Spin Crossover in a Hexaamineiron(II) Complex: Experimental Confirmation of a Computational Prediction

Abstract Single crystal structural analysis of [FeII(tame)2]Cl2⋅MeOH (tame=1,1,1‐tris(aminomethyl)ethane) as a function of temperature reveals a smooth crossover between a high temperature high‐spin octahedral d 6 state and a low temperature low‐spin ground state without change of the symmetry of the crystal structure. The temperature at which the high and low spin states are present in equal proportions is T 1/2=140 K. Single crystal, variable‐temperature optical spectroscopy of [FeII(tame)2]Cl2⋅MeOH is consistent with this change in electronic ground state. These experimental results confirm the spin activity predicted for [FeII(tame)2]2+ during its de novo artificial evolution design as a spin‐crossover complex [Chem. Inf. Model. 2015, 55, 1844], offering the first experimental validation of a functional transition‐metal complex predicted by such in silico molecular design methods. Additional quantum chemical calculations offer, together with the crystal structure analysis, insight into the role of spin‐passive structural components. A thermodynamic analysis based on an Ising‐like mean field model (Slichter–Drickammer approximation) provides estimates of the enthalpy, entropy and cooperativity of the crossover between the high and low spin states.


Abstract:
Single crystal structural analysiso f [Fe II (tame) 2 ]Cl 2 ·MeOH (tame = 1,1,1-tris(aminomethyl)ethane)a safunctiono ft emperaturer eveals as mooth crossover between ah ight emperature high-spino ctahedral d 6 state and al ow temperature low-spin ground state withoutc hange of the symmetry of the crystal structure. The temperature at which the high and low spin states are present in equal proportions is T 1/2 = 140 K. Single crystal, variable-temperature optical spectroscopy of [Fe II (tame) 2 ]Cl 2 ·MeOH is consistent with this changei n electronic ground state. Thesee xperimental results confirm the spin activity predictedf or [Fe II (tame) 2 ] 2 + during its de novo artificial evolution design as as pin-crossover complex [Chem. Inf. Model. 2015, 55,1 844],o ffering the first experimental validation of af unctionalt ransitionmetal complexp redicted by such in silico molecular design methods. Additional quantum chemical calculations offer,t ogether with the crystal structure analysis, insight into the role of spin-passive structural components. At hermodynamic analysis based on an Ising-like mean field model (Slichter-Drickammer approximation) provides estimates of the enthalpy,e ntropya nd cooperativity of the crossover between the high andlow spin states.
Spin crossover (SCO) in molecular materials is associatedw ith achange of spin state of acentral 3d metal under external perturbation such as temperature, pressure and irradiation. [1] The spin state instability is expected for predominantly octahedral complexes of 3d metals where the LS ground state and the metastable HS exciteds tate are close in energy but differ in termso fe ntropy contributions in their free energies. [2] All macroscopic properties conjugated with the spin state, such as a colour, magnetization,d ensity,a nd sometimes even shape and size of the crystals may be altered by externalp erturbation. Therefore, SCO compounds may be candidatem aterials for applications including sensors, imaginga nd information storage. [3] The potential applicationsm aker ational and computational designo fS CO materials attractive, albeit very challenging, goals due to the delicate energetics of the spin states involved. [4] Spin instability in an isolated complex is well understood already within ligand field theory [5] and the collective behaviour of spin-activem olecules can be parameterized with Ising-like models [6] and ag eneric phase diagram is provided by Landaut heory. [7] Although quantum chemical methods such as density functionalt heory (DFT) may predict the relative stability of the spin states with useful accuracy,t hese methods are computationally too expensive for materials design. In contrast, even simple correlationsb ased on geometry parameters mays ometimes accurately predict the relative spin state stabilityo faspinactive complex, [8] albeit only within relatively narrow ligand classes. Ab roader applicabilityd omain is offered by ligandfield molecular mechanics (LFMM). [9] LFMM is molecular mechanics augmented by empiricalt erms to treat the d-orbital splitting and interelectronic repulsion, and may predictr elative spin-state stabilityw ith an accuracya pproachingt hat of DFT, only orders of magnitude faster. [10] This accuracy has been used to demonstrate that LFMM can be applied in structural screening to identify SCO candidate complexes [4b] and as the fitness function( figure of merit, or scoring function) in de novo design of candidate Fe II SCO complexes. [11] Although the most promising candidates were predicted by DFT to be bistable, computationally designedS CO compounds have yet to be tested experimentally.
Experimental validation will help determine to what extent molecular computational design including the local spin-active complex alonem ay be useful in development of SCO materials. In this report we present the case of [Fe II (tame) 2 ]Cl 2 ·MeOH [Scheme1] (tame = 1,1,1-tris(aminomethyl)ethane). The complex cation [Fe II (tame) 2 ] 2 + emerged as the most promisingc andidate predicted by the above-mentioned de novo design study. [11] The properties of solid SCO materials are far more complex than those of isolateds pin-activef ragments.F or example, spin-passive structuralc omponents not directly bondedt ot he transition-metal complex, such as solventm olecules or counter ions, may affect the thermodynamics of the transition. [12] This level of complexity is currently beyondr outine theoretical predictions. Screening [4b] and de novo design [11] based on the central complex alone thus serve as ag uide toward the Fe II complexes most likely to lead to SCO behaviour subjectt oo uter-sphere perturbations. Therefore, experimental studies of crystal structures and the empirical correlations between variousp acking modes and collective responses are essentialt ou nderstand the mechanismo fp ropagation of as pinswitch throughout the crystal. In this work spin crossover is reflected not only in the FeÀNb ond distances but also in the size and shape of the unit cell, unit cell volume and optical properties. These data are used to estimate basic thermodynamics parameters, such as enthalpy, entropy, and cooperativity,based on the mean-field free energy expression. [13] There are relativelyf ew ligand systemst hat have been shown to form hexaamine complexesw ith Fe II (Fe II N 6 ). As election of these are shown in Scheme 2. Iron(II)c omplexes of organic amines are typically air sensitive and once oxidizedt o their ferric form usually undergo Fe-catalysed oxidative ligand dehydrogenation to generate unsaturated imine ligands. [14] This ligand degradation reaction can be avoided if the geometric constraints of the ligand prevent dehydrogenationo ft he Fe-N(H)-C(H) bond as in the macrocyclic complexes [Fe II (tacn) 2 ] 2 + and [Fe II (trans-diammac)] 2 + (Scheme 2) where stable ferrous [15] and ferric [16] complexes have been isolated and structurally characterized.
This presented ac hallenge in isolating the [Fe II (tame) 2 ] 2 + complex in ap uref orm. Both oxygen and water were problematic.I fo xygen is present,i mmediate oxidation of the mixture to give an insolublep recipitate of ferric oxide results, probablyc oupled with ligand dehydrogenation. Secondly, unless anhydrous MeOH is employed, addition of the tame free base leads to formation of metal hydroxides. Thirdly, the desired product [Fe II (tame) 2 ]Cl 2 ·MeOH must be crystallised by slow addition of chloride to as olution of ferrous triflate and tame free ligand followed by slow evaporation of the MeOH solution.I fc hloride is presenti nas toichiometrica mount duringF e II /tame complexation then the reaction does not yield ac rystalline product. For these reasons, high yields of pure compound were not pursued and the crystalline product was separated from other material manually under am icroscope. In spiteo fi ts instability in solution and susceptibility to oxidation, in the solid state the crystals are robust (for more than ay ear) if protected from oxygen and free from solvent. Furthermore, crystalso f[ Fe II (tame) 2 ]Cl 2 ·MeOH can be manipulated in air for periodso fh ours without any noticeable degradation.
The crystal structure of [Fe II (tame) 2 ]Cl 2 ·MeOH was determined at an umber of temperatures from 10 Kt o3 33 K. In all cases at rigonal structure was defined (space group R3 m), which is isomorphous with [Ni II (tame) 2 ]Cl 2 . [17] The complex cation occupies as ite with 3 m( D 3d )s ymmetry.T he methyl and quaternary C-atomsl ie on at hreefold axis (coinciding with three vertical mirror planes). An initial refinementf ound the methylene and amineg roups of the ligand on ac rystallographic mirror plane, as reported for the Ni II analogue. [17] However,a bnormal thermalp arameters and an unreasonablys hort H 2 N-CH 2 bond length (1.433 (5) )i ndicated that the N-atom was actually displaced from the mirror plane and disordered either side as are the methylene protons. Indeed, DFTcalculations (see Supporting Information) confirm that the [Fe II (tame) 2 ] 2 + cation gains stabilityb yt wisting each tripodal "cap" abouti ts C 3 axis in either direction, thus removing the mirror plane symmetry elements. The disorderi sf ixed by symmetry and there seems to be no ordering process correlated with spin crossover at least at the level of the average structure. So the structure amountst oad isordered mixture of S 6symmetric complexc ations related by a C 2 symmetry operation (see Supporting Information, Figure S1). The crystal structure is built from layerso fc ationic spin-activeF ec omplexes and Cl À counter-ions, the layers are separatedb yl ayers of disordered MeOH molecules;a ll the layers are orthogonal to the (threefold) c-axis (Figure 1, right).
Spin inversion (LS!HS) is associated with elongation of FeÀ Nb onds by % 0.15 .T his deformation mainly affects the c-axis of the unit cell, whichi se longated by more than 3.5 %, while the a-axis is less affected ( % 0.5 %e longation). This reflects an anisotropy of the packing forces with stronger Coulomb interactions within the anion-cation layers and weaker between the layers.
The elongation of c,i ncreasei nu nit cell volume, and increase in the apparent FeÀNb ond length are all correlated. The two extreme FeÀNb ond lengths at 333 K( 2.189(3) )a nd 10 K( 2.035(3) )a re consistentw ith high spin [18] and low spin [15] hexaamineiron(II) complexes,r espectively.I nt he temperaturer ange 250-333 Kt here is very little variation in the FeÀNc oordinate bond lengths or angles. Similarly,t he structures in the lowt emperature range 10-60 Ka re essentially the same and consistentw ith a1 00 %l ow spin Fe II structure. In the intermediate range (60-250 K) as igmoidal variation in the apparent FeÀNb ond length as af unction of temperaturei s seen, whichr eflects aw eighted average of the relative proportions of HS and LS Fe II in the crystal (Supporting Information Eqn S2).
The LS!HS spin-transition-induced elongationa long the C 3 axis is qualitatively consistentw ith that predicted by previous molecular-level modeling of [Fe II (tame) 2 ] 2 + . [11] DFT was used to quantify the elongation in models of the [Fe II (tame) 2 ] 2 + cation with and without the chloride counteri ons (see Supporting Information). The resultss uggest that the [Fe II (tame) 2 ] 2 + complex engages in an etwork of NH···Cl interactions that are stronger in the more compactL Sc oordination.T hese interactions thus also limit the LS!HS transition-induced elongationc ompared to that of Fe II N 6 complexes of other ligands than tame.
The polarised single crystal visible spectrum at 11 Ki ss hown in Figure2using light propagating perpendicular to the c axis with the polarisation k c (p spectrum) and ? c (s spectrum) and unpolarised propagating parallel to the c axis (a spectrum). The relationship s = a ¼ 6 p implies that the transitions are electric dipole allowed in this direction andc onsistent with threefold-symmetry of the complex cation (see SupportingI nformation for more details).
The temperature dependentb ehaviour of the spectra (p-polarisation,F igure 3a), illustrates the spin-crossover evolution between 50 Ka nd 295 Ka st he intensity of the LS 1 A 1g ! 1 T 1g , 1 T 2g transitions (O h term symbols) decreasew ith increasing temperature. Figure 3b showst he intensity of the 607 nm peak of the 1 A 1g ! 1 T 1g transition in p-polarisation for several cooling and heatingc ycles.
Entropy,e nthalpy and cooperativity have been estimated on the basis of diffractiond ata using the Slichter-Drickamer approximation [13] (see Figure4 and the Supporting Information   for details). Our estimates reveal that the cooperativity parameter for correlations between neighbouring spin centres is negative, indicating alternation of different states to be preferred over clusteringo fs imilar species.
Single crystal structural analysis and variable-temperature opticals pectroscopy have confirmed the spin-crossover properties of [Fe II (tame) 2 ]Cl 2 ·MeOH. Since no superstructure has been observed, the title compound mays erve as ac andidate for observation of short-range correlations, associated diffuse scattering, and othern on-linear phenomena near T 1/2 .I mportantly,t he central cation [Fe II (tame) 2 ] 2 + was automatically designed from scratch by assembling small and unconstrained structural fragments (in silico de novo molecular design)t o give Fe II N 6 complexes. These complexes were optimized (by artificial evolution) to express spin bistability. [11] To our knowledge, such an automated de novo procedure is here, for the first time, experimentally confirmed to have predictedacompound that reflects the intended, in silico optimised property. This is ap romising step forward for de novo molecular design beyondt raditional drug-like organic molecules.

Experimental Section
Syntheses, instrumental and computation details are in the Supporting Information.