titration points at three different
cis/trans ratios were determined (i.e., in total, 240 NMR
spectra were recorded). The association constants were cal-
culated by using a nonlinear titration curve fitting.[17,32,33]
The most important data to prove our concept of photodis-
sociation by switching of the steric hindrance are the associ-
ation constants of the 1:1 and 1:2 complexes of the porphyr-
in with the trans and the cis ligands (see Table 2; for a com-
Table 1. Photostationary states of 4-substituted 3-phenylazopyridines 2a–
g and 3b, and thermal half-lives of their cis isomers in the absence of Ni–
porphyrin 1.
R
PSS-365[a]
cis [%]
PSS-440[b]
cis [%]
Thermal
half-life [h][c]
2a
2b
2c
H
Me
iPr
iPr
I
63
72
75
75
86
84
90
66
91
20
17
13
14
10
13
13
62
15
228ꢃ5
100ꢃ2
316ꢃ8
249ꢃ6
115ꢃ2
206ꢃ3
915ꢃ4
31ꢃ0.5
370ꢃ6
2c+1[d]
2d[e]
2e
Ph
2 f
2g
3b
OMe
NMe2
Me, 3,5-tBu
[a] Percentage of the cis isomer in the trans/cis photostationary equilibri-
um (PSS) after irradiation at 365 nm in [D8]toluene (determined by
1H NMR spectroscopy). [b] Percentage of the cis isomer in the trans/cis
photostationary equilibrium (PSS) after irradiation at 455 nm in
[D8]toluene (determined by 1H NMR spectroscopy). [c] In [D8]toluene,
258C (determined by 1H NMR spectroscopy). [d] 105 mm 2c and 106 mm
porphyrin 1. [e] Compound 2d decomposed after several switching
cycles.
The cis-isomer concentration of all azopyridines can be in-
creased to more than 60% in the photostationary state
(PSS) upon irradiation with UV light (365 nm). Compound
16362
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 16358 – 16368

Rational Design of Photodissociable Ligands
FULL PAPER
Table 2. Association constants of porphyrin 1 with azopyridine derivates
2a–f and 3b at 298 K in [D8]toluene, given in Lmolꢀ1 (1:1 complexes)
and L2 molꢀ2 (1:2 complexes). For a definition of the association constants
(P: 1; L: 2a–f, 3b), see the footnote.[a]
change in magnetic properties of the solution upon irradia-
tion at 365 (PSS-365) and 440 nm (PSS-440) (see Table 3).
In all experiments, the concentration of the Ni–porphyrin
was 0.106 mm. The optimal concentrations of the ligands
[L]opt were calculated from the association constants (see
Table 2 and the Supporting Information) by using a nonlin-
ear approximation. The photostationary states (PSS-365 and
PSS-440) are moderate for the parent system (64 and 21%
cis) and quite close to the values determined in the absence
of the porphyrin (63 and 20% cis, Table 1). Photochemical
conversion is increasingly efficient as the size of the sub-
stituent R in the 4-position of the pyridine ring increases (72
and 17% cis for R=Me, 75 and 14% for R=iPr). The
reason for this trend becomes clear upon close inspection of
the UV/Vis spectra (see the Supporting Information). The
extinction coefficient of the n–p* band (lmax ꢂ440 nm) of
the cis isomer increases as the size of the substituent R in-
creases, because the distortion from planarity of the phenyl-
azopyridine system increases, and thus the transition is “less
forbidden”. Overall, however, the change in binding of the
ligands upon isomerization is quite low and insufficient for
an efficient switching of magnetic properties. Among the
stable and reversible systems, 2c exhibits the highest switch-
ing efficiency of only about 15%.
2a
2b
2c
2d
2e
2 f
2g[b]
3b
K1,trans
K1,cis
2.80
1.02
2.75
17.0
28.9
0.59
5.83
2.84
2.05
17.0
8.64
1.97
4.90
1.96
2.50
13.6
9.38
1.45
0.81
0.36
2.25
10.1
6.99
1.44
3.14
2.29
1.37
14.0
18.0
0.77
9.49
6.99
1.36
18.7
31.7
0.59
55.3
6.27
1.17
5.36
14.2
4.93
2.88
ratio[c]
K2,trans
K2,cis
18.7
ratio[c]
[a] K1,trans =P+Ltrans!P·Ltrans
;
K
1,cis =P+Lcis!P·Lcis
;
K2,trans =P·Ltrans +
Ltrans!P·(Ltrans)2; K2,cis =P·Lcis +Lcis!P·
G
ACHTUNGTRENNUNG
constants is given in the Supporting Information. [b] K1,cis and K2,cis of 2g
could not be determined because of inefficient photochemical conversion
from trans to cis. [c] The ratios of the association constants between the
cis and trans isomers are a measure of the switching efficiencies [see
Eq. (6)].
plete list of all 56 association constants, see the Supporting
Information).
The 1:1 association constants of the trans ligands (K1,trans
)
follow the expected trend (Table 2). Donor substituents at
the 4-position of the pyridine ring increase the binding
strength. Most efficient is the NMe2 group, which increases
binding by a factor of 20 over the parent system. As stated
in Equation (6), the maximum switching efficiency (SEmax
)
Theoretical calculations: To elucidate the reasons for the
lack of discrimination of our PDLs in bonding to the metal,
we performed further model calculations including those for
the Ni–porphyrin. To save computer time, the pentafluoro-
phenyl substituents at the porphyrin were omitted, and
PBE/SVP calculations were performed to predict the energy
of complex formation (DEf) of the trans-a, trans-b, cis-a,
and cis-b forms of the ligands 2a, 2b, 2c, and 2 f with unsub-
stituted Ni–porphyrins.
The calculated complex formation energies (DEf;
Scheme 4) reveal that the parent system 2a is the only
ligand that preferentially binds in its trans-a conformation.
As the size of the substituent R in the 4-position increases,
the trans-a conformation is disfavored because of the steric
interaction of R with the azo group. Another important and
surprising conclusion that can be drawn from the calcula-
depends on the ratio of the association constants of the
trans and the cis ligand (Ktrans/Kcis) (Table 2). Surprisingly, in-
creasing the steric hindrance at the 4-position does not im-
prove the switching efficiency systematically. The parent
system 2a with a ratio Ktrans/Kcis =2.75 is more efficient than
the methyl-substituted system 2b (2.05) or even the isoprop-
yl system 2c (2.50). The 1:2 association constants (K2,trans
,
K2,cis) follow the previously observed trend.[17] K2 generally
is larger than K1. In contrast to K1, donor substituents at the
4-position of the pyridine ring have little effect on K2. In the
case of the very strong donor ligand 2g (R=NMe2), K1 is
larger than K2.
In addition to the association constants of the trans and
cis isomers, we also quantified the photochemically induced
conversions of ligands 2a–g in the presence of 1 and the
Table 3. Photochemical and magnetic switching properties of the Ni–porphyrin/phenylazopyridine systems (1/2a–g) at the photostationary states PSS-
365 and PSS-440. A concentration of 0.106 mm 1 in [D8]toluene at 258C was used in all experiments.
[d]
[d]
% cis isomer[a]
% paramagnetic Ni2+[b]
PSS-365 PSS-440
[L]opt
[equiv][c]
SEmax
Lexptl
ACHTUNGTRENNUNG
[equiv][e]
SEexptl
PSS-365
PSS-440
U
2a
2b
2c
2d
2e
2 f
2g
3b
64
72
75
81
84
90
63
91
21
17
14
10
13
23
62
15
28.2 35.9
1230
902
1175
3939
862
182
n.d.
1586
7.3
11.2
15.2
19.1
2.7
1.4
n.d.
40.8
1000
800
1000
2000
800
175
1000
1500
7.7
10.9
14.9
n.d.
2.7
1.2
0.6
40.1
44.7
40.8
55.6
55.7
decomp
36.3
decomp
39.0
18.4
19.6
91.0
90.4
71.9
31.8
[a] Determined by 1H NMR spectroscopy. [b] Determined from the 1H NMR spectroscopic shifts of the pyrrole protons.[18] [c] Theoretically calculated
optimal ligand concentration in equivalents with respect to 1. [d] The experimental values of the switching efficiencies (SEexptl) are compared with theo-
retically calculated maximum switching efficiencies (SEmax) determined from the binding constants from Table 2 and the Supporting Information.
[e] Ligand concentration used in the experiment.
Chem. Eur. J. 2012, 18, 16358 – 16368
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
16363

R. Herges et al.
Scheme 4. DFT (PBE/SVP)-computed complex formation energies (DEf)
of the complexes of phenylazopyridines 2a–c and 2 f with parent Ni–por-
phyrin. DEf is defined as the energy that is released upon formation of
the complexes (structures at the bottom) starting from the most stable
conformation of each configuration. The pentafluorophenyl substituents
at the porphyrin meta positions were omitted to save computational cost.
The energy differences between the strongest binding trans conformation
and the strongest binding cis conformation of each ligand are indicated
by the arrows [kcalmolꢀ1].
Scheme 5. Computed complex formation energies DEf (PBE/TZVP//
B3LYP/TZVP) including a complete basis-set extrapolation (see the Sup-
porting Information) of 2b, 3a, and 3b with Ni–TPFPP (1).
Table 4. Complex formation energies (DEf) of 2a, 3a, and 3b with Ni–
porphyrin 1 calculated at the PBE/TZVP//PBE/1ZVP level of theory.
DEf [kcalmolꢀ1] of 1 with ligand
DDEf[a]
tions is the fact that the cis-a conformations still bind to the
Ni–porphyrin quite well, and that the steric repulsion be-
tween the phenyl ring and the porphyrin is less severe than
initially assumed from simple model considerations. Howev-
er, the cis-b conformations follow the intended trend. This
conformation is strongly disfavored with increasing size of
the substituent R (Scheme 4). From these results we con-
clude that the methyl- and isopropyl-substituted systems 2b
and 2c are good starting points for further improvement of
the switching efficiency. However, to achieve this goal, the
cis-a conformation has to be modified to prevent it from
binding to the porphyrin. Thus, either steric crowding at the
porphyrin unit or at the ligand has to be increased. Unfortu-
nately, the pentafluorophenyl substituents at the meso posi-
tions of the porphyrin cannot be easily replaced by larger
groups (e.g., mesityl) because electron-withdrawing substitu-
ents are required for effective spin switching. The most
straightforward strategy is to increase steric hindrance at the
phenyl ring of the ligand. To find an optimal candidate and
to gain insight into the steric interactions we performed fur-
ther theoretical calculations. At the phenyl ring of our li-
gands, substitution at the ortho, meta, or para position could
provide additional steric repulsion. Ortho substitution is dif-
ficult to achieve synthetically. Therefore, we considered tBu
substituents in the meta and para position for preliminary
model calculations (Scheme 5, Table 4).
trans-a
trans-b
cis-a
cis-b
2a
3a
3b
ꢀ4.24
ꢀ4.28
ꢀ3.84
ꢀ4.68
ꢀ4.81
ꢀ4.29
ꢀ3.87
ꢀ3.34
ꢀ1.45
ꢀ3.72
ꢀ3.73
ꢀ2.87
0.81
1.08
1.42
[a] Energy difference between DEf of the most stable cis and the most
stable trans complex.
with 4-substituted pyridines in previous studies.[17] Even with
the large TZVP basis set, a considerable basis set superposi-
tion error (BSSE) can be expected upon calculation of the
binding energy from the separate components, thus leading
to an overestimation of the complex formation energy. Un-
fortunately, the usual counterpoise method[35,36] cannot be
applied to account for this error because the Ni2+ changes
spin state upon complex formation from singlet to triplet.
Therefore, we performed a complete basis set extrapolation
according to the Helgaker scheme[37] to account for the
BSSE (see the Supporting Information). At this level of
theory, quite accurate complex formation energies can be
expected (Table 4).
The calculated complex formation energies (DEf) confirm
the expected fact that neither the large tBu substituent in
the para position of the phenyl ring in 3a, nor the two tBu
substituents in the meta position in 3b affect the complex
formation of the trans-b conformation. Binding of the cis-a
conformation, however, is strongly disfavored by the two
tBu groups in the meta position in 3b. A single tBu group in
the para position (3a) is considerably less effective in this
respect. Moreover, the tBu groups in 3b also lower the bind-
ing strength of the cis-b conformation by steric interaction
with the methyl group at the pyridine ring. We therefore
chose 3b as the target molecule for the synthesis of an im-
All calculations were performed with the TURBOMOLE
program.[34] The structures were optimized at the Perdew–
Burke–Ernzerhof (PBE)/TZVP level of density functional
theory. Single-point energy calculations were performed at
the PBE-optimized geometries at the B3LYP/TZVP level.
This combination of functionals and the basis set proved to
provide quite accurate binding energies in Ni–porphyrins
16364
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 16358 – 16368

Rational Design of Photodissociable Ligands
FULL PAPER
proved photodissociable ligand. Compound 3b was prepared
by the oxidation of 3,5-di-tert-butylaniline to the corre-
sponding nitroso compound, and subsequent condensation
with 3-amino-4-methylpyridine (for details, see the Experi-
mental Section).
Switching experiments with improved PDL 3b: To check
the success of our design, we performed photochemical
switching experiments with solutions of 1 ([1]total =0.106 mm)
in the presence of the optimized PDL 3b (159 mm) in
[D8]toluene at 258C, and compared the switching efficiency
with the parent system 2a (130 mm). The samples were irra-
diated in an alternate sequence using light-emitting diodes
with wavelengths of 365 (Nichia NC4U133) and 440 nm
(Roithner VL-440-EMITTER) in an NMR spectroscopy
tube until the PSS was reached. Figure 3 shows that the
Figure 4. Speciation plot (relative concentrations with respect to [1]total
)
of 1, and all axial complexes of 1 as a function of the concentration of
3b. For 3b, a ratio of cis and trans configurations corresponding to PSS-
365 was assumed (91% cis, 9% trans). The gray bar indicates the concen-
tration of 3b (158 mm) at which the switching experiments were per-
formed ([1]total =106 mm). The relative concentrations of all Ni-complex
species at this point are given on the right.
Figure 5. Speciation plot (relative concentrations with respect to [1]total
)
of 1, and all axial complexes of 1 as a function of the concentration of
3b. For 3b, a ratio of cis and trans configurations corresponding to PSS-
440 was assumed (15% cis, 85% trans). The gray bar indicates the con-
centration of 3b (158 mm) at which the switching experiments were per-
formed ([1]total =106 mm). The relative concentrations of all Ni-complex
species at this point are given on the right.
Figure 3. Reversible switching of the spin state of 1 in solution ([1]tot
=
0.106 mm) in the presence of an excess amount of azopyridine 3b
(159 mm) in [D8]toluene at 258C by alternate irradiation with 365 and
440 nm. The molar magnetic susceptibility of the solution cPM is given in
10ꢀ6 cm3 molꢀ1. The predominant species at both photostationary states
are given on the right. The corresponding values of the less-efficient
parent system (ligand 2a) are indicated in gray.
porting Information), the concentration of each of the six
Ni2+ complex species can be calculated as a function of the
ligand concentration (speciation plots, Figures 4 and 5). At
the concentration of 3b used in the experiment, which was
close to the optimal ligand concentration ([L]opt
, see
Table 3), the predominant complex at PSS-365 in solution is
the diamagnetic bare Ni–porphyrin 1 (69%; Figure 4). At
PSS-440, the paramagnetic 1 with two trans-3b molecules as
axial ligands prevails (45%; Figure 5).
switching process is fully reversible and does not show any
fatigue. Upon irradiation with UV light at 365 nm, 91%
conversion of 3b to the cis isomer is achieved. At this pho-
tostationary state, 32% of the Ni ions in solution are para-
magnetic. This corresponds to a molar magnetic susceptibili-
ty of cMP =1143ꢃ10ꢀ6 cm3 molꢀ1. Irradiation at 440 nm leads
to a photostationary state with 15% cis and 85% trans
isomer. Of the Ni species, 72% are paramagnetic, and the
solution attains a molar magnetic susceptibility of cPM =
2584ꢃ10ꢀ6 cm3 molꢀ1. Hence the magnetic susceptibility of
the solution can be switched by a factor of more than two!
According to Equation (5) there is an optimum ligand
concentration ([L]opt) for which a maximum switching effi-
ciency (SEmax) is attained. Both [L]opt and SEmax are inde-
pendent of the porphyrin concentration if there is a large
excess amount of the ligand ([L]tot ꢂ[L]). However, an ana-
lytical solution of the problem was only possible for 1:1
complexes. By using the six experimentally determined asso-
ciation constants of 3b with 1 (Table 2, see the Supporting
Information) and a nonlinear treatment of the binding iso-
therms, we now checked if our assumption that [L]opt and
SEmax are independent of the porphyrin concentration were
still correct in more complicated systems with mixtures of
1:1 and 1:2 complexes. The results of our simulation
(Figure 6) show that in a concentration range from 0 to
about 5 mm of 1, both [L]opt and SEmax are constant. Practi-
cal concentrations in switching experiments and the concen-
The experimentally determined switching efficiency (SEexptl
)
of 3b of 40.1% is close to the theoretical value (SEmax
=
40.8%) and is far superior to all other ligands that lack the
tBu groups at the phenyl ring (SEexptl: 2a: 7.7%, 2b: 10.9%,
2c: 14.9%; see Table 3).
From the six 1:1 and 1:2 association constants determined
by NMR spectroscopic titration (see Table 2 and the Sup-
Chem. Eur. J. 2012, 18, 16358 – 16368
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
16365

R. Herges et al.
sponding PDL (3b) has been synthesized and successfully
tested. Upon irradiation at 365 and 440 nm, a switching effi-
ciency of 40.1% was attained for the reversible conversion
of the Ni2+ center from high to low spin. Thus, the magnetic
susceptibility of the homogeneous solution can be switched
by a factor of more than two. Further studies are devoted to
the development of stronger binding PDLs that would ex-
hibit SEmax at lower concentrations, for example, azoimid-
AHCTUNGTRENNUNG
azoles.[38]
Experimental Section
General remarks: Solvents and starting materials were used as received.
Column chromatography was carried out using 0.04–0.06 mm mesh silica
gel from Merck. NMR spectra were recorded at 298 K using a 300 MHz
(75.5 MHz for 13C) Bruker ARX 300, 500 MHz (125 MHz for 13C)
Bruker DRX 500, or a 600 MHz (150 MHz for 13C) Bruker AV 600.
Chemical shifts were calibrated either to the internal standard TMS or
using residual protonated solvent signals (1H (CHCl3): d =7.24 ppm, (tol-
uene): d=2.04 ppm; and 13C (CHCl3): d =77.0 ppm). For the assignment
of the NMR spectroscopic signals and the numbering of the correspond-
ing carbon atoms, see the Supporting Information. Mass spectrometry
was performed using a Finnigan MAT 8230 (EI, 70 eV) and MAT 8200
(CI, isobutane) instrument. Infrared spectra were recorded using
a
Perkin–Elmer ATR spectrometer with a Golden-Gate-Diamond-ATR
A531-G for neat samples. UV-visible absorption spectra were recorded
using a Perkin–Elmer Lambda-14 spectrophotometer with quartz cells of
1 cm path length. All solvents employed in optical spectroscopy were of
spectrophotometric grade. Irradiation experiments were performed in
toluene. For the isomerization of trans- to cis-azopyridine, the samples
were irradiated using an LED (Nichia NC4U133(T), peak wavelength:
(365ꢃ9) nm, three LEDs, power dissipation: 12 W each, luminous flux:
10 lm each, distance ꢂ1 cm). The conversion of cis- to trans-azopyridine
was performed using an LED from Roithner Laser Technik GmbH (VL
440-EMITTER, peak wavelength: (440ꢃ5) nm, three LEDs, power dissi-
pation: 1.3 W each, distance ꢂ1 cm).
Figure 6. Maximum switching efficiency (SEmax; blue curve) and optimal
ligand concentration [L]opt of 3b as a function of [1]total (top: 0–50 mm of
[1]total). The range with 0–1 mm of [1]total of the plot is magnified in the
bottom figure. SEmax and [L]opt are independent of [1]total within a range
of 0 to 5 mm of [1]total. The gray bar indicates the concentration of [1]total
used in our experiments.
trations used in our experiments are well within this linear
range.
Synthesis of starting materials: Nickel–5,10,15,20-tetrakis(pentafluoro-
phenyl)porphyrin (Ni–TPFPP) (1) was prepared as described previous-
ly.[17]
Conclusion
3-Phenylazopyridine (2a): Prepared by the method of Campbell et al. 1H
and 13C NMR spectroscopic and MS analytical data are consistent with
those reported in the literature.[39] trans isomer: 1H NMR (300 MHz,
[D8]toluene, 300 K): d=9.29 (d, J=2.4 Hz, 1H; 1-H), 8.39 (dd, J=4.7,
1.5 Hz, 1H; 5-H), 7.85–7.82 (m, 2H; 7-H), 7.76 (ddd, J=8.2, 2.4, 1.7 Hz,
1H; 3-H), 7.13–7.08 (m, 3H; 8-H, 9-H), 6.68 ppm (dd, J=8.2, 4.7 Hz 1H;
4-H). cis isomer: 1H NMR (300 MHz, [D8]toluene, 300 K): d=8.03 (d,
J=2.5 Hz, 1H; 1-H), 8.01 (dd, J=4.8, 1.5 Hz, 1H; 5-H), 6.74–6.70 (m,
2H; 8-H), 6.66–6.61 (m, 1H; 9-H), 6.54 (ddd, J=8.1, 2.5, 1.6 Hz, 1H; 3-
H), 6.44–6.41 (m, 2H; 7-H), 6.36 ppm (ddd, J=8.0, 4.8, 0.8 Hz, 1H; 4-H).
Photodissociable ligands (PDLs) are an efficient approach
to switch physicochemical properties of metal complexes.
The ligands have to be designed in such a way that one con-
figuration binds to the metal complex and the other isomer
does not. The maximum attainable switching efficiency
(SEmax) only depends on the ratio of the association con-
stants of both configurations of the ligand. There is an opti-
mal ligand concentration ([L]opt) that leads to the SEmax in-
dependent of the concentration of the metal complex. [L]opt
and SEmax are system parameters that characterize the com-
bination of a PDL and a metal complex. The strategy that
can be derived for the molecular design of a PDL is that
more emphasis should be given to reduce the binding of the
low-affinity isomer than to optimize the binding properties
of its high-affinity counterpart. Effective PDLs based on 3-
phenylazopyridine should have both a large substituent at
the 4-position of the pyridine ring to disfavor formation of
the cis-a conformation and two substituents at the 3,5-posi-
tions of the phenyl ring are needed to prevent the cis-b con-
formation from binding to the metal complex. A corre-
4-Methyl-3-(phenylazo)pyridine (2b): 3-Amino-4-methylpyridine (2.29 g,
21.2 mmol), tetramethylammonium hydroxide (57.0 mL, 25%), and pyri-
dine (25 mL) were heated to 808C. Nitrosobenzene (3.00 g, 28.0 mmol)
was dissolved in pyridine (50 mL) and added dropwise. After stirring at
808C for 45 min. the reaction mixture was allowed to cool down to RT.
The crude product was extracted with plenty of toluene. The organic
layer was dried over magnesium sulfate and after removal of the solvent,
column chromatography on silica gel (cyclohexane, ethyl acetate 1:1, Rf =
0.58) afforded
a red solid (146 mg, 0.740 mmol, 3%). M.p. 62.88C;
1H NMR (600 MHz, CDCl3): d=8.73 (s, 1H; 1-H), 8.51 (d, J=5.0 Hz,
1H; 5-H), 7.93 (dd, J=8.2, 1.6 Hz, 2H; 7-H), 7.55–7.48 (m, 3H; 8-H, 9-
H), 7.26 (d, J=5.0 Hz, 1H; 4-H), 2.70 ppm (s, 3H; CH3); 13C NMR
(150.9 MHz, CDCl3): d=152.8 (C-6), 150.8 (C-5), 146.5 (C-2), 145.2 (C-
3), 138.0 (C-1), 131.5 (C-9), 129.1 (C-8), 125.9 (C-4), 123.1 (C-7),
ꢀ
ꢀ
17.1 ppm (CH3); IR (KBr): n˜ =3039 (C Harom), 2960, 2921 (C Haliph),
1593, 1466, 1441, (C=C), 833, 769, 726, 688 cmꢀ1 (C Hdef); MS (EI): m/z
ꢀ
16366
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 16358 – 16368

Rational Design of Photodissociable Ligands
FULL PAPER
(%): 197 (49) [M]+, 120 (3) [MꢀPh]+, 105 (25) [PhN2]+, 92 (45)
[MꢀPhN2]+, 77 (100) [Ph]+; MS (CI): m/z (%): 198 (100) [M+H]+; ele-
mental analysis calcd (%) for C12H11N3 (197.26): C 73.07, H 5.62, N
5.2 Hz, 1H; 4-H), 6.72–6.67 (m, 2H; 8-H), 6.63–6.61 (m, 1H; 9-H), 6.60–
6.58 ppm (m, 2H; 7-H).
4-Phenyl-3-phenylazopyridine (2e): Prepared by the method of Otsuki
1
1
21.30; found: C 73.39, H 5.85, N 21.19. trans isomer: H NMR (300 MHz,
et al. H NMR and MS analytical data are consistent with those reported
in the literature.[23] trans isomer: 1H NMR (500 MHz, [D8]toluene,
300 K): d=9.05 (d, J=0.5 Hz, 1H; 1-H), 8.48 (d, J=5.0 Hz, 1H; 5-H),
7.73–7.70 (m, 2H; 7-H), 7.22–7.18 (m, 2H; 11-H), 7.12–7.07 (m, 3H; 12-
H, 13-H), 7.04–6.98 (m, 3H; 8-H, 9-H), 6.92 ppm (dd, J=5.0, 0.6 Hz, 1H;
4-H). cis isomer: 1H NMR (500 MHz, [D8]toluene, 300 K): d=8.14 (d,
J=5.1 Hz, 1H; 5-H), 7.88 (d, J=0.6 Hz, 1H; 1-H), 6.69–6.65 (m, 3H),
6.63–6.60 (m, 3H), 6.41–6.38 ppm (m, 2H). Missing signals might have
been disguised by signals of the trans isomer or the solvent.
[D8]toluene, 300 K): d=8.96 (s, 1H; 1-H), 8.31 (d, J=4.5 Hz, 1H; 5-H),
7.78 (d, J=7.4 Hz, 2H; 7-H), 7.13–7.02 (m, 3H; 8-H, 9-H), 6.59 (d, J=
4.8 Hz, 1H; 4-H), 2.28 ppm (s, 3H; CH3). cis isomer: H NMR (300 MHz,
[D8] toluene, 300 K): d=7.97 (d, J=5 Hz, 1H; 5-H), 7.49 (s, 1H; 1-H),
6.75–6.70 (m, 2H; 8-H), 6.66–6.59 (m, 1H; 9-H), 6.50–6.48 (m, 2H; 7-H),
6.36 (d, J=4.5 Hz, 1H; 4-H), 1.85 ppm (s, 3H; CH3).
1
4-Isopropyl-3-phenylazopyridine (2c): 2,2-Dimethyl-n-(3-pyridyl)propane
amide was synthesized by reaction of 3-aminopyridine with trimethylace-
tyl chloride (64%).[40] A Grignard reaction with isopropyl magnesium
chloride led to 2,2-dimethyl-n-(4-isopropyl-3-pyridyl)propane amide
(46%).[18,41] Deprotection with sulfuric acid gave 3-amino-4-isopropyl-
pyridine (77%).[18] 3-Amino-4-isopropylpyridine (2.00 g, 14.7 mmol),
sodium hydroxide (50 mL, 25%), and pyridine (25 mL) were heated to
808C. Nitrosobenzene (2.04 g, 19.1 mmol) was dissolved in pyridine
(50 mL) and added dropwise within 45 min. After stirring at 808C for
2 h, the reaction mixture was allowed to cool to RT. The crude product
was extracted with plenty of toluene. The organic layer was dried over
magnesium sulfate, and after removal of the solvent, column chromatog-
raphy on silica gel (cyclohexane, ethyl acetate 1:1, Rf =0.65) afforded an
orange oil (1.20 g, 5.30 mmol, 36%). 1H NMR (500 MHz, CDCl3): d=
8.70 (s, 1H; 1-H), 8.59 (d, J=5.2 Hz, 1H; 5-H), 7.95–7.93 (m, 2H; 9-H),
7.52 (mc, 3H; 10-H, 11-H), 7.36 (dd, J=5.2, 0.4 Hz, 1H; 4-H), 3.99
3-Amino-4-methoxypyridine:[45]
4-Methoxy-3-nitropyridine
(2.00 g,
13.0 mmol) and anhydrous tin(II) chloride (17.3 g, 91.0 mmol) were dis-
solved in ethanol (50 mL) and acetic acid (7 mL, 2 n) and heated to
reflux for 3 h. Ice-cooled water and potassium hydroxide (140 g) were
added. The hot layers could be separated and the water layer was ex-
tracted with dichloromethane. The combined organic layers were dried
over magnesium sulfate. After removal of the solvent, a yellow solid
(1.48 g, 11.9 mmol, 92%) was afforded without further purification. M.p.
828C; 1H NMR (500 MHz, CDCl3): d=7.99 (s, 1H; 1-H), 7.96 (d, J=
5.4 Hz, 1H; 5-H), 6.68 (d, J=5.4 Hz, 1H; 4-H), 3.88 (s, 3H; CH3),
3.72 ppm (brs, 2H; NH2); 13C NMR (125.8 MHz, CDCl3): d=152.9 (C-3),
141.6 (C-1), 136.5 (C-5), 132.8 (C-2), 105.5 (C-4), 55.3 ppm (CH3); IR
ꢀ
ꢀ
ꢀ
(KBr): n˜ =3401, 3331 (N H), 3166, 3089 (C Harom), 2999, 2935 (C Haliph),
1574, 1511, 1426, (C=C), 1232, 1025, 856, 806, 767 cmꢀ1 (C Hdef); MS
ꢀ
(EI): m/z (%): 124 (100) [M]+, 109 (35) [MꢀCH3]+; MS (CI): m/z (%):
125 (100) [M+H]+; elemental analysis calcd (%) for C6H8N2O (124.06):
C 58.05, H 6.50, N 22.57; found: C 57.91, H 6.63, N 21.59.
(septet, J=7.0 Hz, 1H; CHACTHNUTRGNEUN(G CH3)2), 1.35 ppm (d, J=7.0 Hz, 6H; CH3);
13C NMR (125.8 MHz, CDCl3): d=154.9 (C-3), 152.9 (C-8), 151.5 (C-5),
145.9 (C-2), 137.6 (C-1), 129.2 (C-10), 129.0 (C-11), 123.1 (C-9), 121.1 (C-
ꢀ
4), 27.9 (C-6), 23.0 ppm (C-7); IR (KBr): n˜ =3052 (C Harom), 2929, 2870
4-Methoxy-3-phenylazopyridine
(2 f):
3-Amino-4-methoxypyridine
(C Haliph), 1587, 1466, 1402, (C=C), 833, 757, 686 cmꢀ1 (C Hdef); MS
(EI): m/z (%): 225 (29) [M]+, 210 (100) [MꢀCH3]+, 133 (59)
[MꢀPhCH3]+, 105 (7) [PhN2]+; MS (CI): m/z (%): 226 (100) [M+H]+,
133 (48) [MꢀPhCH3]+; elemental analysis calcd (%) for C14H15N3
(225.13): C 74.64, H 6.71, N 18.65; found: C 74.60, H 6.68, N 19.02. trans
isomer: 1H NMR (500 MHz, [D8]toluene, 300 K): d=8.97 (s, 1H; 1-H),
8.44 (d, J=5.1 Hz, 1H; 5-H), 7.81–7.78 (m, 2H; 9-H), 7.14–7.09 (m, 2H;
10-H), 7.07–7.04 (m, 1H; 11-H), 6.81 (d, J=5.1 Hz, 1H; 4-H), 3.83
ꢀ
ꢀ
(1.00 g, 8.06 mmol) was dissolved in pyridine (4.00 mL) and sodium hy-
droxide (6.00 mL, 75.0 mmol, 25%) and heated to 808C. Within 45 min,
nitrosobenzene (1.15 g, 15.8 mmol) dissolved in pyridine (30 mL) was
added dropwise. The reaction mixture was stirred at 808C for an addi-
tional 45 min. After cooling, the water layer was extracted with toluene
and dried over magnesium sulfate. After removal of the solvent, purifica-
tion by column chromatography on silica gel (ethyl acetate/triethylamine
(5%), Rf =0.4) and recrystallization from diethyl ether afforded red crys-
tals (611 mg, 2.87 mmol, 36%). M.p. 80.48C; 1H NMR (500 MHz,
CDCl3): d=8.66 (s, 1H; 1-H), 8.54 (d, J=5.8 Hz, 1H; 5-H), 7.91 (m, 2H;
7-H), 7.50 (m, 3H; 8-H, 9-H), 4.01 ppm (s, 3H; CH3); 13C NMR
(150.9 MHz, CDCl3): d=161.8 (C-3), 153.0 (C-5), 152.97 (C-2), 139.1 (C-
1), 138.3 (C-6), 134.4 (C-9), 129.2 (C-8), 123.1 (C-7), 107.9 (C-4),
56.2 ppm (CH3); IR (KBr): n˜ =1582, 1567, 1491, 1439, 1470 (C=C), 1016,
(septet, J=7.0 Hz, 1H; CHACTHNUTRGNEUN(G CH3)2), 1.07 ppm (d, J=7.0 Hz, 6H; CH3).
cis isomer: 1H NMR (500 MHz, [D8]toluene, 300 K): d=8.08 (d, J=
5.2 Hz, 1H; 5-H), 7.45 (s, 1H; 1-H), 6.79–6.74 (m, 2H; 9-H), 6.69–6.60
(m, 4H; 4-H, 10-H, 11-H), 3.10 (septet, J=6.9 Hz, 1H; CH
ACHTUGNRTNE(NUNG CH3)2),
1.00 ppm (d, J=6.9 Hz, 6H; CH3).
4-Iodo-3-phenylazopyridine (2d):[42] Lithiation of 2,2-dimethyl-n-(3-pyri-
dyl)propane amide and reaction with iodine followed by deprotection of
the amine with sulfuric acid afforded 3-amino-4-iodopyridine.[41,43,44] 3-
Amino-4-iodopyridine (1.00 g, 4.55 mmol) and nitrosobenzene (490 mg,
4.55 mmol) were dissolved in pyridine (12 mL) and sodium hydroxide
(8 mL, 10 n), then stirred at RT for 16 h. The reaction mixture was dilut-
ed with water (30 mL) and the organic layer was extracted with toluene.
The combined organic layers were dried over magnesium sulfate. After
removal of the solvent, purification by column chromatography on silica
gel (ethyl acetate, Rf =0.59) afforded a red solid (900 mg, 2.92 mmol,
64%). M.p. 86.18C; 1H NMR (600 MHz, CDCl3): d=8.65 (s, 1H; 1-H),
8.21 (d, J=5.2 Hz, 1H; 5-H), 8.00 (dd, J=7.9, 1.8 Hz, 2H; 7-H), 7.96 (d,
J=5.2 Hz, 1H; 4-H), 7.56–7.51 ppm (m, 3H; 8-H, 9-H); 13C NMR
(150.9 MHz, CDCl3): d=152.3 (C-2), 150.9 (C-5), 147.4 (C-6), 138.7 (C-
1), 136.6 (C-4), 132.2 (C-9), 129.3 (C-8), 123.7 (C-7), 111.5 ppm (C-3); IR
812 cmꢀ1 (C Hdef); MS (EI): m/z (%): 213 (100) [M]+, 105 (57)
ꢀ
[MꢀC6H6NO]+; MS (CI): m/z (%): 214 (100) [M+H]+; elemental analy-
sis calcd (%) for C12H11N3O (213.09): C 67.59, H 5.20, N 19.71; found: C
67.44, H 5.32, N 19.56. trans isomer: 1H NMR (300 MHz, [D8]toluene,
300 K): d=8.91 (s, 1H; 1-H), 8.28 (d, J=5.7 Hz, 1H; 5-H), 7.91–7.86 (m,
2H; 7-H), 7.00–7.10 (m, 3H; 8-H, 9-H), 5.73 (d, J=5.7 Hz, 1H; 4-H),
3.18 ppm (s, 3H; OCH3); cis isomer: 1H NMR (300 MHz, [D8]toluene,
300 K): d=7.99 (d, J=5.6 Hz, 1H; 5-H), 7.84 (s, 1H; 1-H), 6.78–6.71 (m,
2H; 7-H), 6.68–6.59 (m, 3H; 8-H, 9-H), 5.83 (d, J=5.6 Hz, 1H; 4-H),
2.83 ppm (s, 3H; OCH3).
4-N,N-Dimethylamino(3-phenylazo)pyridine (2g): Prepared by the reduc-
tion of 3-nitro-4-dimethylaminopyridine and reaction of the amine with
nitrosobenzene. 1H and 13C NMR spectroscopy and MS analytical data
are consistent with those reported in the literature.[23] trans isomer:
1H NMR (600 MHz, [D8]toluene, 300 K): d=8.93 (s, 1H; 1-H), 8.17 (d,
J=6.0 Hz, 1H; 5-H), 7.73 (mc, 2H; 7-H), 7.13 (mc, 3H; 8-H, 9-H), 6.11
(d, J=6.0 Hz, 1H; 4-H), 2.58 ppm (s, 6H; N
(600 MHz, [D8]toluene, 300 K): d=7.89 (d, J=5.8 Hz, 1H; 5-H), 7.41 (s,
1H; 1-H), 6.83–6.79 (m, 2H; 7-H), 6.74–6.65 (m, 3H; 8-H, 9-H), 5.94 (d,
J=5.8 Hz, 1H; 4-H), 2.43 ppm (s, 6H; N
ꢀ
ꢀ
(KBr): n˜ =3037 (C Harom), 1544, 1411, (C=C), 1051 (C Iarom), 818, 770,
720, 682 cmꢀ1 (C Hdef); MS (EI): m/z (%): 309 (88) [M]+, 204 (24)
ꢀ
1
[MꢀPhN2]+, 105 (100) [PhN2]+; MS (CI): m/z (%): 310 (100) [M+H]+,
184 (3) [MꢀI+H]+, 105 (7) [PhN2]+; elemental analysis calcd (%) for
C11H8IN3 (309.12): C 42.74, H 2.61, N 13.59; found: C 43.08, H 2.62, N
13.87. trans isomer: 1H NMR (300 MHz, [D8]toluene, 300 K): d=8.64 (d,
J=0.5 Hz, 1H; 1-H), 7.87–7.84 (m, 2H; 7-H), 7.74 (d, J=5.1 Hz, 1H; 5-
H), 7.23 (dd, J=5.2 Hz, 0.5 Hz, 1H; 4-H), 7.11–7.07 (m, 2H; 8-H), 7.05–
7.02 ppm (m, 1H; 9-H). cis isomer: 1H NMR (300 MHz, [D8]toluene,
300 K): d=7.42 (d, J=5.2 Hz, 1H; 5-H), 7.17 (s, 1H; 1-H), 7.02 (d, J=
(CH3)2). cis isomer: H NMR
ACHTUGNRTNE(NUNG CH3)2).
4-Methyl-3(3’,5’di-tert-butylphenyl)azopyridine (3b):
A
solution of
Oxone (6.00 g, 9.76 mmol) in water (50 mL) was added to a solution of
3,5-di-tert-butylaniline (1.00 g, 4.88 mmol) in dichloromethane (20 mL).
After stirring for 4 h at room temperature, the layers were separated.
Chem. Eur. J. 2012, 18, 16358 – 16368
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
16367

R. Herges et al.
The aqueous layer was extracted twice with dichloromethane. The com-
bined organic layers were dried over magnesium sulfate. After removal
of the solvent, column chromatography on silica gel (CH2Cl2, Rf =0.3
pale green fraction) afforded a green solid assigned as 3,5-di-tert-butylni-
trosobenzene (851 mg, 3.89 mmol, 80%). 1H NMR (500 MHz, CDCl3):
d=7.83 (t, J=1.9 Hz, 1H; 4-H), 7.78 (d, J=1.9 Hz, 2H; 2-H), 1.40 ppm
(s, 18H; CH3); 13C NMR (125.8 MHz, CDCl3): d=167.1 (C-1), 152.4 (C-
[13] Y. Hasegawa, S. Kume, H. Nishihara, Dalton Trans. 2009, 280–284.
[14] M. M. Paquette, B. O. Patrick, N. L. Frank, J. Am. Chem. Soc. 2011,
133, 10081–10093.
[15] M. M. Paquette, R. A. Kopelman, E. Beitler, N. L. Frank, Chem.
Commun. 2009, 5424–5426.
[16] For a review on the switching of magnetic properties through exter-
nal stimuli, see: O. Sato, J. Tao, Y.-Z. Zhang, Angew. Chem. 2007,
119, 2200–2236; Angew. Chem. Int. Ed. 2007, 46, 2152–2187.
[17] S. Thies, C. Bornholdt, F. Kçhler, F. D. Sçnnichsen, C. Nꢁther, F.
Tuczek, R. Herges, Chem. Eur. J. 2010, 16, 10074–10083.
[18] S. Thies, H. Sell, C. Schꢀtt, C. Bornholdt, C. Naether, F. Tuczek, R.
Herges, J. Am. Chem. Soc. 2011, 133, 16243–16250.
3), 129.7 (C-2), 115.9 (C-4), 35.1 (CACHTNURTGENNUG(CH3)3), 31.2 ppm (CCAHUTNGTREN(NUGN CH3)3); IR
ꢀ
ꢀ
(KBr): n˜ =2962 (C Harom), 2904, 2869 (C Haliph), 1597, 1535, 1478, 1460,
1449 (C=C), 1363, 1312, 1246, 885, 699 cmꢀ1 (C Hdef); MS (EI): m/z (%):
ꢀ
219 (66) [M]+, 189 (92) [MꢀNO]+, 133 (100) [MꢀNOꢀtBu]+; MS (CI):
m/z (%): 220 (100) [M+H]+.
[19] S. Venkataramani, U. Jana, M. Dommaschk, F. D. Sçnnichsen, F.
Tuczek, R. Herges, Science 2011, 331, 445–448.
[20] F. A. Cotton, G. Wilkinson, C. A. Murillo, M. Bochmann, Advanced
Inorganic Chemistry, Wiley Interscience, New York, 1999, 6th ed..
[21] M. Schmittel, S. De, S. Pramanik, Angew. Chem. 2012, 124, 3898–
3902; Angew. Chem. Int. Ed. 2012, 51, 3832–3836.
[22] J. Otsuki, K. Narutaki, J. M. Bakke, Chem. Lett. 2004, 33, 356–357.
[23] J. Otsuki, N. Narutaki, Bull. Chem. Soc. Jpn. 2004, 77, 1537–1544.
[24] J. Otsuki, T. Akasaka, K. Araki, Coord. Chem. Rev. 2008, 252, 32–
56.
[25] K. Suwa, J. Otsuki, K. Goto, Tetrahedron Lett. 2009, 50, 2106–2108.
[26] C. Bornholdt, Dissertation, CAU Kiel, 2008.
3-Amino-4-methylpyridine (378 mg, 3 mmol) was dissolved in pyridine
(5 mL) and 60% potassium hydroxide (15 mL) and was heated to 1008C.
A solution of 3,5-di-tert-butylnitrosobenzene (851 mg, 3.89 mmol) in pyri-
dine (25 mL) was added dropwise. After stirring for 16 h, the reaction
mixture was allowed to cool, and the layers were separated. The aqueous
layer was extracted once with toluene (100 mL) and twice with toluene
(40 mL). The combined organic layers were dried over magnesium sul-
fate. After removal of the solvent, column chromatography on silica gel
(cyclohexane, ethyl acetate 2:1, Rf =0.5) afforded an orange oil (95.1 mg,
0.310 mmol, 9%). 1H NMR (500 MHz, CDCl3): d=8.69 (s, 1H; 1-H),
8.50 (d, J=5.0 Hz, 1H; 5-H), 7.79 (d, J=1.8 Hz, 2H; 8-H), 7.60 (t, J=
1.8 Hz, 1H; 10-H), 7.27 (d, J=5.0 Hz, 1H; 4-H), 2.69 (s, 3H; CH3),
1.40 ppm (s, 18H; CACHTUNGTRENNUNG
(CH3)3); 13C NMR (125.8 MHz, CDCl3): d=152.0 (C-
´
[27] G. Fꢀchsel, T. Klammroth, J. Dokic, P. Saalfrank, J. Phys. Chem. B
2006, 110, 16337–16345.
1), 152.4 (C-3), 129.7 (C-2), 115.9 (C-4), 35.1 (CACHTNUGTRNEG(UN CH3)3), 31.2 ppm (C-
[28] J. P. Perdew, K. Burke, Y. Wang, Phys. Rev. B 1996, 54, 16533–39.
[29] A. Schꢁfer, C. Huber, R. Ahlrichs, J. Chem. Phys. 1994, 100, 5829–
5835.
ꢀ
ꢀ
(CH3)3); IR (KBr): n˜ =3057, 2960 (C Harom), 2905, 2868 (C Haliph), 1592,
1478, 1461, 1439, 1394 (C=C), 1363, 1246, 1154, 900, 885, 824, 731,
699 cmꢀ1 (C Hdef); MS (EI): m/z (%): 309 (44) [M]+, 189 (100)
ꢀ
[MꢀNNꢀPy]+; MS (CI): m/z (%): 310 (100) [M+H]+. trans isomer:
1H NMR (500 MHz, [D8]toluene, 300 K): d=9.03 (s, 1H; 1-H), 8.33 (d,
J=5.0 Hz, 1H; 5-H), 7.93 (d, J=1.9 Hz, 2H; 8-H), 7.57 (t, J=1.8 Hz,
1H; 10-H), 6.62 (d, J=5.0 Hz, 1H; 4-H), 2.35 (s, 3H; CH3), 1.25 ppm (s,
[30] The trans/cis percentage was determined by integration of the
1H NMR spectroscopic signals of the 4-H (2a), 4-methyl (2b), the
CHMe2 proton of the 4-isopropyl group (2c), the 4-methoxy group
(2 f), and the NMe2 group (2g). In the case of 2d and 2e, we used
the 5-H of the pyridine ring, and in the case of 3b the 4-methyl pro-
tons were integrated to determine the trans/cis percentage.
[31] F. A. Walker, E. Hui, J. M. Walker, J. Am. Chem. Soc. 1975, 97,
2390–2397.
18H; CACHTUNGTRENNUNG
(CH3)3). cis isomer: 1H NMR (500 MHz, [D8]toluene, 300 K): d=
7.98 (d, J=4.9 Hz, 1H; 5-H), 7.56 (s, 1H; 1-H), 7.15 (t, J=1.7 Hz, 1H;
10-H), 6.67 (d, J=1.7 Hz, 2H; 8-H), 6.41 (d, J=5.0 Hz, 1H; 4-H), 1.88 (s,
3H; CH3), 1.00 ppm (s, 18H; CACTHNURGTENUNG(CH3)3).
[32] C. Frassineti, S. Ghelli, P. Gans, A. Sabatini, M. S. Moruzzi, A.
Vacca, Anal. Biochem. 1995, 231, 374–382.
[33] S. del Piero, A. Melchior, P. Polese, R. Portanova, M. Tolazzi, Ann.
Chim. 2006, 96, 29–49.
Acknowledgements
[34] TURBOMOLE, version 6.0, R. Ahlrichs, Universitꢁt Karlsruhe
(Germany), 2009, see http://www.turbomole.com.
[35] S. F. Boys, F. Bernardi, Mol. Phys. 1970, 19, 553.
[36] S. Simon, M. Duran, J. J. Dannenberg, J. Chem. Phys. 1996, 105,
11024–11031.
We gratefully acknowledge support by the Deutsche Forschungsgemein-
schaft through SFB 677 (Function by Switching).
[37] T. Helgaker, W. Klopper, H. Koch, J. Noga, J. Chem. Phys. 1997,
106, 9639–9646.
[38] T. Wendler, C. Schuett, C. Naether, R. Herges, J. Org. Chem. 2012,
77, 3284–3287.
[39] N. Campbell, A. W. Henderson, D. Taylor, J. Chem. Soc. 1953,
1281–1285.
[40] J. A. Turner, J. Org. Chem. 1983, 48, 3401–3408.
[41] V. Bonnet, F. Mongin, F. Trꢄcourt, G. Quꢄguiner, J. Chem. Soc.
Perkin Trans. 1 2000, 4245–4249.
[42] A. Modrow, D. Zargarani, R. Herges, N. Stock, Dalton Trans. 2011,
40, 4217–4222.
[43] J. Malm, B. Rehn, A.-B. Hçrnfeldt, S. Gronowitz, J. Heterocycl.
Chem. 1994, 31, 11–15.
[44] L. Estel, F. Marsais, G. Quꢄguiner, J. Org. Chem. 1988, 53, 2740–
2744.
[45] T. Wieland, C. Fest, G. Pfleiderer, Justus Liebigs Ann. Chem. 1961,
642, 163.
[1] O. Kahn, Molecular Magnetism, Wiley-VCH, New York, 1993.
[2] Dalton Trans. 2010, 39, 4671, themed issue on “Molecular Magnets”.
[3] R. N. Muller, L. V. Elst, S. Laurent, J. Am. Chem. Soc. 2003, 125,
8405–7.
[4] C. Tu, E. A. Osborne, A. Y. Louie, Tetrahedron 2009, 65, 1241–
1246.
[5] C. Rajadurai, M. Ruben, D. Kruk, WO 2009/080138A2.
[6] F. Touti, P. Maurin, L. Canaple, O. Beuf, J. Hasserodt, Inorg. Chem.
2012, 51, 31–33.
[7] U. Andres, Magnetohydrodynamic and Magnetohydrostatic Methods
of Mineral Separation, Wiley, New York, 1976.
[8] K. A. Mirica, S. S. Shevkoplyas, S. T. Phillips, M. Gupta, G. M.
Whitesides, J. Am. Chem. Soc. 2009, 131, 10049–10058.
[9] J. Mater. Chem. 2009, 19, 1670–1695 themed issue on “Molecular
Spintronics and Quantum Computing”.
[10] C. Roux, J. Zarembowitch, B. Gallois, T. Garnier, R. Claude, Inorg.
Chem. 1994, 33, 2273–2279.
[11] M.-L. Boillot, C. Roux, J.-P. Audiꢄre, A. Dausse, J. Zarembowitch,
Inorg. Chem. 1996, 35, 3975–3980.
[12] M.-L. Boillot, S. Chantrine, J. Zarembowitch, J. Y. Lallemond, J.
Prunet, New J. Chem. 1999, 23, 179–184.
Received: May 14, 2012
Revised: August 29, 2012
Published online: October 22, 2012
16368
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 16358 – 16368