Crystallographic and DFT studies on host-guest complexes consisting of zinc bisporphyrinates and 1-phenylethylamine

Abstract We have investigated the chirality transfer from 1-phenylethylamine to a 5-amino-1,3-phthalic acid diamide-linked zinc bisporphyrinate through crystallographic and DFT studies. When the hosts were mixed with optically pure 1-phenylethylamine, CD showed moderate signals in the Soret band region. Single crystals of the corresponding 1:1 and 1:2 host-guest complexes were obtained. We present the first crystallographic structure of a 1:2 host-guest complex consisting of a bisporphyrin host and chiral monoamine guests. The structure reveals that the host-guest interactions are different for two guest molecules. The first guest is involved in a hydrogen bond and π-π interactions, while the second one is only involved in π-π interactions, which has not been observed in previous studies. The corresponding chirality transfer mechanism was also rationalized by DFT calculations.


Introduction
Porphyrins, especially bisporphyrins, have been widely investigated in recent years in chirality transfer, chiral recognition, chirality induction, etc. [1][2][3][4][5][6]. For bisporphyrin systems, monodentate chiral guests without chemical derivation are generally difficult to study due to the limited binding site. There are only a few reports [7][8][9][10][11] dealing with monodentate chiral guests, such as monoamines or monoalcohols. For example, Borovkov et al. [12] developed an ethane-bridged bisporphyrin system and used it to assign the absolute configurations of monoamines or monoalcohols. In their systems, besides the coordination interaction, there are also steric interactions formed between the ethyl groups of the porphyrin and the ligand's substituents, which result in the corresponding supramolecular chirality. Borhan et al. developed a biphenol-bridged metal-free bisporphyrin. In that case, both hydrogen bonding and steric interactions were formed between the monoamines and the host molecule, which led to the chirality transfer from monoamines to the host-guest complexes [13].
These supramolecular interactions, such as hydrogen bonds and p-p interactions, are very important in chirality transfer processes. Investigations on supramolecular interactions can help us not only understand chirality transfer mechanisms, but also design better systems and control their chirality transfer abilities.
To better understand these supramolecular interactions, it is important to find the accurate binding modes of the host-guest complexes. The accurate binding modes cannot only reveal the host-guest interactions, but also help us understand the spectroscopic results and structure-function relationship, and lead to the developments of new materials. The direct binding information can be provided by crystal structures of the host-guest complexes, but it is usually very difficult to obtain the single crystal structures of the complicated host-guest complexes. For bisporphyrin systems, there are only a few crystallographically characterized examples. For instances, Inoue and Borovkov have studied the first crystallographic structure of a host-guest complex between an achiral bisporphyrin and a chiral diamine [14]. Rath et al. recently reported crystal structures of several host-guest complexes between bisporphyrins and diamines/diols [15].
For monodentate guests, no crystallographic studies have been reported until we developed a series of amide linked bisporphyrins from a 2-aminophenyl substituted monoporphyrin [16][17][18][19][20][21]. Both 1:1 and 1:2 host-guest complexes could form between zinc bisporphyrinate and monodentate guests. The structures of the 1:1 zinc bisporphyrinate-monoamine complexes were reported in our previous studies. But no 1:2 complexes have been crystallographically characterized so far, which brought up questions on the accurate binding mode for the second guest.
In this work, we tried to obtain structural information for 1:2 complexes. When R-1phenylethylamine (R-1-PEA) was used as the guest, we successfully obtained the single crystals of the 1:2 host-guest complex, [Zn 2 -AmBis]Á(R-1-PEA) 2 . It is the first crystallographic structure of a 1:2 host-guest complex formed between a bisporphyrin host and chiral monoamine guests. The structure reveals that the second ligand adopts an "inside" binding mode and forms p-p interactions with the linker, but does not form any hydrogen bonds. Such a binding mode has not been reported in previous studies.
Based on the structural data, the chirality transfer mechanism was also rationalized by DFT calculations.

Materials and general methods
All reagents were obtained from Energy Chemical Company and used without further purification unless otherwise noted. NiCl 2 Á6H 2 O was stored in a desiccator before use. N,N'-bis-(2-(10,15,20-triphenylporphyrin-5-yl)phenyl)-5-amino-1,3-phthalic acid diamide (AmBis) was prepared according to the literature [19]. Elemental analyses (C, H, and N) were performed with an Elementar Vario EL III analytical instrument. 1 H NMR spectra were obtained at room temperature using a Bruker AVANCE 400 MHz spectrometer in DMSO-d 6 . Fourier transform infrared (FT-IR) spectra were recorded with a BRUKER VERTEX70 FT-IR spectrometer. Thermal gravimetric analysis (TGA) was performed on a TA SDTQ600 thermal analysis system under a nitrogen atmosphere, ramping at 10 C per min from room temperature to 800 C. UV/Vis spectra were measured with a Shimadzu UV-3150 spectrometer. CD spectra were recorded on a AVIV Model 410 spectropolarimeter at 295 K. Scanning conditions were as follows: wavelength step = 1.00 nm, bandwidth = 2 nm, response time = 0.1 seconds, averaging time = 0.100 seconds, settling time = 0.333 seconds. CD measurements have been performed by adding solution of the optically active monoamines to the [Zn 2 -AmBis] solution in CH 2 Cl 2 at room temperature. Then CD spectra were measured in millidegrees, then normalized based on the concentration of [Zn 2 -AmBis]. AmBis (0.50 g, 0.36 mmol) was dissolved in a mixture of CH 2 Cl 2 (100 mL) and CH 3 OH (50 mL). Solid Zn(CH 3 COO) 2 (0.06 g, 0.34 mmol) was added in one portion to the above solution and the mixture was refluxed. We used TLC to monitor the reaction to obtain optimized conditions for a maximal yield of the [Zn 1 -AmBis]. Then the solution was washed with water, and the organic layer was collected and evaporated to dryness under vacuum. The resulting purple solid was purified by silica gel chromatography. The obtained [Zn 1 -AmBis] (0.40 g, 0.27 mmol) was dissolved in DMF (15 mL). NiCl 2 Á6H 2 O (0.08 g, 0.56 mmol) was added to the above solution. The mixture was refluxed for 2 h, then cooled to room temperature. It was washed with water, and the organic layer was evaporated to dryness. A dark purple product was purified by silica gel chromatography (1:2 CH 2 Cl 2 /petroleum ether) (yield 0.37 g, 90%). 1  [ZnNi-AmBis] (50 mg, 0.03 mmol) was dissolved in a CH 2 Cl 2 /toluene solution (5:1) (1 mL); 1-phenylethylamine (36 mg, 0.3 mmol) was added and the mixture was stirred for $3 min. Then it was transferred into 7 mm Â 250 mm glass tubes. n-Heptane was added slowly to glass tubes as a nonsolvent. Those tubes were flame sealed, and left undisturbed for two months, yielding purple crystals. Then the purple crystals were isolated by filtration, washed with n-heptane, and dried in air (yield 13 mg, 25%

X-ray crystallography
For [ZnNi-AmBis]Á(1-PEA), the measurement of a single crystal was performed on an Agilent diffractometer with graphite monochromated Mo Ka (k ¼ 0.71073 Å) radiation at 293 K. The data reduction and multiscan absorption correction were performed using the CrysAlisPro software (Agilent, 2013) [22] supplied by the manufacturer. For [Zn 2 -AmBis]Á(R-1-PEA) 2 , the measurement of a single crystal of was performed on a Bruker APEX-II CCD X-ray diffractometer at 223 K. the data reduction and multiscan absorption correction were performed using the Bruker SAINT software [23] supplied by the manufacturer. Both structures were solved by direct methods using the program SHELXS-97 [24] and refined on F 2 using the full matrix least-squares method with SHELXTL version 2014 [25]. All non-hydrogen atoms were refined anisotropically; all hydrogens were theoretically added and rode on their parent atoms.
For [ZnNi-AmBis]Á(1-PEA), the guest 1-PEA is disordered over two positions; the (anisotropic) displacement parameters of the disordered atoms were constrained to similar values by EADP and SIMU and the final refinement gave the occupancy of the larger component as 52%.
For [Zn 2 -AmBis]Á(R-1-PEA) 2 , the structure was solved in a chiral space group P1. It was refined as a 2-component inversion twin. One asymmetric unit contains two independent zinc bisporphyrinate molecules (258 non-hydrogen atoms). There are four 1phenylethylamine ligands. One of them was disordered over two positions; their (anisotropic) displacement parameters were constrained by SIMU and EADP. The final refinement gave the occupancy of the larger component as 64%. One phenyl ring of the porphyrins was disordered over two positions. Its (anisotropic) displacement parameters were constrained by SIMU and the final refinement gave the occupancy of the larger component as 62%. The (anisotropic) displacement parameters for other three ligands and several phenyl rings of the porphyrins are a little large probably due to slight disorders, but attempt to refine them as disorder components failed, so these (anisotropic) displacement parameters were constrained by SIMU. The Flack parameter (0.129 (15)) is a little large; the possible reason is that the quality of the crystal was not good.
In both structures, the asymmetric unit contains badly disordered solvent molecules. SQUEEZE [26] was used to model all disordered solvates. For [ZnNi-AmBis]Á(1-PEA), the elemental analysis suggests there are two heptane molecules. According to the TGA measurement (supplementary material Figure S11), the loss of 17.2% weight is caused by the loss of two heptane and one 1-phenylethylamine molecules (calculated lost is 17.4%).
For [Zn 2 -AmBis]Á(R-1-PEA) 2 , the elemental analysis suggests there is one heptane solvent. According to the TGA measurement (supplementary material Figure S12), the loss of 17.6% weight is caused by the loss of one heptane and two 1-phenylethylamine molecules (calculated lost is 18.2%). Details of the crystal parameters, data collection, and refinements are summarized in Table 1.

Computational methods
Based on the crystal structural data, we first performed density functional theory (DFT) calculations on the free host [Zn 2 -AmBis]. Then we performed calculations on the corresponding 1:1 and 1:2 host-guest complexes formed between [Zn 2 -AmBis] and R-1-PEA ( Figure 1). Full optimizations were performed by density functional theory at the level of WB97XD/6-31G Ã using the Gaussian 09 suite of program [27]. We employed density functional theory (DFT) with no symmetry constraints to investigate the optimized geometries.

CD Spectral studies
The CD titration spectra were measured by addition of optically pure 1-PEA into the solution of [Zn 2 -AmBis]. In Figure 2, CD spectra clearly show moderate signals in the Soret band region. For R-1-PEA, there is a positive peak at 434 nm and a negative peak at 426 nm. For S-1-PEA, the spectra show similar shapes but opposite signs.
How do we understand the CD spectra? We also did a comparison experiment by measuring the CD spectra of the mixture between chiral monoamines and metal-free bisporphyrin; no observable signal was obtained. That suggests that the zinc ions are essential for the chirality transfer, and the corresponding host-guest complexes are  responsible for the resulting CD. Since there are two zinc ions in [Zn 2 -AmBis], when the guest molecule is coordinated to [Zn 2 -AmBis], the above equilibria should exist in solution. Then there will be two possible complexes with the host-guest molar ratio as 1:1 and 1:2.

Crystal structures
To get the binding model of the host-guest complexes, we tried to get their crystal structures. Many methods were attempted to grow crystals from the mixture of [Zn 2 -AmBis] and a large excess of R-1-PEA. Suitable single crystals of the 1:2 host-guest complex, [Zn 2 -AmBis]Á(R-1-PEA) 2 , were finally obtained by diffusion of n-heptane into the solution in CH 2 Cl 2 /toluene. We also attempted to grow single crystals , its structure was solved in the space group P i. One asymmetric unit contains one zinc bisporphyrinate molecule in which two porphyrin subunits are linked by a 5-amino-1,3-benzenedicarboxamide group. Zinc is five-coordinate with one 1-PEA as the axial ligand, while nickel is fourcoordinate, just as expected. As we notice, the ligand adopts the "inside" binding mode (the ligand is facing the linker). There is also an interesting hydrogen bond as shown in Figure 3, which is formed between N1 and the carbonyl oxygen O012. The corresponding distance N1ÁÁÁO012 is 3.076(14) Å; the bond angle N1-H1BÁÁÁO012 is 165 . The role of NH 2 is different from Borhan or Borovkov's systems [12,13]. In Borhan's system, the NH 2 is involved in hydrogen bonds. In Inoue and Borovkov's system, the NH 2 is involved in a coordination interaction. In our case, the NH 2 is involved in both coordination and hydrogen bonding interactions. Besides the hydrogen bond, there is also a p-p interaction between the phenyl of the guest and the phenyl of the linker. The corresponding centroid to centroid distance is 3.917 Å; the dihedral angle between two phenyl rings is 7.98 . This type of binding mode has been also observed in previous studies [18][19][20]. These interactions could be the major factors stabilizing the "inside" binding mode for the first ligand.
Since the crystal structure was solved in an achiral space group, the overall crystal is a racemic mixture, which is optical inactive. Even so, the structure does provide useful binding information since a single molecule in an asymmetric unit is chiral. In the structure, when the guest is R-type, two porphyrin subunits form a clockwise twist with the torsional angle C1M3-C1M1-C2M1-C2M3 of 153 . Based on the exciton chirality method [29], the clockwise twist should lead to positive exciton chirality. For the case of the S-type of guest, the corresponding torsion angle is negative, which should lead to negative exciton chirality. These are consistent with our experimental results.
For the 1:2 host-guest complex, the structure of [Zn 2 -AmBis]Á(R-1-PEA) 2 was solved in a chiral space group, P1. One asymmetric unit contains two independent zinc bisporphyrinate molecules (mol A and mol B), shown in Figure 4.
In mol A, each zinc is five-coordinate with R-1-PEA as the axial ligand, which leads to the 1:2 host-guest complex. In this molecule, both ligands adopt the "inside" binding mode. In general, the binding mode for the first ligand is similar to our previous studies [18][19][20]. But the binding modes for two ligands are different. The first ligand is coordinated to Zn1. Besides coordination interactions, the first ligand is also involved in hydrogen bonding and p-p interactions as shown in Figure 4. The corresponding NO distances (2.96 Å) indicate there are hydrogen bonds between the carboxylic oxygens and the coordinated NH 2 . Besides hydrogen bonds, the crystal structure also exhibits intramolecular p-p interactions between the phenyl ring of the linker and the phenyl rings of the first ligand. The corresponding centroid-centroid distance is 3.817 Å; the dihedral angles between phenyl planes is 14.3 .
We are more concerned about the binding mode for the second ligand. Interestingly, it also adopts an "inside" binding mode. Different from the first ligand, there are no hydrogen bonds involved. We noticed that there were weak p-p interactions between the phenyl rings of the second ligand and the phenyl ring of the linker, which could be the reason to stabilize the "inside" mode. The corresponding centroid-  centroid distance is 4.130 Å; the dihedral angles between phenyl planes is 24.1 . Such binding mode for the second ligand has not been reported in previous studies. These p-p interactions also lead to a "sandwich" like structure as shown in Figure 4. This "sandwich" like structure was also found in those host-guest complexes for zinc triporphyrinate [30,31], but in that case both ligands are involving in hydrogen bonds and p-p interactions. It suggests hydrogen bonds could be not essential for this "sandwich" like structure.
In mol B, the basic host-guest bonding interactions are similar to those in mol A. Notably, two host molecules in mol A and mol B form a pair of quasi-enantiomers, but the guest molecule is just a single enantiomer, which lead to two diastereomers of the host-guest complexes in the crystal structure.
For the 1:2 complex [Zn 2 -AmBis]Á(R-1-PEA) 2 , the two diastereomers form different twists, which could lead to different CD signals. It is most likely that one diastereomer is dominated in solution, which leads to the corresponding CD signals. In order to better understand the chirality transfer mechanism for this system, we did further investigation by DFT calculations.

Computational studies
All the optimized structures are displayed in Figure 5. In Figure 5(a), for the host [Zn 2 -AmBis], the optimized structure shows a noncentrosymmetric (chiral) configuration. These conformers exist as a pair of enantiomers (conformers A and B).
We are more concerned about what structural differences cause the difference in energy for these two conformers. For the guest, the chiral carbon is bonded to four different groups: NH 2 , phenyl, methyl and hydrogen. Due to the coordination interactions and p-p interactions, the NH 2 and the phenyl group in both conformers are fixed in similar positions, but the orientations of the methyl group and the hydrogen atom are quite different. In their optimized structures as shown in Figure 5(b), the methyl group is facing the porphyrin plane in conformer BÁ(R-1-PEA), while it is away from the porphyrin plane in conformer AÁ(R-1-PEA). Since methyl is more bulky than hydrogen, such orientation may cause less steric repulsion for conformer AÁ(R-1-PEA) and hence, lower its energy. So, conformer AÁ(R-1-PEA) could be the major contributor to the CD signal.
In conformer AÁ(R-1-PEA), the two porphyrin subunits adopt a clockwise twist with the C5-C15-C5'-C15' torsion angle of 165 . The above conformer will lead to positive exciton chirality according to the exciton chirality method [29], consistent with our crystal structural results.
When the 1:1 host-guest complexes are bonded to the second ligand, the corresponding 1:2 complexes, conformer AÁ(R-1-PEA) 2 and conformer BÁ(R-1-PEA) 2 , are formed. Their optimized structures are shown in Figure 5(c). Compared with the crystal structure, the coordination interaction, hydrogen bonds and p-p interactions are well maintained in the optimized results. The calculations suggest that conformer AÁ(R-1-PEA) 2 is more energetically favorable than conformer BÁ(R-1-PEA) 2 (2.11 kJ/mol lower). Similar to the case of 1:1 complexes, their optimized structures in Figure 5(c) show that the methyl group is facing the porphyrin plane in conformer BÁ(R-1-PEA) 2 , while it is away from the porphyrin plane in conformer AÁ(R-1-PEA) 2 . That may cause less steric repulsion for conformer AÁ(R-1-PEA) 2 and hence, lower its energy. So, conformer AÁ(R-1-PEA) 2 could be the major contributor to the CD signal. In conformer AÁ(R-1-PEA) 2 , the two porphyrin subunits adopt a clockwise twist with the C5-C15-C5'-C15' torsion angle of 167 . Based on the exciton chirality method [29], the above conformer will lead to positive exciton chirality. So DFT calculations suggest that both 1:1 and 1:2 complexes lead to the positive exciton chirality, which is also consistent with our experimental results.

Conclusion
We have prepared both single crystals of 1:1 and 1:2 complexes formed between metallic bisporphyrinates and R-1-phenylethylamine. CD measurements showed moderate signals for the mixture of the guest and the host. In the crystal structure of [Zn 2 -AmBis]Á(R-1-PEA) 2 , there are two diastereomers of the 1:2 host-guest complexes; all the guest molecules adopt the "inside" binding mode. The first guest is stabilized by p-p interactions and intramolecular hydrogen bonds. Importantly, the structure reveals the second ligand is most likely stabilized by p-p interactions. Based on the structural data, DFT calculations further rationalized the CD results. These studies could also help us understand the binding interactions in other amide-linked porphyrin systems and build up new hosts for better chirality transfer abilities.

Disclosure statement
No potential conflict of interest was reported by the authors.