Tris( pyrazolyl)phosphines with copper( I ): from monomers to polymers †

The parent tris(pyrazolyl)phosphine and its 3,5-Me 2 , 3-Ph, and 3- t -Bu derivatives have been prepared by a simple procedure and show modest Lewis basicity of the phosphorus apex as was established by the magnitude of the 1 J P,Se coupling constant of the phosphine selenides. Because of the chelating properties of both the N- and P-sites, neutral phosphorus-centered scorpion ligands allow coordination modes that are unavailable to the abundantly used anionic tris(pyrazolyl)borate scorpionates as we established for Cu( I )-complexation. The substituted P-scorpion ligands only allow for N-coordination, as the P-apex is presumably less accessible. Two X-ray crystal structures were obtained for the Cu-complex of tris(3,5-di-methylpyrazolyl)-phosphine with acetonitrile and triphenylphosphine in the fourth coordination site. The parent P-scorpion ligand can chelate with both its pyrazolyl groups and its P-apex with the product depending on the ratio in which it is mixed with the Cu( I ) complex. Reacting two equivalents of the ligand with [Cu(MeCN) 4 ][PF 6 ] resulted in a complex in which Cu is coordinated to the three pyrazolyl groups of one ligand and to the P-apex of the other ligand as con ﬁ rmed by an X-ray crystal structure determination and a DFT computational analysis. Reacting the ligand and the Cu( I ) complex in an equimolar ratio resulted in a remarkable one-dimensional P-scorpion coordination polymer for which a single crystal X-ray structure could be determined. A detailed analysis of the structural features is presented.


Introduction
Tris( pyrazolyl)borates (A, Fig. 1) were first introduced by Trofimenko 1 in 1966 and became one of the most widely applied polydentate anionic ligands in coordination chemistry and catalysis because of their versatility and stability. [2][3][4][5] Neutral ligands result from changing the boron apex to a carbon one as in tris( pyrazolyl)methane (B) 6 for which ample syntheses have been reported. 7 The neutral phosphoruscentered analogues C and their oxides OC were introduced in the mid-1970s, but hardly used. [8][9][10] Tolman's group explored chiral versions of OC ligands for asymmetric catalysis, [11][12][13][14] while Weigand's group applied tris(3,5-dimethylpyrazolyl)phosphine (C Me2 ) as a synthon for cationic phosphorus compounds. [15][16][17][18][19] Recently, we reported a simple method to synthesize the PO-centered OC ligands and studied their coordination chemistry. 20 Tris( pyrazolyl)phosphines have multiple coordination sites, making them well suited for ligation with more than one metallic complex. Recently, Hazari and coworkers showed C coordinating to Mg in a κ 3 fashion, leaving the phosphorus apex available for additional coordination. 21 We were able to use this coordinating ability in a study on the related tris- (triazolyl)phosphine D to form the bimetallic complex (OC) 5 WDMo(CO) 3 (Fig. 2). 22 We also showed that the two oxidized ligands OC and OD influence the coordinated metal similarly, 23 but OC has more substitution options because of the position of the carbon atoms in its heterocyclic rings.
Herein we present the synthesis of tris( pyrazolyl)phosphines C, their coordination with copper(I), and their Janus property to generate a one-dimensional coordination polymer.

Ligand synthesis
The known tris( pyrazolyl)phosphine C H and tris(dimethylpyrazolyl)phosphine C Me2 , 8,9 and the new, more congested tris(3-phenylpyrazolyl)phosphine C Ph and tris(3-tert-butylpyr-azolyl)phosphine C t-Bu were synthesized in modest to excellent yields using our recently reported protocol (Scheme 1). 20 The products were obtained by slow addition of phosphorus trichloride to a mixture of three equivalents of the appropriate pyrazole and a slight excess of triethylamine in THF cooled to 0°C, followed by prolonged stirring, either at room temperature for C H and C Me2 or under reflux for the bulkier C Ph and C t-Bu . The reaction progress was monitored by 31 P NMR spectroscopy using the characteristic singlet for the phosphorus apex of the products (δ 61.1 (C H ), 72.0 (C Me2 ), 60.2 (C Ph ), and 58.2 (C t-Bu )). After filtering off by-products and removal of the volatiles, colorless to yellow solids were obtained for which additional purification was only needed for C H (recrystallization) and C Ph (washings), reducing their isolated yields significantly. The 1 H and 13 C NMR spectra revealed simple signal patterns, reflecting the expected equivalence of the three pyrazolyl groups; the 13 C{ 1 H} NMR spectra showed the characteristic doublets ( J C,P ∼ 10 Hz) for the C 3 and C 5 carbons.
Relative Lewis basicities of P-centered systems, and thus their ligating ability, can be examined by means of their phosphine selenides since the magnitude of the 1 J P,Se coupling constant is inversely related to the σ donating character of the free phosphine. [42][43][44] To examine the P-donor capacity of compounds C x , we have investigated selenides SeC H and SeC Me2 (Scheme 2). Both selenides were obtained as yellow powders after reaction of the parent compounds with an excess of selenium for two to three days in refluxing toluene, followed by filtration over silica, evaporation of the volatiles, and washing with pentane. Unsubstituted SeC H (∼24%) showed a resonance at δ( 31 P) 37.7 ppm with a 1 J P,Se coupling constant of 1014 Hz with the corresponding values for SeC Me2 (∼88%) of 40.7 ppm and 872 Hz (see ESI p. S-17 and S-19 † respectively). The latter 1 J P,Se coupling is at the high end of those reported for selenophosphoramides, indicating C Me2 to be a weaker donor than, e.g., P(NMe 2 ) 3 (cf., 1 J P,Se = 784 Hz for the corresponding selenide). 45 Unsubstituted C H turns out to be a still weaker donor that compares better with weakly donating phosphonates (cf., 1 J P,Se = 1025 Hz for SeP(OPh) 3 ). 46 The weak σ donating nature of the P apices of C H and C Me2 is also reflected by the harsh conditions required for the formation of their selenides.

Complexation to copper(I)
For tris( pyrazolyl)phosphines (C) to be suitable Janus-type ligands for generating one-dimensional coordination polymers they must be able to ligate in a head-to-tail fashion with Cu(I). Hence, their N-and P-donor sites must have complementary affinities for the copper complex. Earlier, it has been shown that the three pyrazolyl groups of carbon-centered B t-Bu and phosphine oxide-centered OC Ph,Me and OC t-Bu coordinate with Cu(I) with acetonitrile completing the coordination sphere. 20,47 For OC Me2 we have shown that the coordination can also be completed with a phosphorus ligand such as triphenylphosphine. 20 However, the donor ability of PPh 3 differs significantly from that of C (e.g., 1 J P,Se = 735 Hz for SeP(Ph) 3 48 6 ] for one C Me2 . Apparently, the phosphorus apex of the second C Me2 is unable to replace the remaining acetonitrile. Reacting C Me2 and the Cucomplex in an equimolar ratio gave the same product in 67.7% isolated yield. In the crystal, the metal complex is located at a general position and has only a very approximate non-crystallographic C 3v symmetry. The molecular structure shows the C Me2 ligand to be bound in a tridentate fashion via its pyrazolyl rings with Cu-N bond lengths ranging from 2.071(3) to 2.080(3) Å and N-Cu-N angles ranging from 88.49 (11) to 94.52(10)°. The acetonitrile is linearly coordinated to copper with a Cu-N14 bond length of 1.888(3) Å that is typical for tris( pyrazolyl)acetonitrile Cu(I)-complexes. 49 The 1 H and 13 C NMR data are also similar to those of the comparable acetonitrile containing Cu-complexes of B t-Bu , OC Ph,Me and OC t-Bu , but the acetonitrile ligand could not be observed due to a rapid exchange with the CD 3 CN solvent; the solid material was poorly soluble in common NMR solvents. It is interesting to note that, while the 13 6 ] could be exchanged for PPh 3 upon stirring an equimolar mixture in CH 2 Cl 2 for one hour. Following workup, the product (81%) showed NMR spectra with features similar to its precursor, but with additional signals in the aromatic region of both 1 H and 13 C NMR spectra. The 31 P NMR spectrum showed in addition to the resonance for the C Me2 apex at δ 21.4 ppm one at δ 6.5 ppm for the new PPh 3 ligand. Crystals obtained from CH 2 Cl 2 /pentane were suitable for a crystal structure determination by X-ray diffraction. The asymmetric unit showed two independent complexes with two CH 2 Cl 2 molecules. Fig. 4 6 ] complex. Its molecular structure confirms C Me2 as a tridentate ligand with PPh 3 completing the distorted tetrahedral coordination around copper. The Cu-N bond lengths range from 2.079 (2) 6 ] in 56 and 43% isolated yields, respectively. Their 1 H and 13 C NMR spectra confirmed a 1 : 1 ratio of acetonitrile and the scorpion ligand. For both complexes the 31 P NMR resonance of the P-apex is shielded by about 15 ppm compared to the free ligands [ which is much less than 49 ppm observed for the dimethyl derivative C Me2 . This difference is likely due to the methyl groups at the 5-position of C Me2 , which shield the P-apex upon κ 3 -N 3 complexation (Fig. 4).
Next, we turned to tris( pyrazolyl)phosphine C H , because its substituted derivatives are apparently too congested around the phosphorus apex for Cu(I) complexation. Reacting C H and [Cu(NCMe) 4 ][PF 6 ] in a 2 : 1 ratio in CH 2 Cl 2 also gave a colorless solid, isolated in 88% yield, but its spectroscopic properties showed it to differ substantially from the Cu-complexes of the other C ligands. Whereas the 1 H and 31 P NMR spectra recorded in CD 2 Cl 2 at ambient temperature gave only very broad signals that could not be interpreted, 1 H and 13 C NMR spectra at 201 K both showed two sets of signals for the pyrazolyl groups in a 1 : 1 ratio (see ESI p. S-30 and 31 †). The 31 P NMR spectrum at the same temperature demonstrated the different effects Cu coordination has on both sides of the C H ligand. It displayed two singlets for the complex, and a septet at δ −144.7 ppm ( 1 J P,F = 712 Hz) for the PF 6 − anion. The phosphorus at δ 39.0 is shielded compared to the apex of the free ligand (δ 61.1 ppm) due to adoption of the paddlewheel conformation required for κ 3 -N 3 coordination. The other phosphorus, at δ 62.0 ppm, is deshielded, as is common for coordinating phosphines. The small change in the chemical shift reflects the modest interaction of this weakly donating phosphorus with the Cu(I) center. The connectivity for this dimeric Cu-complex [(C H ) 2 Cu][PF 6 ] could be confirmed by a single crystal X-ray structure determination at 110 K. The crystal structure contains two independent metal complex molecules. One metal complex is located at an exact, crystallographic threefold axis, the other molecule is located at a general position with an approximate, non-crystallographic C 3 symmetry (r.m.s. deviation 0.149 Å, see Fig. 5). Its molecular structure confirms κ 3 -coordination of the three pyrazolyl groups of one C H ligand to Cu(I), forming Cu-N bonds and N-Cu-N angles that range from 2.035 (3) 6 ], repetitively recorded solid state 13 C and 31 P NMR data at 297 K are not. Only one singlet is observed at δ 40.3 ppm for the two phosphorus apices and the 13 C NMR spectrum suggests the presence of two sets of signals for the pyrazolyl rings in a 2 : 1 ratio. This may suggest that both C H ligands are bound to Cu(I) with two pyrazolyl groups, leaving the third one uncoordinated. Density functional calculations at B3PW91/6-31G(d) (LANL2DZ for Cu) showed the C 2 symmetric form of the [(C H ) 2 Cu] cation to be only 2.8 kcal mol −1 less stable than the C 3 isomer (Fig. 6). 50 This energy difference may even be less as the C 2 form can be stabilized by π-π stacking in the solid state as has been demonstrated for Ag(I) complexes of tris( pyrazolyl)methane ligands B. 25,26 Since both isomers are so close in energy, it may well be that isomerisation has occurred due to the large difference in the measurement temperature (187 K) or due to loss of the co-crystallized solvent during solid state NMR sample preparation. The geometry of the calculated κ 3 ,κ 1 -N 3 P isomer corresponds reasonably well with the structures found in the X-ray crystal structure, but the gas phase calculations overestimate all bond lengths around the Cu center; the calculated Cu-N and Cu-P bonds are 2.115 Å and 2.204 Å, respectively, whereas in the crystal structure Cu-N bonds of 2.035(3)-2.051(3) Å and Cu-P bonds of 2.074(2)-2.0938(11) Å are found. It seems fair to conclude that the Cu-P interaction is readily disturbed, which concurs with the low P-donor ability of C H and its inability to form the substituted analogues

Coordination polymer
With the promising [(C H ) 2 6 ]) n turned out to be rather difficult. A needle-shaped crystal obtained from dichloromethane was cracked into two fragments. There is severe disorder in the PF 6 anions and co-crystallized solvent molecules. More importantly, we detected "whole molecule" disorder in the one-dimensional Cu coordination polymer. It appears that a chain running in the uvw = [1, −1, 1] direction has very similar packing properties as a chain running in the opposite direction. Least-squares refinement of this crystal structure in the triclinic space group P1 (no. 1) could only be handled by introducing a large number of geometry restraints. Details are given in the ESI. † A first crystal obtained from 1,2-dichloroethane had a trigonal symmetry and was merohedrally twinned (see the ESI †). Fortunately we were able to obtain a non-twinned crystal with the same unit cell parameters. The crystal structure contains two independent polymeric Cu coordination chains which are located on threefold axes, respectively. Both polymers show the "whole molecule" disorder described above. In one of the chains this disorder could be ignored because of rather low residual electron densities (Fig. 7). In the other chain, a disorder model could be found and refined (Fig. 8). Because of the disorder, it was not possible to obtain reliable bond distances and angles, but the coordination mode could be proofed unambiguously. The two independent one-dimen-sional coordination polymers are oriented in the direction of the crystallographic c-axis. Both chains have two monomeric units in the crystallographic unit cell, respectively. Consequently, the axis length of c = 10.6155(4) Å leads to an average Cu⋯Cu distance of c/2 = 5.3078(2) Å. The polymeric nature of the structure is also reflected in the shape of the crystal. The c-direction corresponds to the long dimension of the needle shaped crystal (Fig. S4 in the ESI †).
The fact that tris( pyrazolyl)phosphine C H but not C Me2 gives access to a Cu(I)-coordination polymer highlights that even modest steric factors can be inhibiting.

Conclusions
We have demonstrated the different binding modes of tris( pyrazolyl)phosphine C. It can bind metals as a tridentate nitrogen donor and as a monodentate phosphine donor. Both binding modes of this Janus-type ligand can be combined when choosing the appropriate steric requirements, making it a suitable building block for coordination polymers. The fact that ligands C are readily available and that a great variety of substituted pyrazoles is already known 3,4 make tris( pyrazolyl)phosphines very promising candidates for further studies.

General procedures
All experiments were performed under an atmosphere of dry nitrogen. Solvents were purified, dried, and degassed by standard techniques. 3-Phenyl-1H-pyrazole, 52 3-tert-butyl-1Hpyrazole, 52 (0.0 ppm). Most coupling constants in the 1 H NMR spectra were determined after applying line narrowing. Solid-state CP-MAS NMR spectra were recorded on a Bruker Avance 400 spectrometer ( 31 P: 161.9 MHz) equipped with a 4 mm MAS probe. Solid samples were prepared in standard ZrO 2 rotors. MAS experiments were carried out using spinning speeds between 6 and 15 kHz at 297 K. Cross polarization was applied using a ramp-shaped contact pulse and a mixing time between 3 and 5 ms. 31 P chemical shifts were referenced to 85% H 3 PO 4 as the external standard (Ξ = 40.480747 MHz). The given values for 1 J (P, Cu) refer to the averaged value of couplings to the two isotopes 63 Cu and 65 Cu if not specified otherwise. High-resolution electrospray ionization mass spectrometry (HR ESI-MS) was performed using a Bruker MicroTOFQ, with ESI in positive mode (capillary voltage 4.5 kV). Melting points were determined on a Stuart Scientific SMP3 melting point apparatus using sealed capillaries. Elemental analyses were performed at the Microanalytical Laboratory of the Laboratorium für Organische Chemie, ETH Zürich, Switzerland.
Tris( pyrazolyl)phosphine (C H ). 8,9 A solution of phosphorus trichloride (5.65 g, 41.1 mmol) in THF (40 mL) was added dropwise over 50 min to a stirred solution of pyrazole (8.4 g, 123 mmol) and triethyl amine (17.5 mL, 126 mmol) in THF (200 mL) cooled to 0°C. Immediately a colorless solid (Et 3 N·HCl) started to precipitate. After the addition was complete, the reaction mixture was allowed to attain room temperature while stirring was continued for 24 h. The colorless solids were removed from the reaction mixture by filtration over a cannula fitted with a glass wool filter and the residue was washed twice with THF (20 mL). After concentrating the combined light yellow filtrates to about half the volume, they were cannula filtered again. Taking the resulting filtrates to dryness yielded a light yellow solid (8.17 g, 85.6%). According to 1 H NMR spectroscopy (CDCl 3 ) this solid contained some impurities (less than 2 mol%). Therefore, the solid was recrystallized from CH 2 Cl 2 (19 mL) at −20°C. After removal of the mother liquor, the light yellow crystals were dried in vacuo yielding C H Tris(3,5-dimethylpyrazolyl)phosphine (C Me2 ). 8,9 In a similar procedure as reported for tris( pyrazolyl)phosphine (C H ), phosphorus trichloride (0.9 mL, 10.7 mmol), 3,5-dimethyl pyrazole (3.08 g, 32 mmol) and triethylamine (4.6 mL, 33 mmol) were reacted in a total of 50 mL of THF. The time to complete the reaction was 1 h at room temperature. All volatiles were evaporated from the obtained THF extracts to yield a colorless oil. To remove the residual solvent, pentane (10 mL) was added, which was then removed in vacuo to co-evaporate the last traces of THF. This was done twice to afford C Me2 as a pale yellow solid (3. Tris(3-phenylpyrazol-1-yl)phosphine (C Ph ). In a similar procedure as reported for tris( pyrazolyl)phosphine (C H ), phosphorus trichloride (1.6 g, 11.7 mmol), 3-phenyl-1H-pyrazole (5.05 g, 35.0 mmol) and triethylamine (4.9 mL, 35.3 mmol) were reacted in a total of 50 mL of THF. The time to complete the reaction was 16 h at reflux temperature. All volatiles were evaporated from the obtained THF extracts and the resulting solid was treated with pentane (10 mL) to co-evaporate any remaining THF. 1 H-NMR spectroscopy showed that some 3-phenyl-1H-pyrazole was still present, therefore the solid was washed with diethyl ether. After all volatiles were removed, C Ph Tris(3-tert-butylpyrazol-1-yl)phosphine (C t-Bu ). In a similar procedure as reported for tris( pyrazolyl)phosphine (C H ), phosphorus trichloride (1.46 g, 10.6 mmol), 3-tert-butyl-1H-pyrazole (3.96 g, 31.9 mmol) and triethylamine (4.6 mL, 33 mmol) were reacted in a total of 50 mL of THF. The time to complete the reaction was 16 h at room temperature and 2 h at reflux temperature. All volatiles were evaporated from the obtained THF extracts and the resulting solid was treated with pentane (10 mL) to co-evaporate any remaining THF. Tris( pyrazolyl)phosphine selenide (SeC H ). C H (231 mg, 0.995 mmol) and elemental selenium (232 mg, 2.94 mmol) were stirred in toluene (7 mL) for 38 h at 120°C in a closed Schlenk tube. After cooling down, the reaction mixture was filtered by cannula filtration. All volatiles were removed from the filtrate, yielding 96.5 mg of a brown solid. This solid was a mixture of mainly SeC H and C H (1 : 0.3) according to 1 13  Tris(3,5-dimethylpyrazolyl)phosphine selenide (SeC Me2 ). C Me2 (400 mg, 1.26 mmol) and elemental selenium (2 g, 25 mmol) were stirred in toluene (20 mL) for 3 days at 120°C in a closed Schlenk tube. After cooling down, the reaction mixture was filtered over silica and the residue was washed with toluene until the washings were colorless. After evaporation of the volatiles, the remaining powder was washed with pentane (50 mL) and dried in vacuo to yield 440 mg of a yellow powder, which was not pure according to 31   After stirring overnight, a large amount of solid had precipitated, which was collected by filtration and was washed with pentane. No 31 P NMR signal was observed in the supernatant. Further drying led to the isolation of a colorless powder (171 mg). Vapor diffusion of pentane in a saturated solution of the title compound in CH 2 Cl 2 afforded crystals suitable for an X-ray structure determination, which showed the crystals to be those of [ Hz, 3H; PzH-4); signals of the coordinated MeCN ligand were not observed due to exchange with the solvent. 13  (sept, 1 J P,F = 706.3 Hz; PF 6 ), 23.3 (s; P(Pz)). 31     Bis(tris( pyrazolyl)phosphine)copper(I) hexafluorophosphate ([(C H ) 2