Paramagnetic transition metal ions in molecules continue to interest chemists and physicists. They occur in metalloproteins, and the functional sites of industrial catalysts used for amazing chemical reactions, such as the activation of inert small molecules. The wide gamut of optical and magnetic properties of molecular transition metal centers and clusters stimulates extended research in material sciences and development of molecular devises. The decisive factor in controlling all of the fascinating properties of transition metal centers is the electronic structure, more than merely the spatial arrangement and the molecular structure. Even geometrically very similar systems may display drastically different properties, and undergo very different reactions.

Therefore, in order to make rational use of transition metal ions in future chemistry, one needs to understand the origin of the observed features and functions in terms of the electronic structure. This can be attained with notable success by a combination of spectroscopy and quantum chemistry.

A major technique for the investigation of iron compounds is ⁵⁷Fe-Mössbauer spectroscopy. Measurements of the electric hyperfine parameters isomer shift δ and quadrupole splitting ΔE_Q with or without using applied magnetic fields provide most valuable insight in the oxidation state and the coordination of the specific metal ion. Paramagnetic properties, which in Mössbauer spectroscopy can be probed with elaborate applied-field measurements, are particularly interesting, because the magnetism originates directly from the spin of valence electrons. Therefore magnetic techniques in general are particularly powerful tools of inorganic spectroscopy.

In our lab we combine ⁵⁷Fe-Mössbauer spectroscopy, multi-frequency electron-paramagnetic-resonance spectroscopy (EPR), static magnetic susceptibility measurements, and magnetic circular-dichroism spectroscopy (MCD) for the study of bioinorganic compounds and metalloproteins. Low temperature measurements and magnetic fields (flux up to 7 and 10 Tesla) are regularly applied for the investigation of complex systems.

The spectra obtained from the different techniques can be consistently parameterized within a common theoretical frame, which is provided by spin-Hamiltonian simulations. This phenomenological concept, which is an application of ligand-field theory, yields descriptions of the electronic structure of paramagnetic ions in terms of zero-field splitting (D and E/D values), Zeeman interaction (g values), and spin coupling (J values). Moreover, the Spin Hamiltonian parameters D, E/D, g, J can be derived theoretically from sophisticated quantum chemical electronic structure calculations, and thus provide an eminently useful and convenient link between spectroscopy and quantum chemistry, and their common interface. Chemical information is obtained from the combination of both techniques: about the oxidation states of metals and ligands, the nature of valence orbitals, coordination symmetry and strength, and putative interaction between metal sites. In the following some recent research projects shall be presented.

‘Ordinary’ Iron Compounds Can Have Very Complex Electronic Structures

(A project together with Marat Khusniyarov, Thomas Weyhermüller and Karl Wieghardt, published in Angew. Chem. Int. Ed. 47, 1228 (2008). Dr. Khusniyarov is presently a “Habilitant” at the Department of Chemistry and Pharmacy, Friedrich-Alexander-University Erlangen-Nürnberg).

For many enzymes and technical processes the functionality of the catalytic site is owed to the accessibility of variable oxidation states of transition metal ions, i.e. their ability to accept or donate electrons from or to the reaction partners when needed. Iron, in this respect, is a very versatile element because it exhibits rich redox chemistry with a large range of oxidation states easily available for storing oxidation equivalents. Most abundant in the environment, and best investigated in chemistry, are iron compounds with the oxidation numbers (II) and (III) of the metal. Low-valent iron(I) and iron(0) are known to occur in key enzymes such as hydrogenases, whereas high-valent iron(IV) and iron(V) centers play a decisive role in reaction intermediates of oxygen-activating enzymes. Hexavalent iron occurs also, but it was known only in ferrates; recently in this institute for the first time an iron(VI) molecule was reported. In material sciences, iron with high valence states holds some promises for the development of new batteries for electric energy storage.

Oxidative reactions in biochemistry are often catalyzed by metalloenzymes, for which not only the transition metal ion takes part in redox reactions, but also the organic environment or even the ligands to the metal. The archetypical examples are the heme proteins horseradish peroxidase and the cytochrome P-450 family of enzymes from liver. Their iron porphyrin complexes are oxidized during the catalytic cycle up to the formal oxidation number (V). However, it is long known that this highly reactive intermediate in fact is an iron(IV) and a porphyrin pi-cation radical complex, as shown for horse radish peroxidase 1969 by Mössbauer spectroscopy. Since then a large number of metalloenzymes has been found that contain redox-active ligands taking part in the catalytic process by donating or accepting electrons under formation of organic radicals. Among them are so different examples as ribonucleotide reductase, photosystem II of green plants, galactose oxidase, prostaglandin H synthase and amine oxidase.

Fascinated by the variety of such systems, a major part of our research activity was dedicated to transition metal complexes with small synthetic ligands, which were known or could be suspected to be redox-active. Recently, M. Khusniyarov synthesized two unusual iron compounds with the α-diimine ligands ‘dad’ (= 2,6-_iPr²-C₆H₃-N=C(Me)-C(Me)=N-2,6-ⁱPr₂-C₆H₃) and ‘pda’ (=tetrakis-(3-methylphenyl)-N,N,N′,N′-2,5-phenylene -diamine), which turned out to be paramount examples of complexes with simple basic structure but rich and complex electronic properties. The two compounds crystallized together as cation-anion pairs with molecular structures as shown in Fig. 1. The cation (C) is the six-coordinate complex [(dad)₃Fe]ⁿ⁺, whereas the anion (A) is the four-coordinate pda complex [(pda)₂Fe]^n-. Other ions are not present in the crystal structure. Compound C is quasi octahedral, whereas the structure of A is intermediate between tetrahedral and planar.

Apparently the charges of the two ionic molecules cannot be directly inferred from the structures of A and C, because both iron may be ferric or ferrous, and each of the five ligands may take one of three oxidation states shown.

Figure 1. Molecular structure of the dianion A = [(pda)₂Fe]^2-and the dication C = [(dad)₃Fe]²⁺ recorded at 120 K. The ellipsoids mark the 30% probability level of the distribution of atomic positions due to thermal vibrations.

Figure 2. Schematic structures of the alpha-diimine ligands ‘dad’ and ‘pda’ and metric details of the diimine backbone in neutral, monoanion and dianionform (top row). The distribution of ‘N-C-C-N’ bond lengths (numbers in red) reflects surprisingly well the pattern of (long) single bonds and (short) double bonds predicted from the corresponding Lewis formulae for the three forms.

Therefore, even if the molecular charges were known, the electronic structure of the compounds could not be inferred, because the ligands and the metal ions exhibit that variety of possible oxidation states. The electronic structures can be deciphered only from a careful evaluation of magnetic susceptibility and Mössbauer data, probing the electronic structure of iron, in conjunction with inspection of the metric details of the ligand structure. In this latter program, the values of certain atomic distances may reveal the oxidation state of the ligand, because they reflect the characteristic distribution of single and double bonds expected from the Lewis structure for the different oxidation states, as sketched in Fig. 2.

How to read the structures of A and C

Molecule A: The two ‘pda’ ligands are very similar in details, exhibiting the typical bond length pattern expected for the fully reduced state (pda^2-), having a short C=C bond and two long N-C bonds in the ‘N-C=C-N’ back-bone. Thus, the two ligands provide formally four negative charges and should not contribute to the magnetism of A because of closed-shell character.

Moreover, the Fe-N distances of A are remarkably long, compared to those of related compounds, indicating the presence of iron(II) in high-spin configuration with four unpaired electrons and a local spin of S_Fe= 2. Thus, the structural details predict A to be in fact a dianion of the type [(pda^2-)₂Fe(II)]^2-. The compound should have a quintet ground state with total spin S_A= 2, which arises purely from iron.

Molecule C: Since the anion-cation pair must be neutral, C can only be a dication. The ‘dad’ ligands of the octahedral compound show all the same long C-C bond and two short N=C bonds, that are typical of neutral diimine (dad⁰). Therefore also this iron must have the oxidation number (II), and the corresponding structure for C should be [(dad⁰)₃Fe(II)]²⁺. The Fe-N bonds, however, are short in this case, which is compatible only with a low-spin configuration, S_Fe= 0. Hence, compound C should be diamagnetic with S_C= 0.

Magnetic properties corroborate the conclusions drawn from structure

Apparently the interpretation of the structural details can be examined by magnetic measurements, due to the exact predictions of the spin states of the compounds. Solid samples of (A+C) were found in fact to exhibit an effective magnetic moment of μ_eff≈4.9 μ_B in the temperature range 30 – 235 K, which fits nicely to the spin-onlyvalue for S= 2.                               

Figure 3. Temperature dependence of the effective magnetic moment of solid A+C. The inset shows a hysteresis measurement by cycling the temperature up and down around T = 235 K.

Thus, the actual spin of the system corresponds to the prediction S_A= 2, S_C= 0 (Fig. 3).

The decline of µ_eff below 30 K reveals zero-field splitting, as expected for the 3d⁶ configuration of high-spin iron(II). This results from spin-orbit coupling and depends on the strength of the ligand-field experienced by the metal ion. The temperature range affected by zero-field splitting (ca. 0 – 30 K) provides an estimate for the size of the interaction. A simulation yields an exact value, D = 4.5 cm^-1, which is again typical for iron(II). The value is a measure for the bonding properties of iron and the (pda^2-) ligands, and can be used for further interpretation based on quantum chemical calculations.

Basically, the magnetic susceptibility measurements corroborate the above conclusions, except that there is a minor kink and a little jump in μ_eff(T) at about 235 K. One certainly is attempted to ignore and assign this little glitch to an experimental error. In reality, however, this is a minor response of the magnetic data to a major rearrangement of the whole electronic structure of the A+C ion pair, as could be discovered by ⁵⁷Fe-Mössbauer spectroscopy.

Mössbauer spectra reveal the full truth of A+C

A zero-field Mössbauer spectrum of A+C recorded at 80K yields a superposition of two quadrupole doublets of equal intensities, as shown in Fig. 4 bottom, corresponding to the different iron sites of the two compounds (a weak third subspectrum represents a minor contaminant from deterioration of the very sensitive sample).

A fit with Lorentzian lines yields the isomer shifts δ = 0.68and 0.38 mm/s and the quadrupole splitting ΔE_Q= 5.20 and 0.53 mm/s, for the contributions of A and C, respectively. The large quadrupole splitting and high isomer shift for the dianion A (marked in blue), which are typical for iron(II) high-spin, clearly corroborate the 3d⁶ configuration of the four-coordinate iron and the spin state S_Fe= 2. The low isomer shift and small quadrupole splitting of C on the other hand, also support clearly the assignment of iron(II) low-spin with S_Fe= 0. This was also quantitatively established from quantum chemical calculations by usingt he spectroscopy-based DFT package ORCA by F. Neese.

Figure 4. Zero-field Mössbauer spectra recorded with a powder sample of A+C at 80K (bottom) and 260 K (top). The subspectra marked in green are a small contaminant from some decay of the sensitive sample.

Apparently, the 80 K Mössbauer spectra readily support the model suggested above for the electronic structures prevailing for A and C at low temperatures.

However, the spectra change drastically when measured at 260 K, which is just above the little step in μ_eff(T) at 235 K (Fig. 4, top). Both iron sites, apparently, experience a major rearrangement of their electronic structures, as seen from the changes in δ and ΔE_Q, compared to those obtained at 80 K (with δ= 0.76 mm/s, ΔE_Q= 1.80 mm/sfor C’; δ= 0.16 mm/s, ΔE_Q= 4.19 mm/s for A’). The new state C’ of the cation shows a significantly increased isomer shift (component in red), whereas the value for the new anion A’ is clearly lower than that for C’ (component in blue). The quadrupole splittings also show reversed trends for both iron sites. From a detailed analysis, one can infer that the high-spin iron(II) of A has been oxidized and is now iron(III) in A’, having the unusual intermediate spin state, S_Fe= 3/2, whereas iron in the cation C’ has kept its oxidation state (II), but flipped its spin state from low-spin to high-spin, S_Fe= 2.

This magnetic transition observed for A+C at 235 K apparently involves an electron transfer from A to C. The apparent question is, where the transferred electron resides on C’, or whether one can prove the corresponding reduction of C’, which is obviously not metal-centered. This was possible from X-ray diffraction data measured at room temperature. One of the ligands in fact showed the metric details predicted for the (dad^1-) state. It is apparently singly charged and therefore has changed to an organic radical. Since this must carry a ligand spin, S_rad= 1/2, the change is reflected in the magnetic data.

The paramagnetic moment of both molecules A’ and C’ prevailing above the transition temperature results from intermediate spin iron(III) in A’, S_A’= S_Fe(A’)= 3/2, and the contribution from high-spin iron(II) in C’ with S_Fe (C’) = 2,as well as that of the radical spin S_rad(C’)= 1/2. Since the radical is coordinated to iron(II), both are very strongly antiferromagnetically coupled to a quartet state, S_C’= 3/2,due to the vector addition of angular moment, S_C’= S_Fe– S_rad= 2 – 1/2. The sum of the resulting effective magnetic moments for both molecules, having S_A’= 3/2 andS_C’= 3/2 yields , which is amazingly close to the experimental result measured in the temperature range 235 – 300 K. The general features of the electronic structure and magnetic properties of the ion pair A and C below 235 K and in the new states A’+C’ above 235 K are summarized in Fig. 5.

Figure 5. Schematic presentation of the electronic structure and spin of the ion pair complexes A+C.

Attempt of an interpretation

In summary, structural and spectroscopic data consistently reveal a very unusual and interesting transition of the electronic structure of the A+C ion pair at 235 K, which involves an inter-molecular electron transfer as well as changes of the local spin states of iron. The question arises what may be driving the processes. Spin transitions are well known for iron and have been studied in depth, also because of the potential applications in molecular devices and switches. In this case here, however, the dianion A switches from high-spin at low temperature to low-spin at high temperature, which has never been observed before. Usually the transition is opposite, as shown here by the cation C. It is generally accepted that the increase of enthalpy related to the formation of the high-spin state (concerning the number of possible spin vibronic states) is the major driving force for ‘regular’ spin transitions as shown by C. At first glance, the behavior of A seems to violate this rule.

On the other hand, electron transfer between molecules is the basis of redox chemistry, and particularly in biochemistry transfer over remarkable distances is also known. The direction of such electron hopping is always determined by the potential of the associated sites and always down-hill. The transfer observed here,  however, is temperature-dependent and reversible. The observation of a hysteresis of the magnetic data at the transition indicates a ‘collective’ phenomenon. For such transitions, a ‘signal’ must exist (like a change in volume, etc.) in order to synchronize the process within large domains of particles in the solid. Since something like this is not known at all for usual redox transfer processes, but as many spin transitions are collective, we may speculate that the reversible transition A+C ↔ A’+C’ is triggered by the ‘regular’ spin transition observed for C ↔ C’, for which iron(II) switches between S_Fe= 0 at low temperature and S_Fe= 2 above 235 K. An electron transfer may follow this, swapping a charge from iron(II) in A to the ligand of C, caused by the structural changes of the primary spin transition. The ‘inverse’ spin flip of iron(III) in A from S_Fe= 1/2 to S_Fe= 3/2 then would be the consequence of the charge transfer.

In summary, we have presented an example of a pair of two charged complexes that crystallize together and exhibit amazing and challenging electronic structures, which could be deciphered from high-resolution structural data, combined with magnetic susceptibitlity and Mössbauer measurements. For the first time a complex transition in the solid could be detected and described which involves coupled electron transfer and two spin transitions.