Methoxycarbonylation of Methyl Oleate

With the new complexes 51-(OTf)2 to 57-(OTf)2 in hand, their catalytic properties in the isomerizing methoxycarbonylation of methyl oleate (MO) were studied (Scheme 25).

Scheme 25. Methoxycarbonylation of methyl oleate (MO) (59) to the linear diester (60) with the novel catalyst precursors 51-(OTf)2 to 57-(OTf)2.

A constant pressure of CO (20 bar), n(Pd) = 0.048 mmol, n(MO) = 6 mmol, 10 mL of methanol, 90 °C and a reaction time of 120 hours were chosen as experimental parameters to evaluate their catalytic performance. In order to study ‘real-life’ catalytic performances, technical grade Dakolub MB 9001 with a methyl oleate content of 92.5 % was used^[60] rather than highly purified methyl oleate (Table 5).

Table 5. Results of the isomerizing methoxycarbonylation of methyl oleate (Dakolub MB 9001, 92.5 % MO) using Pd(II) complexes 51-(OTf)2-57-(OTf)2, and the benchmark complex 58-(OTf)2.

Entry Pd(II) complex Conversion of

reaction conditions: n(Pd) = 0.048 mmol, 2 mL Dakolub MB 9001 (92.5 % MO), 10 mL of MeOH, 20 bar CO, 90 °C, 120 h.

Catalysts 51-(OTf)2 to 57-(OTf)2 convert methyl oleate into diesters in a broad range of 18-98 %.

Surprisingly, the selectivity towards the desired 1,19-diester is rather similar (70-80 %) for all catalysts studied, while the conversion varies strongly. Complex 56-(OTf)2 showed only very little conversion (1.3 %) of methyl oleate which is in line with the observation that no hydride formation

occurred upon addition of methanol to this complex (vide infra). The differences in the conversions of 51-(OTf)2 to 57-(OTf)2 can be related to the steric hindrance of the substituents at phosphorus.

To this end, we quantified the available space around the metal center from the available X-ray crystal structure data (vide supra). Note that the angles were calculated from the corresponding X-ray structures without considering the rotation of alkyl substituents around the P-C bond in solution. The Pd-P bond lengths were not normalized but taken as given by the structure determination. Larger values describe increasing available space around the metal center and thus a smaller steric constraint as imposed by the diphosphine ligand. In Figure 12, the conversion of methyl oleate and the selectivity towards the -diester are plotted versus the increasing steric constraint imposed by the diphosphine ligand. The increasing steric demand (56-(OTf)2 < 52-(OTf)2 < 53-(OTf)2 < 55-(OTf)2 < 54-(OTf)2 < 51-(OTf)2) goes along with an increased selective conversion. The most active unsymmetrical catalysts 51-(OTf)2 and 54-(OTf)2 (89° and 90.8°, respectively) have similar half-cone angles φ as 58-(OTf)2 (86.8°). Complex 57-(OTf)2 is not discussed or in Figure 12, although data in Table 5 suggest it is very active. Unfortunately, all grown crystals of 57-(OTf)2 were unsuitable for X-ray diffraction.

Figure 12. Plot of the calculated half cone angles φ versus the conversion of methyl oleate ( ) and selectivity towards the -diester ( ), respectively.

It is concluded from Figure 12 and Table 5 that the difference in the ligand substitution pattern has a greater influence on the catalytic productivity of the new complexes than on the selectivity towards the 1,19-diester. In contrast to the catalyst productivity, the catalyst selectivity does not correlate well with a greater steric demand of the alkyl groups on the phosphorus atoms. For example complex 51-(OTf)2 (entry 1, 94 %) is more productive than 54-(OTf)2 (entry 4, 77 %), as well as 55-(OTf)2 (entry 5, 22 %). However, the selectivities of complexes 54-(OTf)2 (entry 4, 79 %) and 55-(OTf)2 (entry 5, 80 %) are significantly higher than for complex 51-(OTf)2 (entry 1, 65 %). It might be possible that below a certain threshold steric congestion induces phosphine labilisation leading to catalyst instability. Independently from the steric bulk and consequently from the half cone angle of the complexes, the one with an adamantyl group on phosphorus (54-(OTf)2, 55-(OTf)2 and 58-(OTf)2) are more selective for the -diester formation than the one with a P^tBu2 rest (51-(OTf)2, 52-(OTf)2, 53-(OTf)2). Regarding both terms, selectivity and productivity, also the formation and the stability of the catalytically active hydride species must be taken into consideration (vide infra).

In the course of a cooperation with Lucia Caporaso from the University of Salerno (Italy) topographic steric maps were calculated, which represent another approach to visualize the steric bulk of the ligand (Figure 13).^[61] The program (modified version of SamVca) calculates the buried volume of a given ligand, which is a number that quantifies the amount of the first coordination sphere of the metal occupied by this ligand. The buried volume VBur (in percent) is evaluated in the single quadrants around the palladium center of each complex. The four different quadrant contributions visualize the asymmetry in a way ligands wrap around the palladium and allow to understand how changing the substituents on the phosphorus atom influence the shape of the reactive pocket.

Focusing on the steric maps in Figure 13 in more detail, the quadrants show significant differences in their shape. Common for all complexes is a quite asymmetric folding of the steric bulk around the palladium center. The differences in the steric bulk of the ligands can be clearly assigned on the surface of all maps. For complexes with the most bulky substituents (tert-butyl and adamantyl) (51-(OTf)2 and 54-(OTf)2) on phosphorus the steric bulk is more or less distributed in all four quadrants. This finding not only confirms the high steric bulk around all of the palladium center, but also indicates that those two complexes are quite rigid and have less possibility to be flexible than the complexes with less sterically demanded substituents on phosphorus. Consequently, the

catalytic pocket of the Pd center gets smaller. This favors dissociation of bulky substrates and offers less span for other coordinating substrates in the catalytic cycle. For less efficient complexes the distribution of steric bulk around the metal is remarkably different compared to 52-(OTf)2 and 55-(OTf)2 and the hindrance is localized into only two quadrants, which clearly shows that these complexes are sterically more flexible, thus making them more accessible for many incoming substrates.^[61]

Figure 13. Calculated steric maps of 51-(OTf)2-57-(OTf)2.

The values for the steric bulk of the phosphine substituents as calculated from the topographic steric maps go along with our half cone angle measurements.

3.2.3.1 Relative Distribution of Branched Esters

As an example for a correlation of diphosphine ligand structure with selectivity a detailed analysis of the products of the isomerizing methoxycarbonylation with [(dtbpx)Pd(OTf)2] (58-(OTf)2) was made by Philipp Roesle (Scheme 26).^[25,62]

Beside the high selectivity towards the linear -diester (60), during the reaction a spectrum of other ester by-products (shorter B1, B2, B3 and longer branched diesters B4-B16) can in principle be formed. In these by-products the newly formed ester functionality is not at the ω position, but in all positions from ω-1 (B1) to the α position (B16). GC analysis of the crude reaction mixtures showed that at least seven branched esters are formed. To identify the by-products, these were enriched by removing the linear 1,19-dimethyl nonadecanedioate (60) from the crude reaction mixture by crystallisation from methanol and column chromatography of the supernatant solution to further remove the starting material. By NMR analysis (¹H; ¹³C; ¹H,¹H-COSY; ¹H,¹³C-HSQC;

1H,¹³C-HMBC) of this purified reaction mixture all ester by-products could be clearly identified.

Scheme 26. Product spectrum of the isomerizing methoxycarbonylation of methyl oleate.

Table 6 gives the product distribution as determined by GC. The analysis of the relative distribution of the branched ester by-products shows that catalysts with similar productivity (51-(OTf)2, 54-(OTf)2, 57-(OTf)2 and 58-(OTf)2 versus 52-(OTf)2, 53-(OTf)2 and 55-(OTf)2) also show similar distribution of branched products. The less active catalysts have a stronger preference for the methyl branched product B1, which is the predominant compound in all cases, while the more

bulky and more active catalysts (51-(OTf)2, 54-(OTf)2, 57-(OTf)2 and 58-(OTf)2, respectively have a lower preference for B1 versus other branched by-products.

Table 6. Relative distribution of branched ester by-products found with different catalysts in%.

Pd(II) complex B1 B2 B3 B4-B16

51-(OTf)2 (tBu-tBu) 31 3 3 63

52-(OTf)2 (tBu-Cy) 49 5 5 41

53-(OTf)2 (tBu-iPr) 49 4 4 43

54-(OTf)2 (Ad-tBu) 28 3 3 66

55-(OTf)2 (Ad-Cy) 51 4 4 41

57-(OTf)2 (Ad-Ad) 24 3 3 70

58-(OTf)2 (dtbpx) 34 9 6 51