1,2,4-Triphospholyl anions – versatile building blocks for the formation of 1D, 2D and 3D assemblies† †Dedicated to Professor Ekkehardt Hahn on the occasion of his 60th birthday. ‡ ‡Electronic supplementary information (ESI) available: Experimental part, crystallographic data and additional figures.

The potential of K[P3C2R2] (R = tBu, Mes) as building blocks in metallo-supramolecular chemistry was investigated and self-assembly processes with Cu(i) halides resulted in the formation of a large variety of unprecedented one-, two- and even three-dimensional aggregates.


Experimental Part
All reactions were performed under an inert atmosphere of dry nitrogen or argon with standard vacuum, Schlenk, and glove-box techniques. Solvents were purified, dried and degassed prior to use by standard procedures. [K(P3C2Mes2)] 1 and [K(P3C2 t Bu2)] 2 were synthesized following reported procedures. Commercially available chemicals were used without further purification. Solution NMR spectra were recorded on either Bruker Avance 300 or 400 spectrometer. The corresponding ESI-MS spectra were acquired on a ThermoQuest Finnigan MAT TSQ 7000 mass spectrometer, while elemental analyses were performed on a Vario EL III apparatus.

Synthesis of 2
In a Schlenk tube a solution of [K(P3C2Mes2)] (66 mg, 0.17 mmol) in dme (10 mL) is layered with a solution of CuCl (50 mg, 0.5 mmol) in CH3CN (10 mL). At the phase boundary a color change to deep red can be observed. After complete diffusion and precipitation of a beige powder the red solution is decanted. While storing at 8°C the formation of red rods of 2 can be observed within a few days. The mother liquor is decanted, the crystals are washed with hexane (3 x 5 mL) and dried in vacuo. By concentrating the mother liquor and layering with Et2O, a second crop of crystals can be obtained.

Synthesis of 3
In a Schlenk tube a solution of [K(P3C2Mes2)] (50 mg, 0.12 mmol) in dme (9 mL) is first layered with a solvent mixture of dme and CH3CN (1:1; 3 mL), afterwards with a solution of CuCl (51 mg, 0.52 mmol) in CH3CN (12 mL). After complete diffusion and precipitation of a beige powder the red solution is filtered and concentrated to 5 mL. While storing at -28°C the formation of yellow prisms of 3 can be observed within a few days. The mother liquor is decanted, the crystals are washed with hexane (3 x 5 mL) and dried in vacuo. By concentrating the mother liquor, a second crop of crystals can be obtained.

Synthesis of 4
In a Schlenk tube a solution of [K(P3C2Mes2)] (50 mg, 0.12 mmol) in dme (9 mL) is first layered with a solvent mixture of dme and CH3CN (1:1; 3 mL), afterwards with a solution of CuBr (93 mg, 0.52 mmol) in CH3CN (12 mL). After complete diffusion and precipitation of a beige powder the red solution is stored at -28°C. Within one day the formation of yellow-orange blocks of 4 can be observed. The mother liquor is decanted, the crystals are washed with hexane (3 x 5 mL) and thf (2 x 8 mL) and dried in vacuo. By concentrating the mother liquor and storing it at -28°C, a second crop of crystals can be obtained. 4 can also be synthesized by using thf instead of dme, however due to the lower solubility of [K(P3C2Mes2)] in thf a larger volume is needed (20 mL).

Synthesis of 5
In a Schlenk tube a solution of [K(P3C2Mes2)] (16 mg, 0.042 mmol) in dme (5 mL) is layered with a solution of CuI (25 mg, 0.13 mmol) in CH3CN (5 mL). After complete diffusion the red solution is concentrated to 5 mL and layered with hexane. The formation of small red-brown blocks of 5 at the phase boundary could be observed once. The mother liquor was decanted, the crystals were washed with hexane (3 x 5 mL) and dried in vacuo. Attempts to reproduce 5 failed every time.

Synthesis of 8
In a Schlenk tube a solution of [K(P3C2 t Bu2)] (30 mg, 0.11 mmol) in dme (15 mL) is layered with a solution of CuCl (55 mg, 0.55 mmol) in CH3CN (10 mL). Already after several hours the formation of big dark red blocks of 8 at the red phase boundary is observed. After complete diffusion the mother liquor is decanted, the crystals are washed with hexane (3 x 5 mL) and dried in vacuo.

Synthesis of 9
In a Schlenk tube a solution of [K(P3C2 t Bu2)] (30 mg, 0.11 mmol) in dme (15 mL) is layered with a solution of CuBr (80 mg, 0.55 mmol) in CH3CN (10 mL). Already after several hours the formation of big dark red blocks of 9 at the red phase boundary is observed. After complete diffusion the mother liquor is decanted, the crystals are washed with hexane (3 x 5 mL) and dried in vacuo.
Analytical data of 9:

Synthesis of 10
In a Schlenk tube [K(P3C2 t Bu2)] (50 mg, 0.18 mmol) and CuI (105 mg, 0.55 mmol) are dissolved in thf (12 mL). An immediate colour change to deep red can be observed. The solution is stirred for 2 hours, before the solvent is removed in vacuo. The red solid is dissolved in as less CH3CN as possible and layered onto toluene. After complete diffusion the mother liquor is almost colourless and dark red blocks of 10 have formed. The mother liquor is decanted, the crystals are washed with toluene (3 x 5 mL) and dried in vacuo. Attempts to reproduce 10 were successful only once.
Analytical data of 10:

X-ray Structure Analysis
The compounds including co-crystallized solvent molecules characterized with X-ray structure analysis are Crystals of 2-10 were taken from a Schlenk tube under a stream of argon and covered with mineral oil. The single crystal was taken to the pre-centered goniometer head with CryoMount ® and directly attached to the diffractometer into a stream of cold nitrogen. The data for 6 were collected on an Agilent Technologies diffractometer equipped with Titan S2 CCD detector and a SuperNova CuK microfocus source using 1 ω scans. The data for 2, 3, 4, 5, 9 were collected using 1 ω scans and for 7, 8 and 10 using 0.5 ω scans on an Agilent Technologies diffractometer equipped with Atlas CCD detector and a SuperNova CuK microfocus source. All measurements were performed at 123 K. Crystallographic data and details of the diffraction experiments are given in Table S 1-Table S 3. The structures of 2-8, 10 were solved by direct methods with SIR97, 4 SHELX97 or SHELX2013. 5 The structures were refined by fullmatrix least-squares method against F  2 in anisotropic approximation using either SHELXL97 or the multiprocessor and variable memory version SHELXL2013. All non-hydrogen atoms were refined anisotropically, while the hydrogen atoms were refined riding on pivot atoms.

5:
The diffraction pattern of crystal 5 was featured by diffuse scattering (see Section 4) that is described below.

Space Group Determination and Refinement
In the tetragonal I‾ 42d space group (best R(sym)=0.03, current model) the quality factors are R1=0.095, wR2=0.248 (for observed reflections), GooF=1.043 providing that racemic twinning batch refined to 0.20 (10). In this space group no other twinning by merohedry is possible. All atoms have high displacement parameters, and the splitting of their positions does not improve the geometry of the complex.
The attempts to find another better solution failed with both SHEXL and SIR2014. The attempts to solve the structure in I41md did not return a more reasonable model. It is also in contradiction with the refined twinning batch. The attempts to solve the structure in the t-subgroup I‾ 4 and I212121 with corresponding twinning models did not yield better quality factors.
In addition, a second X-ray diffraction experiment of crystals of 2, crystallized from a different sample, shows the same phenomenon as described above.

Diffuse scattering in 5
The diffraction pattern of 5 shows quite strong diffuse scattering (Fig. S 13) that is visible even during the routine diffraction experiment. The diffraction pattern in the reciprocal space reconstructed with CrysAlisPro (Agilent Technologies) software shows that in addition to the Braggs peaks, two types of diffuse spots appear.

Fig. S 13:
Typical frame in the diffraction experiment of 5 (correlated 1° ω-scan with total exposure of 20 sec). The black round dots correspond to the Bragg peaks.
The diffuse spots of the first type lie on the planes perpendicular to c* with the l = integer, near to the points where h and k values are both half-integer. If h or k is integer, the diffuse scattering in the planes is not observed. It is also true for the area near to the systematically absent Bragg reflections with hk0, h+k=2n+1 (due to the presence of the glide plane nc). The spots of this type have quite a complex shape with two close maximums (Fig. S 10). The distribution of the diffuse scattering intensity is quite narrow but noticeably wider than those for the Bragg peaks (Fig. S 14c-e). The spots of the first types are observed up to high l values (Fig. S 15). The spots of the second type are cross-shaped and lie on the planes perpendicular to c*, but on the contrary, with the half-integer l and near to the points with integer h and k values (Fig. S 16). These spots are observed for the whole range -10.5  l  7.5. The 'arms' of the 'crosses' are directed along a*±b* diagonals. The reconstruction of the detailed profile for these spots appeared to be impossible due to the artefacts resulting from the wide ω-scans taken during a routine diffraction experiment. The symmetry of the diffuse scattering spots of both types is in a good agreement with the symmetry of the Bragg diffraction (point group 4/m).
The diffuse scattering obviously originates from the disorder of the polymeric chains and/or {Cu5I4(MeCN)4(P3C2Mes2)} repeating units by 90° rotation around the 4-fold axis parallel to c. At that, each orientation of the {Cu5I4(MeCN)4(P3C2Mes2)} repeating unit may require a preferable orientation either of the units of the neighboring chains or within the single chain due to the sterical hindering between the bulky Mes groups. Therefore, there is a correlation between the neighboring orientations despite their equal probability in the averaged crystal structure imposed by the 4-fold axis.
The correct reduction of diffuse intensities for the modelling of the real crystal would require special diffraction techniques (using a strong monochromatic radiation source, noise-free detector etc.). The data from a routine structural determination we possess are obviously not enough to make final conclusions. Nevertheless, preliminary consideration shows that we can propose two possible disorder models. One model assumes coexistence of differently oriented chains, but the orientation of all repeating units in every chain is identical. Another more complicated model assumes that the repeating units of the same chain can be rotated around the Cu-P bond in respect to each other. The results of preliminary calculations prove that only the latter model, sophisticated with local relaxation of atomic groups, explains the presence of both types of the diffuse scattering spots. Qualitatively the best result corresponds to the models where the orientation of the closest repeating units alternate both within the polymeric chain and, less strictly, between them.