Supramolecular structures of NiII and CuII with the sterically demanding Schiff base dyes driven by cooperative action of preagostic and other non-covalent interactions

This work shows the importance of counterintuitive ultra long distances, intramolecular preagostic attractive C—H⋯Ni/Cu and C—H⋯O, C—H⋯H—C contacts and a bunch of other extremely strong non-classic mainly London dispersion driven intermolecular interactions involving C—H bonds: C—H⋯X (X = H—C, N, O, π) in the crystal structures of the NiII and CuII based complexes fabricated from the sterically crowding cyclohexyl-containing Schiff base dyes. Photophysical properties of the reported complexes are also discussed.

Another peculiar type of non-covalent interaction, namely anagostic interaction (Brookhart et al., 2007;Sundquist et al., 1990), is of ever-growing interest owing to its presence in many catalytic processes. This type of interaction is inherent to square-planar d 8 -metal complexes, and sometimes anagostic interactions are speculatively claimed as agostic interactions (Castro et al., 2005;Thakur & Desiraju, 2006). However, agostic and anagostic interactions differ significantly from the structural point of view. In particular, the former interactions are characterized by the MÁ Á ÁH distance of $1.8-2.2 Å and C-HÁ Á ÁM bond angles of $90-140 , while the latter interactions exhibit long MÁ Á ÁH distances of $2.3-3.0 Å and C-HÁ Á ÁM bond angles of $110-170 (Brookhart et al., 2007). While agostic bonds are attractive, it is still under debate as to whether anagostic bonds are attractive or repulsive.
With all this in mind and in continuation of our investigations in the field of non-covalent interactions, as well as studying their influence on the structure stabilization, we have directed our attention to molecules containing several synthons that can produce non-covalent interactions. Thus, we have addressed Schiff base dyes. The main advantage being the ease of synthesis by condensation of corresponding aldehydes with primary amines under mild conditions. In particular, we have selected bulky cyclohexylamine and salicyaldehyde/2-hydroxy-1-naphthaldehyde. The resulting Schiff bases o-HOC 6 H 4 -CH=N-cyclo-C 6 H 11 (HL I ) and o-HOC 10 H 6 -CH=N-cyclo-C 6 H 11 (HL II ) ( Fig. 1  Diagrams of the applied Schiff base dyes. aliphatic cyclohexane rings and metal-containing chelate rings. Theoretical studies are then applied to shed light on the origin of their photophysical properties. Although the crystal structures of [Ni(L I ) 2 ] (Bhatia et al., 1983), [Cu(L I ) 2 ] (Jain & Syal, 1988;Kashyap et al., 1975;Tamura et al., 1977) and [Cu(L II ) 2 ] (Ferná ndez-G et al., 1997) were reported recently, we have decided to redefine the structures with a higher precision as well as identify classic and unintuitive non-covalent interactions responsible for the formation of their supramolecular structures.

Results and discussion
A reaction of an equimolar amount of cyclohexylamine and salicyaldehyde or 2-hydroxy-1-naphthaldehyde in ethanol under reflux yielded the Schiff bases HL I,II as yellow viscous oil. HL I,II were involved in the reaction with a half molar amount of M(CH 3 COO) 2 (M = Ni, Cu) in ethanol. As a result, discrete mononuclear homoleptic complexes [Ni(L I,II ) 2 ] and [Cu(L I,II ) 2 ], respectively, were isolated with high yields.
Complexes [Ni(L I ) 2 ] and [Cu(L I ) 2 ] were found to be isostructural, as shown by the single-crystal X-ray diffraction data (see the Experimental Section). Their crystal structures were best solved in the triclinic space group P-1 (No. 2), while the crystal structures of [Ni(L II ) 2 ] and [Cu(L II ) 2 ] were solved in the monoclinic space group P2 1 /n with a half of the complex molecule in the asymmetric unit for all the complexes. In complexes, the metal cation is coordinated by two molecules of the deprotonated ligand L I,II via imine nitrogen atom and phenoxy oxygen atom affording a tetracoordinate environment with the formation of a perfect square-planar coordination geometry as shown by the 4 descriptor ( Fig. 2, Table 1) (Yang et al., 2007). The ligands are linked in a trans-configuration with the six-membered chelate rings adopting an envelope conformation in the structures of [Ni(L I,II ) 2 ] and [Cu(L I ) 2 ], while they are much more planar in the structure of [Cu(L II ) 2 ] (Fig. 2, Table 1). The cyclohexyl fragments are in a typical chair conformation (Fig. 2). The M-N bond lengths are $1.9-2.0 Å , while the M-O bonds are $0.1 Å shorter ( Table 1). The C=N and C-O bonds in the structures of the complexes are very similar and are $1.3 Å (Table 1) Table 1).
The angles between planes formed by the benzene or naphthyl and cyclohexyl rings corresponding to the same ligand in the structures of [Ni(L I,II ) 2 ] and [Cu(L I ) 2 ] are $35 , while the same angles in the structure of [Cu(L II ) 2 ] are $45 . The same angles between the planes formed by the benzene or naphthalene and chelate rings, and cyclohexyl and chelate rings are $7-11 and 45 , respectively ( Table 1).
The crystal structures of the complexes are stabilized by a set of intramolecular interactions (Fig. 2, Table 2). In particular, the hydrogen atom of the cyclohexyl tertiary carbon is involved in the C-HÁ Á ÁO interaction with the oxygen atom of a second ligand (Fig. 2, Table 2). In the structures of [Ni(L I ) 2 ] and [Cu(L I,II ) 2 ] the same oxygen also forms the second C-HÁ Á ÁO bond with one of the hydrogen atoms from one of the secondary carbons linked to the tertiary carbon ( Fig. 2, Table  2). However, the latter non-covalent bond is significantly longer than the former one because of the formation of an anagostic bond by the same hydrogen atom (Fig. 2, Table 2). The same anagostic bond was found in the structure of [Ni(L II ) 2 ], which formation, together with a coordination geometry of chelate cycles, prevents the formation of the second intramolecular C-HÁ Á ÁO bond. Notably, all crystal structures are further stabilized by intermolecular non-covalent interactions of the C-HÁ Á Á(benzene/naphthalene) and C-HÁ Á Á(chelate) nature (Fig. 3, Table 2).
The bulk samples of all the complexes are free from phase impurities as shown by comparison of the experimental X-ray powder patterns with calculated powder patterns generated from the single-crystal X-ray data (see Fig. S1    Supporting information), as well as from the elemental analysis data (see the Experimental Section).
We further applied a Hirshfeld surface analysis (Spackman & Jayatilaka, 2009) to study intermolecular interactions in the crystal structures of both complexes. As a result, a set of 2D fingerprint plots (Spackman & McKinnon, 2002) were generated using CrystalExplorer 3.1 (Wolff et al., 2012). In order to estimate the propensity of two chemical species to be in contact, we calculated the enrichment ratios (E) (Jelsch et al., 2014) of the intermolecular contacts.
It was found that the intermolecular HÁ Á ÁH and HÁ Á ÁC contacts occupy an overwhelming majority of the molecular surfaces of all the complexes (Table 3). There is a clear splitting of the HÁ Á ÁH fingerprint of [Ni(L I,II ) 2 ] and [Cu(L I ) 2 ], which is caused by the shortest contact being between three atoms, rather than being a direct two-atom contact (Figs. S2-S4) (Spackman & McKinnon, 2002). The HÁ Á ÁC contacts are shown in the form of 'wings' (Figs. S2-S4), with the shortest d e + d i ' 2.7 Å , and are recognized as characteristic of C-HÁ Á Á nature (Spackman & Jayatilaka, 2009). The structures of [Ni(L I,II ) 2 ] and [Cu(L I ) 2 ] are also characterized by significantly smaller proportions of the HÁ Á ÁN and HÁ Á ÁO contacts (Table 3). Furthermore, the proportions of these contacts are even smaller in the structure of [Cu(L II ) 2 ], while the proportions of the CÁ Á ÁC, CÁ Á ÁN, CÁ Á ÁO and CÁ Á ÁCu contacts are quite distinct (Table 3, Fig. S5). This is explained by the formation of (chelate)Á Á Á(naphthalene) intermolecular interactions (Table 2). Notably, the molecular surface of all the structures is also described by HÁ Á ÁM intermolecular contacts ( Table 3, Figs. S2-S5), which are assigned to the abovementioned intermolecular C-HÁ Á ÁM and C-HÁ Á Á(chelate) interactions ( Table 2). All the HÁ Á ÁX contacts are favoured in the structures of [Ni(L I,II ) 2 ] and [Cu(L I ) 2 ], since the corre-sponding enrichment ratios E HX are close to or even higher than unity (Table 3). However, only HÁ Á ÁH and HÁ Á ÁC intermolecular contacts are favoured in the structure of [Cu(L II ) 2 ], while remaining contacts are impoverished (Table 3).
In order to complement the above structural and Hirshfeld surface analyses, and to determine which contacts stabilize/destabilize the obtained crystals, we performed in-depth bonding studies based on the two complementary approaches, namely the chargeand energy-decomposition scheme ETS-NOCV (Mitoraj et al., 2009) as well as the Interacting Quantum Atoms (IQA) scheme (Blanco et al., 2005). The former approach is well suited for the description of intermolecular interactions, whereas the latter approach is more convenient for analyses of various intramolecular contacts and is particularly useful since it can determine whether still-controversial long-distance intramolecular C-HÁ Á ÁM contacts could be repulsive (anagostic) or attractive (agostic). We have recently discovered (Mitoraj et al., 2019b), contrary to intuition and the recent state of knowledge (Scherer et al., 2015), that longer C-HÁ Á ÁNi distances ($3 Å ) can stabilize the complex structure. However, the shortening of C-HÁ Á ÁNi contacts up and London dispersion terms (Lu et al., 2018), overall might bring the repulsive C-HÁ Á ÁNi interactions owing to overwhelming positive (destabilizing) Coulomb constituent (Mitoraj et al., 2019b).
The selected IQA/MP2/6-311 + G(d,p) diatomic interaction energies ÁE int for the discussed structures are gathered in Fig.  4 and Table 4. Notably, despite a long NiÁ Á ÁH distance of 2.885 Å in [Ni(L I ) 2 ], a very efficient intramolecular instantaneous stabilization is gained with ÁE int (NiÁ Á ÁH) = À11.36 kcal mol À1 . It is mainly owing to the attractive Coulomb contribution ÁE Coulomb = À10.00 kcal mol À1 and slightly stabilizing exchange-correlation term ÁE XC = À1.36 kcal mol À1 (Fig. 4, Table 4). It is important to note that for the Ni II square-planar complex previously studied by us based on N-thiophosphorylated thiourea ligands, where exactly the same NiÁ Á ÁH distance was noticed (formed by a hydrogen atom of the methyl unit with nickel), the Coulomb term appeared to be positive, which led to the overall repulsive (anagostic) C-HÁ Á ÁNi interactions (Mitoraj et al., 2019b). This clearly demonstrates different electron-density distribution within the methyl and methylene groups, which in turn is reflected in the opposite values of the Coulomb terms. The origin of such intriguing behaviour will be more carefully studied in the future in order to obtain a more general overview of the nature of long-distance intramolecular C-HÁ Á ÁM interactions.
It was further found that there are two less important stabilizing intramolecular interactions than NiÁ Á ÁH: ÁE int (CÁ Á ÁH) = À6.99 kcal mol À1 and ÁE int (OÁ Á ÁH) = À5.89 kcal mol À1 (Fig. 4, Table 4)  belonging to the family of C-HÁ Á Á contacts, is electrostatically dominated with the major attractive ÁE Coulomb = À6.41 kcal mol À1 , whereas, interestingly, in the latter case, the Coulomb term appears to be repulsive and the sole prevailing attractive constituent is the exchange-correlation energy ÁE XC = À7.23 kcal mol À1 (Fig. 4, Table 4). Notably, the second longer OÁ Á ÁH contact leads to the overall complex destabilization owing to the strongly unfavourable Coulomb contribution, ÁE Coulomb = 11.59 kcal mol À1 , and the weaker exchange-correlation constituent (Fig. 4, Table 4). It is a very intriguing physical outcome since C-HÁ Á ÁO contacts are considered in the literature as rather purely stabilizing interactions (Grabowski, 2011;Grabowski & Lipkowski, 2011;Tsuzuki, 2012). We have shown here that intramolecular C-HÁ Á ÁO interactions might be both attractive and repulsive depending on distance variation (Fig. 4, Table 4). The existence of a stabilizing charge-delocalization channel (XC) for such ultra long distance OÁ Á ÁH is also an important observation. It has been additionally supported by the ETS-NOCV results where the mentioned intramolecular charge-delocalization channels in addition to C-HÁ Á ÁH-C (Cukrowski et al., 2016;Liptrot & Power, 2017;Mitoraj et al., 2020;Wagner & Schreiner, 2015) have been discovered (Fig. S6). Recently, the latter has been of particular attention in terms of reconsidering the real nature of steric crowding in bulky species (Cukrowski et al., 2016;Liptrot & Power, 2017;Mitoraj et al., 2020;Wagner & Schreiner, 2015). Notably, substitution of L I by L II leads to a similar picture of the already discussed intramolecular non-covalent interactions (Fig. 4, Table 4). Interestingly, in complex [Ni(L II ) 2 ] the second NiÁ Á ÁH contact, with quite similar length to the first, was revealed, which, however, destabilizes the overall structure, although quite insignificantly owing to an unfavourable Coulomb term and negligible stemming from the exchange-correlation constituent (Fig. 4, Table 4). As far as the copper-containing complex [Cu(L I ) 2 ] is concerned, quite similar stabilizing intramolecular interactions CÁ Á ÁH and OÁ Á ÁH were obtained (Fig. 4, Table 4). It is particularly interesting that the CuÁ Á ÁH contact is associated   with the significant stabilization ÁE int (CuÁ Á ÁH) = À14.16 kcal mol À1 despite a very long distance of 3.065 Å (Fig. 4, Table 4). Furthermore, the same close contact in [Cu(L II ) 2 ] results in even more efficient preagostic attraction ÁE int (CuÁ Á ÁH) = À14.67 kcal mol À1 owing to a shorter distance of 3.015 Å (Fig. 4, Table 4). Interestingly, the stabilization in the same complex is further augmented by the second preagostic contact with the corresponding ÁE int (CuÁ Á ÁH) = À8.81 kcal mol À1 (Fig. 4, Table 4). Finally, we briefly analyzed the intermolecular interactions in the example dimeric model of [Ni(L II ) 2 ] using the ETS-NOCV scheme (Fig. 5). It was found that the monomers are extremely strongly bonded to each other, with the overall binding energy ÁE total = À61.80 kcal mol À1 mostly owing to C-HÁ Á Á, C-HÁ Á ÁO, C-HÁ Á ÁN and C-HÁ Á ÁNi contacts. In line with the literature (Grabowski, 2011;Grabowski & Lipkowski, 2011;Tsuzuki, 2012), the London dispersion constituent is indeed the major contributor with $45% of the overall stabilization (Fig. 5). We have complemented herein that the charge-delocalization contribution ÁE orb = À28.76 kcal mol À1 is also a crucial cofactor (36% of the overall stabilization) as opposed to the literature claims on insignificance of this constituent (Grabowski, 2011;Grabowski & Lipkowski, 2011;Tsuzuki, 2012). The electrostatic term ÁE elstat = À15.52 kcal mol À1 appears to be the least important (Fig. 5). Quite similar sets of intermolecular non-covalent interactions, but significantly weaker, are valid in the counterpart [Ni(L I ) 2 ] (Fig. S7).
The Fourier transform infrared (FTIR) spectra of the complexes are pairwise very similar and each contain characteristic bands for the C=C and C=N bonds at 1500-1650 cm À1 (Fig. 6). The C-H groups of the cyclohexyl fragments are shown as bands at 1325-1340 and 1450 cm À1 , and a set of bands at 2800-3000 cm À1 . The aromatic and imine C-H functions are shown as a set of weak bands at 3000-3100 cm À1 . Notably, the IR spectra of the complexes do not exhibit a characteristic band for the OH group in the range 3200-3400 cm À1 (Fig. 6). This testifies to the deprotonated form of the parent ligands in the structures of the complexes.
Dissolving crystals of [Ni(L I,II ) 2 ] and [Cu(L I,II ) 2 ] in CH 2 Cl 2 yields yellow and reddish yellow solutions, respectively. In the UV-Vis absorption spectra of the complexes, three regions can be clearly defined. The first region, ranging from 200 to $300 nm, contains a set of high intense bands corresponding to intraligand ! * and n ! * transitions (Fig. 7). The second range at $300-440 nm exhibits significantly less intense bands for the metal-to-ligand charge transfer (MLCT) transitions (Fig. 7). Finally, the weak shoulder in the longerwavelength region of the spectra is caused by ligand field (dd) transitions (Fig. 7).
In order to shed light on the electronic transitions, we reoptimized all four complexes followed by modelling of the absorption spectra with the TDDFT/B3LYP/TZVPP/ PCM(CH 2 Cl 2 ) calculations. Since the qualitative picture of the electronic transitions is similar for all the complexes, we briefly discuss the data for [Ni(L I ) 2 ]. All the complexes remain a square-planar geometry in CH 2 Cl 2 and, in line with the experimental data, the analogous three absorption regions were obtained for all species (Fig. 7). The absorption bands at 300-400 nm are indeed predominantly characterized as MLCT, d xz (M) ! *, as indicated by the dominant transition #13 with the oscillator strength f = 0.208 a.u. (Fig. 8). However, the latter two less intense transitions, #12 ( f = 0.106 a.u.) and #5 ( f = 0.088 a.u.), are additionally described by both the ligand-to-ligand and ligand-to-metal charge transfers (Fig. 8).  Table 3 Hirshfeld contact surfaces, derived 'random contacts' and 'enrichment ratios' for [Ni(L I,II ) 2 ] and [Cu(L I,II ) 2 ]. Fingerprint plots of the observed contacts are available in the Supporting information.
[Ni(L I ) 2 ] [Cu (L I ) Importantly, it was found that all the complexes are emissive in CH 2 Cl 2 ; however, complex [Cu(L II ) 2 ] is remarkably more emissive (Fig. 9). The emission spectra of [Ni(L I,II ) 2 ] and [Cu(L II ) 2 ] exhibit a broad intense band centred at $435-450 nm, while the spectrum of [Cu(L I ) 2 ] exhibits a broad band with two maxima at $375 and 430 nm (Fig. 9). Assignment of these bands was made based on the excitation spectra (Fig. 9). As evident from comparison of the excitation and UV-Vis spectra of the complexes, the emission bands arise from the MLCT emission.

Conclusions
In summary, we studied structural and photophysical properties of the Ni II and Cu II discrete mononuclear homoleptic complexes [Ni(L I,II ) 2 ] and [Cu(L I,II ) 2 ], fabricated from the Schiff base dyes o-HOC 6 H 4 -CH=N-cyclo-C 6 H 11 (HL I ) and o-HOC 10 H 6 -CH=N-cyclo-C 6 H 11 (HL II ), respectively, each containing a bulky aliphatic fragment, namely cyclohexyl.
Single-crystal X-ray diffraction revealed that all the structures exhibit a trans-square-planar geometry. Remarkably, the six-membered metallocycles adopt a clearly defined envelope conformation in [Ni(L I,II ) 2 ] and [Cu(L I ) 2 ], while they are much more planar in the structure of [Cu(L II ) 2 ]. This was found to be clearly associated with the formation of different intra-and inter-molecular contacts, which were deeply characterized by the charge-and energy-decomposition scheme ETS-NOCV as well as the IQA approach. In particular, London dispersion dominated intramolecular C-HÁ Á ÁO, C-HÁ Á ÁN and C-HÁ Á ÁH-C interactions were identified and, predominantly, the attractive, mostly Coulomb driven, C-HÁ Á ÁNi/Cu preagostic (not repulsive anagostic) bonds were discovered despite their long distances ($2.8-3.1 Å ). Interestingly, despite the long distances, non-negligible charge-delocalization constituent was discovered. Notably, all the crystal structures are further stabilized by very efficient (the interaction energy is >60 kcal mol À1 ) intermolecular C-HÁ Á Á(benzene) and C-HÁ Á Á(chelate) interactions, which are responsible for their high stability as seen from the thermogravimetric (TG) analyses. Although they contain the prevailing dispersion constituent, the charge-delocalization contribution is only slightly less important followed by the Coulomb term. Our results, clearly showing that the bulky cyclohexyl groups are the sources of London dispersion stabilization, are in line with the recent discoveries outlining the true character of steric effects in small and sizable species (Cukrowski et al., 2016;Liptrot & Power, 2017;Mitoraj et al., 2019dMitoraj et al., , 2020 Table 4 IQA energy decomposition of the selected diatomic interactions obtained at the MP2/6-311 + G(d,p) level of theory for the crystal monomers of [Ni(L I,II ) 2 ] and [Cu(L I,II ) 2 ]. ÁE int = ÁE Coulomb + ÁE XC , where ÁE int is the overall diatomic interaction energy, ÁE Coulomb is the Coulomb constituent and ÁE XC is the exchangecorrelation contribution (Blanco et al., 2005).   Table 4 for details).
C-HÁ Á ÁO interactions can be both attractive and repulsive depending on the distance. Finally, dissolving crystals of the complexes in CH 2 Cl 2 yielded yellow and reddish yellow solutions for the Ni II and Cu II derivatives, respectively. The UV-Vis absorption spectra exhibit three clearly defined regions, corresponding to intraligand ! * and n ! * transitions, MLCT transitions and ligand field (d-d) transitions, as indicated by the time-dependent density functional theory (TDDFT) computations. Importantly, all the complexes were found to be planar and photoluminescent in CH 2 Cl 2 , with [Cu(L II ) 2 ] exhibiting the most pronounced emission, mostly owing to MLCT transitions.

Materials
All reagents and solvents were commercially available and used without further purification.

Physical measurements
Nuclear magnetic resonance (NMR) spectra in CDCl 3 were obtained on a Bruker AVANCE II 400 MHz spectrometer at 25 C. Chemical shifts are reported with reference to SiMe 4 . Infrared spectra (KBr) were recorded with a FT-IR FSM 1201 spectrometer in the range 400-3400 cm -1 . UV-Vis and fluorescent spectra from the freshly prepared solutions (5 Â 10 À5 M) in freshly distilled CH 2 Cl 2 were recorded on an Agilent 8453 instrument and a RF-5301 spectrofluorophotometer. TG analyses were performed by a NETZSCH STA 449 F5 Jupiter instrument in a dynamic air or argon atmosphere (100 ml min À1 ) from laboratory temperature to 1000 C with a 10 C min À1 heating rate. Microanalyses were performed using a ElementarVario EL III analyzer.

Synthesis of HL I,II
A solution of an equimolar amount of salicylaldehyde or 2hydroxy-1-naphthaldehyde (10 mmol; 1.221 and 1.722 g, respectively) and cyclohexylamine (10 mmol, 0.992 g) in ethanol (50 ml) was stirred for 1 h under reflux. For a solution of HL I , the solvent and non-reacted starting materials were removed in vacuo. The resulting yellow viscous oil was analyzed and used as is. The resulting solution of HL II was allowed to cool to room temperature to give crystals, which were filtered off.
(a) HL I . Yield = 1.809 g (89% To a solution of HL I,II (2 mmol; 0.407 and 0.507 g, respectively) in ethanol (10 ml) was added a solution of Ni(CH 3-COO) 2 4H 2 O (0.249 g, 1 mmol) or Cu(CH 3 COO) 2 (0.182 g, 1 mmol) in a mixture of water (1 ml) and ethanol (50 ml). The mixture was stirred at room temperature for 1 h. The resulting precipitate was filtered off, washed with ethanol (3 Â 50 ml) and dried in vacuo. Then the product was dissolved in CH 2 Cl 2 . X-ray suitable crystals were formed during the next few days upon slow evaporation of the solvent. X-ray powder diffraction for a bulk sample was carried out using a Rigaku Miniflex X-ray powder diffractometer ( = 1.54059 Å ). Data for all the structures were collected on a Stoe IPDS II two-circle diffractometer with a Genix Microfocus tube with mirror optics using Mo K radiation ( = 0.71073 Å ). The data were scaled using the frame-scaling procedure in the X-AREA program system (Stoe & Cie, 2002). The structures were solved by direct methods using the program SHELXS (Sheldrick, 2008, 2015) and refined against F 2 with full-matrix least-