Heterobimetallic Complexes

As the copper complex 33 demonstrated, metal exchange with monovalent metals is possible, too. It was now of interest to prepare complexes of the ligand {Ph2PCH2S(NSiMe3)2}^- and the heavier alkali metals to investigate the possible changes in coordination.

Compound 34 was prepared from reacting equimolar amounts of [(tmeda)Li{Ph2PCH2S(NSiMe3)2}] (11) and [Rb{N(SiMe3)2}] at -78 °C in pentane (Equation 5-6). Interestingly, the metal exchange was not complete. Only half an equivalent of lithium was exchanged with rubidium, resulting in the formation of a heterobimetallic dimeric complex with the molecular structure shown in Figure 5-12.

As could have been expected, the lithium cation coordinated to four nitrogen atoms is persistent in the complex. In contrast, the rubidium cation is complexed by two nitrogen atoms of the diimido moieties, each of the phosphorus side-arms and the TMEDA molecule which completes the coordination. Consequently, both alkali metals have different coordination geometries and environments. The structure is bisected by a mirror plane which includes both metal ions.

Figure 5-12: Molecular structure of [(tmeda)Rb{Ph2PCH2S(NSiMe3)2}2Li] (34). Hydrogen atoms are omitted for clarity.

Li1 has a distorted tetrahedral geometry with Li–N bond lengths between 2.053(2) Å (Li1–N1) and 2.262(2) Å (Li1–N2). The Li1–N2 bond is slightly elongated because N2 is also coordinating Rb1. The rubidium atom Rb1 is six fold coordinated in a severely distorted octahedral manner. It is interesting that one of the TMEDA molecules remains in the complex but has switched its coordination to rubidium. The Rb–N and Rb–P bonds are longer than 3.0 Å which is normal and can be found in other complexes like [Rb(thf)P(SiMe3)2]∞ [199], [RbP(H)(dmp)] ^[200] (dmp = 2,6-dimesitylphenyl) or [Rb{((Me3Si)2C)P(C6H4-2-CH2NMe2)2}]n [201]. Bond lengths and angles can be found in Table 5-6.

A CCDC search reveals that heterobimetallic complexes of lithium and rubidium are virtually unknown. There are only five examples of structurally characterised compounds of which three should be discussed. Mulvey et al. describe a heptalithium tetrarubidium mixed alkoxide peroxide wherein the clusters are linked by Rb–

(TMEDA)–Rb bridges.^[202] The reported mean N–Rb bond length of 3.197 Å is close to the value for Rb1–N3 (3.0458(16) Å). Another compound with the constitution [(tBuO)8Li4Rb4] contains neither nitrogen nor phosphorus atoms.^[203] This is also the case in the third complex where lithium and rubidium are bridged by oxygen donors.^[204]

Table 5-6: Selected bond lengths [Å] and angles [°] in 34 and 35

34 35 34 35

S1–N1 1.6024(15) 1.602(2) N1–S1–N2 103.69(8) 103.62(12) S1–N2 1.6076(15) 1.604(2) S1–C13–P1 113.78(9) 113.92(15) S1–C13 1.8272(18) 1.827(3) N1–Li1–N2 71.43(6) 71.07(9) P1–C13 1.8515(18) 1.853(3) N1–Li1–N1’ 143.1(3) 144.0(6) N1–Li1 2.053(2) 2.063(4) N2–Li1–N2’ 130.4(2) 128.2(5) N2–Li1 2.262(2) 2.263(5) N2–M1–P1 59.15(3) 60.43(5) N2–M1 3.0646(15) 2.934(2) N2–M1–P1’ 97.97(3) 101.35(5) N3–M1 3.0458(16) 2.936(3) N2–M1–N2’ 84.11(6) 87.87(9) P1–M1 3.5996(5) 3.5613(7) N3–M1–N3’ 59.03(6) 62.83(12) N1–Si1 1.7106(16) 1.712(2) P1–M1–P1’ 150.818(18) 156.23(4)

Interestingly, there are two signals in the ⁷Li{¹H} NMR spectrum of 34. The signal at 1.81 ppm can be associated with the Rb/Li heterobimetallic complex 34; the other one at 1.02 ppm seems to belong to the starting material [(tmeda)Li{Ph2PCH2S(NSiMe3)2] (11). This is certainly not due to contamination of the sample as crystals of 34 were dissolved for the NMR spectra and the sample was sealed airtight. In addition, the signal for the starting material is quite high. An impurity of such a high concentration should have been detected in the elemental analysis. However, this was not the case. When looking at the ³¹P{¹H} spectrum the presence of the starting material in the sample gets even more obvious. There is a small broad signal at -39 ppm, whereas the heterobimetallic complex 34 shows a signal at -33 ppm. The integration reveals a ratio of 1 to 0.2. All these analytical results indicate that the Rb/Li complex is not retained completely in solution. Part of it seems to loose the rubidium. However, this is impossible because the charges would not be balanced anymore. Thus, another possibility has to be taken into account. The whole system is flexible in solution like the other structures discussed so far.

Therefore, the TMEDA molecule is switching positions in solution and is also binding to the lithium cation. The shift of this new compound would then be very similar to 11.

In essence, 34 represents the first heterobimetallic lithium/rubidium complex with nitrogen and phosphorus donor atoms. The coordination of phosphorus in such complexes is unprecedented. The compound is soluble in polar and unpolar organic solvents which is a great advantage and is due to the ligand periphery. Complexes of

the heavier alkali metals usually tend to form larger aggregates which are poorly soluble. However, in the case of 34 this is averted by the ligand.

A reaction according to Equation 5-6 with [K{N(SiMe3)2}] yielded colourless crystals in the space group C2/c. The compound is the heterobimetallic lithium/potassium complex [(tmeda)K{Ph2PCH2S(NSiMe3)2}2Li] (35) with the same structural features as 34. The molecular structure is shown in Figure 5-13.

Figure 5-13: Molecular structure of [(tmeda)K{Ph2PCH2S(NSiMe3)2}2Li] (35). Hydrogen atoms are omitted for clarity.

Unfortunately, the crystals were of poor quality and there still was some electron density unaccounted for in the refined model. The structure is shown here nevertheless for comparison reasons. Bond lengths and angles are discussed to give a general idea of their magnitude and the difference to [(tmeda)Rb{Ph2PCH2S(NSiMe3)2}2Li] (34).

The Li–N bond lengths are 2.069(4) Å (Li1–N1) and 2.254(4) Å (Li1–N2). The K1–P1 bond of 3.5613(7) Å is almost the same as in the rubidium derivative (3.5996(5) Å).

This is probably due to the fact that the coordination of the lithium cation as a structural anchor already predetermines the position of the phosphorus side-arm.

The potassium-nitrogen bond lengths almost have the same value (K1–N2: 2.933(2) Å, K1–N3: 2.937(2) Å). The angles differ very little from the rubidium derivative. It is

striking though, that the P1–K1–P1' angle of 156.23(4)° is almost six degrees wider than in 34.

There are only a few examples of lithium/potassium heterobimetallic complexes. Several examples with butoxy ligands are supposed to be novel superbases; there are even Li/Na/K heterotrimetallic compounds.^[146,147]

Westerhausen et al. synthesised mixed phosphanediide/silanolate heterotrimetallic aggregates with Li/K/Sr or Li/K/Ba.^[205] The average K–P bond length of 3.365 Å is shorter than in 35 though the structural features are also not very alike.

In 1998, our group presented a lithium/potassium heterobimetallic complex based on the {S(NtBu)3}^2- ligand.^[206] The compound [(thf)2Li4K2(OtBu)2{(NtBu)3S}2] was the first example of such a complex with a triazasulfite dianion. The metal ions are sandwiched between two {S(NtBu)3}^2- caps with different coordination geometries. Thus, the arrangement is very different to 35 because there is no side-arm donation.

It would be interesting to see the results from metal exchange with [Na{N(SiMe3)2}] as it is not clear if this reaction would also be incomplete or if a monometallic compound could be obtained due to the smaller cation size.