Building responsive materials by assembling {Fe4Co4} switchable molecular cubes

Article abstract of DOI:10.1039/d1tc01825f

Responsive materials that can answer to chemical or physical external stimuli offer numerous prospects in material science. Here, we have elaborated a two-step synthetic approach that allows incorporating molecular cubic switches into a polymeric material

Full text of DOI:10.1039/d1tc01825f

Journal of
Materials Chemistry C
View Article Online
View Journal | View Issue
PAPER
Building responsive materials by assembling
{Fe4Co4} switchable molecular cubes†
Cite this: J. Mater. Chem. C, 2021,
9, 8882
a
a
Qui Pham Xuan,
Jana Glatz,a Amina Benchohra,a Juan-Ramon Jimenez,
´ ´
Remi Plamont,a Lise-Marie Chamoreau,
Alexandrine Flambard,a Yanling Li,
a
a
´
a
bc
Laurent Lisnard,
Damien Dambournet,
Olaf J. Borkiewicz,d
e
e
f
Marie-Laure Boillot,
Laure Catala,
Antoine Tissot
and
a
Rodrigue Lescouezec
*
¨
Responsive materials that can answer to chemical or physical external stimuli offer numerous prospects
in material science. Here, we have elaborated a two-step synthetic approach that allows incorporating
molecular cubic switches into a polymeric material. Firstly, a preformed half-capped, Cs+-templated
{Fe4Co4} cyanido-polymetallic cubic unit (‘‘pro-cube’’) is obtained and proven to be stable in solution, as
demonstrated by paramagnetic NMR. Secondly, the reaction of the pro-cube with a ditopic scorpionate
ligand enables the precipitation of a polymeric network containing the cubic unit. Furthermore, the
adequately chosen ditopic ligand that coordinates the Co ions of the pro-cube allows us to preserve the
switchable properties of the cubic unit. Indeed, the magnetic properties of the polymeric material
compare well with those of the molecular cubic model that is obtained by reacting a non-bridging
scorpionate ligand, and that was prepared as a reference. Both the polymeric material and the molecular
model cube show a thermally-induced metal–metal electron transfer near room temperature. Interestingly,
the magnetic state of the polymeric material is shown to depend on its hydration state, indicating its
capability to act as a chemo-sensor.
Received 20th April 2021,
Accepted 20th June 2021
DOI: 10.1039/d1tc01825f
rsc.li/materials-c
class of molecular switches are the spin-crossover (SCO) com-
plexes that undergo a spin-state change under the application
Introduction
The design of responsive materials capable of changing their of different stimuli (light, temperature, pressure, etc.).5–9 In
physical properties upon exposure to external, chemical or particular, octahedral Fe(II) SCO complexes showing radical
physical stimuli is an active topic in materials science with changes in their optical, magnetic and dielectric properties
numerous potential applications such as in sensors, information due to the reversible conversion between the 1A1g (t26g) and 5T2g
storage and medical science.1–4 A possible approach for synthesizing (t24ge2g) states, have attracted strong attention.8,10,11 The common
such materials consists in using metal-based complexes that strategy to design responsive materials based on SCO complexes
can be reversibly switched between two different electronic consists in reacting metal ions with bridging ligands that induce
states, associated with different physical behaviours. A prominent an adequate ligand field in order to observe a spin-crossover
phenomenon. For example, the simple reaction of Fe(II) metal
salts with the triazolyl ligand (Tz) leads to the one-dimensional
a
´
´
Sorbonne Universite, CNRS, UMR 8232, Institut Parisien de Chimie Moleculaire,
(1D) polymer material [FeII(Tz)3]n showing magnetic and
optical bistability near room temperature.12 More sophisticated
strategies leading to functional materials have also been
explored. For instance, the reaction of poly-cyanometallates in
conjunction with pyridyl-type ligands and Fe(II) salts was used to
design 2D or 3D frameworks showing switchable properties.13
Interestingly, Real et al. showed that this strategy allows
synthesizing porous materials that show interesting chemo-
sensing properties thanks to the remarkable sensitivity of the
SCO centres to their chemical surrounding.14–17 The cyanido-
bridged charge transfer (CT) complexes that can be switched
between two electronic states represent another interesting class
F-75005 Paris, France. E-mail: rodrigue.lescouezec@sorbonne-universite.fr
b
c
´
´
Sorbonne Universite, CNRS, UMR 8234, Physico-chimie des electrolytes et
`
nano-systemes interfaciaux, PHENIX, F-75005 Paris, France
´
Reseau sur le Stockage Electrochimique de l’Energie (RS2E), FR CNRS 3459,
80039 Amiens, France
d X-ray Science Division, Advanced Photon Source, Argonne National Laboratory,
9700 South Cass Avenue, Argonne, Illinois 60439, USA
e
´
´
Universite Paris-Saclay, CNRS, UMR 8182, Institut Chimie Moleculaire et
´
Materiaux d’Orsay, 91405, Orsay, France
f
´
´
Institut des Materiaux Poreux de Paris, Ecole Normale Superieure, ESPCI Paris,
CNRS, PSL University, 75005, Paris, France
† Electronic supplementary information (ESI) available. CCDC 2072191. For ESI
and crystallographic data in CIF or other electronic format see DOI: 10.1039/
d1tc01825f
8882 | J. Mater. Chem. C, 2021, 9, 8882–8890
This journal is © The Royal Society of Chemistry 2021

View Article Online
Paper
Journal of Materials Chemistry C
of molecular switches, which share many features with SCO synthesis38,40–43 by reacting all the subcomponents together at
complexes.18–21 The FeCo Prussian blue analogues (PBA) were room temperature: the [FeIII(Tp)(CN)3]À complex, a divalent
the first switchable charge transfer systems to be reported.22 metal salt of M, an alkali salt and the RTpÀ anionic blocking
In these coordination polymers, light irradiation, temperature ligand. However, such a one pot reaction is not always successful
II
LS
III
LS
change or pressure variation allow converting Fe –Co diamag- or does not give the best yield. For example, in the case of the
III
LS
II
HS
netic pairs into Fe –Co paramagnetic ones, leading to drastic CsC{Fe4Mn4} cubic complex one needs to form a pro-cube
changes in the magnetic properties of the solids.23–26 These AC{[FeIII(Tp)(CN)3]4[Mn(S)3]4} (S = solvent) that is further reacted
purely inorganic coordination polymers are readily obtained by with a RTp blocking ligand.40 The nature of the solvent was
reacting hexa-cyanometallates with metal salts. Overall, in the proven to be critical in the pro-cubes and cube’s synthesis as well
above examples, the responsive materials are obtained by using as the reactants’ concentrations. Lower concentrations were
monometallic cyanido complexes to lead to 2D or 3D coordina- shown to favour the cube’s dissociation and the precipitation
tion polymers. The use of preformed polymetallic cyanido-based or crystallization of undesired polymeric coordination phases.
building units (BU) has been scarcely explored. However, some For instance, at low concentration the known cyanide-bridged
works showed that this strategy could lead to the preparation of chain compound of formula {[FeIII(Tp)(CN)3]2[MII(S)2]} (S =
cyanide coordination polymers with large voids or porosity.27–30 solvent)44 can be obtained as a side-product. Interestingly, our
For example, Long et al. showed that the reaction of octahedral previous studies showed that the introduction of Cs+ as a
cyanido clusters, [Re6Se8(CN)6]4À, leads to three-dimensional templating ion confers a higher stability to the {Fe4Co4} cube
coordination polymers described as ‘‘expended PBA’’ and and allows its synthesis with higher yield.38 Considering these
showing enhanced ion intercalation properties.31 Interestingly, previous results, a step-by-step strategy was adopted in the
Nihei, Oshio et al. recently used an elegant supramolecular present work. Thus, Cs-based pro-cubes were formed in a first
approach to design responsive molecular materials by assem- stage by reacting all precursors (except the blocking ligand, TpÀ)
bling {Fe2Co2} switchable CT complexes through donor–acceptor in an adequate solvent. This should guarantee the stability of the
hydrogen-bond interactions.32–34 However, in these cases the cubic nodes during the network assembly. The stability of the
1
resulting responsive materials are fragile as the switchable FeCo octametallic building units was checked by H and 133Cs NMR
complexes are linked by weak supramolecular interactions. The spectroscopy (Fig. 3 and ESI†). Once formed, these pro-cubes
use of polymetallic building units is much more common in the were connected to each other by using a polytopic organic linker
field of metal–organic frameworks (‘‘MOFs’’), where the so-called of the TpÀ family, the Tp–Ar–Tp ligand depicted in Scheme 1.
‘‘secondary building units’’ (SBU) are generally formed in situ This strategy allows avoiding the precipitation of undesired
during the reaction.35,36 Inspired by these examples and in polymeric species such as the neutral {Co(Tp–Ar–Tp)}n. The
search of new routes to form robust responsive materials that reaction leads to the desired material of formula
are able to interact with their surroundings, we decided to CsC{[Fe(Tp)(CN)3]4[Co4(Tp–Ar–Tp)1.9]}Á10DMF (see characterizations
explore an original synthetic approach based on the assembly below) as a highly insoluble amorphous precipitate. All our
of polymetallic FeCo CT complexes that could be covalently attempts to slow down the reaction and to obtain crystalline
linked to each other through rigid organic linkers in order to products failed (e.g. H-tubes and crystallization in silica gels).
obtain stable networks. Among the large variety of FeCo switchable It is worth noticing that a significant colour change occurs
charge transfer complexes, we selected the cubic cyanide-bridged during the reaction of the pro-cube with the bridging Tp–Ar–Tp
complexes of general formula AC{[Fe(Tp)(CN)3]4[Co(RTp)]4} (abbre- ligand (detailed in experimental part), which is also observed
viated AC{Fe4Co4}). In these cubic molecules, A is an inserted
alkali ion, the edges are built of cyanide bridges while tris(pyrazol-1-
yl)borates anions (Tp and RTp, R being a functional group) act as
blocking ligands to prevent polymerization. Beside showing both
photo- and thermo-induced electron transfer, these cubes also
show remarkable redox properties: they exhibit up to nine
reversibly accessible redox states associated with electrochromic
properties.37–40 More importantly, these cubes are soluble
molecular models of the FeCo PBA and can be extremely stable
in solution, making them ideal building units for the design of
high dimensionality polymers. We thus present here our first
results in the design of cube-based responsive polymers.
Results and discussion
Synthesis of the molecular cubic model and cube-based network
Most of the cyanido AC{Fe4M4} cubic complexes (M = Mn, Fe,
Scheme 1 Synthetic pathways of (a) the molecular cubic switch, (b) the
cube-based network and, (c) a Prussian blue analogue.
Co, Ni) that we previously reported can be obtained in a one pot
This journal is © The Royal Society of Chemistry 2021
J. Mater. Chem. C, 2021, 9, 8882–8890 | 8883

View Article Online
Journal of Materials Chemistry C
Paper
when reacting the pro-cube with the Tp blocking ligand
forming the new molecular model CsC{[Fe(Tp)(CN)3]4[Co(Tp)]4}
(see ESI†). This colour change from red to deep green accounts
for the occurrence of a metal–metal electron transfer during the
synthesis where some FeIII–CN–CoII pairs are being converted
into FeII–CN–CoIII ones. Indeed, FeIICoIII pairs present an
intense charge transfer transition near 800 nm while CoIIFeIII
pairs present a charge transfer transition near 500 nm.45 Here,
the substitution of the DMF solvent molecules by the Tp anionic
ligand induces a stronger ligand-field that is expected to favour
the CoIII oxidation state. Such phenomenon was already reported
in a previous study where the reaction of pzTp blocking ligand
(pzTp = tetra(pyrazolyl)borate anionic ligand) with the red
{FeI4IICo4II} pro-cube led to a deep green CsC{Fe4IICoI3IICoII} cubic
complex.38 In this peculiar case, one of the cobalt remained in
the +II oxidation state and the charge was balanced by the
inserted alkali ion to yield a neutral cubic complex. Here, we
observe a noticeable difference with our previous report. If the
reaction of the pro-cube with TpÀ does initially lead to the same
CsC{Fe4IICo3IIICoII} electronic state (as shown by the NMR data
below), the cubic complex, which is more soluble in DMF
Fig. 1 Perspective view of the XRD structure of the model paramagnetic
cubic complex CsC{[FeII(Tp)(CN)3]4[CoIII(Tp)]3[CoII(Tp)]}. C: grey, N: light
blue, B: yellow, Fe: orange, Co: blue, Cs: violet. (H atoms are omitted for
clarity).
than the pzTp-based analogue cubic complex, shows a slow value for CoII–N bond lengths).46,47 The average FeÁÁÁCo distance,
oxidation over time in air. Therefore, depending on the crystal- 5.000(2) Å, is slightly longer than the expected FeII–CoIII value and
lization conditions, the cube can be isolated in its neutral form coherent with that hypothesis.24 The presence of the CoII ion, is
CsC{Fe4IICoI3IICoII} or in the fully oxidized CsC{Fe4IICoI4II}ClO4 unambiguously proved by the solution 133Cs, 1H NMR spectra and
form as a perchlorate salt (see below).
This observation indicates that the electronic state of the the absence of counter-anion in the crystal structure (Fig. 1).
polymeric material obtained by the reaction of the {Fe4Co4} Concerning the polymeric compound, the SEM images (see
magnetic measurements (see below and ESI†) and coherent with
pro-cube and the bridging ligand Tp–Ar–Tp should be carefully ESI†) reveal that it precipitates as small anisotropic particles
analysed. Thus, we first describe here the structural features of with long edges of ca. 0.2 mm. X-ray diffraction analysis
both the cubic model compound and the polymeric network. performed on the precipitate revealed the absence of long-
Then, by using a multi-technique approach, the electronic state range order with diffuse scattering characteristic of amorphous
of the materials is carefully characterized by the use of FT-IR, compounds. We used high-energy (l
= 0.2116 Å) X-ray
NMR, UV-vis spectroscopy and elemental analysis (based on synchrotron experiments to measure total scattering data,
atomic absorption and Energy Dispersive spectrometry). which were Fourier transformed to obtain the pair distribution
Finally, in the last section, we investigate to which extent the function (PDF). PDF represents the probability of finding a pair
switchable properties of the molecular cubes can be transmitted of atoms separated from a distance r. As a total scattering
to the polymeric material. In particular, we study the changes of analysis, it can probe the short- and long-range order whatever
the magnetic properties upon intercalation and deintercalation the crystalline state of the material.48,49 For example, it was
of water molecules into the amorphous framework.
shown to be well-suited to study the local structure of
MOFs.50,51 Fig. 2 shows the PDFs of the polymeric compound
and the cube model for comparison purpose along with the
Structural characterization
Suitable crystals for X-ray diffraction of the molecular model assignments of the inter-atomic distances r. Note that as an
CsC{[FeII(Tp)(CN)3]4[CoIII(Tp)]3[CoII(Tp)]}Á14CH3CNÁH2O were X-ray based technique, the signal scales with the atomic mass
obtained by slow evaporation of an acetonitrile solution (cubic of the scattered element. At low r, the PDFs include the inter-
model). The material crystallizes in the Pbca space group and atomic distances related to light organic moieties. The only
the asymmetric unit is composed of only 1/2 of the cube, which substantial difference between the two samples is observed in
corresponds to a face of the cubic molecule. In other words, the that region with a broad and intense peak centred at 1.33 Å
Fe and Co are not discernible in the present case, neither are for the cube model that is barely observed for the polymeric
the C and N atoms of the cyanide bridges. The experimentally network. This peak corresponds to N–N, C–N and C–C distances
measured metal–cyanide bond lengths, 1.927(4) Å, thus represent and could arise in part from the acetonitrile solvent molecules
an averaged value of the Fe–C and Co–N distances. This averaged within the crystal lattice. The polymeric network shows a
value is coherent with the occurrence of only one CoII ion per distinct peak at 1.6 Å enlisting B–N and C–C and C–N single
cube, which implies the presence of 21 metal–cyanide distances of bonds. Thereafter, two interatomic distances characteristic for
approx. 1.9 Å (the expected value for the CoIII–N and FeII–C bond metal–ligand bonds can be clearly distinguished: (i) one
lengths)46,47 and 3 other distances close to 2.09 Å (the expected intense peak at 1.93 Å assigned to Fe–C/N and Co(III)–N bonds
8884 | J. Mater. Chem. C, 2021, 9, 8882–8890
This journal is © The Royal Society of Chemistry 2021

View Article Online
Paper
Journal of Materials Chemistry C
the metal ions (Co/Fe) and the second neighbour atoms (C/N)
of the cyanide ligand; the boron atom and the transition
metals. The peak at 2.62 Å could account for the interaction
between C/N atoms and the transition metal from the perpendicular
cubic edges. As we moved toward higher r values, some peaks
can be seen as the fingerprints of the cubic cage. The presence
of an encapsulated Cs+ ion is visible through its interaction
with the CN bridges of the cubic edge that falls in the range of
small peaks from 3.5 to 3.7 Å (as observed for the molecular
model cube). The distances between Cs+ and the transition
metals are in the range of 4.0–4.4 Å. As previously shown,
the Cs+ atom is not necessarily in the exact centre of the
cube.38,40,52 Moreover, FeÁ Á ÁCo distances corresponding to the
edges of the cubic units are particularly visible with the intense
peak located at 4.95 Å, while those arising from the cube face
diagonal observed at around 7 Å, are barely detectable within
the two samples. Finally, at higher r, the PDF fades as expected
for short-range ordering. In conclusion, the pair distribution
function clearly demonstrates that the cube integrity is
maintained during the network self-assembly.
Fig. 2 Pair distribution function analysis of the polymeric network (red)
and of the model cube (blue); r represents the interatomic distance.
The chemical composition of the polymeric network was
from the pyrazolyl ligands;46,47 (ii) one smaller peak at 2.22 Å, checked by EDX analysis that indicates an Fe/Co ratio of
which clearly attests for the presence of Co(II)–N bonds. This approximately 1/1, which is in agreement with a polymeric
indicates that within the synthesis conditions, some cobalt ions structure made of cubic units and an ideal composition of
are not oxidized. The broad intense peak centred at 3.03 Å is (CsC{[Fe(Tp)(CN)3]4[Co4(TpArTp)2]})n. However, the combined
assigned to different contributions: e.g. the interaction between results of elemental analysis and thermo-gravimetric experiments
Fig. 3 (a) FT-IR spectra (b) UV-vis spectra (c) 133Cs-NMR of 1. The cubic model 2. The pro-cube and 3. The polymetallic network. Notes: (b1) diamagnetic
cube in dotted line and paramagnetic cube in continuous line. b1 and b2 are solution UV-vis measurements while b3 is a solid-state measurement; c1 and
c2 are solution NMR spectra while c3 is a solid-state MAS NMR spectrum.
This journal is © The Royal Society of Chemistry 2021
J. Mater. Chem. C, 2021, 9, 8882–8890 | 8885

View Article Online
Journal of Materials Chemistry C
Paper
suggest that the DMF solvent molecules of the Co centres are conversion CsC{FeI4ICo3IIICoII} (paramagnetic) - CsC{Fe4IICo4III}
not all replaced by the Tp–Ar–Tp bridging ligand. The material (diamagnetic) is the same as that observed with spectra
seems to precipitate before a complete substitution occurs. recorded at controlled potential (UV-vis electrochemistry
The elemental analysis is better in line with the following coupled experiments) in an analogue cubic complex bearing
composition: (CsC{[Fe(Tp)(CN)3]4[Co4(TpArTp)1.9]}Á10DMF)n pzTp ligand instead of the Tp on the cobalt ions.38 Finally, the
that has been used in the magnetic study below.
UV-vis spectrum of the polymeric material recorded in the solid-
state shows a MLCT band characteristic of the {FeII(Tp)(CN)3} unit
and a broad band centred at 630 nm (Fig. 3b3), similar to that
observed in the cubic model complex, which is coherent with the
Characterization of the electronic state of the cubic units inside
the polymeric network
In order to elucidate the electronic state of the metal ions in presence of FeII–CN–CoIII pairs.
the cubic units contained in the polymeric material, various
The electronic state of the pro-cube, the cubic model and the
spectroscopic data have been collected and compared to the polymeric material were also probed using 133Cs NMR (Fig. 3c)
results obtained for the pro-cube and for model cubic units. and 1H NMR (see ESI†). Although the 133Cs is a quadrupolar
Relevant FT-IR, UV-vis and NMR spectroscopic data are nucleus (I = 7/2), the quadrupolar constant is very weak so
gathered in the Fig. 3.
that well-resolved thin peaks can be obtained in solution. As
FT-IR spectroscopy is a very common tool to probe the demonstrated in previous studies, the inserted cation is highly
electronic state of cyanide materials, as the stretching cyanide sensitive to small amounts of spin-density brought by the cubic
vibrations strongly depend on the coordination mode of the cages.38,40,42 Thus, 133Cs NMR can be used to probe the
cyanide ligand (bridging or terminal) and the oxidation state of electronic state of the molecule. Moreover, the presence of
the linked metal ions.18,53 FT-IR spectra of the pro-cube, the one Cs+ ion per cubic unit makes the analysis quite straightforward.
model cube and the polymeric material are gathered in Fig. 3a. In the 133Cs NMR a single broad peak at ca. À720 ppm is observed
The stretching vibration located at 2162 cmÀ1 in the pro-cube for the pro-cube in CD3CN solution (Fig. 3c2). This value is far
(Fig. 3a2) is a clear indication of the occurrence of the FeIII–CN– beyond the 133Cs diamagnetic range (+130/À30 ppm for CsNO3
CoII linkages in this species, which is coherent with the red 0.1 M in D2O) and accounts for the presence of only CsC{Fe4IIICoI4I}
colour of the material. In contrast, the cyanide stretching paramagnetic species in solution. Upon formation of the model
vibration is shifted to lower wavenumbers (ca. 2125 and cube by coordinating the TpÀ ligand, the 133Cs NMR chemical shift
2105 cmÀ1) that are characteristic of FeII–CN–CoIII linkages in decreases to À232 ppm (in CD2Cl2) but remains out of the
the diamagnetic cubic model (Fig. 3a1). This is coherent with diamagnetic range (Fig. 3c1). This agrees with the formation of
the occurrence of an electron transfer upon the coordination of paramagnetic cubic cages with only one paramagnetic CoII ion,
the Tp ligand to the pro-cube and with the colour change from CsC{FeI4ICo3IIICoII}, as observed in previously published cubic
red to blue. When the paramagnetic cube is isolated, the IR complexes.38,41 If the solution is left for two days in air, a slight
spectrum slightly differs (see ESI†) because of the presence of colour change is observed (see UV-vis data above), and the 133Cs
an additional FeII–CN–CoII linkage that leads to a small NMR signal shifts to the diamagnetic range at ca. +50 ppm.
shoulder near 2087 cmÀ1. The stretching vibrations in the This result fully agrees with the above-mentioned data, in
polymeric materials allow discarding the occurrence of FeIII– particular the slow crystallization of diamagnetic cubes as seen in
CN–M moieties but they are coherent with the occurrence of the XRD analysis. It is worth noting that the 1H NMR spectra
both FeII–CN–CoIII and FeII–CN–CoII linkages. The peak appearing (shown in ESI†) of the pro-cube and cubic model complex are more
at ca. 2050 cmÀ1 can be ascribed to the FeII–CN–CoII pair while the complicated than the 133Cs ones because of the multiplicity of 1H
peak centre at 2104 cmÀ1 corresponds to FeII–CN–CoIII pair sites. However, relevant assignments have been done and are fully
(Fig. 3a3). The cyanide stretching peak is broad, which could coherent with the 133Cs data: a unique paramagnetic species is
account for the presence of a distribution of environments around observed in the pro-cube solution, which first converts into a
Fe–CN–Co bridges.
CsC{Fe4IICoI3IICoII} cube upon addition of the ligand. As the
The recorded UV-vis absorption spectra of the pro-cube, solution of the model cube is left at air, the signals
cubic model and polymeric material (Fig. 3) are coherent with become diamagnetic and can be assigned to a diamagnetic cube
the above-mentioned FT-IR data. The spectrum of the pro-cube CsC{FeI4ICo4III}+ of higher symmetry.
(Fig. 3b2) is dominated by a very strong Ligand-Metal Charge
Concerning the polymeric material, the 1H signal in the
Transfer (LMCT) band centred at approximately 445 nm and solid-state is broad because of the dipolar coupling and no
typical for the [FeIII(Tp)(CN)3] complex.40,54 The spectrum of the relevant information can be extracted. However, the situation is
model cube is notably different from the pro-cube. In its much better for the solid-state 133Cs NMR spectrum, where
paramagnetic neutral state CsC{FeI4ICoIIICoII}, the spectrum is different signals can be clearly distinguished. A signal at
dominated by two broad absorption bands centred at 413 and +40 ppm indicates the presence of coprecipitated CsClO4.
614 nm that are ascribed to a MLCT band of the {FeII(Tp)(CN)3} Moreover, three signals are observed in the paramagnetic range
subunits and the FeII–CoIII Metal–Metal Charge Transfer band at approximately À205, À285 and À370 ppm (Fig. 3c3), the last
(MMCT), respectively (Fig. 3b1). As the compound gets oxidized one being of weak intensity (the other small signals are
the MMCT band notably increases and shifts to higher energy ascribed to side band signals as shown in ESI†). The regular
(593 nm). This change of the optical signature upon the splitting of the chemical shift of the 133Cs probe could be
8886 | J. Mater. Chem. C, 2021, 9, 8882–8890
This journal is © The Royal Society of Chemistry 2021

View Article Online
Paper
Journal of Materials Chemistry C
explained by the presence of an increasing number of para- ones (Fig. 4(a)). The transition is gradual and remains incomplete
magnetic ions in the cubic units. Such behaviour is actually at 400 K (the highest available temperature in the magnetometer).
quite similar to that observed in our previous studies on the The value obtained at 400 K, 4.3 cm3 molÀ1 K, suggests that
model Fe–Cd Prussian Blue analogues. In this work, we showed approximately one FeCo pair per cube is converted on average.
that increasing the number of paramagnetic species around a This estimation is based on the wMT value of the Co(II) ion
diamagnetic 113Cd probe leads to an almost regular increase of measured at high temperature in a paramagnetic cube model
the chemical shift.55,56 In the present case, the peaks at À205, (3.0 cm3 molÀ1 K)46 and the previously reported wMT value of the
À285 and À370 ppm can thus be tentatively assigned to the [FeIII(Tp)(CN)3]À subunit.58 The transition is irreversible: upon
following cubic units CsC{FeI4ICoIII3CoII}, CsC{Fe4IICoIII2CoII
}
cooling from 400 K to low temperature the wMT value slowly
2
and CsC{FeI4ICoIIICo3II}. It is worth noticing that the presence of decreases, which can be ascribed to the spin–orbit coupling of
a high number of CoII ions in octahedral surrounding is in line both CoII
and Fe ions (see ESI†). More interestingly, a
LS
III
HS
with the magnetic data described below. Obviously, from a photomagnetic effect is observed on the cubic models at low
chemical point of view, the presence of CoII sites should be temperature. Upon irradiating the diamagnetic or paramagnetic
favoured by an incomplete substitution of the DMF solvent samples at 20 K at 808 nm (5 mW cmÀ2), a significant increase of
molecules by the Tp–Ar–TpÀ ligand. The latter may not replace magnetization occurs (see ESI†), which is ascribed to an ETCST as
all the coordinated DMF ligands as the compound readily recently demonstrated on a related compound.39 The overall
precipitates when it forms. The presence of structural defects increase of wMT value obtained after saturation, ca. 6.1 and
is also coherent with the amorphous nature of the material. 5.6 cm3 molÀ1 K, corresponds roughly to the conversion of 3 or
In summary, the complementary structural and electronic 2 FeCo pairs.38 Fig. 4d show the wMT versus T curve of the resulting
analysis allows describing the polymeric material as an photo-induced metastable state upon heating from 2 K to
amorphous network made of robust cubic units in which the high temperature. The variation of the wMT value at very low
electronic state is controlled by the substitution of the DMF temperatures could be ascribed to antiferromagnetic interactions.
solvent molecules by the Tp–Ar–Tp ligands.
The decrease of the wMT value observed upon increasing the
temperature (scan rate of 0.4 K minÀ1), is due to the thermally-
induced depopulation of the metastable state. The relaxation
Switchable electronic properties
The magnetic measurements on the network and on the cubic temperatures measured in the present experimental conditions
molecular model were investigated in the temperature range are given by the inflexion points and correspond to ca. 70 and
between 2–400 K. Fig. 4(b) shows the wMT versus T curve (wM is 190 K for the paramagnetic and diamagnetic cubic models,
the molar magnetic susceptibility per Fe4Co4 formula unit) of respectively.
the paramagnetic model cube (black curve). At 300 K, the
The wMT versus T curve of the polymeric network (Fig. 4(b)
measured wMT value of 2.6 cm3 molÀ1 K, is coherent with the red curve) shows a similar shape as that described for
presence of one octahedral Co(II) ion per cubic unit on average the molecular cube, although slight differences are observed.
(S = 3/2, geff = 2.8). The non-linearity of the wMT curve upon The wMT value measured at 300 K, 3.2 cm3 molÀ1 K, is higher
cooling is due to the spin–orbit coupling of the octahedral CoII than that measured for the model cube, which is coherent with
complexes as it has been well described elsewhere.57 More the presence of a higher number of CoII paramagnetic ions that
interestingly, above 300 K the significant increase of wMT can be would arise from the presence of CsC{FeI4ICoIII2CoII2} and
ascribed to a thermally-induced electron transfer coupled to a CsC{FeI4ICoIIICo3II} units in the polymeric framework, as
II
LS
III
LS
spin transition (ETCST) on the Co(II) ion: some Fe –Co dia- suggested above. However, a sizeable increase of wMT is clearly
III
LS
II
HS
magnetic pairs being converted into Fe –Co paramagnetic observed above 300 K and ascribed to an ETCST. The wMT value
Fig. 4 (a) Scheme of the electronic reorganisation occurring upon ETCST. (b) wMT versus T curves measured on: (i) freshly filtered crystals of the
paramagnetic CsC{FeI4ICo3IIICoII} cubic model (black), (ii) the polymeric network (red), (iii) the network recovered after magnetic measurement and
rehydrated (blue). (c) wMT values measured at 300 K of a sample (polymeric network) successively hydrated and dehydrated sample; (d) wMT versus T
curves of the photo-induced metastable states obtained after irradiation at 808 nm (see text).
This journal is © The Royal Society of Chemistry 2021
J. Mater. Chem. C, 2021, 9, 8882–8890 | 8887

View Article Online
Journal of Materials Chemistry C
Paper
obtained at 400 K, ca. 5.3 cm3 molÀ1 K, corresponds to a partial red solution stirred for 5 minutes. Then, CsI (154 mg, 0.6 mmol) or
conversion of the diamagnetic pairs into paramagnetic ones. CsClO4 (140 mg, 0.6 mmol) in 2 mL DMF was added and the
The variation of wMT value between 300 and 400 K is similar to resulting solution stirred. After 5 minutes the ligand KTp (76 mg,
that observed in the model paramagnetic cube and corresponds 0.3 mmol) was added, and the greenish blue solution was stirred for
roughly to a conversion rate of 35% (considering that three 4 hours. The excess Cs-salt was separated via centrifugation and the
diamagnetic pairs could be converted per cubic units). As for resulting solution was precipitated with 40 mL ether. The resulting
the cubic model, the transition is irreversible and cannot be solid was washed with DMF/ether (1/8) and crystalized in acetonitrile
observed when cooling from 400 K (ESI†). However, when the and characterised by NMR.
sample is dipped into water and quickly dried in air before
Magnetic measurements
measuring again in the magnetometer (Fig. 4(b) blue curve), the
ETCST transition can be recovered. This procedure has been
repeated and Fig. 4(c) show the approximate wMT values
obtained before the magnetic measurement once the sample
is hydrated and after heating at 400 K. This is in striking
contrast with the situation of the cubic model whose transition
cannot be recovered after such treatment. We assume that the
polymeric structure of the network and its insolubility prevents
the cubic units from decoordination. The adsorption of solvents
into the framework would thus lead back to a diamagnetic state.
Such chemo-sensitive spin-state changes have already been
observed in related polymeric spin-crossover compounds.17,59
Moreover, recent studies on FeCo ETCST molecular systems
showed that interaction with solvate or small molecules inside
the crystal lattice can modulate the electronic state of these
systems.60–62 A similar phenomenon is observed here, making
the polymeric material a potential chemo-sensor. Finally, the
photomagnetic properties of the network were also probed,
however it was not possible to observe any increase of
magnetization upon irradiating the sample at 808 nm and
20 K. This might be ascribed to the higher rigidity of the network
that hinders the structural changes accompanying the ETCST
process.63
DC Magnetic susceptibility measurements were carried out
using a Quantum Design SQUID magnetometer. The magnetic
susceptibility values were corrected by subtracting the
diamagnetism of the molecular constituents and of the sample
holder. The samples (packed in a polyethylene bag) were
introduced in the SQUID at 110 K under helium flow to avoid
solvent loss. The magnetic field was set at 10 000 Oe and the
. Photomagnetic
measurements were carried out at 20 K on ca. 0.3 mg samples
deposited as thin films.
temperature sweep rate was 2 K minÀ1
Pair distribution function
High-energy X-ray data were collected at the 11-ID-B station at
the Advanced Photon Source (Argonne National Laboratory).
After corrections (background and Compton scattering), PDFs
were extracted from the data using PDFgetX2 software.64
X-Ray diffraction
Details about X-ray crystal structure determination are given in
ESI,† CCDC 2072191.
Other characterizations
Technical details about FT-IR, solution and solid-state UV-vis
and NMR spectroscopies, mass spectrometry, elemental analyses,
thermogravimetric analyses, electronic microscopy are given in
ESI.†
Experimental
Synthesis of the network
The ditopic ligand and its precursors have been synthesized as
in reported procedures (see details in ESI†).8,11 The network has
been prepared using the following procedure. To a stirred yellow
solution containing [NBu4][FeIII(Tp)(CN)3] (180 mg, 0.3 mmol) in
10 mL DMF CoII(ClO4)2Á6H2O (90 mg, 0.25 mmol) was added.
The colour of the solution turned to light red. After 2 min, CsI
(195 mg, 0.75 mmol) was added. After stirring for 2 min the
ditopic ligand (80 mg, 0.15 mmol) was added and the resulting
mixture was first sonicated and then stirred for 30 min. A green
precipitate forms right away. The mixture was then filtered and
washed with DMF and hot water to yield 240 mg of deep green
solid as a product. Elemental analysis: found: C (%) 38.76 H (%)
3.76 N (%) 24.48. Calculated for the network with 1.9 eq. of
ligand, 10 DMF molecules, coprecipitated CsClO4 C (%) 38.50 H
(%) 3.94 N (%) 24.99 EDX analysis Fe : Co: Cs = 4 : 4 : 2.
Conclusions
In this work, we have shown that polymetallic cyanido-bridged
{Fe4Co4} molecular switches can be integrated into polymeric
materials by using a stepwise molecular chemistry approach.
While the chemical approach based on the use of stable
pro-cube allows us to maintain the structural integrity of the
cubic nodes, the rapid bond formation with the organic linkers
and high insolubility of the resulting network certainly favour
the precipitation of an amorphous material. As for the {Fe4Co4}
molecular model, the cubic nodes in the network material are
in different electronic states. Spectroscopic evidences suggest
the presence of further electronic states besides the
paramagnetic and diamagnetic states observed in the model
cube. This is coherent with the fact that not all solvents on the
Co coordination sphere are substituted by bridging ligands.
Synthesis of model cubic complex
To a stirred solution of [NBu4][FeIII(Tp)(CN)3] in 2 mL DMF (173 mg, However, the thermo-induced ETCST observed in the cube was
0.3 mmol) CoII(ClO4)2Á6H2O (108 mg, 0.3 mmol) was added and the transferred to the network. More interestingly, while the
8888 | J. Mater. Chem. C, 2021, 9, 8882–8890
This journal is © The Royal Society of Chemistry 2021

View Article Online
Paper
Journal of Materials Chemistry C
and H. A. Goodwin, Springer Berlin Heidelberg, Berlin,
Heidelberg, 2004, pp. 229–257.
phenomenon is irreversible upon heating of the molecular
model, it can be retained in the polymeric materials, making
this material a possible chemo sensor. At the present stage, our
current research efforts are focused on (i) the exploration of
related chemical approaches that are susceptible to lead to
crystalline materials; and on the other hand, on (ii) the exploration
of the potentialities of these materials for chemo-sensing or as
intercalation materials.
˜
14 D. Aravena, Z. A. Castillo, M. C. Munoz, A. B. Gaspar,
K. Yoneda, R. Ohtani, A. Mishima, S. Kitagawa, M. Ohba,
J. A. Real and E. Ruiz, Chem. – Eur. J., 2014, 20, 12864–12873.
˜
˜
15 F. J. Munoz-Lara, A. B. Gaspar, M. C. Munoz, A. B. Lysenko,
K. V. Domasevitch and J. A. Real, Inorg. Chem., 2012, 51,
13078–13080.
˜
˜
16 F. J. Munoz-Lara, A. B. Gaspar, M. C. Munoz, M. Arai,
S. Kitagawa, M. Ohba and J. A. Real, Chem. – Eur. J., 2012,
18, 8013–8018.
Conflicts of interest
´
˜
17 M. Ohba, K. Yoneda, G. Agustı, M. C. Munoz, A. B. Gaspar,
J. A. Real, M. Yamasaki, H. Ando, Y. Nakao, S. Sakaki and
S. Kitagawa, Angew. Chem., Int. Ed., 2009, 48, 4767–4771.
There are no conflicts to declare.
`
18 C. Mathoniere, Eur. J. Inorg. Chem., 2018, 248–258.
Acknowledgements
19 Y.-S. Meng, O. Sato and T. Liu, Angew. Chem., Int. Ed., 2018,
57, 12216–12226.
20 D. Aguila, Y. Prado, E. S. Koumousi, C. Mathoniere and
This work was funded by the DIM Respore, the labex MiChem,
ANR MoMa, the CNRS and Sorbonne Universite. In the framework
´
`
`
´
of the CNRS RECIPROCS network, this work has been accepted
´
R. Clerac, Chem. Soc. Rev., 2016, 45, 203–224.
for synchrotron beamtime by the Soleil scientific proposal
committee (BAG proposals 20191503). The authors are grateful
to Pierre Fertey for his help on CRISTAL beamline at SOLEIL.
The work done at the Advanced Photon Source, an Office of Science
User Facility operated for the U.S. Department of Energy (DOE)
Office of Science by Argonne National Laboratory, was supported by
the U.S. DOE under contract no. DE-AC02-06CH11357.
21 G. N. Newton, M. Nihei and H. Oshio, Eur. J. Inorg. Chem.,
2011, 3031–3042.
22 O. Sato, T. Iyoda, A. Fujishima and K. Hashimoto, Science,
1996, 272, 704–705.
23 V. Escax, A. Bleuzen, C. Cartier dit Moulin, F. Villain,
A. Goujon, F. Varret and M. Verdaguer, J. Am. Chem. Soc.,
2001, 123, 12536–12543.
24 J.-D. Cafun, G. Champion, M.-A. Arrio, C. C. dit Moulin and
A. Bleuzen, J. Am. Chem. Soc., 2010, 132, 11552–11559.
25 N. Shimamoto, S. Ohkoshi, O. Sato and K. Hashimoto,
Inorg. Chem., 2002, 41, 678–684.
Notes and references
1 Molecular Switches, ed. B. L. Feringa, Wiley-VCH-Verl, Wein-
heim, 2011.
26 A. Bleuzen, J.-D. Cafun, A. Bachschmidt, M. Verdaguer,
´
P. Mu¨nsch, F. Baudelet and J.-P. Itie, J. Phys. Chem. C,
2008, 112, 17709–17715.
2 M. Irie, Chem. Rev., 2000, 100, 1683–1684.
3 W. Ke, W. Yin, Z. Zha, J. F. Mukerabigwi, W. Chen, Y. Wang,
C. He and Z. Ge, Biomaterials, 2018, 154, 261–274.
27 J.-J. Zhang and A. Lachgar, Inorg. Chem., 2015, 54, 1082–1090.
28 J. Zhang and A. Lachgar, J. Am. Chem. Soc., 2007, 129, 250–251.
29 Y. V. Mironov, N. G. Naumov, K. A. Brylev, O. A. Efremova,
V. E. Fedorov and K. Hegetschweiler, Angew. Chem., Int. Ed.,
2004, 43, 1297–1300.
30 L. G. Beauvais, M. P. Shores and J. R. Long, Chem. Mater.,
1998, 10, 3783–3786.
31 M. V. Bennett, M. P. Shores, L. G. Beauvais and J. R. Long,
J. Am. Chem. Soc., 2000, 122, 6664–6668.
´
4 G. Molnar, S. Rat, L. Salmon, W. Nicolazzi and
A. Bousseksou, Adv. Mater., 2018, 30, 1703862.
5 Spin Crossover in Transition Metal Compounds I, ed.
P. Gu¨tlich, H. A. Goodwin, Springer Berlin Heidelberg,
Berlin, Heidelberg, 2004.
6 P. Gu¨tlich and H. A. Goodwin, Spin Crossover in Transition
Metal Compounds II, Springer Berlin Heidelberg, Berlin,
Heidelberg, 2004.
32 M. Nihei, Y. Yanai, I.-J. Hsu, Y. Sekine and H. Oshio, Angew.
Chem., Int. Ed., 2017, 56, 591–594.
33 Y. Sekine, M. Nihei and H. Oshio, Chem. – Eur. J., 2017, 23,
5193–5197.
34 M. Nihei, Y. Yanai, D. Natke, R. Takayama, M. Kato,
Y. Sekine, F. Renz and H. Oshio, Chem. – Eur. J., 2019, 25,
7449–7452.
7 P. Gu¨tlich and H. A. Goodwin, Spin Crossover in Transition
Metal Compounds III, Springer Berlin Heidelberg, Berlin,
Heidelberg, 2004.
8 Spin-Crossover Materials: Properties and Applications, ed.
M. A. Halcrow, Wiley, Chichester, 2013.
9 A. B. Gaspar and B. Weber, in Molecular Magnetic Materials,
ed. B. Sieklucka, D. Pinkowicz, Wiley-VCH Verlag GmbH &
Co. KGaA, Weinheim, Germany, 2016, pp. 231–252.
10 F. Moreau, J. Marrot, F. Banse, C. Serre and A. Tissot,
J. Mater. Chem. C, 2020, 8, 16826–16833.
´
35 D. J. Tranchemontagne, J. L. Mendoza-Cortes, M. O’Keeffe
and O. M. Yaghi, Chem. Soc. Rev., 2009, 38, 1257.
36 M. J. Kalmutzki, N. Hanikel and O. M. Yaghi, Sci. Adv., 2018,
4, eaat9180.
11 M.-L. Boillot and B. Weber, C. R. Chim, 2018, 21, 1196–1208.
12 O. Kahn, Science, 1998, 279, 44–48.
´
37 D. Garnier, J.-R. Jimenez, Y. Li, J. von Bardeleben,
Y. Journaux, T. Augenstein, E. M. B. Moos, M. T. Gamer,
˜
13 Y. Garcia, V. Niel, M. C. Munoz and J. A. Real, in Spin
¨
F. Breher and R. Lescouezec, Chem. Sci., 2016, 7, 4825–4831.
Crossover in Transition Metal Compounds I, ed. P. Gu¨tlich
This journal is © The Royal Society of Chemistry 2021
J. Mater. Chem. C, 2021, 9, 8882–8890 | 8889

View Article Online
Journal of Materials Chemistry C
Paper
´
˜
38 J.-R. Jimenez, M. Tricoire, D. Garnier, L.-M. Chamoreau, 51 C. Castillo-Blas, J. M. Moreno, I. Romero-Muniz and
¨
J. von Bardeleben, Y. Journaux, Y. Li and R. Lescouezec,
A. E. Platero-Prats, Nanoscale, 2020, 12, 15577–15587.
Dalton Trans., 2017, 46, 15549–15557.
52 J. Glatz, L.-M. Chamoreau, A. Flambard, J.-F. Meunier,
´
´
¨
39 N. Daffe, J.-R. Jimenez, M. Studniarek, A. Benchohra,
A. Bousseksou and R. Lescouezec, Chem. Commun., 2020,
¨
M.-A. Arrio, R. Lescouezec and J. Dreiser, J. Phys. Chem.
56, 10950–10953.
Lett., 2019, 10, 1799–1804.
53 M. Shatruk, C. Avendano and K. R. Dunbar, in Progress in
Inorganic Chemistry, ed. K.D. Karlin, John Wiley & Sons, Inc.,
Hoboken, NJ, USA, 2009, pp. 155–334.
´
40 J. Jimenez, J. Glatz, A. Benchohra, G. Gontard, L. Chamoreau,
¨
J. Meunier, A. Bousseksou and R. Lescouezec, Angew. Chem.,
Int. Ed., 2020, 59, 8089–8093.
54 M. Nihei, M. Ui, N. Hoshino and H. Oshio, Inorg. Chem.,
2008, 47, 6106–6108.
´
41 D. Garnier, J.-R. Jimenez, Y. Li, J. von Bardeleben,
¨
¨
Y. Journaux, T. Augenstein, E. M. B. Moos, M. T. Gamer, 55 A. Flambard, F. H. Kohler and R. Lescouezec, Angew. Chem.,
¨
F. Breher and R. Lescouezec, Chem. Sci., 2016, 7, 4825–4831.
Int. Ed., 2009, 48, 1673–1676.
¨
¨
42 R. Plamont, J. Tami, J.-R. Jimenez, A. Benchohra, O. Khaled, 56 A. Flambard, F. H. Kohler, R. Lescouezec and B. Revel,
¨
G. Gontard, Y. Li and R. Lescouezec, J. Coord. Chem., 2018,
Chem. – Eur. J., 2011, 17, 11567–11575.
´
71, 601–614.
57 F. Lloret, M. Julve, J. Cano, R. Ruiz-Garcıa and E. Pardo,
´
43 A. Benchohra, C. Methivier, J. Landoulsi, D. Kreher and
Inorg. Chim. Acta, 2008, 361, 3432–3445.
¨
¨
R. Lescouezec, Chem. Commun., 2020, 56, 6587–6589.
58 R. Lescouezec, J. Vaissermann, F. Lloret, M. Julve and
44 H.-R. Wen, C.-F. Wang, Y. Song, S. Gao, J.-L. Zuo and
M. Verdaguer, Inorg. Chem., 2002, 41, 5943–5945.
˜
X.-Z. You, Inorg. Chem., 2006, 45, 8942–8949.
59 M. C. Munoz and J. A. Real, Coord. Chem. Rev., 2011, 255,
´
45 S. De, J.-R. Jimenez, Y. Li, L.-M. Chamoreau, A. Flambard,
2068–2093.
¨
Y. Journaux, A. Bousseksou and R. Lescouezec, RSC Adv., 60 C.-Q. Jiao, W.-J. Jiang, Y.-S. Meng, W. Wen, L. Zhao,
2016, 6, 17456–17459.
J.-L. Wang, J.-X. Hu, G. G. Gurzadyan, C.-Y. Duan and
´
´
`
46 D. Li, R. Clerac, O. Roubeau, E. Harte, C. Mathoniere, R. Le
T. Liu, Natl. Sci. Rev., 2018, 5, 507–515.
Bris and S. M. Holmes, J. Am. Chem. Soc., 2008, 130, 252–258. 61 M. Nihei, Y. Yanai, I.-J. Hsu, Y. Sekine and H. Oshio, Angew.
47 A. Mondal, Y. Li, M. Seuleiman, M. Julve, L. Toupet, Chem., Int. Ed., 2017, 56, 591–594.
M. Buron-Le Cointe and R. Lescouezec, J. Am. Chem. Soc., 62 Y. Sekine, M. Nihei and H. Oshio, Chem. – Eur. J., 2017, 23,
2013, 135, 1653–1656. 5193–5197.
48 S. J. L. Billinge and M. G. Kanatzidis, Chem. Commun., 2004, 749. 63 L. Trinh, S. Zerdane, S. Mazerat, N. Dia, D. Dragoe,
¨
´
`
49 T. L. Christiansen, S. R. Cooper and K. M. Ø. Jensen,
C. Herrero, E. Riviere, L. Catala, M. Cammarata, E. Collet
Nanoscale Adv., 2020, 2, 2234–2254.
and T. Mallah, Inorg. Chem., 2020, 59, 13153–13161.
50 K. W. Chapman, D. F. Sava, G. J. Halder, P. J. Chupas and 64 X. Qiu, J. W. Thompson and S. J. L. Billinge, J. Appl. Crystal-
T. M. Nenoff, J. Am. Chem. Soc., 2011, 133, 18583–18585. logr., 2004, 37, 678.
8890 | J. Mater. Chem. C, 2021, 9, 8882–8890
This journal is © The Royal Society of Chemistry 2021