Synthesis of μ2-Oxo-Bridged Iron(III) Tetraphenylporphyrin–Spacer–Nitroxide Dimers and their Structural and Dynamics Characterization by using EPR and MD Simulations

Article abstract of DOI:10.1002/chem.201805016

Iron(III) porphyrins have the propensity to form μ₂-oxo-dimers, the structures of which resemble two wheels on an axle. Whereas their crystal structure is known, their solution structure and internal dynamics is not. In the present work, the st

Full text of DOI:10.1002/chem.201805016

DOI: 10.1002/chem.201805016
Full Paper
&
Step Motors |Hot Paper|
Synthesis of m2-Oxo-Bridged Iron(III) Tetraphenylporphyrin–
Spacer–Nitroxide Dimers and their Structural and Dynamics
Characterization by using EPR and MD Simulations
Dinar Abdullin+,[a] Nico Fleck+,[a] Christoph Klein,[a, b] Philipp Brehm,[a] Sebastian Spicher,[c]
Arne Lꢀtzen,[b] Stefan Grimme,[c] and Olav Schiemann*[a]
Abstract: Iron(III) porphyrins have the propensity to form m2-
oxo-dimers, the structures of which resemble two wheels on
an axle. Whereas their crystal structure is known, their solu-
tion structure and internal dynamics is not. In the present
work, the structure and dynamics of such dimers were stud-
ied by means of electron paramagnetic resonance (EPR)
spectroscopy and quantum chemistry based molecular dy-
namics (MD) simulations by using the semiempirical tight-
binding method (GFN-xTB). To enable EPR investigation of
the dimers, a nitroxide was attached to each of the tetraphe-
nylporphyrin cores through a linear and a bent linker. The
inter-nitroxide distance distributions within the dimers were
determined by continuous-wave (cw)-EPR and pulsed elec-
tron–electron double resonance (PELDOR or DEER) experi-
ments and, with the help of MD, interpreted in terms of the
rotation of the porphyrin planes with respect to each other
around the Fe–O–Fe axis. It was found that such rotation is
restricted to the four registers defined by the phenyl sub-
stituents. Within the registers, the rotation angle swings be-
tween 308 and 608 in the proximal and between 1258 and
1458 in the distal register. With EPR, all four angles were
found to be equally populated, whereas the 308 and 1458
angles are strongly favored to the expense of the 608 and
1258 angles in the MD simulation. In either case, the internal
dynamics of these dimers thus resemble the motion of a
step motor.
triplet spin state.[3] This allows such porphyrins to be studied
by time-resolved electron paramagnetic resonance (EPR) tech-
niques[4] and even to be considered for the role of spin labels
for biomolecules.[5] Furthermore, the porphyrin core can also
Introduction
Porphyrins are heterocyclic organic compounds that display a
large variability of molecular structures and physicochemical
properties.[1,2] The variability of porphyrin structures is achieved
in several different ways, for instance, by metalation of the
macrocycle core, alteration of peripheral substituents, or link-
age of multiple macrocycles together. This provides a large
playground for designing new materials with desired proper-
ties and, hence, causes tremendous interest for porphyrins in
different fields of natural sciences and chemical engineering.[2]
The most common porphyrins include tetraphenylporphyrin
(TPP) and octaethylporphyrin (OEP). Both have a singlet
ground spin state, but can be photoexcited to a paramagnetic
host a number of different metal ions, such as Mg2+, Zn2+
Cu2+, Mn2+/Mn3+, Fe2+/Fe3+, or Co2+/Co3+ [2]
As several of
,
.
these ions are paramagnetic, the corresponding metallopor-
phyrins can be studied by EPR.[6] For instance, the EPR studies
on iron porphyrins and their biological counterparts, known as
hemes, has helped a lot in understanding the biological pro-
cesses taking place in hemoproteins.[7] Because of the increas-
ing importance of pulsed EPR techniques for distance mea-
surements in structural biology,[8] metalloporphyrins[9–14] or
other metal complexes[15–18] are also used as model systems to
test such distance measurements to metal centers.
[a] Dr. D. Abdullin,+ N. Fleck,+ C. Klein, P. Brehm, Prof. Dr. O. Schiemann
Institute of Physical and Theoretical Chemistry
Previous studies on iron porphyrins revealed their interest-
ing tendency to form dimeric species, in which the iron ions in
two TPP rings are connected with each other through an
oxygen atom.[14,19–21] This phenomenon is usually called m2-oxo-
dimerization. Many previous publications have dealt with the
identification and structural characterization of such m2-oxo-
bridged Fe3+ porphyrins owing to their interesting physico-
chemical properties, such as their catalytic activity in oxidizing
alkanes.[22,23] In some cases, the dimerization process was
found to be reversible: treating the monomeric porphyrins
with an aqueous basic solution resulted in formation of m2-
oxo-bridged dimers, whereas the analogous treatment of the
University of Bonn, Wegelerstr. 12, 53115 Bonn (Germany)
E-mail: schiemann@pc.uni-bonn.de
[b] C. Klein, Prof. Dr. A. Lꢀtzen
Kekulꢁ Institute of Organic Chemistry and Biochemistry
Gerhard-Domagk-Str. 1, 53121 Bonn (Germany)
[c] S. Spicher, Prof. Dr. S. Grimme
Mulliken Center for Theoretical Chemistry
University of Bonn, Beringstr. 4, 53115 Bonn (Germany)
[+] These authors contributed equally to this work.
Supporting information and the ORCID identification number(s) for the
author(s) of this article can be found under:
https://doi.org/10.1002/chem.201805016.
Chem. Eur. J. 2019, 25, 1 – 12
1
ꢁ 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
dimers with an aqueous acidic solution led to the breakup into
monomeric species.[14] The presence of m2-oxo-bridged dimers
in solution was proven by several methods, for example, elec-
trospray ionization (ESI) mass spectrometry[23–25] and ultravio-
let-visible (UV/Vis) spectroscopy.[26] The UV/Vis spectra of Fe3+
on the semiempirical tight-binding method GFN-xTB.[28] These
data show that such dimers maybe regarded as molecular step
motors.
porphyrins contain a Soret band, which corresponds to elec- Results and Discussion
tronic transitions within the porphyrin, as well as several Q
Synthesis
bands, which originate from charge transfer between the por-
phyrin and the Fe3+ ion. Upon m2-oxo dimerization, a charac-
teristic shift of the Soret band towards lower wavelengths as
well as significant changes within the Q bands occur. These
changes cause the observable difference in the color of the
monomeric (green) and dimeric (red) Fe3+ porphyrins dis-
solved in organic solvents. Crystallographic studies on m2-oxo-
bridged FeTPPs revealed that the two TPP cores are almost co-
planar, that the dihedral angle between them is only 3.78, and
that they form a staggered configuration with a twist angle of
54.68.[27] The two FeꢀO bonds are 1.763 ꢂ long and form an
almost linear bridge between the porphyrins with an Fe-O-Fe
angle of 1748. Although this crystallographic data provides a
clear picture of the static molecular structure of m2-oxo-bridged
FeTPPs, its structure and dynamics in solution is unclear.
The linkage of the FeTPPCl to a nitroxide through a connecting
bridge was achieved by two alternative synthetic routes de-
picted in Figure 1. The concept of click chemistry[29] was used
to synthesize compound 1·Cl, whereas Sonogashira–Hagihara
coupling[9] was adopted to obtain compound 2·Cl. Both syn-
theses make use of porphyrin 3 as a precursor, which was ob-
tained in accordance with previous publications.[30–33] For com-
pound 1·Cl, zinc porphyrin 4 was generated by treating por-
phyrin 3 with zinc acetate. This step allows the incorporation
of Cu2+ into the porphyrin core to be avoided during the click
reaction performed later on. Next, the copper(I)-catalyzed
azide alkyne cycloaddition (CuAAC)[34] reaction of 4 with 4-
azido-4’-hydroxybiphenyl, obtained from 4-iodo-4’-hydroxy-bi-
phenyl by an Ullmann-type reaction,[35] was carried out to yield
porphyrin 5. Note that it was also possible to perform these
two subsequent reactions in one pot starting from 4-iodo-4’-
hydroxybiphenyl and generating the azide in situ by using CuI
in DMSO as the copper(I) source for the Ullmann step. Subse-
quently, the zinc ion of 5 was removed by hydrochloric acid,
which resulted in porphyrin 6. The Fe3+-chlorido center was in-
corporated into 6 by reaction with Fe2+-chloride. This reaction
was carried out in THF in the presence of oxygen. Note that
these conditions are much milder than the ones proposed by
Adler et al.,[36] who used DMF at 1508C as a solvent. The ob-
tained iron porphyrin 7 was then used for Mukaiyama esterifi-
cation[37] with 2,2,5,5-tetramethyl-3-pyrrolin-oxyl-3-carboxylic
acid, which provided the final compound 1·Cl. An overall yield
of 9% was achieved for 1·Cl starting from 3. For compound
2·Cl, a more divergent approach was realized. First, Sonoga-
shira–Hagihara coupling of 3 with the nitroxide-labeled bi-
phenyl 8[9,10] was carried out to provide porphyrin 9. Finally,
the Fe3+ ion was introduced into 9 by the same procedure as
described above for compound 7. This gave an overall yield of
6% for porphyrin 2·Cl. The m2-oxo-bridged dimers 12·O and
22·O were obtained upon treatment with aqueous sodium hy-
droxide (see Experimental Section for details). The reversibility
of the dimerization process was tested by treatment of these
dimers with aqueous hydrochloric acid, which yielded the
monomeric porphyrins 1·Cl and 2·Cl again. Both monomers
1·Cl and 2·Cl as well as dimers 12·O and 22·O were analyzed by
mass spectrometry, UV/Vis, and EPR spectroscopy.
In the present study, the synthesis of two differently bridged
FeIIITPP/nitroxide pairs 1·Cl and 2·Cl (Figure 1) is described and
their m2-oxo-dimerization is confirmed. Different molecular link-
ers were used for the two porphyrins with the aim of disentan-
gling the effect of these linkers on the dynamics of the m2-oxo-
bridged TPP cores. The TPP rotamer dynamics within the
dimers in solution is unraveled by using a combination of EPR
spectroscopy and molecular dynamics (MD) simulations based
Mass spectrometry
Figure 1. Synthesis of model compounds 1·Cl and 2·Cl. a) Zn(OAc)2, CH2Cl2/
MeOH (1:1), r.t., 16 h, quant. b) 4-Hydroxy-4’-azidobiphenyl, CuSO4, sodium
ascorbate, DMSO/H2O (10:1), 708C, 72 h, 95%. c) HCl, CH2Cl2, r.t., 5 min,
93%. d) FeCl2, THF, 808C, 2 h, 66%. e) 2,2,5,5-Tetramethyl-3-pyrrolin-1-oxyl-3-
carboxylic acid, CMPI (2-chloro-1-methylpyridinium iodide), DMAP, Et3N,
CH2Cl2, r.t., 18 h, 78%. f) [Pd(PPh3)4], Et3N, 708C, 4 h, 59%. g) FeCl2, THF,
708C, 2 h, 91%.
The MALDI mass spectra of compounds 1·Cl and 2·Cl revealed
that they undergo partial dimerization during or after the syn-
thesis, probably upon exposure to air moisture. However, sig-
nificant dimerization of 1·Cl and 2·Cl could be detected only
after base treatment. The MALDI mass spectra of the base-
&
&
Chem. Eur. J. 2019, 25, 1 – 12
www.chemeurj.org
2
ꢁ 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
treated porphyrins are depicted in Figure 2. These spectra dis-
play intense peaks at m/z=2154.6 for 12·O and m/z=2068.6
for 22·O, respectively. The peaks correspond to the dimers of
1·Cl and 2·Cl in which the Fe3+ ions lost their chloride ligands
and are bridged to each other through an oxygen atom. In ad-
dition, both spectra contain two minor peaks, which appear at
m/z=1069.3 and 1104.3 in the case of the sample derived
from 1·Cl and at m/z=1026.3 and 1061.3 in that derived from
UV/Vis spectroscopy
As a means of reference, UV/Vis spectra of the monomers 1·Cl
and 2·Cl were measured first (Figure 3). Both spectra display a
Soret band at 422 nm and Q-bands at 372, 508, 573, and
689 nm. These values are in agreement with those found earli-
er for the Fe(TPP)Cl complex dissolved in toluene.[38] The
UV/Vis spectra of the dimers 12·O and 22·O show the expected
shift of the Soret band to shorter wavelengths, 415 nm for
12·O and 413 nm for 22·O, and the expected change in the Q-
bands, showing now only two bands at 571 and 612 nm in the
12·O sample and at 573 and 613 nm in the 22·O sample. Such
differences were previously reported in the literature and were
assigned to m2-oxo-dimerization of Fe3+ porphyrins.[26,39] Note
that these changes in the Q-bands lead to the noticeable dif-
ference of the color of their solutions. The toluene solutions of
dimeric porphyrins 12·O and 22·O feature a greenish color,
whereas their monomeric counterparts have a reddish color.
2·Cl. These peaks are assigned to the monomeric porphyrins
+
C
with and without the chloride atom occurring as [M ] and
[MꢀCl]+ respectively. Each of the MALDI mass spectra of the
,
acid-treated samples of the dimeric porphyrins shows only the
two peaks of the corresponding monomeric porphyrins listed
above but no peaks of the m2-oxo-bridged dimers. Although
MALDI mass spectra do not provide a quantitative estimate for
the relative amount of detected species, it seems that the acid
treatment leads to quantitative conversion into monomers,
whereas the base treatment forms dimers but monomers can
still be detected. An additional peak at m/z=2133.7 for 22·O
can be assigned to a reduced adduct with H2O and Na+,
which has the composition [(2ꢀCl)2·O+4H2O+Na]+.
Figure 3. UV/Vis spectra of a) 12·O (black) and 1·Cl (red) and b) 22·O (black)
and 2·Cl (red). The insets show the magnified region of the Q-bands.
Continuous-wave (cw)-EPR spectroscopy
Figure 2. MALDI mass spectra of a) m2-oxo-bridged dimer 12·O (black) and
monomer 1·Cl (red) and b) m2-oxo-bridged dimer 22·O (black) and monomer
2·Cl (red). For the sake of comparison, the spectra of 1·Cl and 2·Cl are plot-
ted inverted.
Both porphyrins 1·Cl and 2·Cl contain paramagnetic centers,
the nitroxide and the Fe3+ center, and thus can be investigat-
ed by means of continuous-wave EPR (cw-EPR) spectroscopy.
Chem. Eur. J. 2019, 25, 1 – 12
www.chemeurj.org
3
ꢁ 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
As the relaxation rates of both spin centers differ, the cw-EPR
experiments on the porphyrin samples were performed at two
different temperatures, at 10 K for the Fe3+ center (Figure 4)
and at 100 K for the nitroxide (Figure 5).
Figure 4. X-band cw-EPR spectra of a) 12·O (black) and 1·Cl (red) and
b) 22·O (black) and 2·Cl (red). All spectra were recorded at 10 K.
Figure 4 depicts the cw-EPR spectra of the monomeric (1·Cl
and 2·Cl) and dimeric (12·O and 22·O) porphyrins recorded at
10 K. Both monomers, 1·Cl and 2·Cl display an intense peak at
g=5.92, which proves that the Fe3+ ions are in the high-spin
state.[40,41] The other component of the axial spectrum of high-
spin Fe3+ ions usually appears at gꢁ2, but is overplayed by
the more intense nitroxide spectrum. In the dimer samples
12·O and 22·O, the line at g=5.92 almost disappears, whereas
the nitroxide spectrum preserves the intensity observed in the
acid-treated samples. The disappearance of the high-spin Fe3+
signal in 12·O and 22·O can be attributed to the strong antifer-
romagnetic exchange interaction between two m2-oxo-bridged
Fe3+ ions, which results in a total spin of S=0.[21] By comparing
the integral intensities of the high-spin Fe3+ signal in 1·Cl and
12·O, as well as in 2·Cl and 22·O, the amount of the monomeric
porphyrins in the samples of 12·O and 22·O was estimated to
be 6% in the case of 12·O and 3% in the case of 22·O. These
values reveal that the base treatment led to efficient formation
Figure 5. X-band cw-EPR spectra of the nitroxide at 100 K. a) The reference
spectrum of nitroxide 8 is overlaid with the spectra of 1·Cl and 2·Cl. b) The
spectrum of 12·O is overlaid with the spectrum of 1·Cl. The inset shows the
magnified part of the spectra, in which the strong dipolar couplings of the
nitroxides in 12·O manifest themselves (marked by asterisk). The simulation
of the 12·O spectrum is depicted by a blue dashed line. c) The spectrum of
22·O is overlaid with the spectrum of 2·Cl. The inset shows the magnified
part of the spectra, in which the strong dipolar couplings of the nitroxides
in 22·O manifest themselves (marked by asterisk). The simulation of the 22·O
spectrum is depicted by a blue dashed line.
of the m2-oxo-bridged porphyrins independently of the type of
the molecular bridge between the porphyrin core and the ni-
troxide group.
&
&
Chem. Eur. J. 2019, 25, 1 – 12
www.chemeurj.org
4
ꢁ 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
The 100 K nitroxide spectra of all four porphyrins samples as
well as the spectrum of nitroxide 8 are depicted in Figure 5.
The nitroxide spectra of the monomeric porphyrins 1·Cl and
2·Cl have, at the same concentration and within the same sol-
vent, very similar shapes and linewidths as the spectrum of the
uncoupled nitroxide 8 (Figure 5a). This reveals that the high-
spin Fe3+ in 1·Cl and 2·Cl, which is roughly 25 ꢂ apart from
the nitroxide center (see computational results below), has a
negligible effect on the nitroxide spectrum at the given tem-
perature. The nitroxide spectra of 12·O and 22·O are similar to
1·Cl and 2·Cl in shape and linewidth, but they have additional
broad signals on both sides of the nitroxide spectrum (asterisk
in the insets in Figures 5b,c). These signals result from the
pairs of nitroxides, which experience a strong dipolar coupling
between each other owing to their proximity in the structure
of some of the m2-oxo-bridged dimers (see below). The strong
dipolar coupling between the pairs of nitroxide centers leads
to an additional splitting of the corresponding EPR lines. The
value of this splitting is determined by the inter-spin distance r
in two ways.[42] First, the dipolar splitting is proportional to rꢀ3.
Second, the dipolar splitting is 3/2 times larger for the strong
coupling regime, which is realized for r<7 ꢂ, than for the
weak coupling regime, which corresponds to r>15 ꢂ. The dis-
tances in the range 7 ꢂ<r<15 ꢂ result in an intermediate di-
polar coupling, which is not adequately described neither by
the weak coupling nor by the strong coupling model. In the
weak coupling regime, the dipolar splitting cannot be resolved
in the cw-EPR spectrum but contributes to the EPR linewidth.
To estimate the corresponding inter-nitroxide distance distri-
butions and the percentage of the dimers with such short dis-
tances, the spectra of 12·O and 22·O were simulated. In the
simulations, both spectra were considered as a superposition
of unbroadened and dipolar broadened nitroxide spectra. The
spectra of 1·Cl and 2·Cl were used for the unbroadened part.
The dipolar broadened spectrum was calculated by convolu-
tion of the unbroadened spectra with a dipolar spectrum cor-
responding to a normal distribution of the inter-nitroxide dis-
tances (see Experimental Section for details).[42,43] A good fit to
the nitroxide spectrum of 12·O was obtained with 25% m2-oxo-
bridged dimers having a mean inter-nitroxide distance of 6.3 ꢂ
with a standard deviation of 0.4 ꢂ (dashed lines in Figure 5b).
The other 75% of m2-oxo-bridged dimers in 12·O should then
have inter-nitroxide distances above 15–20 ꢂ, yielding the
almost unbroadened nitroxide spectrum owing to the absence
of a strong dipolar coupling between the two nitroxides. In
the case of 22·O, good agreement with the experimental spec-
trum was obtained with 15% m2-oxo-bridged dimers having a
mean inter-nitroxide distance of 7 ꢂ with a standard deviation
of 2 ꢂ (dashed lines in Figure 5c).
PELDOR spectroscopy
To test whether longer nitroxide–nitroxide distances are also
present in the dimers, pulsed electron–electron double reso-
nance (PELDOR or DEER)[44,45] measurements were performed.
The results of these measurements are summarized in Figure 6
(see Experimental Section for details). The PELDOR time traces
of 1·Cl and 2·Cl display only a monotonous decay, which is
due to the inter-molecular dipolar interactions, but no dipolar
modulation (Figure 6a, dotted curves). This indicates that the
acid-treated samples contain only the monomeric porphyrins,
as also found by MALDI, and that the used PELDOR settings
Figure 6. PELDOR measurements of the acid-treated and base-treated samples of 1·Cl and 2·Cl. a) Original PELDOR time traces. The background fits for the
time traces of 12·O and 22·O are shown as red dashed lines. b) Background-corrected PELDOR time traces of 12·O and 22·O and their fits obtained by Tikhonov
regularization. c) Inter-nitroxide distance distributions of 12·O and 22·O. The gray shades depict the estimated errors of the distance distributions, which were
obtained by means of the validation toolbox of the DeerAnalysis program.
Chem. Eur. J. 2019, 25, 1 – 12
www.chemeurj.org
5
ꢁ 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
exclude the measurement of Fe3+/nitroxide distances. In con-
trast, the PELDOR time traces of 12·O and 22·O show a promi-
nent dipolar modulation with a modulation depth of 32% (Fig-
ure 6a, solid curves). The obtained value of modulation depth
is in agreement with the expected value for the used pump
pulse lengths of 14 ns (see Figure S1 in the Supporting Infor-
mation), which provides further evidence for quantitative di-
merization of the porphyrin molecules into 12·O and 22·O. To
obtain the inter-nitroxide distance distributions, the PELDOR
time traces were analyzed by means of the program DeerAnal-
ysis (see Experimental Section for details).[46] The optimal fits to
the PELDOR time traces and the corresponding inter-nitroxide
distance distributions are shown in Figure 6. Interestingly, the
inter-nitroxide distance distributions of both m2-oxo-bridged
dimers consist of two regions of non-zero probability with zero
probability between them. The first region is between 15 and
35 ꢂ, the second between 40 and 55 ꢂ. The short distances
around 6–7 ꢂ seen in the cw-EPR experiments are not seen in
the PELDOR experiment. This is related to the distance de-
pendence of the modulation depth factor, leading to a lower
distance limit for PELDOR of around 15 ꢂ.[42,46] In more detail,
in the case of 12·O, the distance distribution displays an asym-
metric peak with a maximum at 19.9 ꢂ and a shoulder at
about 27 ꢂ, as well as a symmetric peak centered at 46.6 ꢂ.
The distance distribution of 22·O exhibits an asymmetric peak
with a maximum at 27.7 ꢂ and a bi-modal peak with maxima
at 44.7 and 47.4 ꢂ. The high signal-to-noise ratio, the length of
the PELDOR time traces, and the clearly resolved oscillation pe-
riods render the mean values and the shape of the inter-nitrox-
ide distance distributions as being reliable (see the validation
results in Figure 6).
Figure 7. A schematic representation of the m2-oxo-bridged porphyrin con-
formers. The coplanar mean porphyrin planes of the adjacent porphyrins are
drawn as red and gray polygons. The dihedral angle between two porphy-
rins planes, C1-Fe1-C2-Fe2, is denoted as f.
A possible explanation for the multiple peaks in the experi-
mental inter-nitroxide distance distributions can be provided
in terms of the relative orientation of two porphyrins in the m2-
oxo-bridged dimers. If the two porphyrin ring planes could
rotate freely around the Fe-O-Fe axis, which would lead to a
uniform distribution of the dihedral angle f (C1-Fe1-Fe2-C2 in
Figure 7) between the two porphyrin planes, a smooth nitrox-
ide–nitroxide distance distribution covering a large distance
range would have to be expected. This is clearly not the case.
In contrast, in a very simple geometric model, steric clashes
between the orthogonal phenyl substituents are minimized if
f takes values around (458+908·n), where n=0, 1, 2, 3. This
defines four registers, of which two are symmetry related in
the case of 12·O and 22·O, leading to two principally different
orientations. Interestingly, an available crystal structure[27] of
[Fe(TPP)]2O reveals that the porphyrin planes do indeed adopt
a staggered configuration but with a twist angle f of 54.68
(Figure 7). The crystallographic value of f deviates by 108 from
458, which might be due to attractive interactions between the
phenyl rings of two adjacent TPPs. Because of the symmetry
within a register, an angle of 35.48 is equivalent to an angle of
54.68. Thus, the twist angle f can adopt eight possible values
in 12·O and 22·O, ꢂ35.48 and ꢂ54.68 in the proximal register
and ꢂ125.48 and ꢂ144.68 in the distal register (Figure 7). The
proximal and distal register are defined by the two nitroxide
arms being close to or far away from each other, respectively.
If these f values would indeed be occupied, this would lead to
a rough distance distribution with relatively short inter-nitrox-
ide distances for the proximal register with f= ꢂ35.48 and
ꢂ54.68 and significantly longer inter-nitroxide distances for the
distal register with f= ꢂ125.48 and ꢂ144.68. To test this hy-
pothesis and to correlate the predicted inter-nitroxide dis-
tances with the experimentally determined values, in silico
structure simulations were performed for the m2-oxo-bridged
dimers 12·O and 22·O.
Molecular modeling and dynamics simulations
The m2-oxo-bridged porphyrin systems are difficult to describe
by ab initio electronic structure theory. The bi-radical ground
state cannot be properly described in conventional density
functional theory (DFT) owing to the partial multireference
nature of such states. More importantly, the system’s size of
about 250 atoms introduces an impractical computational
price at the DFT level of theory, as the desired simulation tech-
niques are MD simulations and the calculation of energetic
barriers. Owing to the non-trivial electronic structure and the
molecular size of the investigated systems, the semiempirical
GFN-xTB/GBSA[47,48] method was applied and solvation effects
were only treated implicitly.
First, two initial structures with a dihedral angle f of 54.68
and 144.68, corresponding to the conformers in the proximal
&
&
Chem. Eur. J. 2019, 25, 1 – 12
www.chemeurj.org
6
ꢁ 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
and the distal registers were generated and optimized at the
GFN-xTB/GBSA(toluene) level of theory. Second, a conformer/
rotamer ensemble (CRE) was generated for each initial struc-
ture by using the MF-MD-GC/GFN-xTB algorithm[49] and the en-
ergetically lowest conformer was determined for each CRE.
This procedure yielded two conformers for each of the dimers:
12·Oa and 12·Ob for the dimer 12·O, and 22·Oa and 22·Ob for the
dimer 22·O. GFN-xTB analysis of the rotation profile around the
dihedral angle f was carried out to obtain a qualitative ener-
getic overview. Energies were computed at the GFN-xTB/
GBSA(toluene) level of theory as well as the thermostatistical
contribution for Gibbs free energy within the rigid-rotor-har-
monic-oscillator (RRHO) approximation. The rotation profile
around f as well as the calculated relative Gibbs free energies
of the two conformers for dimers 12·O and 22·O are shown in
the energy diagram in Figure 8. Our semiempirical model pre-
dicts the a-conformer to be more stable than the correspond-
ing b-conformer by approximately 2 kcalmolꢀ1 for both dimers
12·O and 22·O. The difference in energy is due to attractive dis-
persion and p–p interactions between the linker, which is only
present in the a-conformer of each dimer. Optimization of the
rotation path from the a- to b-conformer around the dihedral
angle f produced a low activation barrier profile passing
through the transition states TS1 and TS2. The associated com-
puted Gibbs barrier for rotation DG° at 298.15 K amounts to
approximately +19 kcalmolꢀ1 relative to 12·Oa and 22·Oa and
approximately +17 kcalmolꢀ1 relative to 12·Ob and 22·Ob. To
get deeper insight into the mechanism of rotation, the rotation
path optimization and transition state search was repeated for
an artificial porphyrin dimer system where the residues at-
tached to the porphyrin core (see Figure 7) were replaced by
methyl groups. GFN-xTB analysis of the rotation path leads to
the same energetic barrier of approximately +17 kcalmolꢀ1 as
for the untruncated system, revealing the steric hindrance of
the phenyl substituents upon rotation as the cause for the
high rotation barrier.
For the energetically lowest conformers, MD simulations
were carried out for 1 ns at the GFN-xTB/GBSA(toluene) level
of theory (see Experimental Section for details) and the inter-
nitroxide distances were calculated to determine the corre-
sponding distance distributions. The obtained inter-nitroxide
distance distributions are plotted together with the corre-
sponding experimental distributions in Figure 9. Note that two
distance distributions were calculated for each dimer, one with
the starting structure 12·Oa or 22·Oa and the other with the
Figure 8. Energy diagram for the rotation along the angle f and the corre-
sponding structures of the m2-oxo-bridged dimers 12·O and 22·O calculated
at the GFN-xTB level of theory. The molecules are drawn as ball-and-stick
models with carbon atoms colored gray, oxygen atoms colored red, nitrogen
atoms colored blue, and iron atoms colored orange. The inter-nitroxide dis-
tances and the f angles are listed below the names of the corresponding
conformers.
Figure 9. Comparison of the experimental and calculated inter-nitroxide dis-
tance distributions in the m2-oxo-bridged dimers a) 12·O and b) 22·O.
Chem. Eur. J. 2019, 25, 1 – 12
www.chemeurj.org
7
ꢁ 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
starting structure 12·Ob or 22·Ob. This was necessary, because
the phenyl substituents pose high-energy barriers that the
conformer dynamics cannot overcome. This in turn means that
the conformers are confined to their registers at normal tem-
perature. Accordingly, the two MD derived distributions do not
overlap with each other: the distributions that correspond to
the starting structures 12·Oa and 22·Oa display a non-zero prob-
ability for the distance in the range 0–30 ꢂ, whereas the distri-
butions that correspond to the starting structures 12·Ob or
22·Ob have a non-zero probability between 30 and 55 ꢂ, show-
ing that the MD simulations exclude the transition between
the two starting structures of the dimers. It should be men-
tioned that the MD-derived distributions calculated with differ-
ent starting structures are weighted equally in Figure 9, which
is justified because the a- and b-conformers of each dimer are
similar in energy (Figure 8).
Comparison of EPR and MD results
Comparison of the in silico derived distance distributions with
the experimental ones reveals several similarities. First, the MD
simulations reproduce the distance peak between 40 and 50 ꢂ
of the PELDOR-derived distributions for both dimers, although
the shape of the distributions is different. Second, the distance
peak below 10 ꢂ, which was found from cw-EPR experiments
for both dimers, is also confirmed by the MD simulations. Both
methods, MD and EPR, also agree on the width of the short
distance peak for 22·O, but in the case of 12·O the experimental
distance peak is significantly narrower than the MD result. The
most prominent discrepancy between the in silico and experi-
mental distance distributions is observed in the range 15 to
35 ꢂ. The PELDOR-derived distributions of 12·O and 22·O con-
tain an intense distance peak in this range. In contrast, the
MD-derived distributions predict almost zero probability for
these distances.
Figure 10. Correlation between the inter-nitroxide distances and the twist
angle f as determined from MD simulations on the m2-oxo-bridged dimers
a) 12·O and b) 22·O. The arrows denote the correspondence of the correla-
tion peaks to the starting structures used for the MD simulations.
mental PELDOR data (Figure 9). The similar intensity of the two
long-distance maxima in the PELDOR-derived distance distribu-
tion of 22·O reveals that the two conformers with fꢁ1258 and
1458 are equally abundant. In contrast, the conformer with f
ꢁ1458 is favored against the conformer with fꢁ1258 by the
MD simulation. This discrepancy causes the difference in the
shape of the PELDOR- and MD-derived inter-nitroxide distance
distributions in the range 40–50 ꢂ (Figure 9).
To interpret the outlined similarities and differences between
the MD- and EPR-derived distance distributions, the correlation
between the inter-nitroxide distance and the twist angle f was
investigated by using the MD data (Figure 10). Remarkably,
only few distinct correlation peaks were obtained for both
dimers. The MD simulations on the a-conformers of both
dimers (proximal register) yielded an intense correlation peak
around f=308 and only a very weak correlation peak around
f=608, which is almost absent for 12·O. The MD simulations
on the b-conformers led to an intense correlation peak around
f=1458 and again a weaker correlation peak around f=1258.
All four values of the f angle are in good agreement with the
f angles determined from the crystal structure of [Fe(TPP)]2O
(see above). The correlation peaks at fꢁ1258 and 1458 (distal
register) correspond in the MD simulation to inter-nitroxide
distances of 47.7 and 49.5 ꢂ for 12·O, and 42.3 and 49.5 ꢂ for
22·O. These distances are very close to the distance observed
by PELDOR (46.6 ꢂ for 12·O, 44.7 and 47.4 ꢂ for 22·O). More-
over, the larger distance difference for the two conformers in
the distal register for 22·O (7.2 ꢂ) compared with 12·O (1.8 ꢂ)
mirrors itself in the bi- (resolved distances) versus uni-modality
(unresolved distances) of the long-distance peak in the experi-
Another pair of correlation peaks, around f=308 and 608, is
associated with inter-nitroxide distances in the range 5 to
35 ꢂ. The correlation peak at fꢁ308 corresponds to an inter-
nitroxide distance of 8.5 ꢂ for both dimers. This prediction is
close to the distances determined by cw-EPR (6.3 ꢂ for 12·O,
7 ꢂ for 22·O). The correlation peak around f=608, correspond-
ing to the distance range of 15 to 30 ꢂ, is almost absent in the
MD of 12·O (Figure 10a) and very weak and smeared for 22·O
(Figure 10b). In contrast to the MD simulations, the PELDOR-
derived distance distributions contain a clear peak in this dis-
tance region for both 12·O and 22·O. Thus, it seems that al-
though both methods, EPR and MD, show that both registers
are populated, the weight of the conformers within each regis-
ter differs. This might be attributed to the simulation of the dy-
namics of 12·O and 22·O having been performed in liquid solu-
tion, whereas the corresponding EPR results were obtained on
glassy frozen solutions. Usage of an implicit solvation model
may also introduce a deviation between theory and experi-
ment, as steric effects are not taken into account.
&
&
Chem. Eur. J. 2019, 25, 1 – 12
www.chemeurj.org
8
ꢁ 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
In addition, MD predicts a non-zero probability for the f each other around the Fe-O-Fe axis within each of the regis-
angles between 1258 and 1458, as well as between 308 and
608 (can be seen only for 22·O). This shows that the two por-
phyrin planes can rotate relative to each other within each of
the registers in a “scissors-like” motion. For the 22·O molecule,
in which the linker is linear and fairly rigid, such a “scissors-
like” motion should result in two bimodal inter-nitroxide dis-
tance peaks, corresponding to proximal and distal registers.
Indeed, the bimodal peaks between 0 and 35 ꢂ as well as be-
tween 40 and 50 ꢂ can be seen in the EPR-derived distance
distributions of 22·O (Figure 9b). In the case of 12·O, the kinked
linker makes it difficult to resolve the bimodality of the peaks.
Nevertheless, for both dimers, EPR and MD show clearly a zero
probability for angles/distances between registers. As shown
by GFN-xTB analysis, the conformations with f=908·n, where
n = 0, 1, 2, 3, have high energetic barriers posed by steric
clashes of the phenyl rings, which prevents an exchange be-
tween registers. This also means that different registers can
only be occupied during dimer assembly. By using the distance
distributions of the EPR data, one can estimate the relative oc-
cupation of the proximal versus distal register. For 12·O, this
yields 59% and 41% for the proximal and the distal register,
respectively, and for 22·O, 44% and 56% for the proximal and
distal register, respectively. Thus, both registers are, within ex-
perimental error, equally populated.
ters. Such a “scissor-like” motion corresponds to an exchange
of the favorable conformers within each register. Importantly,
both synthesized porphyrin dimers display very similar dynam-
ics of the central TPP core, which hints towards a minor effect
of the linker on the intrinsic dynamics of the TPPs and, thus,
allows us to extend the derived conclusions to other FeTPP-
based m2-oxo dimers. With respect to using such dimers within
molecular machines, one might view them as molecular step
motors. Furthermore, the present study provides a basis for
using these compounds as models for testing pulsed EPR-
based distance measurements to high-spin Fe3+ ions with
large zero-field splitting constants.
Experimental Section
Synthesis
The details of the synthesis of 1·Cl and 2·Cl and their precursors, as
well as the corresponding analytics (MS, NMR, elemental analysis,
etc.), are given in the Supporting Information. The dimerization of
1·Cl and 2·Cl to the m2-oxo-bridged dimers 12·O and 22·O, respec-
tively, was achieved by treatment of the monomers with aqueous
sodium hydroxide. In particular, a 200 mm solution of 1·Cl or 2·Cl in
toluene was prepared and mixed with 0.05m aqueous NaOH in a
volume ratio of 1:1. The obtained solution was shaken for 2 min by
means of a Vortex Genie 2 (Scientific Industries Inc.) and then incu-
bated for 1 min to achieve separation of water and toluene layers.
After this, the toluene layer was extracted and the entire procedure
was repeated again to ensure efficient dimer formation. Subse-
quently, the toluene layer was transfused into a clean tube and the
solvent was evaporated from the samples under high vacuum. The
reversibility of the dimerization process was tested by treatment of
the dimers with aqueous hydrochloric acid. This procedure was
done in complete analogy to the treatment with NaOH except for
the fact that 0.05m aqueous HCl was used instead of 0.05m aque-
ous NaOH. This procedure again gave rise to the monomers 1·Cl
and 2·Cl.
Conclusions
Two synthetic approaches, one based on click chemistry and
the other on Sonogashira–Hagihara coupling, were successfully
applied to synthesize two new Fe3+ porphyrins, 1·Cl and 2·Cl,
each linked to a nitroxide group. Treating the monomeric sam-
ples with an aqueous basic solution, the monomers were con-
verted almost quantitatively into m2-oxo-bridged dimers 12·O
and 22·O. The structure and dynamics of the m2-oxo dimers
was then investigated by means of cw-EPR and PELDOR spec-
troscopy. The cw-EPR experiments were used to determine the
short inter-nitroxide distances below 15 ꢂ, whereas the
PELDOR experiments allowed the determination of long inter-
nitroxide distances above 15 ꢂ. Thus, both methods comple-
ment each other nicely. To interpret the obtained distance con-
straints in terms of structure and dynamics of the dimers, the
combined cw-EPR/PELDOR derived inter-nitroxide distance dis-
tributions were compared with the results of the MD simula-
tions. This comparison yielded several fruitful results. It was
shown that the conformers of the m2-oxo-bridged dimers
occupy four registers that are defined by the phenyl substitu-
ents. The dimer rotamers cannot exchange their register by ro-
tation owing to the high energetic barrier of 17–19 kcalmolꢀ1
posed by the phenyl substituents. Within each register, two fa-
vorable conformers were identified. They correspond to the
twist angle between the two porphyrin cores of 308 and 608 in
the case of the proximal register and 1258 and 1458 in the
case of the distal registers. The population of these two con-
formers in each register was weighted differently by EPR and
MD. Additionally, it was proposed based on MD simulations
that m2-oxo-bridged porphyrin planes can rotate relative to
Mass spectrometry
Matrix-assisted laser desorption/ionization (MALDI) mass spectra
were measured by using an Autoflex II TOF/TOF (Bruker) spectrom-
eter.
UV/Vis spectroscopy
UV/Vis spectra were recorded by using SPECORD 200 (Analytik
Jena AG) and Cary 100 UV/Vis (Agilent Technologies) spectrome-
ters. All spectra were taken in the range from 300 nm to 800 nm
with a 1 nm step. All UV/Vis samples were prepared with the con-
centration of 10 mm in deuterated toluene.
cw-EPR spectroscopy
cw-EPR experiments were carried out with an X-Band EPR spec-
trometer EMXmicro (Bruker) equipped with super-high-quality res-
onator (SHQ). cw-EPR spectra were collected at two different tem-
peratures, 10 K and 100 K. To obtain 100 K, the variable-tempera-
ture accessory ER 4131VT (Bruker) was employed. For 10 K, the res-
onator was mounted inside a continuous-flow helium cryostat
ER4112HE (Bruker), the temperature of which was controlled by
Chem. Eur. J. 2019, 25, 1 – 12
www.chemeurj.org
9
ꢁ 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
the Mercury iTC system (Oxford Instruments). The cw-EPR measure-
ments at 100 K were performed with a microwave power of
0.1927 mW (30 dB), a modulation frequency of 100 kHz, a modula-
tion amplitude of 0.2 mT, and a time constant of 10.24 ms. The
10 K cw-EPR spectra were acquired by using a microwave power of
17.22 mW (10 dB), a modulation frequency of 100 kHz, a modula-
tion amplitude of 0.1 mT, and a time constant of 10.24 ms. All cw-
EPR samples were prepared with a nitroxide spin concentration of
200 mm in deuterated toluene. The sample volume was kept con-
stant at 200 mL. Moreover, the cw-EPR measurements at 10 K were
performed with the same resonator Q factor (ꢁ500) and the same
receiver gain (40 dB) for all samples.
ried out with an ELEXSYS E580 (Bruker) spectrometer using a Flex-
Line probehead with a Q-band resonator ER5106QT-2 (Bruker). All
microwave pulses were amplified by a 150 W TWT amplifier (model
187Ka). To achieve the working temperature of 50 K, a continuous-
flow helium cryostat CF935 (Oxford Instruments) and a tempera-
ture control system iTC 503S (Oxford Instruments) were employed.
The PELDOR experiments were performed with the standard four-
pulse sequence p/2(ndet)–t1–p(ndet)–(t1 +t)–p(npump)–(t2ꢀt)–p(ndet)–
t2–echo. The frequency of the pump pulse and the magnetic field
were adjusted to be in resonance with the maximum of the nitrox-
ide spectrum, whereas the frequency of the detection pulse was
set 100 MHz lower than the frequency of the pump pulse. All
PELDOR measurements were performed with the lengths of detec-
tion p/2- and p-pulses of 16 and 32 ns, respectively, and the pump
pulse was 14 ns long. The p-pulse was phase-cycled to eliminate
receiver offsets. The t1 interval was set to a starting value of 250 ns
and was incremented during each experiment eight times with a
step of 16 ns, to suppress the deuterium ESEEM (electron spin
echo envelope modulation). The t2 interval was set to 10 ms. The
position of the pump pulse relative to the primary echo was incre-
mented with a step of 8 ns. All PELDOR spectra were recorded at
50 K with a repetition time of 3 ms. The signal was averaged over
250 runs to achieve a good signal-to-noise ratio.
Determination of spin–spin distances from cw-EPR spectra
The nitroxide spectra of 2·Cl and 22·O were simulated by using the
convolution method.[42,43] Both spectra were considered as a super-
position of unbroadened spectra fu(B) and dipolar broadened
nitroxide spectra fb(B):
fðBÞ ¼ w fbðBÞ þ ð1 ꢀ wÞ fuðBÞ
where w is the weight of the dipolar broadened part in the total
spectra. The nitroxide spectra of 1·Cl and 2·Cl were used for the
unbroadened part fu(B). The dipolar broadened spectrum was cal-
culated by convolution of the unbroadened spectra with a dipolar
spectrum fdd(B):
Determination of spin–spin distances from PELDOR data
The inter-nitroxide distance distributions were determined from
the experimental PELDOR time traces by means of the program
DeerAnalysis.[46] First, the time traces were divided by an exponen-
tial decay to remove the background resulting from the inter-mo-
lecular interactions. Then, Tikhonov regularization was applied to
the background-free time traces, yielding the desired distance dis-
tributions. The best regularization parameter was determined
through the L-curve approach. The validation of the inter-nitroxide
distance distributions was done by varying the background start-
ing time from 2 ms to 6 ms with a step of 100 ns and the back-
ground dimension from 3.0 to 3.5 with a step 0.1.
fbðBÞ ¼ fuðBÞ ꢃ fddðBÞ
The dipolar spectrum was calculated in the strong coupling ap-
proximation for a normal distribution P(r) of the inter-nitroxide dis-
tances with a mean distance hri and a standard deviation sr:
p=2
1
Z
Z
fddðBÞ ¼ PðrÞdr Dðr; qÞsinqdq
0
0
in which:
Computational details
3 m0 gb
4 4p r3
Dðr; qÞ ¼
ð1 ꢀ 3cos2qÞ
For all four initial structures, the generation of the conformer/rota-
mer ensemble (CRE) was performed by using the semiempirical
tight-binding method GFN-xTB and the CRE search algorithm.[49]
The MF-MD-GC/GFN-xTB algorithm consists of three steps: normal
mode following (MF), molecular dynamics (MD) simulations, and
“pseudo-genetic” structure crossing (GC). In all steps, GFN-xTB is
used as the underlying electronic structure method. The CRE gen-
eration was conducted in toluene as the solvent, simulated by the
implicit GBSA solvation model.[47,48] The energetically lowest con-
former found for each initial structure was fully optimized at the
GFN-xTB/GBSA(toluene) level of theory and molecular dynamics
(MD) simulations were carried out with implicit solvent GBSA (tol-
uene) on each structure at 298 K for 100 ps, 500 ps, and 1 ns with
a time step of 4 fs and an equilibration phase of 10 ps. The
SHAKE[50] algorithm was used to constrain all bonds. For geometry
optimization and MD simulation, the stand-alone program xtb[51]
was used. Evaluation of the trajectories was performed with the
program TRAVIS.[52]
ꢀ
ꢁ
1
ðr ꢀ hriÞ
pffiffiffiffiffi
PðrÞ ¼
exp ꢀ
2s2r
2psr
Here, m0 is the vacuum permeability, g is the g-factor of the nitrox-
ide, b is the Bohr magneton, r is the inter-nitroxide distance, and q
is the angle between the distance vector and the direction of the
external magnetic field. The integral of fdd(B) was solved by means
of Monte-Carlo method using 106 samples.
In the present analysis, the values of the mean distance hri, its
standard deviation sr, and the weight w were optimized manually
until the convolution of the unbroadened cw-EPR spectra fu(B)
with the dipolar spectrum provided the best fit to the cw-EPR
spectra of 2·Cl and 22·O. The MATLAB-based source code of the op-
timization program is available at https://github.com/dinarabdul-
lin/DipolarBroaderingFitting.
The calculation of free energies was performed as follows. Single-
point energies E were computed at the GFN-xTB level of theory.
Harmonic vibrational frequencies were calculated at the same level
to obtain the thermostatistical corrections GRRHO. The solvation free
energy DGsolv was calculated by GBSA at 298.15 K in the respective
PELDOR spectroscopy
PELDOR experiments were performed on the same samples as the
cw-EPR experiments. All samples had the nitroxide spin concentra-
tion of 200 mm in deuterated toluene. The measurements were car-
&
&
Chem. Eur. J. 2019, 25, 1 – 12
www.chemeurj.org
10
ꢁ 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
[19] E. B. Fleischer, T. S. Srivastava, J. Am. Chem. Soc. 1969, 91, 2403–2405.
[20] E. B. Fleischer, J. M. Palmer, T. S. Srivastava, A. Chatterjee, J. Am. Chem.
Soc. 1971, 93, 3162–3167.
solvent (toluene). The free energy is then the sum of all contribu-
tions:
[21] K. S. Murray, Coord. Chem. Rev. 1974, 12, 1–35.
[22] E. Tabor, J. Połtowicz, K. Pamin, S. Basa˛g, W. Kubiak, Polyhedron 2016,
119, 342–349.
G ¼ E þ GTRRHO þ DGsTolvðXÞ
Reaction paths were calculated for all reactions to get a good
guess of the transition state structure. To begin from reasonable
geometries, the start and end structures were optimized with GFN-
xTB/GBSA. The reaction path was then calculated with the growing
string method GSM[53] by using GFN-xTB/GBSA as the underlying
electronic structure method. An additional frequency calculation
was performed to check for a true transition state with only one
imaginary mode.
[23] E. V. Kudrik, P. Afanasiev, L. X. Alvarez, P. Dubourdeaux, M. Clꢃmancey,
J. M. Latour, G. Blondin, D. Bouchu, F. Albrieux, S. E. Nefedov, A. B. Soro-
kin, Nat. Chem. 2012, 4, 1024–1029.
[24] T. Gozet, L. Huynh, D. K. Bohme, J. Mass Spectrom. 2010, 45, 35–42.
[25] A. M. N. Silva, A. Aguir, S. S. Balula, A. M. G. Silva, M. Rangel, J. Mass
Spectrom. 2014, 49, 763–765.
[26] S. K. Ghosh, R. Patra, S. P. Rath, Inorg. Chim. Acta 2010, 363, 2791–2799.
[27] A. B. Hoffman, D. M. Collins, V. W. Day, E. B. Fleischer, T. S. Srivastava, J. L.
Hoard, J. Am. Chem. Soc. 1972, 94, 3620–3626.
[28] S. Grimme, C. Bannwarth, P. Shushkov, J. Chem. Theory Comput. 2017,
13, 1989–2009.
[29] H. C. Kolb, M. G. Finn, K. B. Sharpless, Angew. Chem. Int. Ed. 2001, 40,
2004–2021; Angew. Chem. 2001, 113, 2056–2075.
Acknowledgments
[30] R. Luguya, L. Jaquinod, F. R. Fronczek, M. G. H. Vicente, K. M. Smith, Tet-
rahedron 2004, 60, 2757–2763.
[31] J. Rochford, E. Galoppini, Langmuir 2008, 24, 5366–5374.
[32] P. Brodard, S. Matzinger, E. Vauthey, O. Mongin, C. Papamicaꢄl, A. Gossa-
uer, J. Phys. Chem. A 1999, 103, 5858–5870.
This work was funded by the German Research Foundation
(DFG) within the collaborative research center SFB813 “Chemis-
try at Spin Centers” (projects A6 and Z2).
[33] T. Kengthanomma, P. Thamyongkit, J. Gasiorowski, A. M. Ramil, N. S. Sar-
iciftci, J. Mater. Chem. A 2013, 1, 10524–10531.
Conflict of interest
[34] M. Meldal, C. W. Tomøe, Chem. Rev. 2008, 108, 2952–3015.
[35] J. Andersen, U. Madsen, F. Bjçrkling, X. Liang, Synlett 2005, 2209–2213.
[36] A. D. Adler, F. R. Longo, F. Kampas, J. Kim, J. Inorg. Nucl. Chem. 1970, 32,
2443–2445.
The authors declare no conflict of interest.
[37] T. Mukaiyama, Angew. Chem. Int. Ed. Engl. 1979, 18, 707–721; Angew.
Chem. 1979, 91, 798–812.
[38] F. Paulat, N. Lehnert, Inorg. Chem. 2008, 47, 4963–4976.
[39] N. Sadasivan, H. I. Eberspaecher, W. H. Fuchsman, W. S. Caughey, Bio-
chemistry 1969, 8, 534–541.
[40] H. Aissaoui, R. Bachmann, A. Schweiger, W. D. Woggon, Angew. Chem.
Int. Ed. 1998, 37, 2998–3002; Angew. Chem. 1998, 110, 3191–3196.
[41] F. Trandafir, P. Heerdt, M. Fittipaldi, E. Vinck, S. Dewilde, L. Moens, S. Van
Doorslaer, Appl. Magn. Reson. 2007, 31, 553–572.
[42] J. E. Banham, C. M. Baker, S. Ceola, I. J. Day, G. H. Grant, E. J. J. Groenen,
C. T. Rodgers, G. Jeschke, C. R. Timmel, J. Magn. Reson. 2008, 191, 202–
218.
[43] H. J. Steinhoff, N. Radzwill, W. Thevis, V. Lenz, D. Brandenburg, A.
Antson, G. Dodson, A. Wollmer, Biophys. J. 1997, 73, 3287–3298.
[44] A. D. Milov, K. M. Salikhov, M. D. Shchirov, Fiz. Tverd. Tela 1981, 23, 975–
982.
[45] G. Jeschke, M. Pannier, A. Godt, H. W. Spiess, Chem. Phys. Lett. 2000,
331, 243–252.
[46] G. Jeschke, V. Chechik, P. Ionita, A. Godt, H. Zimmermann, J. Banham,
C. R. Timmel, D. Hilger, H. Jung, Appl. Magn. Reson. 2006, 30, 473–498.
[47] W. Clark Still, A. Tempczyk, R. C. Hawley, T. Hendrickson, J. Am. Chem.
Soc. 1990, 112, 6127–6129.
Keywords: complex formation · DEER spectroscopy · metal
centers · nitroxide · pulsed dipolar spectroscopy
[1] L. P. Cook, G. Brewer, W. Wong-Ng, Crystals 2017, 7, 223.
[2] Handbook of Porphyrin Science, Vol. 1–44 (Eds.: W. Kadish, K. M. Smith,
R. Guilard), World Scientific, Singapore, 2010.
[3] C. W. M. Kay, J. Am. Chem. Soc. 2003, 125, 13861–13867.
[4] S. Richert, C. E. Tait, C. R. Timmel, J. Magn. Reson. 2017, 280, 103–116.
[5] M. Di Valentin, M. Albertini, M. G. Dal Farra, E. Zurlo, L. Orian, A. Polime-
no, M. Gobbo, D. Carbonera, Chem. Eur. J. 2016, 22, 17204–17214.
[6] A. Abragam, B. Bleaney, Electron Paramagnetic Resonance of Transition
Ions, Oxford University Press, Oxford, 1970.
[7] A. Messerschmidt, R. Huber, T. Poulos, K. Wieghardt, Handbook of Metal-
loproteins, Vol. 1–2, Wiley, Chichester, 2001.
[8] Structural Information from Spin-Labels and Intrinsic Paramagnetic Cen-
tres in the Biosciences, Structure and Bonding, Vol. 152 (Eds.: C. R.
Timmel, J. R. Harmer), Springer, Heidelberg, 2013.
[9] B. E. Bode, J. Plackmeyer, T. F. Prisner, O. Schiemann, J. Phys. Chem. A
2008, 112, 5064–5073.
[10] B. E. Bode, J. Plackmeyer, M. Bolte, T. F. Prisner, O. Schiemann, J. Organo-
met. Chem. 2009, 694, 1172–1179.
[11] J. E. Lovett, A. M. Bowen, C. R. Timmel, M. W. Jones, J. R. Dilworth, D.
Caprotti, S. G. Bell, L. L. Wong, J. Harmer, Phys. Chem. Chem. Phys. 2009,
11, 6840–6848.
[12] A. M. Bowen, M. W. Jones, J. E. Lovett, T. G. Gaule, M. J. McPherson, J. R.
Dilworth, C. R. Timmel, J. R. Harmer, Phys. Chem. Chem. Phys. 2016, 18,
5981–5994.
[48] S. Grimme, P. Shushkov, unpublished results.
[49] S. Grimme, C. Bannwarth, S. Dohm, A. Hansen, J. Pisarek, P. Pracht, J.
Seibert, F. Neese, Angew. Chem. Int. Ed. 2017, 56, 14763–14769; Angew.
Chem. 2017, 129, 14958–14964.
[50] J. P. Ryckaert, G. Ciccotti, H. J. C. Berendsen, J. Comput. Phys. 1977, 23,
327–341.
[51] Please contact E-mail: xtb@thch.uni-bonn.de to obtain the program.
[52] M. Brehm, B. Kirchner, J. Chem. Inf. Model. 2011, 51, 2007–2023.
[53] P. Zimmerman, J. Chem. Theory Comput. 2013, 9, 3043–3050.
[13] M. H. Rakowsky, K. M. More, A. V. Kulikov, G. R. Eaton, S. S. Eaton, J. Am.
Chem. Soc. 1995, 117, 2049–2057.
[14] M. H. Rakowsky, A. Zecevic, G. R. Eaton, S. S. Eaton, J. Magn. Reson.
1998, 131, 97–110.
[15] D. Akhmetzyanov, P. Schçps, A. Marko, N. C. Kunjir, S. T. Sigurdsson, T. F.
Prisner, Phys. Chem. Chem. Phys. 2015, 17, 24446–24451.
[16] A. Meyer, O. Schiemann, J. Phys. Chem. A 2016, 120, 3463–3472.
[17] A. Potapov, Y. Song, T. J. Meade, D. Goldfarb, A. V. Astashkin, A. Raitsimr-
ing, J. Magn. Reson. 2010, 205, 38–49.
Manuscript received: October 4, 2018
Revised manuscript received: November 13, 2018
Accepted manuscript online: November 23, 2018
[18] P. Lueders, G. Jeschke, M. Yulikov, J. Phys. Chem. Lett. 2011, 2, 604–609.
Version of record online: && &&, 0000
Chem. Eur. J. 2019, 25, 1 – 12
www.chemeurj.org
11
ꢁ 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
FULL PAPER
&
Step Motors
D. Abdullin, N. Fleck, C. Klein, P. Brehm,
S. Spicher, A. Lꢀtzen, S. Grimme,
O. Schiemann*
&& – &&
Like wheels on an axle: Two iron(III)
tetraphenylporphyrin–spacer–nitroxide
complexes were synthesized and their
monomers and m2-oxo-bridged dimers
characterized. Continuous-wave and
pulsed EPR spectroscopy in combina-
tion with molecular dynamics calcula-
tions show that the porphyrin moieties
in the dimers rotate like a step motor
around the FeꢀOꢀFe axis (see figure).
Synthesis of m2-Oxo-Bridged Iron(III)
Tetraphenylporphyrin–Spacer–
Nitroxide Dimers and their Structural
and Dynamics Characterization by
using EPR and MD Simulations
&
&
Chem. Eur. J. 2019, 25, 1 – 12
www.chemeurj.org
12
ꢁ 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!