Molecular Magnetic Materials Based on {CoIII (Tp*)(CN)3 }- Cyanidometallate: Combined Magnetic, Structural and 59 Co NMR Study.

The cyanidocobaltate of formula fac-PPh4 [CoIII (Me2 Tp)(CN)3 ] ⋅ CH3 CN (1) has been used as a metalloligand to prepare polynuclear magnetic complexes (Me2 Tp=hydrotris(3,5-dimethylpyrazol-1-yl)borate). The association of 1 with in situ prepared [FeII (bik)2 (MeCN)2 ](OTf)2 (bik=bis(1-methylimidazol-2-yl)ketone) leads to a molecular square of formula {[CoIII {(Me2 Tp)}(CN)3 ]2 [FeII (bik)2 ]2 }(OTf)2  ⋅ 4MeCN ⋅ 2H2 O (2), whereas the self-assembly of 1 with preformed cluster [CoII 2 (OH2 )(piv)4 (Hpiv)4 ] in MeCN leads to the two-dimensional network of formula {[CoII 2 (piv)3 ]2 [CoIII (Me2 Tp)(CN)3 ]2  ⋅ 2CH3 CN}∞ (3). These compounds were structurally characterized via single crystal X-ray analysis and their spectroscopic (FTIR, UV-Vis and 59 Co NMR) properties and magnetic behaviours were also investigated. Bulk magnetic susceptibility measurements reveal that 1 is diamagnetic and 3 is paramagnetic throughout the explored temperature range, whereas 2 exhibits sharp spin transition centered at ca. 292 K. Compound 2 also exhibits photomagnetic effects at low temperature, selective light irradiations allowing to promote reversibly and repeatedly low-spin⇔high-spin conversion. Besides, the diamagnetic nature of the Co(III) building block allows us studying these compounds by means of 59 Co NMR spectroscopy. Herein, a 59 Co chemical shift has been used as a magnetic probe to corroborate experimental magnetic data obtained from bulk magnetic susceptibility measurements. An influence of the magnetic state of the neighbouring atoms is observed on the 59 Co NMR signals. Moreover, for the very first time, 59 Co NMR technique has been successfully introduced to investigate molecular materials with distinct magnetic properties.

Introduction

Cyanide coordination chemistry has been widely used for decades to yield polynuclear coordination clusters and extended networks with adjustable structural and physical features. In particular, in the quest for building magnetic systems, the cyanide ligand has appeared advantageous because: (i) it can act as an efficient magnetic relay between paramagnetic spin carriers and thus promotes magnetic exchange interaction, (ii) its ambidentate nature allows accessing to a variety of M−CN−M’ bimetallic systems, some of them showing valence tautomerism, and (iii) the linear nature of the cyanide bridge can provide directionality, which allows reliable prediction of the magnetic exchange interaction in the resulting compound by means of simple orbital‐based model of the magnetic exchange interaction. Moreover, a variety of cyanidometallates building blocks [M(L)(CN)x]n− whose electronic properties and geometry can be modulated are achievable (L stands for a blocking ligand). This versatility makes possible the design of original molecular materials whose magnetic, optical, and electronic properties can be tuned in a systematic fashion through the self‐assembly of preformed cyanido building blocks.[ , , , ] In our group, we have been particularly interested in the use of the fac‐[FeIII(Tp)(CN)3]− complex (Tp=tris(pyrazol‐1‐yl)borate) that we first reported in 2002 and which we have used as a building block for designing molecular magnetic systems or redox active materials. Over the years, other groups have successfully used this building‐block and related Tp‐based tris‐cyanido complexes for designing original molecular magnetic materials such as Single Molecule Magnets (SMMs), Single Chain Magnets (SCMs), or stimuli‐responsive systems.[ , , ] In particular the [FeIII(RTp)(CN)3]− complexes have been broadly used in the last decade to design switchable polynuclear complexes where temperature, pressure change or light irradiation can trigger a metal‐metal charge transfer inside the FeII/III‐CN‐MII/III unit (M=Co, Mn).[ , , , , , ] When combined with {FeII(bik)2(S)2}2+ or {FeII(Tp)(S)3}+ building blocks (bik=bis(1‐methylimidazol‐2‐yl)ketone; S=labile solvent), we have also shown that tetra‐ or octa‐metallic mixed valence spin‐crossover (SCO) complexes can be obtained.[ , ] In continuation of this work, we have recently reported the synthesis and structure of a diamagnetic analogue: the fac‐PPh4[CoIII(Me2Tp)(CN)3] ⋅ CH3CN (1) building block featuring Me2Tp (hydrotris(3,5‐dimethylpyrazol‐1‐yl)borate) ligand. As for its paramagnetic counterpart, this diamagnetic complex can also be used as a building‐block to prepare polynuclear complexes with novel topology. Its diamagnetic nature does not prevent the observation of interesting magnetic properties as for example, its use as metalloligand toward {FeII(bik)2(S)3}2+ units should lead to switchable systems. Actually, the main objective of this contribution is two‐fold: (a) to explore the coordination chemistry of new cyanidometallate‐derived molecular magnetic materials and (b) to investigate if 59Co NMR ‐thanks to the NMR active nature of the Co(III) nuclei‐ can be used to probe the magnetic properties of these compounds. Actually, 59Co is an NMR‐active high sensitivity nucleus. It is a quadrupolar spin 7/2 nucleus with 100 % natural abundance. 59Co has also the broadest reported chemical‐shift range, about 18 000 ppm, which stresses its sensitivity to the local structure. So far, 59Co solid‐state NMR spectroscopy has been widely employed to probe various small molecules, battery materials, biological model systems, homogeneous catalysts and octahedral basic inorganic complexes. In most cases, the cobalt ion exhibits a +III oxidation state (in some cases it is +I) and the complexes are diamagnetic. When the cobalt atom carries unpaired electrons, the relaxation becomes too fast and no signal can be detected by standard NMR techniques. In our present work, we intend to probe the 59Co NMR chemical shifts of the diamagnetic Co(III) ions belonging to the {CoIII(Me2Tp)(CN)3}− building block engaged in paramagnetic polymetallic species. As the cyanide is efficient in transmitting spin density, one can expect these signals to be sensitive to the magnetic state of the neighbouring metal ions. With these aims, we report here the use of the fac‐[CoIII(Me2Tp)(CN)3]− as metalloligand: (a) toward the in situ formed mononuclear [FeII(bik)2(S)2]2+ complex to obtain a molecular square complex exhibiting spin transition, (b) toward a preformed dinuclear [CoII 2(μ‐OH2)(piv)4(Hpiv)4] complex (where Hpiv=pivalic acid or trimethylacetic acid) that results in a high nuclearity two‐dimensional (2D) paramagnetic network. The 59Co NMR chemical shift of these complexes and their temperature dependency are compared to the temperature dependent magnetic susceptibility (obtained from SQUID magnetometry) to probe the sensitivity of 59Co chemical shift to the magnetic properties of the materials. The study of the SCO based compound is particularly suited for carrying out this study as the Fe(II) spin state can change from diamagnetic to paramagnetic upon increasing temperature.

Results and Discussion

Similarly to previously reported square complexes, the tetrametallic complex, {[CoIII(Me2Tp)(CN)3]2[FeII(bik)2]2}(OTf)2 ⋅ 4MeCN ⋅ 2H2O (2) can be obtained by reacting 1 with the in situ prepared [FeII(bik)2(CH3CN)2](OTf)2 complex in acetonitrile at room temperature. 2 is isolated as dark red plate‐like crystals after slow evaporation (yield: 66 %). Compound 3, {[CoII 2(piv)3]2[CoIII(Me2Tp)(CN)3]2}∞, is obtained by self‐assembly of 1 with preformed dinuclear complex [CoII 2(μ‐OH2)(piv)4(Hpiv)4] in CH3CN. It is isolated as purple prismatic crystals by slow evaporation of an acetonitrile solution (yield: 81 %). It has already been observed that [CoII 2(μ‐OH2)(piv)4(Hpiv)4] can recombine in solution to yield homo‐ and hetero‐valent species with nuclearities ranging from three to fourteen Co ions. This recombination is very much dependent on temperature, which has a strong influence on the nuclearity of the final compound. In the present case, we performed the reaction at room temperature, which maintains the dinuclear structure of the complex. The purity of all three complexes was determined by elemental analyses and thermal stability of these complexes was investigated by performing TGA (see Supporting Information).

Structure Analysis

{[Co(Me2Tp)(CN)3]2[Fe(bik)2]2}(OTf)2 ⋅ 4MeCN ⋅ 2H2O (2) crystallizes as dark red plate‐like crystals in the triclinic P‐1 space group. Its structure consists of {Co2Fe2} square motifs, triflate anions, crystallisation water and acetonitrile molecules. In this centrosymmetric square complex the two {Co(Me2Tp)(CN)3}− building‐blocks act as bis‐monodentate ligands toward two cis‐{Fe(bik)2}2+ units through two cis cyanide groups. The remaining terminal cyanide ligands adopt an anti‐orientation relative to the {Fe2(μ‐NC)4Co2} skeleton (Figure 1). The crystal data obtained at 200 K show that the Co−CN−Fe edges are almost identical and their distance, ca. 4.972(4) Å, is typical for CoIII−CN−FeII LS. linkage (LS=low spin). The angles at each corner of the square slightly deviate from orthogonality [88.41(1)° and 90.60(1)° at Co1 and Fe1, respectively] and slightly distorted octahedral CoN3C3 and FeN6 surroundings result from the contribution of three nitrogen atoms of the tridentate (Me2Tp)− ligand and three cyanide‐carbon atoms [at Co] and four nitrogen atoms from two bidentate bik ligands plus two cyanide‐nitrogen atoms [at Fe]. The distortions of the metal coordination sphere were analysed by continuous shape measurements using the SHAPE program. The reported output (shape factor) allows assessing the matching between an idealized polyhedron and the actual coordination sphere: the lower the shape factor is, the better the matching between the actual coordination sphere and the idealized polyhedron. Here the lowest values calculated by SHAPE are obtained for the octahedron in both cases (0.059 for the Fe and 0.099 for the Co) and point to slightly distorted coordination spheres. The cyanide bridges are slightly bent on the Fe(II) side [Fe1−N1−C1=176.9(2)° and Fe1−N2−C2=173.9(2)°] and remain close to linearity on the Co(III) side [178.6(2)–179.3(2)°]. The CoIII−Ccyano bond lengths in the [Co(Me2Tp)(CN)3]− unit [values in the range 1.871(2)–1.873(2) Å] are in agreement with a low‐spin cobalt(III). The FeII−Nbik [1.952(2)–1.982(2) Å] and FeII−Ncyanide distances [1.947(2) and 1.951(2) Å] at 200 K are coherent with the occurrence of a low‐spin Fe(II) in a N6 surrounding.[ , , , ] Indeed, these distances are comparable with those observed in previously reported cyanide‐bridged tetra‐nuclear {FeIII 2FeII 2} and {MoV 2FeII 2} complexes containing a similar {Fe(bik2)(NC)2}2+ entity.[ , ] The spectroscopic and magnetic data also support the occurrence of a low‐spin {CoIII LS−CN−FeII LS} pair below room temperature (see below). Furthermore, the molecular squares are connected to each other via hydrogen bonding between water molecule, terminal CN, and imidazolyl hydrogen atom (see Supporting Information). The shortest inter‐square metal‐metal bond distance is 9.317(1) Å. The terminal cyanide groups are connected to the triflate anions through two hydrogen‐bonded water molecules along a CN(terminal)⋅⋅⋅O(water)⋅⋅⋅O(OTf−) path (see Supporting Information). The NCN⋅⋅⋅O(water) and O(water)⋅⋅⋅O(OTf−) contacts amount to 2.843(3) and 2.882(2) Å respectively, and the angle O(OTf−)⋅⋅⋅O(water)⋅⋅⋅N(CN) is 128.64(8)°. Compound 3 crystallizes in a monoclinic P2 space group. Its structure consists of a neutral 2D network, which is made of [CoIII(Me2Tp)(CN)3]− complexes that are connected to three different [CoII 2(piv)3] units through three CN ligands (Figure 1). The 2D layer can be described as a network of hexa‐nuclear {CoIII 2CoII 4} motifs (see Figure 1 and Supporting Information) that are linked to each other by cyanide bridges (no oxidation state change is detected between the building block and the final material as described below). The {CoIII 2CoII 4} motifs are made of two fac‐[Co(Me2Tp)(CN)3]− units, each of which coordinates through cis‐cyanide ligands to a Co(II) ion of the two [CoII 2(piv)3] bimetallic units in a square‐like fashion. In the bimetallic unit, one of the Co(II) ions is tetra coordinated and the other is penta‐coordinated. The SHAPE analysis performed on 3 reveals that the ideal structure of the Co(III) centre is close to an octahedron, whereas the penta‐coordinated Co(II) centre exhibits a trigonal bipyramid (shape factor: 0.947) and the tetra‐coordinated Co(II) centre has a distorted tetrahedral coordination sphere (shape factor: 1.229). The penta‐coordinated Co(II) centre coordinates to two nitrogen atoms from the CN bridges (forming the square motif) and three oxygen atoms belonging to three μ‐κO:κO’ bridging pivalate (Me3CCOO−) groups, whereas the tetrahedral Co(II) centre coordinates to one nitrogen atom from the [Co(Me2Tp)(CN)3]− and three oxygen atoms from three μ‐κO:κO’ bridging pivalate groups (see Supporting Information). In the {CoIII 2CoII 4} core unit, the CoIII−CN−CoII edges are quasi identical [4.962(1) and 5.040(1) Å] and the angles at the corners slightly deviate from orthogonality [CoIII−CoII−CoIII=91.39(1)° and CoII−CoIII−CoII=88.61(1)°]. Although the cyanide bridges are bent on the Co(II) side [CoII‐N−C angles: 163.3(2) and 167.2(2)°], they remain close to linearity on the Co(III) side [CoIII−C−N angles: 172.4(2) and 179.5(2)°]. In the [Co(Me2Tp)(CN)3]− subunit, the third cyanide ligand (CoIII−C3−N3) acts as a linker between neighbouring {CoIII 2CoII 4} units. The CoIII−Ccyano bond distances range from 1.871(2) to 1.889(2) Å. These values are slightly higher than those observed in the related low‐spin [Co(Me2Tp)(CN)3]− building block and coherent with a low‐spin Co(III) ion. The angles subtended by the tripodal ligands at the Co(III) site slightly deviate from orthogonality [86.90(9) to 92.99(10)°]. Overall, X‐ray analysis supports the presence of Co(III) and Co(II) ions at 200 K, and this observation is also coherent with the magnetic data (see below). The layers are quite well isolated from each other, the shortest intermolecular CoIII−CoIII and CoIII−CoII distances between two layers being 9.033(1) and 10.055(1) Å, respectively. Relevant data regarding the Co(III) coordination spheres in these three complexes are given in Table 1. To sum up, the Co(III) centres in all these three compounds are low spin in nature and they hold an almost ideal octahedral geometry (shape factor S ≈0.1).

Figure 1

Perspective views of (a) the [CoIII(Me2Tp)(CN)3]− complex in 1; (b) the {CoIII 2FeII 2}2− square complex in 2; (c) the 2D network in 3. Hydrogen atoms and solvent molecules are omitted for clarity. Light orange: CoIII, yellow : CoII, light green : FeII. (d) Zoom of the FT‐IR spectra of 1 (orange), 2 (green) and 3 (violet) showing the cyanide stretching vibrations.

Table 1

Selected structural data and geometry of 1, 2 and 3. Distances in Å and angles in degrees.

	1	2	3
d[CoIII(CN)bridging]av	–	1.872(1)	1.872(2)
d[CoIII‐N(Tp)]av	1.993(6)	1.989(9)	1.975(7)
Σ for CoIII centre	11.5	13.7	14.0
Shape factor S (OC‐6) for CoIII centre	0.119	0.099	0.099
Shortest intermolecular CoIII⋅⋅⋅CoIII distance	9.830(1)	9.779(1)	9.033(1)
[(NC‐CoIII‐CN)bridging]av angle	–	88.23(7)	91(2)

–

–

FTIR Analysis

FTIR spectroscopy has been employed to characterize all three complexes and to get detailed information about the cyanide coordination mode involved in these complexes. Similarly to the cyanido‐Fe(III) counterparts, the CN stretching vibration is also indicative of the redox state of the metal ion linked to the cyanide carbon atom, and of the bridging mode of the cyanide ligand. The infrared spectrum of 1 obtained at room temperature exhibits characteristic B−H stretching vibration at 2531 cm−1 that is similar to that observed in the related Fe(III) building blocks. A unique cyanide stretching vibration is observed at 2131 cm−1, close to that observed in [Co(CN)6]3− (Figure 1d). This indicates that all cyanide ligands adopt a terminal coordination mode. The infrared spectrum of 2 exhibits B−H vibration around 2534 cm−1 and two cyanide vibrations are located at 2166 and 2136 cm−1 which are characteristic of bridging [{CoIII LS(μ‐CN)FeII HS}] units (HS=high spin) and terminal cyanide ligands, respectively (Supporting Information).[ , ] The infrared spectrum of 3 also displays strong B−H vibrations at 2553 and 2535 cm−1. The stretching cyanide vibrations at 2180 and 2166 cm−1 are clearly shifted compared to 1, and unambiguously indicate that all cyanide ligands adopt a bridging mode. The small stretching vibration around 2252 cm−1 can be ascribed to the CH3CN molecules trapped in the crystal lattice.

Magnetic behaviour

Bulk molar magnetic susceptibility (χ M) measurements were carried out on the fresh samples of 1, 2 and 3. The χ curves shown in Figure 2a reveal drastically different magnetic behaviours for the three compounds.

Figure 2

(a) χ T curves of 1–3. (b) scheme of the electronic configuration change occurring in the {FeII(bik)2(NC)2} SCO subunit in compound 2. (c) χ irradiation time at 635 and 808 nm in 2. Red and grey arrows indicate when the 635 and 808 lights are switched on and off. The small jump after switching off the light are due to the thermalization. (d) χ curve of the photoinduced paramagnetic state (λexc=635 nm) upon heating in darkness in 2. (e–f) Solid‐state UV‐vis spectra of 2 upon heating between 100 and 300 K (e) and 300–450 K (f).

While compound 1 remains diamagnetic throughout the explored temperature range, as expected for a low‐spin Co(III) complex (S=0), compound 3 is paramagnetic in the same temperature range. In 3, the χ value for each {CoIII 2CoII 4} repeating unit of the 2D network amounts to 9.31 cm3 mol−1 K at 300 K. This value lies within the expected range for the following set of magnetically isolated ions: two low‐spin Co(III) ions (S=0), and four high‐spin Co(II) ions (S=3/2). Two of them are tetrahedral (2.1<χ M T<3.1 cm3 mol−1 K per CoII at 300 K) and the other two exhibit a trigonal bipyramidal geometry (2.4<χ M T<3.4 cm3 mol−1 K per CoII at 300 K). Upon lowering the temperature, the χ M T value remains almost unchanged down to 45 K, which support the absence of significant magnetic interaction between the ions. Below this temperature, a slight decrease down to χ M T=8.13 cm3 mol−1 K at 2 K is observed. This decrease is probably associated to weak antiferromagnetic exchange interactions between high‐spin CoII sites in 3 or to the effect of magnetic anisotropy (Figure 2a). In contrast, 2 shows a strong temperature dependence of its χ M T product (Figure 2a): the χ M T curve of a fresh sample of 2 exhibits a plateau up to ca. 270 K with a χ M T value close to zero and then sharply increases around room temperature to reach a pseudo plateau, as the χ M T value increases smoothly from 6.01 cm3 mol−1 K at 298 K to 7.02 cm3 mol−1 K at ca. 358 K. This behaviour is ascribed to a spin‐state change of the two {FeII(bik)2(NC)2} subunits with a transition temperature of T 1/2=292 K. The non‐zero χ M T value measured at low temperature (ca. 0.34 cm3 mol−1 K) is twice that of the value measured for the Co(III) monometallic precursor and it is likely due to the temperature independent paramagnetism. The overall variation of the χ M T value is close to that expected for a spin‐crossover from the low‐spin (t2g)6 electronic configuration to the high spin (t2g)4(eg)2 one (Δχ M T=7.26 cm3 mol−1 K with g≈2.2).[ , , , , ] The net increase of ca. 6.68 cm3 mol−1 K corresponds to an almost complete spin‐crossover of the two Fe(II) ions (ca. 92 % of LS – HS conversion). This represents a larger spin‐state conversion than the one previously reported in the related tetrametallic {[MoV(CN)8]2[FeII(bik)2]2}2− cyanido‐bridged complex (T 1/2 ∼350 K) containing the same {FeII(bik)2(NC)2} SCO subunit. However, it compares well with those reported in the analogous mixed‐valence {[FeIII(Tp)(CN)3]2[FeII(bik)2]2}2+ and {[FeIII(Me2Tp)(CN)3]2[FeII(bik)2]2}2+ square complexes.[ , , ] The sharp spin‐transition in 2 is strongly affected by heating. Indeed, the transition becomes gradual after heating at 400 K and is shifted toward lower temperature (see Supporting Information). This can be associated to the loss of crystallization solvent molecules, which occurs inside the magnetometer in the high temperature range. The shift of T toward lower temperature upon desolvation was already observed in the {FeIII 2FeII 2} and {MoV 2FeII 2} square complexes mentioned above (table 2). As previously suggested in another study, one could assume that the removal of the crystallization solvent molecules, would release the steric constraints imposed by the crystal lattice so that the complex would easily enters the HS state. Finally it is worth noticing that the transition temperature measured on the desolvated sample (T =213 K) is close to the value observed in the related {FeIII 2FeII 2} square complex where [FeIII(Me2Tp)(CN)3] acts as metalloligand (223 K). In contrast, the square complexes based on the non‐methylated metalloligand [FeIII(Tp)(CN)3] showed spin‐transition well above 400 K. This comparison suggests that both [FeIII(Me2Tp)(CN)3]− and [CoIII(Me2Tp)(CN)3]− metalloligands are weaker N‐donating ligands than the [FeIII(Tp)(CN)3]− complex.

Table 2

Selected data on switchable properties of SCO complexes containing {FeII(bik)2(N‐)2} unit. Data in parentheses relate to desolvated phases. Ø indicate no LIESST effect.

	T 1/2 [K]	TLIESST [K]	most efficient wavelength (nm)	Ref.
[Fe(bik)3](BF4)2	316	ca 50	635	[21]
[Fe(bik)2(NCS)2]	260	<20	900	[19]
[Fe(bik)2(NCSe)2]	326	<20	900	[19]
{[Fe(Tp)(CN)3]2[Fe(bik)2]2}	330 (240)	Ø (ca 45)	Ø (750)	[6a]
[Fe(Tp)(CN)3]2 ⋅ 18H2O,4CH3OH
{[Fe(Me2Tp)(CN)3]2[Fe(bik)2]}2	223	35	635 and 900	[7a]
(ClO4)2 ⋅ 2H2O
{[Fe(Tp)(CN)3]2[Fe(bik)2]}2	>400	Ø	Ø	[7a]
(ClO4)2 ⋅ 2CH3CN
{[Fe(Tp)(CN)3]2[Fe(bik)2]}2	>400	<20	808[a]	[7a]
(BF4)2 ⋅ 2CH3OH
{[Co(Me2Tp)(CN)3]2[Fe(bik)2]}2	ca 292	ca 55	405, 532, 635 or 900	This work
(OTf)2 ⋅ 4MeCN ⋅ 2H2O (2)
{[Mo(CN)8]2[Fe(bik)2]2}	ca 350 (300)	ca 48	405	[6b]
(HMeIm)2 ⋅ 5H2O,CH3CN
(HNBu3)2{[W2(CN)8]2[Fe(bik)2]2} ⋅ 6H2O ⋅ CH3OH	ca 370 (305)	Ø (ca 63)	Ø (532)	[22]

[a] only 808, 405 and 532 nm wavelengths were tested in that case.

T [K]

T [K]

Photomagnetic behaviour and thermochromism in 2

The switchable properties of 2 were investigated in more depth and reveal both photomagnetic and thermochromic effects beside the thermally‐induced spin transition observed in magnetometry (Figure 2b‐f). In fact, the magnetization of 2 strongly increases when exposed to light irradiation at low temperature (20 K). Note that these measurements were done on fresh samples of 2 and no photomagnetic effects were observed on the desolvated sample. Different laser‐diodes were tested, λ=900, 808, 635, 532 and 405 nm (with the respective powers: ca. 5, 11, 11, 12 and 5 mW cm−2), all of them inducing a rapid increase of the χ M T value upon irradiation (see Supporting Information). Such photomagnetic effect in Fe(II) SCO complex is ascribed to a Light‐Induced Excited Spin‐State Trapping (LIESST) effect, the low‐spin Fe(II) (S=0) state being converted into a metastable high‐spin Fe(II) (S=2) through the photo‐induced population of transient excited state(s). The maximum conversion is observed at 635 nm, where the χ M T value increases from ca. 0.21 to 4.51 cm3 mol−1 K. This increase corresponds approximately 66 % of the thermally induced LS‐HS conversion. All other wavelengths lead to slightly lower conversion rate, except that at 808 nm, which leads to a much lower value of 1.6 cm3 mol−1 K (ca 21 % of the thermally‐induced conversion). The distinct efficiency of the 635 and 808 nm wavelengths in promoting LIESST effect could be related to different photoconversion pathways, or to a “reverse‐LIESST” effect. Actually, the irradiation at 808 nm of the photoinduced metastable state obtained at 635 nm induces a decrease of the χ M T value down to ca. 1.6 cm3 mol−1 (see Figure 2c). This demonstrates that the 808 nm wavelength is also efficient in promoting a partial back‐conversion from the metastable HS state to the diamagnetic LS state. The photo‐stationary state (χ M T ∼1.6 cm3 mol−1 K) obtained when irradiating at 808 nm thus represents the equilibrium state between the LS→HS and the HS→LS photoconversion. Interestingly, the magnetisation can thus be modulated by successive light irradiation, demonstrating that the system can be repeatedly switched on and off by selective irradiations at 625 and 808 nm. Finally, the thermal stability of the photo‐induced metastable state was then probed by measuring the χ versus T upon heating (at 0.4 K/min) after irradiation at 635 nm. The light‐induced metastable paramagnetic state persists up to ca. T LIESST=55 K (determined as the inflexion point of the χ versus T curve) in the present experimental conditions (Figure 2d). Note that the change of χ value below ca. 10 K can be ascribed to the zero‐field splitting effect of the pseudo‐octahedral Fe(II) high‐spin metastable state. In compound 2 the T LIESST temperature (determined as the inflexion point of the χ relaxation curve) is somehow higher than that measured under the exact same experimental conditions in the related tetrametallic {[FeIII(Me2Tp)(CN)3]2[FeII(bik)2]2}2+ (T LIESST=35 K) {[FeIII(Tp)(CN)3]2[FeII(bik)2]2}2+ (T LIESST=45 K) and {[MoV(CN)8]2[FeII(bik)2]2}2− (T LIESST=48 K) complexes, which all contain the {FeII(bik)2(NC−)2} subunits.[ , , ] It is worth noticing that the excitation wavelength allowing the highest LIESST effect in the {FeII(bik)2(NC−)2} subunits clearly depends on the nature of the cyanido‐metalloligand as shown in table 2.[ , , , ]

Compound 2 also shows a significant thermochromic effect associated to the thermally‐induced spin‐state change. Indeed, the UV‐vis spectrum of 2 recorded at low temperature (ca 100 K) exhibits a strong band centred at 612 nm. This band can be ascribed to the metal to ligand charge transfer (MLCT) band of the {FeII(bik)2(NC−)2} chromophores (Figure 2e and Figure 2f) as already observed in the previously reported molecular {FeIII 2FeII 2} and {MoV 2FeII 2} square complexes containing the same {FeII(bik)2(NC−)2} subunit.[ , , ] The intensity of the band notably decreases upon heating, which is coherent with a spin transition as observed in the related square complexes.[ , , , ] More specifically, the intensity decrease is due to the larger metal‐ligand bond length in the high‐spin state, which results in a smaller overlap between metal‐centred and ligand‐centred orbitals as compared to the low‐spin state. It is worth noticing that the change in the optical properties cover a broader temperature range than the one measured in SQUID magnetometry. This is related to the different experimental conditions used for both measurements (see Supporting Information).

Solid‐state 59Co NMR analysis

In previous works, we showed that NMR active nuclei of diamagnetic ions such as 113Cd could be used as local probe to access accurate information on the local structure and electronic states of paramagnetic cyanide‐based compounds. For example, we showed that 113Cd NMR was efficient in revealing the electronic state changes occurring upon reversible Li+ intercalation in nanoporous CdFe Prussian blue analogues (PBAs). This information is made accessible in a M−CN−Cd pair as the changes of electronic/magnetic state on the M ion can be felt by the diamagnetic NMR active Cd(II) ion. In a similar manner, the present study offers the opportunity of investigating the efficiency of 59Co NMR in sensing the magnetic state of cyanide‐based complexes containing the Co−CN−M linkage, where M is the paramagnetic source and the low‐spin Co(III) located on the C cyanide atom acts as the NMR probe. The situation is more challenging as the 59Co nucleus bears a 7/2 nuclear spin with a rather large quadrupole moment that can lead to very broad NMR features in the solid‐state (whose signal width increase with the asymmetry of the environment). However, the 59Co nucleus has a 100 % abundancy and its receptivity is about 1500 times better than that of 13C. Moreover, among all NMR active nuclei, the 59Co nucleus exhibits the broadest chemical shift range, of ca. 18000 ppm (−4000 to 14000 ppm). This has been attributed to the occurrence of an important contribution from the temperature independent paramagnetism (“TIP”) that is related to the 3d nature of the Co atom. Briefly, the TIP contribution to the experimental chemical shift, δTIP, which is to be added to the usual diamagnetic contribution to the chemical shift, δexp=δdia+δTIP, arises from the mixing of the t2g 6 ground state and the excited electronic states (such as the (t2g)5(eg)1 configuration). The mixing is inversely related to the energy difference between these states and it is thus sensitive to the ligand field and the nature of the ligand in the coordination sphere. Besides the TIP contribution, the experimental chemical shift bears in the case of 2 and 3 a paramagnetic contribution that arises from the hyperfine interaction of the 59Co nuclear spin with the neighbouring paramagnetic sources. The hyperfine interaction can be decomposed into a through‐space dipolar contribution (δdip) and a through‐bond contact contribution (δcon). This last term is due to the delocalization of the spin density from the neighbouring paramagnetic centres to the probed nucleus through the cyanide bridge. In our previous study on CdFe PBA materials, we observed that the dipolar contribution remained moderate in comparison to the contact term. A simple calculation based on a point dipole model (see Supporting Information) would lead to dipolar contribution of ca. 5–30 ppm in the case of a CoIII dia−CN−CoII pair (depending on the orientation of the magnetic axis). In summary, in 2 and 3 the experimental chemical shift sums up the four contributions: δexp=δdia+δTIP+δcon+δdip

As the first coordination sphere of the CoIII ions are identical in all three compounds, which exhibit the same fac‐[Co(Me2Tp)(CN)3]− units (see structural data), one can assume that the TIP contributions are similar in 1–3. This assumption seems to be confirmed by the magnetic measurements. In fact, at low temperatures, the non‐zero χ T values of the compounds 1 and 2 are similar (ca 0.17 cm3 mol−1 K per Co units) and can be ascribed to similar TIP contributions as these fresh materials are in a diamagnetic ground state (the TIP contribution cannot be estimated in 3 because the χ T value at low temperature is dominated by the paramagnetic contribution). In first approximation, compound 1 can thus be used as a reference for estimating the diamagnetic and TIP contribution δdia+δTIP that are present in 2 and 3. By subtracting the experimental shift of 1 to that of 2 and 3, one can thus estimate the hyperfine contribution in 2 and 3, δcon+δdip, that is related to their paramagnetic nature.

59Co NMR spectra of powder samples of 1–3 were recorded at 16.4 T in static echo conditions (Figure 3) and under Magic Angle Spinning (MAS) conditions for 1 (see Supporting Information). The NMR line shapes in 1–3 suffer from an expected quadrupolar broadening. Only the compound 1 has been simulated and estimates of the quadrupolar coupling parameters were obtained.[ , ] Note that these simulations cannot be carried out for 2 and 3. To our knowledge, there is at the moment no software that can extract the different diamagnetic and paramagnetic contributions from solid‐state 59Co spectra.

Figure 3

59Co NMR experimental spectra of stationary samples (fresh) of 1, 2 and 3 acquired at 21.10 T, 260 K. Inset represents static NMR spectra of a previously desolvated (in TGA) sample of 2 at 260 K.

The 59Co isotropic chemical shifts (δiso) is thus only accessible for the diamagnetic compound 1: δiso=+3590 ppm (see Supporting Information). The comparison of δiso with the gravity center of this 59Co signal, δG = 3495 ppm allows estimating the shift induced by the quadrupolar coupling, δQIS=−95 ppm. For 2 and 3, we only access the δG which is expected to be close to the isotropic shift. For 2 at 260 K δG=3593 ppm, while for 3 δG=4754 ppm (260 K). The chemical shift of the 59Co nucleus for 3 is thus significantly shifted compared to that of the diamagnetic reference 1. Its hyperfine contribution that arises from the presence of the neighbouring Co(II) ions is of the order of 1250 ppm. In contrast, 2 shows a smaller hyperfine positive shift of approx. +100 ppm when compared to 1. It could come from the occurrence of a small fraction of neighbouring high‐spin Fe(II) ions (as shown in the magnetic measurement in Figure 2), but such a small shift can also be ascribed to the differences observed in the second coordination sphere of the Co(III) ions in 1 and 2. Indeed, the coordination of Fe(II) Lewis acid to the nitrogen cyanide of the {Co(Me2Tp)(CN)3} units in 2 decreases the electron density onto the cyanide ligand and should reduce the ligand field on the Co(III) ion. This can lead to a positive shift of the 59Co chemical shift as generally observed in Co(III) octahedral complexes. In contrast with our previous 113Cd NMR studies on paramagnetic cyanide‐based complexes where strongly shifted NMR signals were observed, here the measured 59Co chemical shifts remain in the usual diamagnetic range for cyanido Co(III) complexes. Actually, the 59Co chemical shift of cyanido complexes varies from 0 ppm for the K3[CoIII(CN)6] with very strong ligand field (ca. 38000 cm−1) to 6700 ppm for the [CoIII(NH3)5(CN)]Cl2 complex that contains only one cyanide group. The fac‐[CoIII(NH3)3(CN)3] complex that is closely related to 1 shows a similar chemical shift of 3315 ppm. In order to clarify if the displacements in the chemical shifts are due to the hyperfine contribution or to ligand field effects, we have investigated the temperature dependence of the chemical shift. In 3, the chemical shift value decreases of ca. 150 ppm upon heating the sample from 260 to 320 K (details in Supporting Information). This observation is in line with the paramagnetic nature of the compound. Indeed, according to the Curie law, the magnetization of a paramagnetic complex is inversely proportional to the temperature. Chemical shift should thus roughly follow the same trend with temperature as chemical shift can be visualised as local magnetic susceptibility. Hyperfine contribution seems therefore to be responsible for the chemical shift in 3. In contrast compound 2 shows a noticeable increase of the 59Co chemical shift value of ca. +440 ppm from 260 to 320 K (table S3). This shift is well beyond the usual increase that is generally observed in diamagnetic Co(III) complexes, and which is ascribed to the lengthening of the Co‐Ligand distance upon temperature increase and weakening of the ligand field (ca. +1.4 ppm/K for the K3[Co(CN)6)]). Here, the strong positive shift is to be ascribed to the thermally‐induced spin‐transition occurring on the neighbouring Fe(II) ions upon heating. As the local magnetic susceptibility of the compound increases with temperature, the chemical shift value also increases.

At this stage, two points are worth commenting when comparing the present results on CoIII−CN−MII systems with the previous one on FeIII−CN−CdII systems. In contrast with the previous 113Cd NMR studies, the NMR 59Co chemical shifts are relatively small and remain in the diamagnetic range, indicating that only a small amount of spin density is transferred from the FeII/CoII paramagnetic source to the Co(III) ion through the cyanide bridge. Their positive sign also contrasts with the negative sign observed for the 113Cd in FeIII−CN−CdII pairs. This also hints to a smaller amount of spin density transferred from the paramagnetic source to the NMR active nucleus in the CoIII−CN−MII linkage than in FeIII−CN−CdII one. This may be related to the nature of the FeIII−CN and the CN−FeII/CoII bonds. The spin density delocalization from the paramagnetic Fe(III) centre to the Cd ion through the CN ligand was shown to be efficient as the CN ligand acts as both σ‐donor and π‐acceptor ligand when coordinated via its carbon atom. This ensures efficient delocalization of the spin density from the paramagnetic ion into the π‐orbital of both C and N atoms. Then, polarization mechanism leads to negative spin density onto the Cd ion. In contrast, the N donor atom is involved in σ‐donation with the paramagnetic metal ion in CoIII−CN−MII. The delocalization of spin density from the Co(II) in 3 to the CN bridge may thus be weaker and mainly involve the N atom. A polarization mechanism through the CoII−N−C pathway should then lead to negative and positive spin density onto the C and CoIII centers, respectively.

Conclusions

The Co(III) tricyanido building block, [Co(Me2Tp)(CN)3]−, was used as metalloligand for the generation of a switchable molecular {CoIII 2FeII 2} square (2) and high nuclearity 2D {CoIII 2CoII 4}n network (3). The {CoIII 2FeII 2} square complex behaves as a molecular switch: a thermally induced cooperative spin transition occurs on the FeII sites near room temperature. This induces drastic changes both in the magnetic state of the material and in its optical properties. In particular, strong changes are observed in the intensity of the charge transfer bands located in the visible region. The cooperative nature of the spin transition is lost upon heating the material above room temperature, likely because of the loss of intermolecular interaction associated with the removal of solvent crystallization molecules. A photomagnetic effect (“LIESST effect”) is also observed at low temperature. The photoconversion is efficient in a broad energy range that corresponds to the MLCT band but also at 900 nm as observed in related complexes. Interestingly, a reverse‐LIESST effect is observed at 808 nm, which permits to reversibly and repeatedly control the magnetic state of the system by selective irradiation. In contrast, the two‐dimensional {CoIII 2CoII 4}n compound remains paramagnetic throughout the explored temperature range and shows a Curie‐like magnetic behaviour in a broad temperature range that accounts for very weak magnetic exchange coupling between the CoII paramagnetic sources. The three compounds – the diamagnetic precursor (1), the molecular switch (2) and the paramagnetic network (3) – were used as model systems to investigate the use of 59Co NMR as a probe for analysing the electronic and magnetic states of cyanido‐based materials. The 59Co NMR study at variable temperature showed that the chemical shifts are clearly correlated to the magnetic state of the material: “the more paramagnetic, the more shifted”. This allows the indirect detection of the spin‐state change occurring on the neighbouring Fe(II) SCO centre in the CoIII−NC−FeII linkage. To our knowledge this study represents the first report of solid state 59Co NMR spectroscopic characterization of a spin‐crossover compound. Interesting differences of sign and intensity are also observed in the paramagnetic chemical shifts of the Co and Cd diamagnetic probes in the M→C−N→Cd and M→N−C→Co linkages, where M is a paramagnetic source. These differences hint that a smaller spin density is transferred in the second case. This is likely related to the different bonding scheme between the paramagnetic source and the cyanide ligand depending on the nature of the donor atom C/N. Further experimental 13C and 15N NMR studies and theoretical calculation are now necessary to deepen our knowledge on the differences in spin density extension mechanisms.

CCDC2117924 (2) and 2117925 (3) contain the supplementary crystallographic data for this paper. These data are provided free of charge by The Cambridge Crystallographic Data Centre

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Click here for additional data file.