Received 11 July 2013
Spin-coupling in dimers of 2,3-dicyano-5,6-dichlorosemiquinone radical anions in the crystalline state
Kresimir Molcanov,a Darko Babic,a Biserka Kojic-Prodic,a* Jernej Stare,b Nadica Maltar-Strmeckia and Lidija Androsa
A crystal engineering approach is used to stabilize a radical anion in the crystalline state and to modulate the separation distance within -stacks of anion radicals. Alkali metal salts of 2,3-dicyano-5,6-dichlorosemiquinone (C8Cl2N2O2, DDQ) radical anions were prepared and their crystal structures determined: LiDDQ·2H2O·(CH3)2CO, RbDDQ·2H2O and CsDDQ·2H2O. In these structures, stacked dimers of radical anions are formed within -stacked columns. Within the stacked dimers, interplanar separation distances are significantly shorter than the sum of the van der Waals radii for two C atoms; the shortest is 2.812 Å for the Li salt and the longest is 2.925 Å for the Cs salt. Diamagnetic character, observed by electron paramagnetic resonance spectroscopy, indicates spin-coupling of the unpaired electrons within the radical anion dimer. The electron-rich cyano substituents on DDQ influence the electron redistribution within the ring skeleton. The crystalline compounds are also characterized by IR spectroscopy, complemented by quantum-chemical calculations based on both isolated and periodic models.
Unique magnetic and conducting properties have been discovered among organic radicals, making them attractive in the design of functional materials (Podzorov, 2010; Sanvito, 2011a,b; Novoa et al., 2011; Ratera & Veciana, 2012). An especially promising class of radicals are semiquinones, both due to their stability and their role in life-important processes. Over the last decade, intensive research has been carried out on perhalogenosemiquinones (Lü et al., 2003; Rosokha et al., 2009; Min et al., 2006, 2007; Molcanov et al., 2011, 2012), 7,7,8,8-tetracyano-p-quinodimethane (Garcia-Yoldi et al., 2009) and 2,3-dicyano-5,6-dichlorosemiquinones (DDQ). Due to the strong inductive effect of two very electronegative cyano groups, DDQ is more stable than similar semiquinones, and it has a very strong permanent dipole. However, research related to this radical in the crystalline state is so far rather limited.
Due to conjugation of their electrons, semiquinones are capable of -interactions which may include charge transfer (Gholami & Tykwinski, 2006; Michinobu et al., 2006; Yi et al., 2009; Verma et al., 2011) and/or spin coupling (Miller, 2011; Ratera & Veciana, 2012). A wide variety of different magnetic properties is possible in such systems (Shultz et al., 1997; Decken et al., 2011; Constantinides et al., 2012; Molcanov et al., 2011, 2012). In a few cases a reversible magnetic transition has been observed (Fujita & Awaga, 1999; Salunke-Gawali et al., 2005; Hicks, 2011; Molcanov et al., 2012; Yu et al., 2012).
The present paper deals with crystallographic and magnetic studies of three hydrated alkali metal salts of DDQ: LiDDQ·2H2O·Me2CO, RbDDQ·2H2O and CsDDQ·2H2O. In all of these compounds, the total number of electrons in a formula unit (MDDQ) is odd, and therefore an electron must be unpaired, characteristic of a radical. The systematic variation of the alkali metal cation represents an approach to modify the crystalline structure, and therefore potentially to modify the magnetic properties in the solid state. From a crystal engineering perspective, it is interesting to compare these compounds with alkali metal salts of perhalogenoquinones (Molcanov et al., 2011, 2012) and to assess the role of: (i) steric effects of the cyano groups; (ii) electronic effects of the more negative cyano groups (compared with Cl); (iii) the lower molecular symmetry of the DDQ anion (having a strong dipole) compared with the perhalogenosemiquinones; (iv) the role of hydrogen bonds in the crystal packing, due to the presence of water molecules and strong hydrogen-bond acceptors (i.e. the cyano groups).
All samples were prepared by slow evaporation from acetone solutions containing 2,3-dicyano-5,6-dichloroquinone (Sigma-Aldrich, p.a. grade) and an excess of the relevant alkali-metal iodide (Sigma-Aldrich, Merck, Kemika, all p.a. grade), with a few drops of water added to the solution.
Thermal analysis was performed on a Shimadzu DTG-60H analyzer in the range from room temperature (293 K) to 1053 K, in a stream of synthetic air and in N2, at a heating rate of 10 K min-1. Additional analysis in the temperature range 293-573 K was also conducted for the Rb salt (563 K for the Cs salt) with a heating rate of 3 K min-1.
IR spectra were recorded as KBr pellets on a Bruker Alpha-T spectrometer in the 4000-350 cm-1 region. Single crystals for recording the spectra were selected under a stereo microscope.
Electron paramagnetic resonance (EPR) spectroscopy was used to determine the magnetic properties of the single crystals and polycrystalline samples. The investigation was carried out on a standard Bruker Elexys 580 X-band EPR spectrometer, equipped with an Oxford continuous-flow He cryostat, in the range 4-293 K. EPR spectra in the range 293-473 K were recorded on a Varian E-109 X-band spectrometer using a Bruker ER 4111 VT variable-temperature unit with a flow of cold N2 gas.
Single-crystal measurements were performed on an Oxford Diffraction Xcalibur Nova R (microfocus Cu tube) equipped with an Oxford Instruments CryoJet liquid N2 cooling device. H atoms of the water molecules were located in difference-Fourier maps and either refined (in the case of the Li structure) with the restraints d(O-H) = 0.95 (2) Å, d(HH) = 1.50 (4) Å, or constrained to ride on the parent O atom (for the Rb and Cs structures) with d(O-H) = 0.95 Å and Uiso(H) = 1.5Ueq(O). Crystallographic and refinement data for the structures are reported in Table 1. Different crystals were used for each measurement. For the non-centrosymmetric cases, the absolute structure was determined satisfactorily only for CsDDQ·2H2O measured at 293 K. For CsDDQ·2H2O measured at 120 K and for RbDDQ·2H2O measured at 120 and 293 K, the crystals were inversion twins, and were refined using the TWIN and BASF commands in SHELXL97 (Sheldrick, 2008). Further details of the refinement procedures are included in the supporting information.1 For CsDDQ·2H2O at 120 K, the absolute structure of the major component was identical to that for the crystal of CsDDQ·2H2O measured at 293 K. For RbDDQ·2H2O at 293 K, the Flack parameter did not differ significantly from 0.5, so the choice of absolute structure for the model is arbitrary. For RbDDQ·2H2O at 120 K the major component is inverted compared with CsDDQ·2H2O measured at 293 K, but the structure is presented here as the minor component in order that all of the structures can be immediately compared with each other [and to the density functional theory (DFT) optimized structures].
In the structure of LiDDQ·2H2O·Me2CO, the DDQ radicals are disordered over two orientations, rotated by approximately 180° about the O=CC=O axis. While each anion component has an approximate C2v symmetry, the disorder overall emulates D2h symmetry (Fig. 1). The two positions for the Cl atoms and cyano groups have refined site occupancy factors 0.658 (6):0.342 (6), respectively. Atoms of the disordered groups were refined as free anisotropic entities, with no restraints. Data were collected at 120 K only. The sub-optimal refinement results (Table 1) appear to be a result of the disorder, the small size of the crystal and its poor diffraction. Attempts to refine the structure in the space group P1 or to introduce any type of twinning model were not successful. Although the crystallization experiments were repeated, crystals of better quality could not be obtained. During the repeated crystallization trials, it was observed that a slight modification of crystallization conditions led to the formation of differently hydrated crystals of somewhat better quality. However, the crystal packing of that compound is of quite a different type and does not fit into the frame of the present work. The validity of the structure presented for LiDDQ·2H2O·Me2CO is supported by the periodic DFT calculations presented below.
Gas-phase calculations were performed with GAUSSIAN09 (Frisch et al., 2009), using the B3LYP functional (Lee et al., 1988; Becke, 1993) except for nucleus-independent chemical shifts, which were calculated with the Hartree-Fock Hamiltonian and GIAO (gauge independent atomic orbital) method. Vibrational frequencies were computed for antiferromagnetically coupled dimers in the crystal structures of RbDDQ·2H2O and CsDDQ·2H2O (293 K) with the symmetry-broken wavefunction. These fragments taken from the crystal structures involved the pair of semiquinone radical anions at the smallest distance in a cage of the surrounding molecular and ionic species. Full structural details are given in the supporting information . The geometry of the innermost pair of radical anions was optimized with all other atoms being frozen. Optimization and vibrational calculations of crystal structure fragments were carried out with the 6-31G(d,p) basis set and SDD pseudopotential with accompanying basis (Andrae et al., 1990) on the Cs and Rb atoms. The larger basis set 6-311+G(2d) was used in all other calculations.
Periodic DFT calculations were carried out for LiDDQ·2H2O·Me2CO and RbDDQ·2H2O, employing either the localized or plane-wave basis set approach, as implemented in the CRYSTAL06 (Dovesi et al., 2006) and VASP, Version 5.2 packages (Kresse & Hafner, 1993; Kresse & Furthmüller, 1996). With CRYSTAL06 the B3LYP functional was used, with the 6-31G(d,p) basis set modulated by Bloch functions over the periodic crystal lattice. For Rb, a semi-core basis set was used together with a pseudopotential from the CRYSTAL06 basis set library (Schoenes et al., 2008). VASP calculations utilized the PBE functional together with the projector augmented wave atomic pseudopotentials (Blöchl, 1994; Kresse & Joubert, 1999) and plane-wave basis set with a cutoff of 400 eV. Optimization was followed by the calculation of harmonic frequencies and normal modes. For RbDDQ·2H2O, the periodic model was constructed on the basis of the crystal structure at 120 K, and optimized in a fixed unit cell following the symmetry constraints of the P21212 space group. The calculations utilized a 4 × 1 × 2 mesh of k-points according to the Monkhorst-Pack scheme (Monkhorst & Pack, 1976). Both CRYSTAL06 and VASP were applied, and it was established that the two approaches gave closely comparable results. For LiDDQ·2H2O·Me2CO, two structure models were optimized, one in space group , keeping only the majority component of the disordered anion, and one in space group P1, keeping the major component for one anion in the unit cell, and the minor component for the other. These models were optimized using the CRYSTAL06 program package only, with the B3LYP functional used in conjunction with the standard Pople's 6-31G(d,p) basis set on C, H, O, N and Cl, and the 5-11G(d) basis set on Li (Mérawa et al., 2004). The experimental unit cell was used for both models, and kept fixed during optimization. The calculations in reciprocal space were performed on a 4 × 2 × 2 Monkhorst-Pack k-point mesh. The optimized crystal structures are provided in CIF format in the supporting information .
The hydrated alkali salts of DDQ turned out to be surprisingly stable, and could survive for months in air at room temperature. This is unlike the acetone solvates of tetrahalogenosemiquinones (Molcanov et al., 2011, 2012), which are unstable in air and rapidly decompose above 330-350 K. The thermal behaviour of RbDDQ·2H2O and CsDDQ·2H2O is very similar (Fig. 2). The TG/DTA experiments indicated that they are also quite stable at higher temperatures, both salts surviving heating up to 523 K. The decomposition process is triggered by release of water molecules that occurs in the temperature range 523-573 K for both the Rb and Cs salts. After dehydration, decomposition of the DDQ radical occurs, accompanied by three exothermic peaks in the DTA curve at 595, 740 and 842 K for Rb (576, 767 and 861 K for Cs) that can be attributed to the release of chlorine, cyano groups and finally the disintegration of the (semi)quinone ring.
| || Figure 2 |
TGA/DTA curves of (a) RbDDQ·2H2O and (b) CsDDQ·2H2O in a synthetic air and in an N2 atmosphere.
Additional thermal analysis in the temperature range 293-573 K for the Rb salt (563 K for the Cs salt) with a heating rate of 3 K min-1 was conducted to determine the stability of the compounds after release of the crystallization water molecules. There was no evidence of phase transitions on the DTA curve in this range. During the cooling there was no increase in mass, so it can be concluded that water molecules do not resorb into the structure.
The experimental values of bond lengths in DDQ match very well with our calculations (Fig. 3, Table 2). They also agree well with reported literature values, both by X-ray diffraction (Zanotti et al., 1980, 1982; Miller & Dixon, 1987; Marzotto et al., 1988; Tachikawa et al., 1993) and quantum-chemical methods (Miller & Dixon, 1987). The semiquinoid skeleton is similar to the skeletons of tetrahalogenosemiquinone radicals in their simple alkali salts (Lü et al., 2003; Rosokha et al., 2009; Molcanov et al., 2011, 2012) and it may be considered as an intermediate between aromatic and quinoid.
| || Figure 3 |
The DDQ anion in RbDDQ·2H2O (293 K) showing displacement ellipsoids at 50% probability. The same atom-numbering scheme is applied to all structures.
The DDQ radical anion is located in a general position in all structures, but its molecular symmetry is approximately C2v. The local, approximate, C2 axis passes through the midpoints of the C2-C3 and C5-C6 bonds. Two sides of the radical anion can be distinguished: one with Cl substituents and one with cyano groups (Rosokha & Kochi, 2007). There are slight geometrical differences between them. The bond C2-C3 (cyano group substituted sites) is longer than C5-C6 (Cl atom substituted sites) by 0.03 Å (Table 2); bond-length differences can also be observed in neutral DDQ (Table 3; Zanotti et al., 1980), but they are less pronounced. Four other C-C bonds differ by 0.01-0.02 Å (Table 2). These findings are in accordance with structures of previously determined DDQ radicals (Zanotti et al., 1980, 1982; Miller & Dixon, 1987; Marzotto et al., 1988, Rosokha & Kochi, 2007) and supported by our quantum-chemical calculations.
+Assignment according to Miller & Dixon (1987).
§Our assignment according to the B3LYP calculations with GAUSSIAN09.
##Assignment according to Zhang et al. (2013).
++Our assignment according to the B3LYP calculations with CRYSTAL06.
§§Assignment according to Boesch & Wheeler (1997).
###Assignment according to Boesch & Wheeler (1995).
The DDQ radical anions in the salts studied here are not perfectly planar, but have a puckered conformation, bent around the O=CC=O axis. The Cremer-Pople puckering parameter (Cremer & Pople, 1975) is in the range 2.0-3.3° (the largest value being for the Li salt), similar to other DDQ radical anions. Neutral DDQ (Zanotti et al., 1980; Table 2) exhibits a twist-boat conformation.
While the gas-phase calculations on the DDQ entity in various charge and spin states already provide a reasonable match with the experimental data, the agreement is improved by considering periodicity (Table 2). It can be seen that the B3LYP/6-31G(d,p) optimization of RbDDQ·2H2O (CRYSTAL06) provides a slightly better match than the PBE/PW method (VASP), and only the CRYSTAL06 method was applied for LiDDQ·2H2O·Me2CO.
To examine further the effect of the electron-rich cyano groups on the radical anion electron redistribution, the aromaticity of the 2,3-dicyano-5,6-dichlorosemiquinone radical anion was calculated. IR spectroscopy was used to observe the effect of the cyano-group substitution on the characteristic anion radical vibrations (Table 3). Vibrations were assigned by using literature data on neutral DDQ (Zhang et al., 2013) and DDQ (Miller & Dixon, 1987) and generally compared with tetrachlorosemiquinone radicals (Girlando et al., 1973, 1978; Hiroma & Kuroda, 1974; Boesch & Wheeler, 1995, 1997). DFT calculations (§2.6) were used to resolve ambiguities and assign bands which were not observed previously. Due to the lack of symmetry in the DDQ anion (approximate C2v only) a larger number of bands appear in the spectrum. Therefore, direct comparison with tetrachlorosemiquinone (D2h symmetry) is difficult.
Assignment of the bands in the region 1500-1700 cm-1 was complicated, since both C=O and C=C stretching bands are severely red-shifted due to electron delocalization in the radical anion. Literature data on these bands are unfortunately ambiguous due to a paucity of reliable experimental data. Our calculations in this region agree best with the assignment made by Miller & Dixon (1987). Other bands in the region 450-1350 cm-1 correspond to C-Cl and C-C stretching vibrations and various bending, ring stretching and ring distortion modes. Similar bands were previously observed by Boesch & Wheeler (1995, 1997) and Zhang et al. (2013).
While our gas-phase calculations did not reliably reproduce the experimental values, the agreement improved with the periodic approach, particularly with the CRYSTAL06 program. In line with the four RbDDQ units in the unit cell, a total number of 16 modes have been calculated for the C=O and C=C bond stretching vibrations. According to CRYSTAL06, 12 of these are IR-active and 4 are not. The IR-active bands corresponding to the C=O stretching vibration are predicted at 1703, 1657, 1656, 1655, 1576 and 1572 cm-1, while the C=C stretching modes are at 1626, 1625, 1616, 1531, 1506 and 1505 cm-1. This offers a plausible tentative assignment of the experimental peak located at 1697 cm-1 and the strong doublet at 1587 and 1571 cm-1 as (C=O), and the small shoulders around the doublet as (C=C) (Table 3). Accordingly, the peaks at 1453 and 1397 cm-1 could be representative of the ring deformation modes, calculated to be 1416 cm-1 and below by CRYSTAL06, but the agreement is somewhat worse. The modes computed by VASP are located at notably lower frequencies than their CRYSTAL06 counterparts [at roughly 1550 cm-1 for (C=O) and 1450 cm-1 for (C=C), with some mixed C=O + C=C modes in between], leaving the experimental band at 1697 cm-1 unexplained. However, despite the differences between CRYSTAL06 and VASP, both calculations make a clear distinction between the C=O and C=C stretching modes in that the former are located at higher frequencies than the latter, which supports the above assignment.
The strong red shift of the C=O stretching bands is related to a bond order of approximately 1.5, while C=C (C2-C3 and C5-C6 bonds) stretch bands are also red-shifted, as the corresponding bond order is considerably less than 2. The stretching bands C=O are also affected by hydrogen bonding through an acceptor function (Table 4). However, the single C-C bonds (C1-C2, C3-C4, C4-C5 and C6-C1) of the semiquinoid ring are blue-shifted, due to electron delocalization and the fact that their bond order is greater than 1. These results are in accordance with geometrical data (Table 2) and similar semiquinone salts (Lü et al., 2003; Rosokha et al., 2009; Molcanov et al., 2011, 2012). However, some C-C stretching bands are not blue-shifted relative to the neutral DDQ and they can be assigned to cyano groups (C2-C7 and C3-C8 bonds).
The aromaticity of the DDQ radical anion was assessed in the same way as was applied previously for the tetrahalogenosemiquinone radical anion (Molcanov et al., 2011), by calculation of the component of a shielding tensor (Stanger, 2006; nucleus independent chemical shift, NICS) orthogonal to the ring, at the distance 0-4 Å from the ring centroid (Fig. 4). For comparison, the HOMA (harmonic oscillator measure of aromaticity) index (Krygowski & Cyranski, 1996) was also calculated from bond distances in the DDQ rings (Table 2). NICS values were calculated for antiferromagnetically coupled dimers of DDQ radical anions on geometries taken from the crystal structures of RbDDQ·2H2O and CsDDQ·2H2O (on 293 and 120 K) with the symmetry-broken wavefunction. For comparison, NICS were also computed for the neutral, radical anion and dianion of an isolated DDQ, whose structures were obtained by geometry optimization. Fig. 4 shows that aromaticity of isolated DDQ strongly depends on charge, varying from a mild aromatic dianion to nonaromatic neutral molecule, with the radical anion in between. The NICS profiles for antiferromagnetic dimers taken from the RbDDQ·2H2O structure almost coincide with the single DDQ radical anion, while the pairs taken from the CsDDQ·2H2O salt deviate towards the dianion. In estimating the significance of the NICS results, it should be kept in mind that the exact amount of charge transfer between cations and anions in the crystal structure is not known. It could be easily less than 1, as assumed in the NICS calculation. It seems safe to conclude that the DDQ radical anions in the Rb and Cs salts are between aromatic and quinoid (nonaromatic). The HOMA values listed in Table 2 support this conclusion. In comparison with tetrahalogenosemiquinones (Molcanov et al., 2011), the aromaticity of DDQ seems to be reduced.
| || Figure 4 |
NICS values as a function of distance from the ring centroid, for dimers taken from the crystal structures of RbDDQ·2H2O and CsDDQ·2H2O, for single DDQ (neutral, radical anion and dianion) and for benzene.
The stability of the DDQ salts can be assigned to crystal packing. Water molecules may play a crucial role, stabilizing the packing by hydrogen bonds. In the isostructural RbDDQ·2H2O and CsDDQ·2H2O salts, hydrogen bonds and cation-anion interactions form a three-dimensional network (Fig. 5, Table 4). These two salts survive heating up to 523 K. In tetrahalogenosemiquinone salts, only cation-anion interactions are present, whereas acetone molecules are held by dispersion interactions only (Molcanov et al., 2011, 2012). However, tetrahalogenosemiquinone radicals are known to survive in air after decomposition of the crystals; the presence of radicals was detected by measuring the EPR signal of the amorphous samples (Molcanov et al., 2011, 2012). In LiDDQ·2H2O·Me2CO, acetone molecules complete the hydrogen-bonding network, resulting in two-dimensional arrays held together by dispersion interactions (Fig. 6, Table 4). The crystals are stable in air at room temperature.
| || Figure 5 |
Crystal packing of RbDDQ·2H2O (data collected at 120 K) viewed in the  direction. Rb+ cations are depicted as violet spheres of arbitrary radius. CsDDQ·2H2O is isostructural.
| || Figure 6 |
Crystal packing of LiDDQ·2H2O·Me2CO (data collected at 120 K) viewed in the  direction. Alkali cations are depicted as violet spheres of arbitrary radius. The C=O group of Me2CO acts as a proton acceptor in hydrogen bonding to water molecules, whereas the methyl groups of the acetone molecules form a hydrophobic channel between hydrogen-bonded chains of radical anions stabilized by Li+.
The main difference between the DFT-optimized and observed structures of RbDDQ·2H2O and CsDDQ·2H2O is not in the geometry of the DDQ entity, but rather in the orientation of the O3 hydrate water molecule involved in hydrogen bonds (§3.3, Table 4). The crystal structures at 120 and 293 K indicate slight rotational reorientation of the O4 water molecule at the higher temperature, suggesting that the energy barriers for such reorientations are reasonably low. One of the H atoms (H4A) is clearly involved in a hydrogen bond to the neighbouring water molecule O3, while the other H atom (H4B) points between atoms Cl1 and N2. The orientation of the water molecule in the optimized geometry matches more closely with the experimental structure at 120 K than at 293 K, which is in line with the formal zero-temperature concept of the geometry optimization.
For LiDDQ·2H2O·Me2CO, two ordered models were optimized. In the space group , including only the major disorder component for both DDQ molecules in the unit cell, the intermolecular interactions between DDQ radical anions are centrosymmetric (opposing dipoles), while in the P1 model, including the major disorder component in one DDQ molecule in the unit cell and the minor component in the other, the intermolecular interactions between DDQ radical anions are non-centrosymmetric (aligned dipoles). The calculated energy difference between these two models is 2.37 kJ mol-1 in favour of the centrosymmetric space group; however, the difference is very small, as expected for polymorphs of similar structures. In both models, the only significant change in the geometry upon optimization is reorientation of the methyl groups of the acetone solvent molecule, which cannot be of any importance for the properties of the systems studied. Comparison of the geometrical parameters of the optimized models with the X-ray structures reveals that the majority of the differences are within 3 standard deviations. The water molecule O5, that acts as the 2 bridge between two Li+ cations, displays a slightly larger movement on optimization so that the coordination geometry around Li+ becomes closer to regular tetrahedral: O1-Li-O5 = 116.2° (X-ray structure in ), 110.8° (optimized in ) and 108.6° (optimized in P1). Small differences in the geometric parameters of the hydrogen bonding involving O5 are also observed, but there is no change that is clearly significant.
The most interesting feature of the crystal structures is the -stacking of DDQ anions. Strong -interactions with spin coupling are not uncommon in crystal structures of semiquinone radicals (Rosokha et al., 2009; Molcanov et al., 2011). The structures described herein comprise stacks of closely located radical dimers (Fig. 7, Table 5) similar to those observed in alkali salts of tetrahalogenosemiquinones (Molcanov et al., 2011, 2012). Within the dimers, the interplanar separation distance is 2.9 Å (Table 5) and such a close contact implies coupling of electronic spins. The dimers are stacked into a column with larger separation (exceeding 3.4 Å, Fig. 7, Table 5).
| || Figure 7 |
-Stacks of DDQ in (a) LiDDQ·2H2O·Me2CO and (b) RbDDQ·2H2O. For (b), interplanar separation distances in RbDDQ·2H2O are indicated in black, whereas those of CsDDQ·2H2O are in red.
Due to a considerable offset (Table 5), the distances between C atoms in contiguous rings may be larger than the interplanar distance. Therefore, we also report CC distances in a closely interacting dimer: they range between 2.902 (9) Å (in LiDDQ·2H2O·Me2CO) to over 3.3 Å, which is still considerably shorter than the sum of van der Waals radii for C (3.5 Å; Cordero et al., 2008). These unusual and unconventional bonding interactions involving radical dimers with delocalized charge over the entire rings can be described as multi-centred bonding rather than conventionally localized and atom-centred (Miller & Novoa, 2007; Jose & Datta, 2011). Similar interactions have been found in a few other known DDQ salts. In the tetraethylammonium (Et4N+) salt (Zanotti et al., 1982) there are stacks of dimers with respective interplanar distances of 2.906 (6) and 3.626 (6) Å. Closely bound radical dimers with interplanar distance of 3.006 (4) Å [and offset of 2.007 (5) Å] were also found in the tetrapropylammonium salt (Pr4N+; Marzotto et al., 1988); however, these dimers are isolated, being surrounded by large tetrapropylammonium cations.
The DDQ anions are tethered by cation-anion interactions (Rosokha et al., 2009). In RbDDQ·2H2O and CsDDQ·2H2O, however, the more distant pairs within a stack are tethered (Fig. 8, Table 5), rather than the more closely bound pairs as observed previously (Rosokha et al., 2009; Molcanov et al., 2011). The shift in closely interacting dimers is transversal (i.e. perpendicular to the O=CC=O axis) in the Rb and Cs salts (Fig. 9a), while it is longitudinal in the Li salt (Fig. 9b). These findings, different from previous studies on semiquinone stacking (Rosokha et al., 2009; Molcanov et al., 2011), can help to recognize the character of the interactions between the rings.
| || Figure 8 |
Cation-tethered DDQ radical anions involving more distant rings (interplanar separation distance of 3.490 Å, Table 5) in RbDDQ·2H2O (room-temperature data). Symmetry operators: (i) -x, 1-y, z; (ii) x, y, -1+z; (iii) -x, 1-y, 1+z.
| || Figure 9 |
Closely interacting pairs of DDQ radicals within a stack: (a) lateral offset in RbDDQ·2H2O and CsDDQ·2H2O; (b) longitudinal offset in LiDDQ·2H2O·Me2CO (only the major component of the disorder is shown).
The open question concerns the influence of dipoles on -interactions. All quinones have strong local dipoles but in symmetrically substituted ones (such as tetrachloroquinone) the total dipole moment is zero. In DDQ cyano groups generate an extremely strong dipole of about 6.8 D (our calculation). The interplanar distance in stacked DDQ anion radicals is considerably shorter than in tetrahalogenosemiquinones (2.8-3.0 versus 3.1-3.3 Å, respectively), indicating much stronger attractions. Also, the orientation of anion dipoles in closely interacting dimers (Fig. 9) is antiparallel. Therefore, we may conclude that dipolar interactions make a significant contribution to the total interaction between the two DDQ anions.
The EPR measurements showed that both RbDDQ·2H2O and CsDDQ·2H2O were EPR-silent, which proves that two radicals associate as spin-paired dimers in the solid state, producing a complete antiferromagnetic spin pairing. This leads to the diamagnetic behaviour of the samples over the whole measured temperature range (4-473 K).
Stabilization of the 2,3-dicyano-5,6-dichloro-semiquinone (DDQ) radical anion in the crystalline state was achieved using alkali metal salts through polar interactions between cations and radical anions. Cation-tethered anion radicals occur between dimers (not within dimers), involving the more distant ring pairs within a -stack (Table 5, Fig. 8). The experimental values of bond lengths in the DDQ match very well our gas-phase and periodic DFT calculations (Table 2), and also the quantum-chemical calculations reported by Miller & Dixon (1987). The effect of substitution of the cyano groups on the semiquione ring was detected by (non)aromaticity of DDQ rings (NICS approach) and shifts of characteristic bands in IR spectra for the studied Rb and Cs salts. According to NICS calculations (Fig. 4) and HOMA values (Table 2), the DDQ radical anions in the Rb and Cs salts are between aromatic and quinoid (nonaromatic). Nevertheless, the best agreement with the experimental structure of RbDDQ is achieved by periodic DFT calculations, indicating the importance of intermolecular interactions within the crystal structures.
The striking interactions in the crystal packing are - interactions, generating columns of stacked radical anions. These columns are built of stacked dimers that are characterized by planar separation distances being significantly shorter than the sum of C-C van der Waals radii [2.812 (3) Å for the Li salt; 2.867 (2) Å for the Rb salt; 2.901 (4) Å for the Cs salt; values for 120 K data]. These short distances, and the lack of signal in electron paramagnetic resonance spectra, correlate with the diamagnetic behaviour of these compounds and spin coupling of unpaired electrons of radical anions within dimers. The extensive O-HO and O-HN hydrogen bonding involving crystal water molecules and radical anions contributes to stabilization of the radical anions within the crystal lattice. The Li salt, with a small ionic radius for Li+, achieves additional stabilization through incorporation of acetone solvate molecules.
This work was supported by Ministry of Science, Education and Sports of Croatia (grant Nos. 098-1191344-2943, 098-0982915-2939 and 098-0982915-2942 and a Croatian-Slovenian bilateral collaboration).
Agilent (2012). CrysAlisPRO. Agilent Technologies Ltd, Abingdon, England.
Andrae, D., Haussermann, U., Dolg, M., Stoll, H. & Preuss, H. (1990). Theor. Chim. Acta, 77, 123-141.
Becke, A. D. (1993). J. Chem. Phys. 98, 5648-5652.
Blöchl, P. E. (1994). Phys. Rev. B, 50, 17953-17979.
Boesch, S. E. & Wheeler, R. A. (1995). J. Phys. Chem. 99, 8125-8134.
Boesch, S. E. & Wheeler, R. A. (1997). J. Phys. Chem. A, 101, 8351-8359.
Constantinides, C. P., Koutentis, P. A. & Rawson, J. M. (2012). Chem. Eur. J. 18, 7109-7116.
Cordero, B., Gómez, V., Platero-Prats, A. E., Revés, M., Echeverría, J., Cremades, E., Barragán, F. & Alvarez, S. (2008). Dalton Trans. pp. 2832-2838
Cremer, D. & Pople, J. A. (1975). J. Am. Chem. Soc. 97, 1354-1358.
Decken, A., Mailman, A., Passmore, J., Rautiainen, J. M., Scherer, W. & Scheidt, E. W. (2011). Dalton Trans. 40, 868-879.
Dovesi, R., Saunders, V. R., Roetti, C., Orlando, R., Zicovich-Wilson, C. M., Pascale, F., Civalleri, B., Doll, K., Harrison, N. M., Bush, I. J., D'Acro, P. & Llunell, M. (2006). CRYSTAL06 User's Manual. University of Torino, Italy.
Farrugia, L. J. (2012). J. Appl. Cryst. 45, 849-854.
Flack, H. D. (1983). Acta Cryst. A39, 876-881.
Frisch, M. J. et al. (2009). GAUSSIAN09, Revision A.2. Gaussian, Inc., Wallingford, CT, USA.
Fujita, W. & Awaga, K. (1999). Science, 286, 261-263.
Garcia-Yoldi, I., Miller, J. S. & Novoa, J. J. (2009). J. Phys. Chem. A, 113, 7124-7132.
Gholami, M. & Tykwinski, R. R. (2006). Chem. Rev. 106, 4997-5027.
Girlando, A., Morelli, L. & Pecile, C. (1973). Chem. Phys. Lett. 22, 553-558.
Girlando, A., Zanon, I., Bozio, R. & Pecile, C. (1978). J. Chem. Phys. 68, 22.
Hicks, R. G. (2011). Nat. Chem. 3, 189-191.
Hiroma, S. & Kuroda, H. (1974). Bull. Chem. Soc. Jpn, 47, 3014-3020.
Jose, D. & Datta, A. (2011). Cryst. Growth Des. 11, 3137-3140.
Kresse, G. & Furthmüller, J. (1996). J. Phys. Rev. B, 54, 11169-11186.
Kresse, G. & Hafner, J. (1993). J. Phys. Rev. B, 47, 558-561.
Kresse, G. & Joubert, D. (1999). Phys. Rev. B, 59, 1758-1775.
Krygowski, T. M. & Cyranski, M. (1996). Tetrahedron, 52, 1713-1722.
Lee, C., Yang, W. & Parr, R. G. (1988). Phys. Rev. B, 37, 785-789.
Lü, J. M., Rosokha, S. V. & Kochi, J. K. (2003). J. Am. Chem. Soc. 125, 12161-12171.
Macrae, C. F., Edgington, P. R., McCabe, P., Pidcock, E., Shields, G. P., Taylor, R., Towler, M. & van de Streek, J. (2006). J. Appl. Cryst. 39, 453-457.
Marzotto, A., Clemente, D. A. & Pasimeni, L. (1988). J. Crystallogr. Spectrosc. Res. 18, 545-554.
Mérawa, M., Labeguerie, P., Ugliengo, P., Doll, K. & Dovesi, R. (2004). Chem. Phys. Lett. 387, 453-459.
Michinobu, T., Boudon, C., Gisselbrecht, J. P., Seiler, P., Frank, B., Moonen, N. N. P., Gross, M. & Diederich, F. (2006). Chem. Eur. J. 12, 1889-1905.
Miller, J. S. (2011). Chem. Soc. Rev. 40, 3266-3296.
Miller, J. S. & Dixon, D. A. (1987). Science, 235, 871-873.
Miller, J. S. & Novoa, J. J. (2007). Acc. Chem. Res. 40, 189-196.
Min, K. S., DiPasquale, A. G., Golen, J. A., Rheingold, A. L., Arif, A. M. & Miller, J. S. (2007). J. Am. Chem. Soc. 129, 2360-2368.
Min, K. S., Rheingold, A. L., Dipasquale, A. & Miller, J. S. (2006). Inorg. Chem. 45, 6135-6137.
Molcanov, K., Kojic-Prodic, B., Babic, D., Pajic, D., Novosel, N. & Zadro, K. (2012). CrystEngComm, 14, 7958.
Molcanov, K., Kojic-Prodic, B., Babic, D., Zilic, D. & Rakvin, B. (2011). CrystEngComm, 13, 5170.
Monkhorst, H. J. & Pack, J. D. (1976). Phys. Rev. B, 13, 5188-5192.
Novoa, J. J., Deumal, M. & Jornet-Somoza, J. (2011). Chem. Soc. Rev. 40, 3182-3212.
Podzorov, V. (2010). Nature Mater. 9, 616-617.
Ratera, I. & Veciana, J. (2012). Chem. Soc. Rev. 41, 303-349.
Rosokha, S. V. & Kochi, J. K. (2007). J. Am. Chem. Soc. 129, 3683-3697.
Rosokha, S. V., Lu, J., Rosokha, T. Y. & Kochi, J. K. (2009). Phys. Chem. Chem. Phys. 11, 324-332.
Salunke-Gawali, S., Rane, S. Y., Boukheddaden, K., Codjovi, E., Linares, J., Varret, F. & Bakare, P. P. (2005). J. Therm. Anal. Calorim. 79, 669-675.
Sanvito, S. (2011a). Chem. Soc. Rev. 40, 3336-3355.
Sanvito, S. (2011b). Nature Mater. 10, 484-485.
Schoenes, J., Racu, A.-M., Doll, K., Bukowski, Z. & Karpinski, J. (2008). Phys. Rev. B, 77, 134515.
Sheldrick, G. M. (2008). Acta Cryst. A64, 112-122.
Shultz, D. A., Boal, A. K., Driscoll, D. J., Farmer, G. T., Kitchin, J. R., Miller, D. B. & Tew, G. N. (1997). Mol. Cryst. Liq. Cryst. Sci. Technol. Sect. A, 305, 303-310.
Spek, A. L. (2003). J. Appl. Cryst. 36, 7-13.
Stanger, A. (2006). J. Org. Chem. 71, 883-893.
Tachikawa, T., Izuoka, A. & Sugawara, T. (1993). J. Chem. Soc. Chem. Commun. pp. 1227-1229.
Verma, P., Weir, J., Mirica, L. & Stack, T. D. (2011). Inorg. Chem. 50, 9816-9825.
Yi, C., Blum, C., Liu, S. X., Keene, T. D., Frei, G., Neels, A. & Decurtins, S. (2009). Org. Lett. 11, 2261-2264.
Yu, X., Mailman, A., Lekin, K., Assoud, A., Robertson, C. M., Noll, B. C., Campana, C. F., Howard, J. A., Dube, P. A. & Oakley, R. T. (2012). J. Am. Chem. Soc. 134, 2264-2275.
Zanotti, G., Bardi, R. & Del Pra, A. (1980). Acta Cryst. B36, 168-171.
Zanotti, G., Del Pra, A. & Bozio, R. (1982). Acta Cryst. B38, 1225-1229.
Zhang, Y., Zhang, F., Ma, K. & Tang, G. (2013). Spectrochim. Acta A, 105, 352-358.