Quenched Lewis Acidity: Studies on the Medium Dependent Fluorescence of Zinc(II) Complexes

Abstract Three new zinc(II) coordination units [Zn(1–3)] based on planar‐directing tetradentate Schiff base‐like ligands H2(1–3) were synthesized. Their solid‐state structures were investigated by single crystal X‐ray diffraction, showing the tendency to overcome the square‐planar coordination sphere by axial ligation. Affinity in solution towards axial ligation has been tested by extended spectroscopic studies, both in the absorption and emission mode. The electronic spectrum of the pyridine complex [Zn(1)(py)] has been characterized by MC‐PDFT to validate the results of extended TD‐DFT studies. Green emission of non‐emissive solutions of [Zn(1–3)] in chloroform could be switched on in the presence of potent Lewis‐bases. While interpretation in terms of an equilibrium of stacked/non‐fluorescent and destacked/fluorescent species is in line with precedents from literature, the sensitivity of [Zn(1–3)] was greatly reduced. Results of a computation‐based structure search allow to trace the hidden Lewis acidity of [Zn(1–3)] to a new stacking motif, resulting in a strongly enhanced stability of the dimers.


Introduction
Switchable materials are of great interest for applications as molecular sensor materials. [1] These materials need to combine a high sensitivity toward external physical or chemical stimuli with an easily detectable change of discrete properties such as photoluminescence. For this purpose, metal complexes with photoluminescent ligands are an interesting class of materials, as they can combine an easily tunable photoluminescence based on the ligand with a switching behavior based on the metal center. Various switching processes are possible that depend on the metal center and its electronic configuration. One possibility is the switching based on externally induced spin transitions. Two well-known phenomena are spin crossover, which is mostly investigated for iron(II) complexes, [2] and coordination-induced spin state switching that can be observed with nickel(II) complexes. [3] In recent years, many workgroups focused on combining one of these two switching mechanisms with a photoluminescent ligand to obtain emissive sensor materials. [4] However, electronic coupling of metal and ligand often results in predominant non-radiative excited-state decay into low-lying d-d states, independent of the spin state. [5] Due to these drawbacks, recently much emphasis was put on metal centers with closed d 0 or d 10 shells such as Zr(IV), Cu(I), or Zn(II). As the d orbitals of the metal center are not directly electronically involved in the photophysical processes, the emission is not quenched and a wide range of photoluminescent complexes has been reported. [5b,6] Especially, zinc(II) complexes are interesting for the application as molecular sensor materials, as their emission is often medium dependent. In the last decade many zinc(II) complexes have been reported that show aggregation-induced emission enhancement (AIEE). [7] Interestingly, the opposite case -an emission quenching due to stacking -has been observed as well with zinc(II) complexes. Many studies focus on zinc(II) complexes based on a Schiff base tetradentate ligand equipped with nitrile substituents. [8] Coordination of additional ligands resulted in a strong increase of the emission intensity. Experimental and computational studies suggested that the zinc(II) complexes stack into dimers or oligomers in solvents that do not support axial coordination. [8d,f] The high emission intensity increase in coordinating solvents was assigned to de-stacking of the zinc(II) complexes. [8b,d] This effect was used previously for bioimaging and biosensing in living cells. [8c,g,9] Furthermore, derivatives of this zinc(II) complex type were investigated for their mechanochromic luminescence behavior, and as emitters for optical temperature sensing via thermally activated delayed fluorescence (TADF). [10] Herein, we present a concerted experiment-theory approach to investigate the influence of the molecular structure on the fluorescence of zinc(II) complexes. A family of neutral complexes of Schiff base-like ligands with appended nitrile groups [Zn(L)X] (for the nature of L, see Scheme 1; X = H 2 O, EtOH, THF, py) was synthesized. Please note that the simplified notation [Zn(1-3)] is used whenever the nature of the axial ligand cannot be specified. In agreement with the established stacking/destacking hypothesis put forward by Di Bella and others for analogue Schiff base derived complexes, [8d,f] turn-on emission behavior is observed in spectroscopic titrations with Lewis-bases, but requires much higher base loads. The seemingly weakened Lewis-base affinity of [Zn(L)X] was found to be due to quenched Lewis-acidity based on the formation of highly stable dimers in solvents that do not support axial coordination. Results from (time-dependent) density functional theory are benchmarked through complete active space selfconsistent field methods. CASSCF, followed by the multiconfigurational pair-density functional theory correction (MC-PDFT) accurately captures correlation effects for the relevant states involved in the diagnostic ILCT process.

Complex syntheses and characterization
The zinc(II) coordination units [Zn(1)] and [Zn(2)] were synthesized in two steps as shown in Scheme 1; [Zn(3)] was synthesized in a one-pot reaction as the synthesis of the free ligand H 2 (3) failed. In the case of H 2 (1) and H 2 (2) the free Schiff base-like ligands were obtained directly through reaction of diaminomaleonitrile with the respective keto-enol ether in the presence of p-toluenesulfonic acid (p-TSA); H 2 (2) has been reported previously. [11] H 2 (1) was received from MeOH as an orange solid in 44 % yield. The identity and purity of the ligands was confirmed by 1 H NMR spectroscopy, mass spectrometry, and elemental analysis. It is noted that resonances of the NH hydrogen atom could not be observed spectroscopically, neither in IR nor 1 H NMR. This finding indicates fast exchange in solution and strong involvement of the NH hydrogen atoms in a hydrogen bond network in the solid.
From the ligands the respective zinc(II) complexes were obtained through conversion with stoichiometric amounts of Zn(OAc) 2 1 H NMR spectroscopy and mass spectrometry. Elemental analysis indicated the presence of stoichiometrically defined amounts of solvent molecules in the samples as indicated by the above empirical formulae. X-ray diffraction of single-crystalline samples indeed identified the complexes to be five (N 2 O 3 ) and six-coordinate (N 2 O 4 ) in the solid, besides additional solvent molecules in the lattice.

X-ray diffraction analysis
Molecular structures and packing pattern were addressed by single-crystal X-ray diffraction. The crystallographic data of all crystal structures are summarized in Table S1/S2 in the Supporting Information. Orange block-like crystals of [Zn(1)(MeOH)]·MeOH were obtained directly from the mother liquor. The material crystallises in the orthorhombic space group Pbca. The asymmetric unit consists of the five-coordinate zinc(II) complex and one solvent MeOH molecule as shown in Figure 1A (see Figure S1 for a fully labelled representation of the asymmetric unit). The zinc(II) center is enclosed in a closeto-ideal tetragonal pyramidal N 2 O 3 coordination sphere (S p (SqPy) = 1.18), [12] wherein an axial methanol ligand adds to the N 2 O 2 chelate of the Schiff base-like ligand. Selected bond lengths are given in Table 1 Figure 1B

Chemistry-A European Journal
Full Paper doi.org/10.1002/chem.202102086 for a fully labelled representation of the asymmetric unit). The zinc(II) center is enclosed in a N 3 O 2 coordination sphere due to an axially coordinating pyridine molecule. The bond lengths (2.01 Å (Zn1À O eq ), 2.06 Å (Zn1À N eq ), 1.99 Å (Zn1À N ax ) are very similar to the ones of [Zn(1)(MeOH)]·MeOH. However, the square-pyramidal coordination sphere features a slightly higher distortion with bond angles of 94-110°(S p (SqPy) = 2.07). [12] A very similar asymmetric unit with five-coordinate zinc(II) com-  were obtained that crystallize in the triclinic space group P � 1. The asymmetric unit consists of a five-coordinate zinc(II), where one water molecule acts as the axial ligand leading to a N 2 O 3 coordination sphere (see Figure 1C and Figure S3 in the Supporting Information for a fully labelled representation of the asymmetric unit). The average bond lengths are with 2.00 Å (Zn1À O eq ), 2.06 Å (Zn1À N eq ), and 2.06 Å (Zn1À O ax ) very similar to the ones of [Zn(1)(MeOH)]·MeOH. However, similar to [Zn(1)(py)], the square-pyramidal coordination sphere shows a slightly higher degree of distortion (S p (SqPy) = 1.97). [12] Similar to [Zn(1)(MeOH)]·MeOH, the zinc(II) center in [Zn(2)(H 2 O)] 2 ·H 2 O is enclosed in a N 2 O 3 coordination sphere where water ligates axially. Orange block-like single crystals of [Zn(2)(H 2 O)] 2 ·H 2 O obtained from the mother liquor were analysed in the triclinic space group P � 1. The asymmetric unit consists of two five-coordinate zinc(II) complexes and one water solvent molecule (see Figure 1D and Figure S4 for a fully labelled representation of the asymmetric unit). The bond lengths and angles (2.00 Å (Zn1À O eq ), 2.06 Å (Zn1À N eq ), 1.99 Å (Zn1À O ax ), 99-104°) of the largely undistorted square-pyramidal coordination sphere (S p (SqPy) = 1.28/1.36) [12] closely match the data of [Zn(1)(MeOH)]·MeOH. Interestingly, the metrics of all these complexes closely mimic structures which have been reported of five-coordinate zinc(II) complexes deriving from [13] and salophen (2,2'-[1,2-phenylenebis (nitrilomethyl-idyne)]diphenol) ligands. [14] Finally, red block-like crystals of [Zn(3)(H 2 O)(EtOH)] were obtained from the mother liquor. Different from [Zn(1/2)], the fragment [Zn(3)] exhibits electron withdrawing CF 3 substituents. This increase in Lewis's acidity of the zinc(II) metal center is directly reflected in a coordination of two axial ligands leading to a six-coordinate zinc(II) complex. In the monoclinic space group P2 1 /c, the asymmetric unit consists of one mononuclear six-coordinate zinc(II) complex, where the axial positions are occupied by one water and one EtOH ligand (see Figure 1E and Figure (16) and cis-angles of 85-87°i ndicate a distorted octahedral coordination sphere (S p (O h ) = 1.83). [12] In the crystal, π-π interactions between the chelate Zn1À N1À C4À C5À N2 of neighboring complexes and towards the zinc(II) metal center result into stacked dimers, in Figure 2 and Figure S6 and S9 in the Supporting Information; Table S3 in the Supporting Information for distances and angles of the π-π and M-π interactions). These stacked dimers interact with neighboring molecules through a hydrogen bond network (see Figure S7, S8, and S10 in the Supporting Information; Table S4 in the Supporting Information for hydrogen bond and angles). The supramolecular packing in [Zn(1)py] clearly differs from the above cases, as no stacked dimers are observed due to the coordination of pyridine. The packing is similar to the above described cases dominated by non-classical hydrogen bonds forming columns (see Figure S11 in the Supporting Information; Table S3/4 in the Supporting Information). In the case of [Zn(3)(H 2 O)(EtOH)], where the zinc(II) metal centers are sixcoordinated, the packing is likewise dominated by hydrogen bonds resulting in a 3D hydrogen bond network (see Figure S12 in the Supporting Information; Table S4 in the Supporting Information for hydrogen bond and angles).

Optical properties of the zinc(II) complexes
Steady-state absorption and emission: Due to the closed shell character and a large net nuclear charge, zinc(II) centers are usually not directly involved in optical processes such as absorption or emission. In particular, radiative transitions with predominating MLCT character that are the basis of the rich photophysics and photochemistry of copper(I) are prohibited in isoelectronic zinc complexes. Instead, optical excitation and relaxation are largely dominated by the ligand. Nevertheless, coordination of zinc(II) by the ligands H 2 (1) and H 2 (2) after deprotonation clearly affects the optical properties in solution as they are strongly red-shifted going from a yellow to an orange solution.

Chemistry-A European Journal
Full Paper doi.org/10.1002/chem.202102086 and the zinc complexes [Zn(1/2)], respectively. The leading absorption in [Zn(3)] bearing CF 3 substituents is located at 451 nm (see Figure 3A and Figure S14/15 in the Supporting Information). Exemplarily, this distinct red shift upon coordination by Δν = 2800 cm À 1 is shown for ligand H 2 (1) and [Zn(1)] in Figure 3A; similar observations hold for H 2 (2)/[Zn(2)] (Figure S14B and S15 A in the Supporting Information). The strong UV band in the ligand spectrum at 305 nm is likewise shifted to smaller energy by Δν = 2100 cm À 1 . Spectral shifts of ligandcentered bands upon coordination are not unique. For a topologically related couple deriving from salicylic aldehyde, a coordination-dependent red shift has been recorded from 374 nm (in acetonitrile) for the ligand [15] to 560 nm for the zinc complex (in DMSO); [16] that is, excitation energy shifts by as much as Δν = 8800 cm À 1 .
[Zn(1)] shows weak fluorescence in CHCl 3 , that is in a noncoordinating environment (λ Em = 495 nm; Φ F � 0.004; all reported quantum yields are relative to [Zn(1)] in pyridine by the comparison of the absorbance corrected integrated emission intensity). Similar observations hold for [Zn(2)] and [Zn(3)], with somewhat smaller quantum yield Φ F of the latter. By contrast the free ligands are essentially non-emissive under the same conditions (Φ Em � 10 À 4 ); it is noted that the salicylic analogue of ligand H 2 (1) supports fluorescence with Φ Em = 0.017. [15] The spectra of all zinc(II) complexes indicate partly resolved vibrational structure, both in absorption and in emission. An energy spacing of the vibrational progression by Δ vib E � 1100 cm À 1 points to a leading role of chelate skeletal modes. This assignment is supported by analysis of DFT-derived harmonic frequencies (see Figure S16 in the Supporting Information and animated gif).
Quite generally zinc complexes of planar-directing N 2 O 2 ligands show significant solvatochromism in both, absorption and emission. [8b,e] Different from these literature precedents, a solvent scan shows that the absorption spectra of unit [Zn(1)] are largely indifferent to solvent variation with respect to energy and band shape ( Figure 3A). It is only in neat pyridine that absorption is affected. Notwithstanding the almost invariant absorption spectra, however, emission properties of [Zn(1)] vary substantially with solvent (see Figure 3B). While emission is very weak in chloroform and acetonitrile (Φ Em (CHCl 3 ) = Φ Em -(MeCN) � 0.004), it is greatly enhanced in EtOH and THF (Φ Em (EtOH) � 0.07 and Φ Em (THF) � 0.10) and reaches its maximum in neat pyridine (Φ Em (Py) � 0.15). As a matter of fact, fluorescence of [Zn(1)] which is silent in acetonitrile, can be increased successively through addition of water (see Figure S17 in the Supporting Information). Increasing the water content up to 50 vol-% results in a quantum yield increase up to Φ Em (MeCN/H 2 O 1/1) � 0.013. Coordination of Lewis-basic solvent molecules to the Lewis-acidic zinc center must be considered as the underlying molecular factor that switches on fluorescence in [Zn (1)]. This notion is supported by the continuous increase of Φ Em with the donor number of the solvent (see Figure S18 in the Supporting Information).
Shown in Figure 4A is a titration of equally concentrated solutions of [Zn(1)] in CHCl 3 with pyridine, which yields a continuous red-shift by up to 15 nm (see Figure S15A/B in the Supporting Information for titrations of [Zn(2)] and [Zn(3)], see Figure S19 for photographs), coupled to an intensity variation between 3.6 × 10 4 cm À 1 M À 1 < ɛ M < 4.6 × 10 4 cm À 1 M À 1 . (see Figure 4A). Fluorometric titration with pyridine shows a continuously enhanced emission for [Zn(1)] and [Zn(2)] which saturates only at the highest pyridine loads, which equals neat pyridine (see Figure 4B and Figures S15C in the Supporting Information). Fluorescence excitation spectra are indifferent to the observation wavelength and qualitatively map the absorption spectrum across the entire UV-Vis range (red in Figure 4C, Figure S15E/F in the Supporting Information). Thereby, the pyridine-dependent red shift in absorption is compensated by a parallel red shift in emission so that the Stokes-shift of Δ Stokes E � 1100 cm À 1 is preserved. Likewise, the coincidence of absorption and fluorescence excitation spectra (deviation at λ < 330 nm is due to uncompensated absorption by excess pyridine) and the vibrational structure of the diagnostic bands are maintained nearly constant (see Figure 4D and Figure S15E in the Supporting Information).
DFT and wave-function theory analysis of the electronic structure: The nature of the leading optical transition of the zinc complexes has been addressed by means of (time-dependent) density-functional theory and wave function-based ap-

Chemistry-A European Journal
Full Paper doi.org/10.1002/chem.202102086 proaches. Complete active space self-consistent field, CASSCF, calculations were performed for the [Zn(1)py] complex, with an active space comprising the entire system of π-orbitals and their electrons, CAS (18,16). This represents the main set of orbitals responsible for the luminescence process. The lowest two singlet spin states (S 0 and S 1 states) and the lowest triplet spin state (T 1 ) wave functions have been optimized at the CASSCF level of theory, followed by a multiconfiguration pairdensity functional theory approach (MC-PDFT) for treating dynamic correlation effects outside the chosen active space. [17] The singlet state calculations were performed at the DFT optimized geometry (pertinent metrics in Tables S5/6, Supporting Information). Triplet spin state calculations were performed both at the optimized triplet geometry (adiabatic excitation) and at the singlet geometry (vertical excitation).
According to theory, the photophysics of [Zn(1)X] is invariably centered at the chelate ligand. Variation of the axial ligand, X, does not considerably affect the transition frequency, neither in experiment nor theory (see Table 2). MC-PDFT calculations on [Zn(1)py] predict the S 0 !S 1 transition at 512 nm, while DFT predicts the same transition at 440 nm, both in fair agreement with the experimental value (481 nm) (see Table 2). TD-DFT tends to systematically overestimate the energy of the leading Vis transition in the entire ligand-

Chemistry-A European Journal
Full Paper doi.org/10.1002/chem.202102086 bound donors, whereas the LUMO is more localized on the dinitrile moiety. As a consequence, optical excitation in [Zn(1)py] leads to a charge shift toward the dinitrile site, rendering the excited state intra-ligand CT-like. Similar conclusions had been drawn previously in studies of the closely related congener [Zn(sal)py], which derives from an enole/ imine ligand based on salicylic aldehyde. In keeping with conserved transition energies, the axial ligand X does not feature significantly in the frontier molecular orbitals, as exemplarily shown in Figure 5 for [Zn(1)py] (data of the entire X-series is summarized in Figure S20 in the Supporting Information). The optimized CASSCF natural orbitals for all states are similar to each other and to the Kohn-Sham (KS) orbitals derived from DFT. They are reported in Figure S21 in the Supporting Information. As for the KS-orbitals, the CAS natural orbitals do not show any significant admixture from the axial pyridine ligand. This indifference is reflected by a small decrease in the transition energies upon axial coordination. MC-PDFT and TD-DFT predict a red-shift of~1000 cm À 1 and 1400 cm À 1 , respectively, of the S 0 !S 1 transition upon adding pyridine to the hypothetical four-coordinate model system [Zn(1)], with no axial ligands. Even smaller red-shifts of 700-900 cm À 1 are induced by the weaker Lewis-bases (see Table 2). N 2 O 3 coordination inherent to the dimeric complexes [Zn(1)] 2 likewise results in a red-shifted S 0 !S 1 transition. It is predicted at � 430 nm, irrespective of dimer structure (see also Discussion below).
Overall, both methods yield consistent results concerning the nature and energies of electronic states and transitions in the five-coordinate species. It is not uncommon that KS-DFT and the subsequent TD-DFT are able to describe excited states (here, the S 1 state) with a reasonable level of accuracy. Neither a bi-configurational wave function nor spin-adaptation is enforced by the method often leading to unphysical eigensolutions to the Schrödinger equation (not spin eigenstates). Nonetheless, KS-DFT can predict accurate energies of electronic states, in virtue of the fact that DFT relies on correct local spindensities and not to the detailed structure of the wave function. Given the widespread utilization of KS-DFT also beyond its inherent single-reference limits, this aspect is discussed in the following. A more detailed treatment is given in Ref. [17c].
Inspection of the CAS wave functions of the S 0 and S 1 states of [Zn(1)py] clearly corroborates the ILCT nature of the S 0 !S 1 transition involving a HOMO to LUMO single excitation, adding value to our DFT computations. Both S 0 and S 1 wave functions are predominantly single-reference. The leading contribution to the S 0 state (79 % weight) is the closed-shell configuration (single-reference) found also by KS-DFT for the same state. In the lowest excited singlet state an electronic configuration corresponding to the HOMO-to-LUMO one-electron excitation dominates with a weight of 73 %. It is important to point out here that while the S 1 wave function is predominantly singleconfigurational (in terms of spin-adapted configuration state functions, CSFs), it is inherently bi-configurational. That is, the two unpaired electrons residing in the HOMO and the LUMO are precisely coupled to a singlet spin state, À � y (here H and L refer to HOMO and LUMO respectively). Thus, while both states can be considered single-reference, the S 1 excited state requires a more careful theoretical treatment, to ensure the correct spin-symmetry and to correctly capture electron correlation effects. These aspects are explicitly considered in the CASSCF/MC-PDFT methodology, due to the multi-configurational character of the method, and their formulation in a spin-adapted basis (via the graphical unitary group approach, GUGA, algorithm). [18] The MC-PDFT-derived vertical and adiabatic singlet-triplet gaps of [Zn(1)py], Δ S/T E, are reported in the Jablonski diagram in Figure 6a; they are given in terms of the S 0 !T 1 transition energies at the relaxed geometry of S 0 (vertical excitation) and T 1 (adiabatic excitation). Close-lying excited singlet and triplet states are generally taken as a basic requirement of rapid and efficient intersystem crossing, which would competitively limit the fluorescence quantum yield. A vertical gap of Δ S/T E = 2.14 eV places the triplet state only 0.28 eV below the

Chemistry-A European Journal
Full Paper doi.org/10.1002/chem.202102086 computed energy of the Franck-Condon state, that is, relative to S 1 in an excited vibronic state. In the relaxed geometry of T 1 the somewhat smaller adiabatic gap of Δ S/T E = 1.97 eV arises, concomitant with a slightly widening gap between S 1 and T 1 . Nevertheless, these calculations indicate an efficient ISC to the triplet state as the reducing factor of the fluorescence quantum yield. In the related zinc(II) system [ZnTPP] (TPP: tetraphenylporphine), fluorescence and phosphorescence have been recorded at 600 nm and ca. 780 nm (77 K), respectively, giving an upper limit for the S 1 -T 1 gap of 0.48 eV. [19] In keeping with the narrow S 1 -T 1 gap, a large triplet quantum yield of 0.88 was observed.
For comparison we plot in Figure 6b the data that derive from KS-DFT and TD-DFT. These are in qualitative agreement with the MC-PDFT values, but indicate larger S 1 -T 1 splitting. We believe that a gap of 1.3 eV is hardly in agreement with efficient intersystem crossing.
In agreement with experiment the Vis transition of [Zn(sal) py] (λ DFT = 565 nm; λ exp = 560 nm [16] ) is predicted at significantly smaller energy than in [Zn(1)py] (λ DFT = 440 nm; λ exp = 481 nm). This tendency is maintained in the (hypothetical, see discussion) planar CN 4 species and is reflected by the shrinking HOMO-LUMO gap of [Zn(sal)py]. The presence of additional phenolateborne π-donor states in the latter gives rise to further Vis transitions at higher energy that are clearly absent in [Zn(1)py]. Nevertheless, both complex families share similar optoelectronic properties, in a qualitative sense; that is, conserved nature and intensity of the leading excitation and significant CT-state emission with minor Stokes-shift. As is detailed below, however, they differ substantially in quantitative terms, with respect to their affinity for Lewis bases.
Quenched Lewis-acidity and emission properties: To discuss the optical phenomena in more quantitative terms, the concentration dependence in the spectroscopic titrations with pyridine has been analyzed. Plots of the integrated emission intensity and the differential changes of absorptivity (abstracted in terms of speciation; that is, reporting the fraction CN 5/6) as a function of pyridine concentration are shown in Figure 7.
As a first result, the above Lewis-base hypothesis is corroborated by the higher affinity for pyridine expressed by  Figure S22 in the Supporting Information). They further support the ILCT character of the underlying transitions. As a more diagnostic effect of enhanced Lewis acidity, the pyridine-dependent spectral evolution of [Zn(3)] saturates above approximately 590 keq, whereas much larger amounts are necessary with [Zn(1/2)] (see Figure 7D). Similarly, enhanced affinity for Lewis-bases has been recently reported for nickel(II) complexes of CF 3 -decorated ligands. [20]

Chemistry-A European Journal
Full Paper doi.org/10.1002/chem.202102086 The experimentally observed order of affinity is well captured by DFT model calculations. Reaction energies of the Lewis acidbase association of pyridine with (hypothetical) planar monomeric zinc complexes (CN 4) are all highly negative, indicating a significant thermodynamic driving force even if unfavorable entropies are considered (equation 1-3). Association is predicted to be more exothermic in [Zn(3)] than in [Zn(1)] by ca. 8 kJ mol À 1 .
There is general agreement that four-coordinate planar monomeric zinc complexes (CN 4) will add one or two additional axial ligands to saturate the Lewis-acidic center as is implied by equation 1-3. In the absence of potent donors selfcomplementary stacking of such zinc(II) units occurs to yield dimers with a Zn 2 O 2 core. [8] Crystal structures including stacked dimers with a Zn 2 O 2 core have been reported for several derivatives of the salophen type and dimers have been generally suggested as the dominant species in solution under non-coordinating conditions. Such dimeric structures are generally found to have reduced fluorescence quantum yields, whereas the on-switch of fluorescence in the presence of Lewisbases has been interpreted in terms of base-induced deaggregation. Stacking into dimers in non-coordinated media was likewise suggested to reduce fluorescence in malnant derivatives with maleodinitrile backbones. 2 ½ZnðsalÞ� Ð ½ZnðsalÞ� 2 ; K dim � 10 8 M À 1 (4) ½ZnðsalÞ� þ py Ð ½ZnðsalÞpy�; K py � 10 5 À 10 6 M À 1 (5) Different from [Zn(sal)], there is no such simple correlation for [ Zn(1-3)]. Closer inspection of the low-concentration sections of the plots in Figure 7A/B reveals that the concentration dependence of the absorption and the emission is not identical. In particular, the plot in Figure 8 of the normalized integrated emission intensity vs. the CN 5/6 molar fraction of [Zn(1-3)] reflects pyridine-dependent transformations among (at least) four different species, from which (at least) the first two are non-emissive. The close similarity of [Zn(1-3)] in the plot of Figure 8 points to an overall conserved phenomenology among all three compounds.
The first chemical step connects two non-fluorescent species; notably almost one third of the overall absorption change is involved in this step, whereas there is hardly any increase in emission. In accordance to previous findings of nonemissive stacked dimers, [8b,d] it appears reasonable to ascribe the lack of fluorescence to the dimeric or oligomeric nature of both non-fluorescent species. This assignment is supported by means of dynamic light scattering (DLS) in MeCN. Solutions in CHCl 3 are expected to behave similarly, but could not be studied by DLS due to limited solubility. Non-emitting, filtered solutions of [Zn(1)] in neat MeCN (3 · 10 À 4 M < c < 1 · 10 À 3 M) all contain nanoparticles with broad particle size distributions (d max � 120 nm) (see Figure S23 in the Supporting Information). Mean particle size scales inversely with the concentration of the complex. The presence of pyridine drastically reduces the amount of nanoparticles, whereas in neat pyridine, DLS gives no response (see Figure S24 in the Supporting Information). In keeping with the destacking of the dimers/oligomers in the   (Figure 1). While we cannot judge with absolute certaintythe decreased emission efficiency of six-coordinate species may be associated with the higher flexibility of the octahedral coordination sphere due to the intrinsically longer ZnÀ N/O bonds leading to a higher rate of non-radiative decays. The coordination of the second pyridine might also result in a decreased singlet-triplet gap leading to emission quenching, which was found indeed in DFT calculations (see Figure S25 in the Supporting Information).

Chemistry-A European Journal
Kleij et al. had previously observed a diminished sensitivity toward Lewis-bases of some dinuclear zinc(II) complexes. [22] Akin to our observations with [Zn(1-3)], > 10 5 molar equivalents of pyridine were necessary to drive the destacking. The authors associated the markedly reduced sensitivity with strong stacking into large oligomers. Based on the observation of nanoparticles in dilute solutions via DLS, we likewise associate the seemingly reduced Lewis-acidity of [Zn(1-3)] with the formation of stable oligomers of unusual stability. That is, the strong Lewis-acidity of [Zn(1-3)] which is indicated by DFT calculations must be overridden by an additional ligand-borne effect not accessible to mononuclear salophen derivatives. Indeed, a DFT structure search served to identify plausible dimer structures of high stability for [Zn(1)] 2 . For sake of comparison, stacked structures of [Zn(sal)] 2 were studied on the same level of theory. The computed metrics of the Zn 2 O 2 core of [Zn(sal)] 2 shown in Figure 9A are similar to the reported crystal structures of salophen-type complexes. [21a] Details of all optimized structures are summarized in Table S7 in the Supporting Information. In particular, the short axial ZnÀ O distance of 2.15 Å indicates tight binding within the dimers; electronic reaction energies of the dimerization equilibrium amount to À 16 kJ mol À 1 . Similar structures could be extracted from optimization of Zn 2 O 2 -bound dimers of [Zn(1)], with similar energies of dimerization.
Most importantly, however, these Zn 2 O 2 -core dimers are not the minimum structures in the case of [Zn(1-3)]. By contrast, DFT calculations identify a completely different stacking motif at significantly lower energy, as shown in Figure 9B. In these (ZnO) 2 macrocycles, ligand-appended carbonyls saturate the N 2 O 3 coordination sphere of zinc to establish fully relaxed square-pyramidal environments on both centers. A somewhat related constellation has been observed by Weber et al. in the solid-state structure of some iron(II) complexes of this general ligand type. [23] In the cited system, the axially arranged carbonyls rather assist coordination, given the long FeÀ O distance of 2.70 Å. In [Zn(1-3)], tight axial binding is indicated by short ZnÀ O ax bond lengths of 2.10 Å, which are even shorter than in the reported and computed dimers of [Zn(sal)]. Favorable dispersive interactions between the chelate planes of the stacked complexes support the laterally shifted dimer structure. Both aspects sum up in a strongly increased stability compared to the Zn 2 O 2 -core dimer. Computed energies of dimerization are in favor of the macrocyclic (ZnO) 2 dimer by À 20 kJ mol À 1 . ] support an alternative, very stable macrocyclic stacking motif which involves ligand-appended, remote carbonyl moieties. As the stability of the non-emissive dimers or oligomers obviously controls the sensitivity of the sensor material, tuning through introduction of sterically demanding substituents might enable the design of sensor materials with consciously adjustable sensitivity areas.

Chemistry-A European Journal
Full Paper doi.org/10.1002/chem.202102086 MC-PDFT has been largely demonstrated a good and cheap alternative to CASPT2 and it has been utilized as main method of choice in this investigation to recover dynamic correlation outside the active space. Within MC-PDFT the tPBE translated functional has been chosen, that in our experience outperforms the other translated functional available to date.
A basis set of generally contracted atomic natural orbital (ANO-RCC) type has been used for all atoms, obtained from the C,N,O(14 s,9p,4d), H(8 s,4p) and Zn(21 s,15p,10d,6 f) primitive functions, contracted to C,N,O(3 s,2p,1d), H(2 s,1p), Zn(5 s,4p,2d,1 f) functions, giving a basis set of split-valence double-ζ plus polarization quality (VDZP). This choice of basis set leads to a total of 577 basis functions for the [Zn(1)py] model system. Intermediate calculations in a minimal basis set (MB), with contraction scheme C,N,O(2 s,1p), H(1 s), Zn(4 s,3p,1d), have been performed as aid in the choice of the active space for all model system and spin states. The basis set has been subsequently expanded to the VDZP and the MB optimized orbitals augmented and used as starting orbitals for the VDZP CASSCF (18,16) optimization. C1 point group symmetry has been utilized in line with the preceding DFT method. The evaluation of the electron repulsion integrals has been greatly simplified by means of the resolution-of-identity Cholesky decomposition technique, with a decomposition threshold of 10 À 4 a.u.