Solid-State Nonlinear Optical Properties of Mononuclear Copper(II) Complexes with Chiral Tridentate and Tetradentate Schiff Base Ligands

Salen-type metal complexes have been actively studied for their nonlinear optical (NLO) properties, and push-pull compounds with charge asymmetry generated by electron releasing and withdrawing groups have shown promising results. As a continuation of our research in this field and aiming at solid-state features, herein we report on the synthesis of mononuclear copper(II) derivatives bearing either tridentate N2O Schiff bases L(a−c)− and pyridine as the forth ancillary ligand, [Cu(La−c)(py)](ClO4) (1a–c), or unsymmetrically-substituted push-pull tetradentate N2O2 Schiff base ligands, [Cu(5-A-5′-D-saldpen/chxn)] (2a–c), both derived from 5-substituted salicylaldehydes (sal) and the diamines (1R,2R)-1,2-diphenylethanediamine (dpen) and (1S,2S)-1,2-diaminocyclohexane (chxn). All compounds were characterized through elemental analysis, infrared and UV/visible spectroscopies, and mass spectrometry in order to guarantee their purity and assess their charge transfer properties. The structures of 1a–c were determined via single-crystal X-ray diffraction studies. The geometries of cations of 1a–c and of molecules 2a–c were optimized through DFT calculations. The solid-state NLO behavior was measured by the Kurtz–Perry powder technique @1.907 µm. All chiral derivatives possess non-zero quadratic electric susceptibility (χ(2)) and an efficiency of about 0.15–0.45 times that of standard urea.


Introduction
Since the generation of optical harmonics by the interactions of a ruby optical maser [1] with crystalline quartz [2], the research in the nonlinear optics (NLO) field [3] has grown exponentially [4][5][6][7][8][9]. One of the most extensively applied features of NLO materials is the generation of second harmonic (SHG) radiation, that is, the process where the frequency ω of an incident radiation becomes 2ω when passing through the medium [3]. Molecular compounds able to show and maximize such an NLO effect in the solid state have to possess high molecular quadratic hyperpolarizabily (β) values together with a non-centrosymmetric crystal packing [10], able to guarantee the final non-zero quadratic electric susceptibility (χ (2) ) of the bulk material [7,8].

General Information
All chemicals were reagent grade purchased from Sigma Aldrich (purity ≥ 97%) and solvents (methanol, MeOH; diisopropyl ether, iPr2O) were used as received. Elemental analyses were performed at the Mycroanalytical Laboratory of the Università degli Studi di Milano. Electrospray ionization mass spectrometry (ESI-MS) experiments were performed with an LCQ Advantage Thermofluxional Instrument (Thermo Scientific, Waltham, USA) on MeOH solutions of the compounds ( Figure S1 in Supplementary Materials, SM). Infrared (IR) spectra were recorded as KBr disks using a Jasco FT-IR 410 spectrophotometer (Jasco, Tokyo, Japan) with a 2 cm −1 resolution ( Figure S2 in SM). UV/visible spectra were recorded at 298 K with a Jasco V-570 UV/Vis/NIR spectrophotometer (Jasco, Tokyo, Japan) both in reflectance (solid samples) and in transmittance (solution samples) modes in the 200-900 nm range; solution studies were performed on 1b and 1c from 10 −3 to 4 × 10 −5 mol L −1 MeOH solutions and then by addition of increasing pyridine amounts to the 4 × 10 −5 mol L −1 MeOH solutions, and on 2a-c from 10 −3 to 2 × 10 −5 mol L −1 CHCl3 solutions; λ values are accurate to ±1 nm and spectra are reported as wavenumbers, in cm −1 (ε, L mol −1 cm −1 ).

General Information
All chemicals were reagent grade purchased from Sigma Aldrich (purity ≥ 97%) and solvents (methanol, MeOH; diisopropyl ether, iPr 2 O) were used as received. Elemental analyses were performed at the Mycroanalytical Laboratory of the Università degli Studi di Milano. Electrospray ionization mass spectrometry (ESI-MS) experiments were performed with an LCQ Advantage Thermofluxional Instrument (Thermo Scientific, Waltham, USA) on MeOH solutions of the compounds ( Figure S1 in Supplementary Materials, SM). Infrared (IR) spectra were recorded as KBr disks using a Jasco FT-IR 410 spectrophotometer (Jasco, Tokyo, Japan) with a 2 cm −1 resolution ( Figure S2 in SM). UV/visible spectra were recorded at 298 K with a Jasco V-570 UV/Vis/NIR spectrophotometer (Jasco, Tokyo, Japan) both in reflectance (solid samples) and in transmittance (solution samples) modes in the 200-900 nm range; solution studies were performed on 1b and 1c from 10 −3 to 4 × 10 −5 mol L −1 MeOH solutions and then by addition of increasing pyridine amounts to the 4 × 10 −5 mol L −1 MeOH solutions, and on 2a-c from 10 −3 to 2 × 10 −5 mol L −1 CHCl 3 solutions; λ values are accurate to ±1 nm and spectra are reported as wavenumbers, in cm −1 (ε, L mol −1 cm −1 ).

X-ray Data Collection and Structure Determination
Crystals suitable for X-ray diffraction (XRD) experiments were obtained directly by slow evaporation of the MeOH reaction mixture for 1a and 1b, and by diffusion of n-hexane into an acetone solution for 1c. Crystal data and details of data collection are summarized in Table 1. Intensity data for 1a were collected on the XRD1 beamline of the Elettra Synchrotron Light Laboratory (Trieste, Italy), using radiation λ = 0.8 Å, owing to the very small dimensions of the crystals. For 1b and 1c, data collection was performed on a Bruker Apex II diffractometer (Bruker AXS Inc., Madison, WI, USA) using graphite monochromatic Mo-Kα radiation (λ = 0.71073 Å). During data collections, no crystal decay was observed, so no time-decay correction was needed. Data reductions were performed with MOSFLM version 6.11c [66] and SCALA [67] (for 1a) and with SAINT and SADABS [68] (for 1b and 1c). All the structures were solved by direct methods and refined with SHELXL-2016/6 [69] implemented in WinGX-Version 2014.1 system [70]. The program Mercury was used for graphics [71]. Anisotropic thermal parameters were used for all non-hydrogen atoms. The isotropic thermal parameters of H atoms were fixed at 1.2 (1.5 for methylene groups) times those of the atom to which they were attached. All H atoms were placed in calculated positions and refined by a riding model. CCDC 1954460-1954462 contains the supplementary crystallographic data for 1a-c, respectively. These data can be obtained free of charge via http://www.ccdc.cam.ac.uk/conts/retrieving.html, or from the Cambridge Crystallographic Data Centre, 12 Union Road, Cambridge CB2 1EZ, UK; fax: (+44) 1223-336-033, or e-mail: deposit@ccdc.cam.ac.uk.

Computational Details
Geometry optimizations of the cations [Cu(L a-c )(py)] + of 1a-c and of 2a-c were performed at UM06/6-311++G(d,p) level of theory. For 1a-c, the respective single crystal X-ray structures were used as a starting point, while for 2a-c, the initial guess for optimization was build up from the optimized geometries of 1b and 1c with the proper substitutions. SHG hyperpolarizabilities, that is, the β(-2ω;ω,ω) tensors, were computed in CHCl 3 within the coupled perturbed Kohn-Sham (CPKS) approach, at the same frequency as used as in the Kurtz-Perry experiments. The CAM-B3LYP functional [72], which has been recommended for hyperpolarizability calculations of mid-size organic chromophores [73], was adopted. The solvent was described as a continuum dielectric according to the polarizable continuum model in its integral equation formalism variant (IEFPCM algorithm) [74]. A pruned (99,590) grid was selected for computation as well as the use of two-electron integrals and their derivatives. As already pointed out in our previous studies on the NLO properties of [Cu(5-A-5 -D-salen/saltn)] [38] and [Ni(5-A-5 -D-saltn)] [39] complexes, two main and almost orthogonal components of the dipole moment and hyperpolarizability vectors can be singled out in these compounds, one along the largest molecular extension and the other across the ON(py)/O 2 and N 2 atoms of the donor set of compounds 1 and 2.
In order to meaningfully compare these components among the examined compounds, a common reference frame was adopted, whereby the origin coincides with the copper ion, the x axis passes through C4 (the carbon atom of the six-membered chelate ring on the acceptor side) and points from the acceptor towards the donor group, the y axis lies in the N 3 O/N 2 O 2 plane and points from the ON(py)/O 2 towards the N 2 atoms of the donor set, and the z axis completes the right-handed reference system. This frame strictly recalls that adopted for the [Cu(5-A-5 -D-salen/saltn)] [38] and [Ni(5-A-5 -D-saltn)] [39] complexes.

Kurtz-Perry Powder Measurements
The measurements of SHG intensity were carried out by the Kurtz-Perry powder technique [47] using a nanosecond Nd:YAG pulsed (10 Hz) laser. The fundamental beam (1.064 µm) was focused in a hydrogen cell (1 m long, 50 atm) and the outcoming Stokes-shifted radiation generation at 1.907 µm was used as the fundamental beam for SHG. The SHG signal coming from capillary tubes containing grinded microcrystalline powders (50-80 µm) of the samples was detected by a photomultiplier.

Synthesis and IR/MS Characterization
Even if it is reported that tridentate HL ligands, especially with chxn, can be obtained without the support of a metal ion [56,75], because of our interest in copper(II) derivatives, we decided to directly employ the template method that was revealed to be efficient with aliphatic diamines en and tn [38,[76][77][78], as well as with the aliphatic diamines chxn and dpen. The series of mononuclear derivatives 1a-c with tridentate Schiff base ligands was then synthesized by reaction of NO 2 -salH:Cu(ClO 4 ) 2 :diamine in a 1:1:1 ratio in the presence of excess py acting as the forth ligand, so as to avoid the double condensation on both NH 2 groups of chxn and dpen. The cheapest (1S,2R)-meso-dpen was initially employed with the aim of studying the best reaction and isolation conditions, and then extending the synthetic protocol to the most expensive optically pure (1R,2R)-(+)-dpen and (1S,2S)-(+)-chxn. It is curious to note how we could instead observe a moderate different reactivity of (1S,2R)-meso-dpen compared with (1R,2R)-(+)-dpen. In fact, the species [Cu(L b )(H 2 O)](ClO 4 ), 1b', could be isolated with (1R,2R)-(+)-dpen, where the crystal packing of cations and anions in the solid state seems to confer a lower solubility to the aquo complex with its preferential precipitation. A similar complex with NO 2 -salH, en and water as the forth coordinated ligand was also isolated by us in our previous work [77]. Only in the presence of a larger excess of pyridine, 1b' was converted into 1b with coordinated py. Intermediates 1b and 1c were then suitably converted to 2a, 2b, and 2c by condensation with the second carbonyl derivative, salH or 5-OMe-salH, in refluxing MeOH; the latter aldehyde, owing to the donor group, is characterized by a higher reactivity with a consequent lower reaction time (30 min) compared with salH (1 h).
All compounds were characterized by infrared spectroscopy, where the N-H stretching bands in the 3350-3250 cm −1 spectral range are diagnostic for the formation of intermediates 1a-c and 1b'.
The disappearance of such bands in the condensation reaction to obtain 2a-c could be used as a probe of the complete conversion of the intermediates to the final tetradentate derivatives. The C=N stretching bands are also subjected to peculiar changes on going from 1645-1647 cm −1 in 1b and 1c to 1632-1635 in 2a-c. In all derivatives, the stretching band of the nitro group at about 1315 cm −1 is detectable. The coordinated water molecule in 1b' can be recognized through the narrow O-H stretching band at 3296 cm −1 in between the symmetric and asymmetric NH 2 bands. Furthermore, the different donating power to copper(II) of the oxygen atom in the water molecule compared with the nitrogen atom of py leads to modifications in the stretching bands of the nitro group, which shifts to 1333 cm −1 , and of the C=N bond, which moves to 1626 cm −1 .
The chemical purity of bulk materials was proved by elemental analyses and all compounds were also subjected to MS-ESI investigations. In particular, mass peaks at m/z corresponding to the [Cu(L a-c )] + and solvated [Cu(L a-c )(MeOH)] + fragments could be detected for 1a, 1b, and 1c, while the most intense peak of 1b' was the [Cu(L b )(H 2 O)] + ion, suggesting a rather strong Cu-O(water) bond. In the case of 2a, 2b, and 2c, mass peaks at m/z corresponding to [M + 1] + and [M + Na] + species, together with the dimeric [2M + 1] + and [2M + Na] + ions, were observed. It is not uncommon to observe such dimeric signals in tetradentate Schiff base metal complexes owing to the establishment of short intermolecular M···O(phenol) interactions [33], especially with copper(II) and its tendency to weakly interact with a fifth donor atom [38,77].

X-ray Structures of 1a·0.5H 2 O, 1b, and 1c
Single crystals suitable for XRD experiments were obtained for 1a·0.5H 2 O, 1b, and 1c bearing the tridentate Schiff base ligands L − . The molecular structures of 1b and 1c are reported in Figure 1, while a fragment of the crystal packing of 1a is depicted in Figure S3 in SM with an atom numbering scheme. Selected bond distances, angles, and other structural parameters of all compounds helpful for the discussion are reported in Table 2. Compound 1a crystallizes in the centrosymmetric P2 1 /n space group, because (1S,2R)-meso-dpen can condense the amino group with NO 2 -salH on either the R or S chiral side, and this gives rise to two enantiomeric complexes, both present in the unit cell and related by the inversion center. When moving to (1R,2R)-(+)-dpen and (1S,2S)-(+)-chxn, only one pure enantiomer is present, so that 1b and 1c crystallize in the orthorhombic P2 1 2 1 2 1 and the chiral tetragonal P4 3 space groups, respectively. these components among the examined compounds, a common reference frame was adopted, whereby the origin coincides with the copper ion, the x axis passes through C4 (the carbon atom of the six-membered chelate ring on the acceptor side) and points from the acceptor towards the donor group, the y axis lies in the N3O/N2O2 plane and points from the ON(py)/O2 towards the N2 atoms of the donor set, and the z axis completes the right-handed reference system. This frame strictly recalls that adopted for the [Cu(5-A-5′-D-salen/saltn)] [38] and [Ni(5-A-5′-D-saltn)] [39] complexes.

Kurtz-Perry Powder Measurements
The measurements of SHG intensity were carried out by the Kurtz-Perry powder technique [47] using a nanosecond Nd:YAG pulsed (10 Hz) laser. The fundamental beam (1.064 µm) was focused in a hydrogen cell (1 m long, 50 atm) and the outcoming Stokes-shifted radiation generation at 1.907 µm was used as the fundamental beam for SHG. The SHG signal coming from capillary tubes containing grinded microcrystalline powders (50-80 µm) of the samples was detected by a photomultiplier.  Table 2. Selected interatomic distances (Å) and angles ( • ), intermolecular contacts (Å), and selected parameters in 1a·0.5H 2 O, 1b, and 1c from single crystal XRD (first line) and UM06/6-311++G(d,p) calculations on cations (second line, in italics; see Figure 1 and Figure S3 in SM for atom labelling).

1a·0.5H 2 O 1b 1c
Cu1-N1/ All complexes show the expected square-planar geometry around the copper(II) atoms, where the basal plane is formed by the N 2 O set of donors from the L a− , L b− , and L c− ligands for 1a, 1b, and 1c, respectively, and the fourth position is occupied by the nitrogen atom of the pyridine ligand. All complexes are characterized by very slight tetrahedral distortion of the donor set, as indicated by the values of the O1-N2-N1-N3 torsion angles, all less than 1.0 • , except for the second molecule of the asymmetric unit of 1b, where the O4-N6-N5-N4 torsion angle measures 7.9 • . The displacement of the copper(II) ions from the N 3 O l.s. planes is very low in all cases (<0.07 Å). Coordination distances and angles are comparable to those of other copper compounds with N 2 O tridentate Schiff base ligands and pyridine as the forth ancillary ligand [77,79].
The asymmetric units of 1a and 1c comprise one independent [Cu(L a/c )(py)] + cation and ClO 4 − anion, together with a co-crystallized half molecule of water in the case of 1a. Analysis of the crystal structure of 1a reveals that the perchlorate anion is disordered over three positions: a majority one (labelled with R, Figure   Short contacts with the perchlorate anions on both sides of the coordination plane can also be detected in 1c (Cu1···O1p = 2.654(5) Å, Cu1···O2p = 2.857(5) Å), even if slightly longer than in 1b, thus producing a zig-zag 1D chain along the crystallographic c axis with an angle between three consecutive copper(II) ions of 159.1 • (Figure 3a). Along this direction, cations are almost parallel (the angle between the coordination l.s. planes of two consecutive cations measures 16.94 • ) and oriented in the opposite direction (see Figure 3b), suggesting a detrimental effect on the NLO response in the crystal phase. The amino group in 1c is involved in a short hydrogen bond with the nitro group of an adjacent molecule (N1-H1A···O3 , r H···O' = 2.28 Å, NH···O' angle = 157.9 • ), which is further engaged in short contacts with the pyridine ring (O3 ···C13 = 3.203(5) Å, O3 ···C14 = 3.155(4) Å), giving rise to the inter-locked mode between parallel chains. This arrangement forms 2D layers parallel to the ab plane, where the molecules of each layer are alternately oriented in the same direction.

Absorption Spectroscopy
The electronic spectra in the solid state of 1b, 1b', 1c, 2a, 2b, and 2c are reported in Figure 4a in the 12,000-50,000 cm −1 (900-200 nm) region. In all cases, it is possible to recognise a well-isolated d-d transition in the 17,300-17,800 cm −1 (562-578 nm) range, with the exception of 1b', in which the band appears to be much broader and weaker. This clearly reflects the different ligand field given by the substitution of a water molecule by a pyridine ligand. Charge transfer (CT) and π → π* transitions start to appear above 20,000 cm −1 . The introduction of the OMe group on going from 2a to 2b leads to the absorption at lower wavenumbers, as expected for the favoured ligand-to-metal charge transfer (LMCT) transition by the presence of the donor group [38,39]. When moving to 2c, even if the pushpull substitution is equal to 2b, the LMCT band moves further to lower energies, which might be caused by the presence of close molecules with short intermolecular interactions [53], favoured by the smaller steric hindrance of the cyclohexyl ring compared with the two phenyl rings on the diamine bridge [60]. The same behaviour is present when comparing 1b and 1c, where the donor sal moiety is absent, which can thus confirm the occurrence of favoured electronic transitions in chxn derivatives. Above 30,000 cm −1 , all spectra appear very similar owing to further π → π* transitions.
Solution studies were performed on non-ionic derivatives 2 in CHCl3 solutions from 10 −3 to 2 × 10 −5 mol L −1 (see Figure 4b). The d-d transition in 2a, 2b, and 2c appears as a relatively weak band at about 17,730-17,860 cm −1 (560-564 nm) with ε ~ 500 L mol −1 cm −1 (see Figure S4 in SM for a zoom of the 12,500-20,000 cm −1 region), as expected for copper(II) complexes with a rigid square planar geometry given by both chiral [51,52,54] and achiral [38] N2O2 salen-type ligands, with no apparent effect given by the diamine bridge or substituents. Position, shape, and intensity of the d-d band also follow the Lambert-Beer law, which suggests the absence of stacked molecules in solution at a concentration lower than 10 −3 mol L −1 . In the 20,000-32,000 cm −1 range, there is an intense absorption with ε coefficients as high as 27,000 L mol −1 cm −1 , which can be attributed to a convolution of CT transitions [52,80]. In particular, the position of the LMCT band depends on the substituents present on the molecular skeleton, as confirmed by its visibility in the lowest energy side as a shoulder at

Absorption Spectroscopy
The electronic spectra in the solid state of 1b, 1b', 1c, 2a, 2b, and 2c are reported in Figure 4a in the 12,000-50,000 cm −1 (900-200 nm) region. In all cases, it is possible to recognise a well-isolated d-d transition in the 17,300-17,800 cm −1 (562-578 nm) range, with the exception of 1b', in which the band appears to be much broader and weaker. This clearly reflects the different ligand field given by the substitution of a water molecule by a pyridine ligand. Charge transfer (CT) and π → π* transitions start to appear above 20,000 cm −1 . The introduction of the OMe group on going from 2a to 2b leads to the absorption at lower wavenumbers, as expected for the favoured ligand-to-metal charge transfer (LMCT) transition by the presence of the donor group [38,39]. When moving to 2c, even if the push-pull substitution is equal to 2b, the LMCT band moves further to lower energies, which might be caused by the presence of close molecules with short intermolecular interactions [53], favoured by the smaller steric hindrance of the cyclohexyl ring compared with the two phenyl rings on the diamine bridge [60]. The same behaviour is present when comparing 1b and 1c, where the donor sal moiety is absent, which can thus confirm the occurrence of favoured electronic transitions in chxn derivatives. Above 30,000 cm −1 , all spectra appear very similar owing to further π → π* transitions.   , 1b', 1c, 2a, 2b, and 2c in the solid state and (b) 2a, 2b, and 2c in 2 × 10 −5 CHCl3 solutions (zoom of the d-d transition region from 12,500 to 20,000 cm −1 can be found in Figure S4 in SM).
Ionic complexes 1 are not soluble in non-coordinating solvents, while they are, for example, in methanol. Nevertheless, they probably undergo substitution of the pyridine ligand once in MeOH [78,81], so that the spectrum corresponds to the solvated species [Cu(L)(MeOH)] + . To overcome this drawback, studies in solution were performed in MeOH (from 10 −3 to 4 × 10 −5 mol L −1 ), adding increasing amounts of pyridine to the most diluted solution. The obtained spectra of 1b and 1c in MeOH (reported in Figure S5 in SM) are not too dissimilar from those of derivatives 2, except for shifted (16,050 cm −1 , 623 nm) and weaker (135 L mol −1 cm −1 ) d-d transitions. Upon addition of pyridine, this transition moves toward higher energies, 17,670 cm −1 (566 nm), and higher intensity, ε = 300 and 235 L mol −1 cm −1 , for 1b and 1c, respectively. This seems to confirm the [Cu(L)(MeOH)] + → [Cu(L)(py)] + back conversion in solution, owing to the stronger ligand field given by pyridine with respect to MeOH. Nevertheless, the LMCT + ILCT transitions are not affected by the coordination of either py or MeOH.

DFT Structural, Electronic, and NLO Properties
In view of theoretically evaluating the molecular nonlinear optical properties of the investigated compounds, DFT geometry optimizations were performed on the cations of 1a-c. The experimental bond lengths were satisfactorily reproduced (see Table 2), with the exceptions of only the Cu-N1 and Cu-N3 distances, which are systematically longer in the optimized structures. This reflects the greater tendency of such bonds to be deformed by packing forces with respect to the other coordinating bonds. For the same reason, the dihedral angles between the N3O and pyridine l.s. planes, which are virtually the same in the optimized structures, are very different from the X-ray values, displaying large variations not only from one structure to another, but also in the same structure, as observed in 1b. The comparison of the coordinating bond lengths in 1a-c does not reveal significant differences, as expected from the analogous electronic environment around the copper ion in the three structures.
DFT geometry optimizations were also performed on 2a-c, for which X-ray data were not available, owing to the well-assessed reliability of DFT in reproducing the molecular features of this class of complexes [38,39]. As we previously observed and theoretically reproduced in the [Cu(5-A-5′-D-salen)] derivatives [38], the NO2 group is more efficient in modulating the coordination geometry with respect to the OMe one. Such influence is mainly manifested in the Cu-O bonds, with elongation of Cu-O1 (A side) and a concomitant decrease of Cu-O2 (D side) bond lengths. The Cu-N bonds are less affected by the presence of the NO2 group, with a slightly more elongated Cu-N2 (A side) with respect to Cu-N1 (D side) bond lengths (Table 3, Figures 5, and Figure S6 in SM). Comparing 2b and 2c, having the same (NO2, OMe) push-pull pattern, but a different diamine (dpen  (a) 1b, 1b', 1c, 2a, 2b, and 2c in the solid state and (b) 2a, 2b, and 2c in 2 × 10 −5 CHCl 3 solutions (zoom of the d-d transition region from 12,500 to 20,000 cm −1 can be found in Figure S4 in SM).
Solution studies were performed on non-ionic derivatives 2 in CHCl 3 solutions from 10 −3 to 2 × 10 −5 mol L −1 (see Figure 4b). The d-d transition in 2a, 2b, and 2c appears as a relatively weak band at about 17,730-17,860 cm −1 (560-564 nm) with ε~500 L mol −1 cm −1 (see Figure S4 in SM for a zoom of the 12,500-20,000 cm −1 region), as expected for copper(II) complexes with a rigid square planar geometry given by both chiral [51,52,54] and achiral [38] N 2 O 2 salen-type ligands, with no apparent effect given by the diamine bridge or substituents. Position, shape, and intensity of the d-d band also follow the Lambert-Beer law, which suggests the absence of stacked molecules in solution at a concentration lower than 10 −3 mol L −1 . In the 20,000-32,000 cm −1 range, there is an intense absorption with ε coefficients as high as 27,000 L mol −1 cm −1 , which can be attributed to a convolution of CT transitions [52,80]. In particular, the position of the LMCT band depends on the substituents present on the molecular skeleton, as confirmed by its visibility in the lowest energy side as a shoulder at about 22,500 cm −1 (ε~4000 L mol −1 cm −1 ) in 2b and 2c bearing the OMe group, while it moves below the strongest intra-ligand charge transfer (ILCT) π → π* transitions in 2a, where there is no donor substituent. In fact, we previously confirmed the nature of the LMCT band on derivatives with non-chiral en and tn bridges, where the HOMO orbital depicted by theoretical calculations is mainly localized on the D-sal moiety [38,39]. In the UV region above 33,000 cm −1 , a strong absorption due to further π → π* transitions appears [80], with noticeable differences among the three complexes. In particular, compounds with dpen possess a stronger absorbance at lower energies, probably given by transitions where the aromatic rings of the diamine bridge are involved. Comparing 2a and 2b, the latter shows a more pronounced maximum at 39,370 cm −1 (254 nm) and ε = 40,600 L mol −1 cm −1 , promoted by the presence of the OMe group.
Ionic complexes 1 are not soluble in non-coordinating solvents, while they are, for example, in methanol. Nevertheless, they probably undergo substitution of the pyridine ligand once in MeOH [78,81], so that the spectrum corresponds to the solvated species [Cu(L)(MeOH)] + . To overcome this drawback, studies in solution were performed in MeOH (from 10 −3 to 4 × 10 −5 mol L −1 ), adding increasing amounts of pyridine to the most diluted solution. The obtained spectra of 1b and 1c in MeOH (reported in Figure S5 in SM) are not too dissimilar from those of derivatives 2, except for shifted (16,050 cm −1 , 623 nm) and weaker (135 L mol −1 cm −1 ) d-d transitions. Upon addition of pyridine, this transition moves toward higher energies, 17,670 cm −1 (566 nm), and higher intensity, ε= 300 and 235 L mol −1 cm −1 , for 1b and 1c, respectively. This seems to confirm the [Cu(L)(MeOH)] + → [Cu(L)(py)] + back conversion in solution, owing to the stronger ligand field given by pyridine with respect to MeOH. Nevertheless, the LMCT + ILCT transitions are not affected by the coordination of either py or MeOH.

DFT Structural, Electronic, and NLO Properties
In view of theoretically evaluating the molecular nonlinear optical properties of the investigated compounds, DFT geometry optimizations were performed on the cations of 1a-c. The experimental bond lengths were satisfactorily reproduced (see Table 2), with the exceptions of only the Cu-N1 and Cu-N3 distances, which are systematically longer in the optimized structures. This reflects the greater tendency of such bonds to be deformed by packing forces with respect to the other coordinating bonds. For the same reason, the dihedral angles between the N 3 O and pyridine l.s. planes, which are virtually the same in the optimized structures, are very different from the X-ray values, displaying large variations not only from one structure to another, but also in the same structure, as observed in 1b. The comparison of the coordinating bond lengths in 1a-c does not reveal significant differences, as expected from the analogous electronic environment around the copper ion in the three structures.
DFT geometry optimizations were also performed on 2a-c, for which X-ray data were not available, owing to the well-assessed reliability of DFT in reproducing the molecular features of this class of complexes [38,39]. As we previously observed and theoretically reproduced in the [Cu(5-A-5 -D-salen)] derivatives [38], the NO 2 group is more efficient in modulating the coordination geometry with respect to the OMe one. Such influence is mainly manifested in the Cu-O bonds, with elongation of Cu-O1 (A side) and a concomitant decrease of Cu-O2 (D side) bond lengths. The Cu-N bonds are less affected by the presence of the NO 2 group, with a slightly more elongated Cu-N2 (A side) with respect to Cu-N1 (D side) bond lengths (Table 3, Figure 5, and Figure S6 in SM). Comparing 2b and 2c, having the same (NO 2 , OMe) push-pull pattern, but a different diamine (dpen vs. chxn, respectively) bridge, the most remarkable difference lies in the dihedral angle between the l.s. planes through the two sal moieties, which is higher in 2c (12.33 • ) with respect to 2a (10.20 • ) and 2b (10.28 • ). The reason for this difference can be probably found in the greater steric demand of the cyclohexyl ring with respect to the dpen bridge. Table 3. Selected interatomic distances (Å) and angles ( • ) in 2a, 2b, and 2c from UM06/6-311++G(d,p) calculations (see Figure 5 and Figure S6 in SM for main atom labelling). vs. chxn, respectively) bridge, the most remarkable difference lies in the dihedral angle between the l.s. planes through the two sal moieties, which is higher in 2c (12.33°) with respect to 2a (10.20°) and 2b (10.28°). The reason for this difference can be probably found in the greater steric demand of the cyclohexyl ring with respect to the dpen bridge. Table 3. Selected interatomic distances (Å) and angles (°) in 2a, 2b, and 2c from UM06/6-311++G(d,p) calculations (see Figure 5 and Figure S6 in SM for main atom labelling).  As expected, the cations of compounds 1a, 1b, and 1c are very similar from an electronic point of view (see Table 4). Using the copper ion as the origin of the Cartesian reference (see computational details), their dipole moments, about 13.5 D, and hyperpolarizability vectors βtot, about 23-24 x 10 −30 cm 5 esu −1 , reflect an almost indistinguishable behavior of the compounds at the molecular level, in both the ground and the excited state . Compounds 2a, 2b, and 2c,   As expected, the cations of compounds 1a, 1b, and 1c are very similar from an electronic point of view (see Table 4). Using the copper ion as the origin of the Cartesian reference (see computational details), their dipole moments, about 13.5 D, and hyperpolarizability vectors β tot , about 23-24 × 10 −30 cm 5 esu −1 , reflect an almost indistinguishable behavior of the compounds at the molecular level, in both the ground and the excited state . Compounds 2a, 2b, and 2c, on the other hand, while having similar dipole moments (13.02, 12.26, and 12.34 D), show increasing values of β tot from 39.7 to 49.2 and 49.6 × 10 −30 cm 5 esu −1 , going from 2a to 2b and 2c, as a consequence of the presence of the donor group OMe in the latter compounds. These values are quite similar to those previously obtained for the [Cu(5-A-5 -D-salen)] analogues with (NO 2 , H) and (NO 2 , OMe) substitutions [38,39], indicating, at the molecular level, the almost irrelevant influence of the ethylene substituents (en vs. dpen vs. chxn) on the electronic properties. Table 4. Computed 1 dipole moments (µ g , µ x , µ y , and µ z , D), SHG (λ = 1.907 µm) hyperpolarizabilities (β tot , β vec , β x , β y and β z , 10 −30 cm 5 esu −1 ) in CHCl 3 , and angle (θ, • ) between µ g and β tot vectors for cations of 1a-c and for 2a-c. Calculations were performed at the (CP)CAM-B3LYP/6-311++G(d,p)/IEFPCM(CHCl 3 ) level of theory on the UM06/6-311++G(d,p) optimized geometries; see computational details section for the adopted reference frame to report the µ i and β i components; hyperpolarizabilities are defined as β tot = (β x 2 + β y 2 + β z 2 ) 1/2 and β vec = Σ i (µ i β i )/|µ|, where β i = (1/3) Σ j (β ijj + β jij + β jji ) [82]. In particular, β tot = (5/3) β and β vec = (5/3) β (z), where β and β (z) are the quantities printed in the Gaussian16 output, according to the Taylor convention [83].
Owing to the absence of the crystal structure of 2a-c, only hypothesis to compare the molecular properties as described by theoretical calculations with the solid-state NLO data can be formulated in the case of derivatives with unsymmetrically-substituted tetradentate ligands. A possible explanation for the observed diminished χ (2) is the unfavorable alignment of the chromophores, which probably still tend to couple in a head-to-tail fashion, in agreement with previous reports [38,39]. Anyway, the chirality conferred to the final complexes by the diamines prevents perfect alignment, resulting in a small, but non-zero solid-state NLO response. In particular, theoretical calculations gave reasonably similar β tot values (49.22 and 49.60 × 10 −30 cm 5 esu −1 ) for 2b and 2c, respectively, bearing the same A-D substitution, while their SHG efficiencies are 0.19 and 0.30 times that of standard urea, which suggests a better performance of the chxn diamine bridge with respect to dpen. This is in opposition to what was observed in the case of [Ni(4,4 -diNEt 2 -sal-(1R,2R)-chxn)] [48] and [Ni(4,4 -diNEt 2 -sal-(1R,2R)-dpen)] [49], where a higher SHG efficiency was achieved by the latter compound (9 vs. 0.3 times that of standard urea for 50-80 µm grain caliber) and ascribed to its higher degree of chirality [85] conferred by the spatial disposition of the two phenyl rings [49]. The opposite chirality of chxn employed in 2c with respect to [Ni(4,4 -diNEt 2 -sal-(1R,2R)-chxn)] does not have an effect because the solid-state NLO response depends on the molecular hyperpolarizability, the orientation of the molecule with respect to the crystalline reference frame, and the nature of the crystal point group [10]-properties that are coincident with those of the enantiomer of 2c, that is, [Cu(5-NO 2 -5 -OMe-sal-(1R,2R)-chxn)].
The same trend in the SHG efficiency can be drawn out for 1b (dpen) and 1c (chxn), where their experimental values are 0.13 and 0.44 times that of standard urea against similar computed β tot values of 22.98 and 24.08 × 10 −30 cm 5 esu −1 , respectively. These values can be interpreted through both the analysis of degree of chirality and the relationship between microscopic β xyz and macroscopic β XYZ , with both pieces of information being affordable for these compounds thanks to their known crystal structure. Quantification of the degree of chirality with the continuous symmetry measures (CSM) formalism proposed by Avnir et al. [86], on a scale from 0 (achiral) to 100, gave a greater value in 1c (7.7948, only the cation: 1.9812) than in 1b (chain A: 4.3929, cation: 1.5796; chain B: 6.4284, cation: 0.5959), considering either the cations [Cu(L)(py)] + or the [Cu(L)(py)](ClO 4 ) assemblies. This would suggest a kind of correlation between solid-state NLO response and the degree of chirality.
On the other side, the relationships between microscopic and macroscopic optical nonlinearities can be derived considering that crystals of 1b belong to the P2 1 2 1 2 1 orthorhombic space group with two cations in the asymmetric unit. For the 222 point group, there is only one phase matchable coefficient, β XYZ = sin2Φ cosθ (β xyy − β xxx sin 2 θ) − β yxx cos2Φ sin2θ [10], where Φ is the projection of the CT axis on the ab plane and a, and θ is the angle between the CT axis and c [87]. The optimal value of the angular factor is 0.192, reached for Φ = 45 • and θ = 54.7 • . On the basis of CP-DFT calculations on the 1b cation, the CT axis lies approximately along the C4-Cu direction (i.e., the x axis, see Figure S6b in SM and Table 4), so that Φ = 89 • and θ = 75 • for molecule 1 of chain A in the asymmetric unit, while Φ = 84 • and θ = 32 • for molecule 2 of chain B. As only β xxx is significantly different from zero for 1b cation, both values of Φ, close to 90 • , result in very small macroscopic susceptibility β XYZ . In the case of 1c (tetragonal space group P4 3 , point group 4), the only non-zero macroscopic nonlinear susceptibilities are β ZZZ = β xxx cos 3 θ and β ZYY = (β xyy +β xxx sin 2 θ) cosθ/2, θ being the angle between the CT axis and the four-fold axis Z. The almost orthogonality between these two axis (θ = 82 • ), far from the optimal value θ = 54.7 • , explains the low NLO susceptibility of 1c.

Conclusions
The synthesis, structural and spectroscopic characterizations, and SHG efficiencies using the Kurtz-Perry technique of mononuclear copper(II) complexes with either tridentate or tetradentate Schiff base ligands are reported here with the aim of investigating their NLO performances to the solid state. The synthetic copper(II)-templated approach for tridentate Schiff base ligands L − derived from salicylaldehydes and aliphatic diamines [77] is efficiently exploited here with the use of the chiral diamines chxn and dpen, allowing also to isolate the intermediate 1b' with coordinated water molecule, further replaced by pyridine. Derivatives [Cu(L a-c )(py)](ClO 4 ) (1a-c) were then converted into compounds [Cu(5-A-5 -D-saldpen/chxn)] (2a-c) with chiral tetradentate Schiff base ligands, obtaining push-pull complexes in which the metal ion promotes an intense LMCT transition, as observed in the UV/vis absorption spectroscopy.
Despite the low solid-state NLO efficiencies of about 0.15-0.45 times that of standard urea for all derivatives, these values are in line with other similar compounds, with the exclusion of only a few exceptional cases [48,49]. X-ray structure determination of 1b and 1c, together with theoretical calculations on their NLO molecular response, allowed to ascribe their modest NLO susceptibilities to the non-optimal alignment of their CT axes with respect to the crystalline reference frame.