Introduction

Porous supramolecular metal–organic structures are prepared by treatment of multidentate organic compounds (ligands) with metal ions. The multidentate organic ligands have specific geometries which determine the shape of the resultant polygons. For example, two bis(terpyridine) ligands are shown in Fig. 1. When ligand I, with a bond angle of 60° between its terpyridine moieties, was treated with Ru(II) and Fe(II), the expected molecular triangles [Fe3(I)3]6+ and [Ru3(I)3]6+ were isolated [1]; similarly, a molecular hexagon, [Ru6(II)6]12+, was obtained when II (120°) was treated with Ru(III) [2].

Fig. 1
figure 1

Two bis(terpyridine) ligands: I reacts with metal ions to give molecular triangles, whereas II yields molecular hexagons [1, 2]

Full size image

Stang and Olenyuk developed a molecular architecture library that allows prediction of the metal–organic polygon that will form based on the geometry of the linker and the metal ion [3]. However, in some cases, polygons form with shapes other than those predicted from the ligands’ geometries. For example, reaction of Pd- and Ru-based “corners” with 4,4′-bpy yields triangles as well as the expected squares [4, 5]. Recent reviews have described progress in supramolecular systems based on multifunctional β-diketone ligands [6, 7]. In our previous work, m-phenylenebis(acetylacetone) ligands (Fig. 2a) reacted with Cu2+ to give exclusively molecular squares instead of the predicted molecular hexagons. The pores of the molecular squares were the right size to accommodate guest molecules such as 4,4′-bipyridine, C60, and C70 [8,9,10].

Fig. 2
figure 2

Structures of: a bis(β-diketones), m-pbaH2, m-pbprH2 and m-pbhxH2 [8, 10]; b triphenylamine and the new triphenylamine-based bis(β-diketone) ligands tpbaH2 (1) and tpbprH2 (2)

Full size image

To make larger molecular polygons, we prepared two new bis(β-diketones), triphenylaminebis(2,4-pentanedione), tpbaH2, 1; and triphenylaminebis(3,5-heptanedione), tpbprH2, 2 (see Fig. 2b), and treated them with Cu2+. We anticipated that the new ligands 1 and 2 would yield molecular squares with larger pores than our previous bis(β-diketones), which means that the squares may be able to accommodate larger guest molecules. Herein, we report syntheses and crystal structures of the ligands 1 and 2, and the conversion of 2 to its Cu complex 3, which has been characterized by ESI–MS and microanalysis.

Experimental

Materials and Methods

DMF was purchased from Sigma-Aldrich and distilled before use; phosphorus oxychloride and triphenylamine were acquired from Sigma-Aldrich and used as received. ESI–MS was performed on an ESI-TOF instrument (Agilent 6210). Elemental analysis was performed by M-H-W Labs, Phoenix, Arizona. NMR spectra were recorded on a Bruker 400 MHz spectrometer with CDCl3 as the solvent. 4,4′-Diformyltriphenylamine [11], 2,2,2-trimethoxy-4,5-dimethyl-1,3,2-dioxaphospholene [12] and 2,2,2-trimethoxy-4,5-diethyl-1,3,2-dioxaphospholene [8, 9] were prepared by literature methods.

Synthesis of Bis(β-Diketones)

These were prepared by mixing 4,4′-diformyltriphenylamine and a phospholene in 1:2 molar ratio in dichloromethane and stirring under N2 at room temperature. When the aldehyde was consumed (as judged by the disappearance of its –CHO peak in NMR), the crude product was isolated by removal of solvent and purified by column chromatography (ethyl acetate–hexane, 1:4 v/v). Crystals of 1 and 2 suitable for X-ray analysis were grown by slow evaporation of solutions in ethyl acetate–hexane.

TpbaH2 (1). 4,4′-Diformyltriphenylamine (1.5 g, 4.9 mmol); 2,2,2-trimethoxy-4,5-dimethyl-1,3,2-dioxaphospholene (2.06 g, 9.8 mmol); dichloromethane (10 mL); reaction complete after 12 days. Yield: white solid, 0.39 g (17%), mp 185–190 °C. 1H NMR: δ 16.67 (s, 2H, OH), 7.31 (t, 3H), 7.15 (d, 2H), 7.10, 7.04 (AB, 8H, aromatic CH), 1.96 (s, 12H). 13C NMR: δ 191.1, 146.8, 131.9, 131.0, 129.5, 125.0, 123.7, 123.6, 114.7, 24.2. Anal. Calcd. for C28H27NO4 (M = 441.52): C 76.17; H 6.16; N 3.17: Found: C 75.87; H 6.17; N 3.09.

TpbprH2 (2). 4,4′-Diformyltriphenylamine (1.2 g, 3.9 mmol); 2,2,2-trimethoxy-4,5-diethyl-1,3,2-dioxaphospholene (1.86 g, 7.8 mmol); dichloromethane (20 mL); complete after 12 days. Yield: white solid, 0.28 g (14%), mp 135–139 °C. 1H NMR: δ 16.72 (s, 2H, OH), 7.32 (t, 3H), 7.16 (d, 2H), 7.09, 7.03 (AB, 8H, aromatic CH), 2.21 (q, 8H), 1.07 (t, 12H). 13C NMR: δ 194.3, 146.8, 132.1, 130.4, 129.5, 125.2, 123.6, 123.5, 113.4, 29.9, 9.6. Anal. Calcd. for C32H35NO4 (M = 497.62): C 77.24; H 7.09; N 2.81: Found: C 77.45; H 6.90; N 2.85.

Synthesis of Copper Complexes

A solution of CuSO4·5H2O in deionized water was converted to [Cu(NH3)4]2+(aq) by treatment with conc. NH3(aq). To this solution was added the bis(β-diketone) (in 1:1 molar ratio with Cu2+) dissolved in dichloromethane.

Cun(tpba)n. [Cu(NH3)4]2+(aq) (from CuSO4·5H2O, 0.24 g, 0.90 mmol); water (50 mL); tpbaH2 (1) (0.30 g, 0.90 mmol) in dichloromethane (40 mL). Immediately a dark green insoluble powder precipitated. This material could not be characterized further, and we were unable to find a solvent that dissolved it without decomposition.

Cu4(tpbpr)4 (3). [Cu(NH3)4]2+(aq) (from CuSO4·5H2O, 0.10 g, 0.40 mmol); water (40 mL); tpbprH2 (2) (0.20 g, 0.40 mmol); dichloromethane (30 mL). The mixture was stirred for 3 h. The dark-green organic layer was dried over Na2SO4 and evaporated to yield a dark-green solid, 0.11 g (50%). Anal. Calcd. for C128H132Cu4N4O16 (M = 2236.62): C 68.74; H 5.95; N 2.50: Found: C 68.92; H 6.16; N 2.60. ESI–MS data for this complex are discussed in the main text.

Safety and Hazards

We experienced no unanticipated reactions or other obvious hazards in performing the syntheses and analyses described here. The phospholenes needed for the ligand syntheses are prepared from α-diketones and trimethyl phosphite, which are volatile and potentially hazardous.

X-ray Analyses

Intensity data were collected at low temperature on Bruker Kappa Apex-II CCD and Nonius KappaCCD diffractometers fitted with Oxford Cryostream chillers. Radiation was CuKα or MoKα, from fine-focus sealed tubes with graphite monochromators. Data reduction included absorption corrections by the multiscan method, with SADABS (for 1) [13] and DENZO (for 2) [14]. The structures were determined by direct methods and difference Fourier techniques and refined by full-matrix least squares, using SHELXL [15]. All non-hydrogen atoms were refined anisotropically. In the final model for tpbaH2 (1), each of the enol H atoms was disordered over two sites (H10H, occupancy 0.39(4); H20H, 0.61(4); H30H, 0.43(5), H40H, 0.57(5)). All other H atoms were placed in idealized positions. For additional details, see Table 1. Data for the structures can be obtained free of charge via https://www.ccdc.cam.ac.uk/structures/ 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.

Table 1 Crystal data and structure refinement for compounds 1 and 2
Full size table

Results and Discussion

Bis(β-Diketones) 1 and 2

Ramirez and co-workers [9, 12] discovered that benzaldehyde reacts with a phospholene to make a phospholane adduct, which then rearranges on refluxing in methanol to produce 3-phenyl-2,4-pentanedione (see Fig. 3).

Fig. 3
figure 3

Phospholene route for converting aromatic aldehydes into β-diketones

Full size image

We have prepared several β-diketones by this route. We found that formation of the final product did not always require heating in methanol: sometimes it proceeds at room temperature in a less polar solvent such as dichloromethane. In this work, two new bis(β-diketones) (1 and 2) are isolated as white solids after stirring 4,4′-diformyltriphenylamine [11] with 2,2,2-trimethoxy-4,5-dimethyl-1,3,2-dioxaphospholene [12] and 2,2,2-trimethoxy-4,5-diethyl-1,3,2-dioxaphospholene [8, 9] respectively (Scheme 1) for 12 days at room temperature under N2. These reactions are slower than that of 4-formamidobenzaldehyde (3 days) [16] and the yields are low (ca. 15%). This may be because of the large size of the 4,4′-diformyltriphenylamine molecule and the difficulty of its motion in the presence of the viscous phospholene.

Scheme 1
scheme 1

Reaction of 4,4′-diformyltriphenylamine with phospholenes affords bis(β-diketones) tpbaH2 (1) and tpbprH2 (2)

Full size image

The formation of β-diketones such as m-pbaH2 (Fig. 2) and phenylacetylacetone via phospholenes involves stirring with the aldehydes at room temperature. The reaction is judged to be complete by the disappearance of the aldehyde − CHO peak, which normally takes about 12 h, followed by refluxing the intermediate in methanol for ca. 3 h [8,9,10, 16,17,18]. It is during the refluxing in methanol step that the β-diketone enol peak appears at ca. 16 ppm. In contrast, in the synthesis of β-diketones 1 and 2, the reaction does not require refluxing in methanol, but it requires more stirring time, i.e. the aldehyde –CHO resonance at 9.90 ppm gradually became smaller as stirring continued and finally disappeared during a period of 12 days; at the same time, the enol peak of 1 (16.67 ppm) started appearing after several days and continued to increase in intensity until no further change was observed after 12 days [19, 20]. The yields were 17 and 14% for 1 and 2 respectively. Prior to this study, a researcher in our group prepared a thiophene-based bis(β-diketone) ligand by a similar procedure. This reaction required ~ 10 days, starting from 2,5-thiophenedicarboxaldehyde and phospholenes at room temperature; it also did not require refluxing in methanol. The yield was also low (~ 20%) in that case [19, 20].

Crystal Structures of tpbaH2 (1) and tpbprH2 (2)

These structures are shown in Figs. 4 and 5. In the solid state, both are in the enol form, which is in agreement with solution NMR data: the enol chemical shifts for 1 and 2 are at 16.67 and 16.72 ppm respectively, and there are no peaks corresponding to the keto form of the molecules.

Fig. 4
figure 4

Crystal structure of the bis(β-diketone) tpbaH2, 1. In this structure, and in the structure of tpbprH2 (2, Fig. 5), the phenyl and phenylene groups are not coplanar, but have a screw-like twist. (Both are shown with right-handed screw sense, often designated P, but the structures are centrosymmetric, so they contain equal numbers of right- and left-handed twists, P and M.) H10H-H40H are disordered hydrogen-bonded enol H atoms; H20H and H40H have majority occupancy. Ellipsoids are at 50% probability in this diagram and in Fig. 5

Full size image
Fig. 5
figure 5

Crystal structure of the bis(β-diketone) tpbprH2, 2. In this structure, N1, C14, C17, and H17 lie along a crystallographically imposed twofold axis. As a result, the two hydrogen-bonded enolic β-diketone moieties are equivalent (O2–H2O 0.999(19), H2O···O1 1.546(19) Å)

Full size image

The majority of β-diketone crystal structures in the Cambridge Structural Database [21] are in the enol form. A search revealed 2482 enol and 459 keto structures; of these, 779 enol and 269 keto structures were acyclic or otherwise chemically appropriate as ligands for metal complexes. Hansen [22] observed that the enol form tends to dominate in β-diketones that contain aromatic substituents, as our two new ligands do.

In the structure of tpbaH2 (1; Fig. 4), the molecule lies on a general position. The two β-diketone moieties are very similar. In each one, the enol H atom is disordered over two sites: H10H/H20H, occupancies 0.39(4)/0.61(4); and H30H/H40H, 0.43(5)/0.57(5). Also, of the two C–O distances in each β-diketone moiety, the one on the side with the higher H population is longer, by ca. 0.02 Å. This is as expected if the higher H occupancy reflects greater C–OH character. The C–C bond distances show a similar slight alternation, with the shorter C–C bond adjacent to the longer C–O bond. There are no unusual intermolecular contacts in the structure.

Molecules of tpbprH2 in crystals of 2 lie on crystallographic twofold axes (see Fig. 5), so there is only one β-diketone moiety in the asymmetric unit. Brock and Dunitz have observed that C2/c is a common space group for molecules that can have twofold symmetry [23]. There is no obvious disorder of the enol H atom: it was modeled with 100% occupancy, and none of the 10 highest peaks in the difference map was near the O atoms. Consistent with this localized enol structure, the differences in C–C and C–O distances are 0.075–0.08 Å, substantially larger than those in 1. This structure also shows no unusual intermolecular contacts.

Like other triphenylamines [24], tpbaH2 (1) and tpbprH2 (2) are nearly planar at N. Their aryl groups are not coplanar, but adopt a twisted or “propeller-shaped” conformation in the solid state (as sketched in Scheme 1). Both structures are centrosymmetric, with equal numbers of right-handed (often called P, for “plus”) and left-handed (M, “minus”) enantiomers. The “pitch” of the propeller can be defined by the angle each phenyl least-squares plane makes with the central NC3 plane. Those angles are 27.28 (C6–C11), 62.10 (C17–C22), and 33.61° (C23–C28) in tpbaH2 (1); and 36.35 (C8–C13, duplicated due to the two-fold symmetry of the molecule) and 43.10° (C14–C17) in tpbprH2 (2). These values fall into the most commonly observed range for twist angles in crystal structures of triphenylamine derivatives (see Fig. 6) [21].

Fig. 6
figure 6

Distribution of twist angles of the phenyl groups (compared to the central NC3 least-squares planes) in the crystal structures of triphenylamine derivatives. Data are from the Cambridge Structural Database [21]: 4241 examples in 2799 structures, for a total of 12,723 measured twist angles. In the search for triphenylamines, substitution at the phenyl 3, 4, and 5 positions was permitted; molecules substituted at the 2 or 6 positions were excluded, to minimize steric influence on conformation

Full size image

Reva et al. have reported a DFT calculation of the barrier for rotation of the phenyl groups in triphenylamine [25]. They estimate the barrier at ca. 20 kJ mol−1. This is in agreement with the fact that simple triphenylamines interconvert too rapidly to permit resolution into their enantiomeric forms (P and M) in solution. There are a few examples of highly hindered derivatives (e.g., tris(2,3,4,5,6-pentachlorophenyl)amine) in which the barrier is large enough to permit separation of the enantiomers [26]. Also, supramolecular helical structures have been prepared that contain many triphenylamine moieties stacked in π fashion [27, 28]. In these helical structures, rotation of individual phenyl groups is inhibited by those in neighboring molecules.

Copper Complexes

Reaction of Cu(NH3)42+(aq) with CH2Cl2 solutions of tpbaH2 (1) and tpbprH2 (2) produced dark green powders. Cun(tpba)n was insoluble in common solvents and could not be studied further. (Its low solubility may be due to a polymeric structure, or it may be molecular but too rigid to be readily soluble.) On the other hand, Cun(tpbpr)n (3), with the more flexible ethyl groups on its β-diketonate moieties, is soluble in chloroform and dichloromethane. Its solutions are olive-green in color, much like other supramolecular Cu(II) β-diketonates [8, 10], as well as simple molecular complexes such as Cu(acac)2. We were unable to obtain 3 in crystalline form for direct structural characterization. However, its microanalysis (C, H, N) agrees with calculated values for Cun(tpbpr)n. Also, we were able to analyze it by ESI–MS. A tetramer is seen, as indicated by the signals observed at 2234.7–2241.7 ([M + H]+) whose isotope distribution pattern matches the theoretical distribution (see Fig. 7). No signal corresponding to the pentamer Cu5(tpbpr)5 (2792–2802 amu) was observed, indicating that 3 is a pure compound and not a mixture of oligomers.

Fig. 7
figure 7

Calculated (top) and experimental (bottom) ESI mass spectra of parent ion of 3 ([M + H]+ (Cu4(tpbpr)4H+)). Calculated values of m/z for the three labeled peaks in the experimental spectrum are 2235.6908, 2237.6917, and 2239.6925 amu, respectively (average error 5.1 ppm)

Full size image

Like the smaller Cu β-diketonate molecular squares we have reported previously, Cu4(tpbpr)4 is expected to have an approximately square geometry, as illustrated by the model in Fig. 8. Coordinates for this model were obtained via HyperChem molecular modeling. Several conformations of the molecular square Cu4(tpbpr)4 (3) are possible, depending on the screw sense of the four bridging triphenylamine moieties. The Cu···Cu distance predicted by molecular modeling varies for different conformers, but overall it is expected to be 20–21 Å, or 6–7 Å larger than that in Cu4(m-pbpr)4. 3 did not form crystals in our experiments, possibly because it consists of a mixture of conformers with different triphenylamine screw senses.

Fig. 8
figure 8

Drawings of Cu β-diketonate molecular squares. a 3 (Cu4(tpbpr)4), as calculated by molecular modeling; b our previously reported molecular square Cu4(m-pbpr)4, from its crystal structure [8], shown at the same scale. The calculated trans Cu–-Cu distances in 3 are ca. 20.7 Å, compared to values of 14.012(1) and 14.661(1) Å in the crystal structure of Cu4(m-pbpr)4

Full size image

Solutions of 3 in CH2Cl2 change from green to turquoise on treatment with multifunctional N donors such as 4,4′-bipyridine, 1,2-bis(4-pyridyl)ethylene (bpe), and 1,2-bis(4-pyridyl)ethane (bpa). This color change is similar to what we have previously observed when diamines bind to our Cu molecular squares. We believe the same reaction is occurring here with 3; however, we were unable to isolate the adducts in pure form.

Conclusions

Two new bis(β-diketones) with triphenylamine-based bridging groups were prepared and their crystal structures determined. The propeller-like conformation of the triphenylamine moieties in the new compounds is similar to that observed in many other crystalline triphenylamine derivatives. One of the bis(β-diketones) forms a molecular square Cu4L4 on reaction with Cu2+, based on ESI–MS data. Molecular modeling indicates that the molecular square is ca. 21 Å in diameter, about 50% larger than our previously characterized squares based on m-phenylene groups. The new bis(β-diketones) offer a greater variety of opportunities for internal functionalization than our previous ligands, due to their two p-phenylene moieties. Such functionalization, and direct coordination to the metal atoms, should enable designed interactions with guest molecules.