Isomerizing-Endfunctionalization Reactions

Other examples of isomerizatioendfunctionalization processes include Nozaki’s highly n-selective three-step one-pot reaction of several internal olefins such as (Z)-2 decene, (Z)-2-tridecene and methyl oleate to the corresponding n-alcohols using a Rh/Ru dual catalyst-system (Scheme 11). The conversion comprise the isomerization of the double bond, hydroformylation to the aldehydes and subsequent hydrogenation to the corresponding alcohols.^[36]

Scheme 11. One-pot isomerization, hydroformylation and hydrogenation.^[36]

Olefin metathesis tandem processes have attracted particular attention in recent years (Scheme 12).

Gooßen et al. described the conversion of oleic acid via isomerizing self- and cross-metathesis reactions into compound mixtures with varying chain length distribution. In analogy to olefin blends available from petrochemical sources the chain length distribution with regard to the mean and span could be adjusted.^[37]

Scheme 12. Isomerizing olefin metathesis.^[37]

An effective catalyst system was identified that continuously equilibrates the position of the double bond in olefinic substrates such as methyl oleate. At the same time the catalyst system metathesize the olefin mixture in order to obtain defined distributions of functionalized products. The dimeric Pd(I) complex [Pd(-Br)tBu3P]2 turned out to be an uniquely active isomerization catalyst, which retains its activity in the presence of state-of-art olefin metathesis catalysts without loosing activity.^[37]

Angelici^[38] et al. explored the behavior of methyl oleate in the selective isomerizing hydroboration reaction in order to produce the boronated ester group exclusively on the terminal end of the carbon chain. Scheme 13 (A) shows the reaction sequence with an in-situ generated catalyst. The catalytic system [Ir(COE)2Cl]2 / dppe promotes both the rapid isomerization of methyl oleate and the selective hydroboration of the terminal isomer in 45 % yield.^[38]

Another example describes the isomerizing hydroformylation of methyl oleate in the presence of a rhodium catalyst (Scheme 13 (B).^[39]

Scheme 13. One-pot isomerizing, hydroboration (A)^[38] and hydroformylation (B).^[6c,39]

Under the given conditions (115 °C, 20 bar syngas) the best yield of the -aldehyde is 26 %.

Unfortunately, hydrogenation of the double bonds occurred as a dominant reaction pathway, which was attributed to an ester group effect. The ester function makes hydroformylation in the immediate vicinity of this group impossible. For this reason hydrogenation predominates over formation of branched aldehydes.^[6c]

1.6.4 ,-Functionalized Long-Chain Alkanes as Building Blocks for Polymers

The use of vegetable oils and fatty acids for polymer applications has a long tradition. Important examples are building blocks for thermoplastic polymers (e. g. dicarboxylic acids such as sebacic acid and oleic acid) as well as polymer additives (e.g. epoxidized soybean oil as plasticizer).^[6a]

A particular efficient route leading to polyesters has recently been developed by the Mecking group (Scheme 14).^[40] Technical plant oils (such as high oleic sunflower oil) can be employed as a substrate without prior separation of the different fatty acids present. The approach is based on readily available methyl oleate and the first step takes advantage of the isomerizing alkoxycarbonylation reaction. The monomer building block (α,ω-diester, dimethyl-1,19-nona-decanedioate) is obtained highly selectively on a 100 g scale in the lab. Further reduction of this monomer either by LiAlH4 or ruthenium catalyzed hydrogenation lead to the corresponding diol, which is used as second monomer partner in the following step-grow polycondensation polymerization reaction leading to polyesters with molecular weights M typically being 2×10⁴ g mol^-1 and a crystallinity of around 75 %. Other chemical reactions include the transformation of the α,ω-diesters to α,ω-diamines and α,ω-acetals which are all obtained in polycondensation grade purities of > 99 %.^[40]

Scheme 14. Thermoplastic polyesters from renewable resources with complete feedstock utilization.

Polyesters generated from 1,19-dimethyl-nondecaneedioate as monomer building block posess useful mechanical properties such as high crystallinity, which are comparable to linear polyethylene (HDPE) their thermal properties being compatible with thermoplastic processing.

Due to their hydrophobicity and crystallinity such long-chain aliphatic polycondensates can be subject to a desirable slow hydrolytic degradation in the environment, e.g. in marine environments.

Although with the dtbpx catalyst system high reaction productivities and selectivities in the isomerizing methoxycarbonylation of olefines can be obtained, the reaction is rather slow.

Especially for long chain molecules such as methyl oleate the reaction times increase greatly (ethylene (TOF: 12,000 h^-1, at 80 °C, 10 bar, CO : ethylene = 1:1) < 2-octene (3 hours, 80°C, 30 bar, TON: 255) < 3/4-octene (16 hours, 80°C, 30 bar, TON : 255) < methyl oleate (18 hours, 90

°C, 20 bar CO, TON: 125).[16b,22,32b,34,40] Furthermore, the dtbpx based catalytic active hydride species is sensitive towards the reaction conditions (e. g. CO pressure). Carbon monoxide can induce the formation of inactive neutral Pd(0) complexes which can couple with other cationic palladium species to form catalytically inactive bridged [(P^P)Pd]22+ complexes or reductive elimination of the Pd(II) hydride complex to the protonated ligand and Pd metal. These processes limit the overall yield of the linear product and decrease the catalytic performance of the catalyst.^[41,42] These problems might be overcome by changing the nature of the diphosphine ligand.

The specific design of the ligand structure and substitution pattern could generate Pd complexes of greater stability, which are more active and more selective when used in the isomerizing methoxycarbonylation of methyl oleate.

2 Scope of this Thesis

In the view of the worldwide constantly growing demand for polymers, a fast and simple access to the plastic products is desirable. Plant oils offer a unique starting material structure with their long methylene sequences. If those can be converted to a linear difunctional chain this provide unique building blocks for semicrystalline polymers. This challenge is achieved by the isomerizing alkoxycarbonylation reaction of fatty acids such as methyl oleate, which is readily available from plant oils. Therby the entire fatty acid chain is converted into the linear -diester as versatile long chain monomer building block for polymers. For this pivotal single-step transformation on a large scale efficient catalysts are required (Scheme 15). So far, no robust process suitable for large-scale production of the long chain 1,19-diester has been developed. On the lab large-scale promising results were achieved with the use of dtbpx derived catalysts.

In this context the goal of the present thesis was to synthesize novel diphosphine based Pd(II) catalyst precursor complexes, which are highly selective in the 1,19-diester production and less sensitive towards deactivation processes. The focus was on unsymmetrical bidentate ligands with an o-tolyl backbone bearing different bulky and electron rich alkyl substituents on the phosphorus donor atom. The orthogonal reactivity pattern of the unsymmetrical o-tolyl backbone (Scheme 15) opens a simple way to a rich library of different unsymmetrical diphosphine ligands. The influence of their steric bulk on the selectivity to the linear 1,19-diester formation and their productivity has not been investigated yet.

Structure productivity relationships can give hints as to how the catalysts’ diphosphine substitution pattern impacts on selectivity and productivity of this remarkable reaction. Consequently, the most effective substitution pattern R and R’ of the unsymmetrical o-tolyl diposphines should be transferred on a second generation of Pd(II) catalysts based on the known -xylene backbone, followed by investigations of their catalytic behavior in the isomerizing methoxycarbonylation of methyl oleate.

Scheme 15. Transformation of plant oil derived methyl oleate into polymers via the Pd(II) catalyzed isomerizing alkoxycarbonylation process.

3 Results and Discussion 3.1 Previous Studies

Important previous studies directly related to the new results reported in this thesis shall be briefly summarized here.

Pd(II) complexes bearing the sterically demanding dtbpx ligand are known to be very productive catalysts in the isomerizing alkoxycarbonylation of plant oils.^[33-35] The favorable properties of the dtbpx ligand (type I wih R = tBu, 1) and its unsymmetrical derivatives lead to the decision to explore the performance of catalysts based on the o-tolyl backbone (II) (Figure 4). o-Tolyl diphosphine ligands II are in a mid-position between the rather rigid diphosphines of type III and the more flexible dtbpx type I ligands. Metal complexes with ligands of this type might be suitable for efficient catalyst precursors. In addition, they could be more resistant towards catalyst decomposition as they are less basic than complexes based on type I and less sensitive towards protonation and Pd black formation.^[30] To investigate whether the coordination chemistry of the unsymmetrical o-phosphino-benzyl-diorgano-phosphine ligands would provide insight into the catalytic performances of its Pd complexes is of further interest.

Figure 4. Diphosphine structures (R = alkyl, e.g. tert-butyl, iso-propyl).

Known syntheses leading to unsymmetrical diphosphines are outlined in the following Schemes 16-18. A suitable starting material for the diphosphine synthesis is o-bromotoluene 2. G. Müller^[43]

et al. developed a photochemical route to convert 2 to 3 (Scheme 16). The starting material is easiliy accessible and inexpensive. Moreover, dihalogene-phosphine 3 is an interesting and versatile substrate that can be used in large scale for transformation into further phosphine ligands.

UV light induces the formation of an exo-methyl radical of 2, which reacts finally with PCl3 or PBr3 to 3. Treatment of di-bromo-phosphine 3 with two equivalents of a Grignard reagent leads to the double substituted benzylphosphine 4, which is lithiated in the presence of n-BuLi (5).

Scheme 16. Photo-catalytic synthesis of unsymmetrical diphosphines by G. Müller et al.^[43]

After subsequent conversion with dialkylphophine chlorides, the desired o-tolyl diphosphines (II) can be obtained. Moreover, derivate 3 even offers the possibility of a stepwise P-alkylation by adding 1 equivalent of a certain Grignard reagent (Cl-Mg-R) followed by the addition of another Grignard reagent (1 equiv., Cl-Mg-R’) to achieve diphosphine ligands carrying different alkyl residues attached on the benzyl phosphorus. In addition, by using Grignard reagents (e.g.

Cl-Mg-Me, Cl-Mg-tBu) as alkylation agents for the phosphorus atom the application of expensive starting materials such as dialkyl-chlorophosphines can be avoided.^[44] The reaction sequence outlined in Scheme 16 clearly shows the advantages of an orthogonal reactivity principle (represented in II). Targeted introduction of both phosphine moieties succeed stepwise, one after another, via two different carbon reactivity patterns. The carbon atom at the benzyl position acts as electrophile and the o-carbon benzene atom operates as a nucleophile attacking the dialkyl-chlorophosphine. However, the use of the photo reaction (2 to 3) is limited due to long reaction times (6 hours) and harsh reaction conditions (e.g. 160 °C) which causes the formation of unwanted side products.

Starting with tolyl-bromide 2 Gosh et al. developed an approach for the synthesis of the tetraphenyl substituted diphosphine 9 in several steps (Scheme 17).^[45] The pivotal step is the substitution

reaction of benzyl bromide 7 and borane-diphenylphosphine complex using highly carcinogenic HMPA (step 3). On the one hand, the broad application of this route is limited due to the use of HMPA, which is essential for the reaction. On the other hand, step 1 (substitution with HP(O)Ph2) as well as step 4 (reduction of phosphine oxide 8) are limited to less electron donating substituents such as phenyl.

Scheme 17. Synthesis of 9 by Gosh et al.^[45]

Alternatively, o-bromo-benzylbromide 10 has been used as starting material to synthesize several unsymmetrical diphosphine ligands (examples shown in Figure 5). Three different initial steps leading to benzyl phosphine derivative 12 were elaborated and published (Scheme 18), respectively, comprising the two-step formation of a Grignard intermediate (a),^[46a,47d] usage of metallated phosphines (b)^[16a,46] as well as the addition of dialkyl phosphine (c).^[47] The straightforward chemistry is characterized by the lithiation (13) using n-BuLi followed by the direct phosphorylation with dialkylchlorophosphines giving rise to the target ligands II.

Scheme 18. Synthesis of diphosphines type II.

According to synthesis (c), Pringle et al. described very recently the manufacture of several target ligands (two examples are shown in Figure 5),^[48] which were used for generating catalysts suitable for the methoxycarbonylation of ethylene. The catalytic performance of the corresponding Pd(II) and Pt(II) complexes illustrates that the results strongly depend on the substitution pattern on the phosphorus atom. Compared to others, diphosphines bearing more sterically demanding groups were highly active.

Figure 5. Diphosphines for the methoxycarbonylation of ethylene.

3.2 Novel Complexes of Bulky Diphosphine Ligands with an Unsymmetric Backbone

At the same time as Pringle’s investigations we have independently chosen route c (Shell patent)^[47]

as basis of own research activities. The Shell route seemed to be most efficient for the introduction of bulky substituents (which are generally unreactive) leading to new unsymmetrical diphosphine ligands (Scheme 19).

2-Bromo-benzylbromide 10 is treated with secondary bulky phosphines such as HP(tBu)2 (14), HP(Ad)2 (15) and HP(Cy)2 (16) giving rise to the corresponding mono-substituted intermediates, which were easily isolated as hydrobromic acid salts 17-19 (white solids and in the case of 18 colorless crystals).

Scheme 19. Efficient route to monophosphines 20, 21 and 22.

The highest yields could be obtained with acetonitrile as solvent, facilitating the nucleophilic attack of the bulky monophosphines at the benzylbromide 10. Treatment of the formed hydrobromides (17-19) with trimethylamine in toluene release the resulting neutral monophosphines 20 (86 %) and 21 (65 %), which could directly be used in the next step without further purification (purity by

1H NMR > 98 %). Crude 22 was received as a pale yellow oil, which was recrystallized from hot ethanol yielding 22 as colorless crystals (45 %).

Scheme 20. Synthesis of unsymmetrical diphosphines 29-34.

Monophosphine 20 was lithiated at room temperature in pentane by adding n-BuLi, resulting in the immediate formation of a white precipitate that has been isolated by simple filtration (Scheme 20).

Further treatment of 23 with chlorophosphines (Cl-tBu2P, Cl-PiPr2 and Cl-PCy2) in THF at -80 °C afforded the desired unsymmetrical diphosphines 29, 30 and 31, respectively, as crude products, which were recrystallized from ethanol or other organic solvents (see Table 1, with detailed crystallization conditions and behavior) as colorless needles suitable for X-ray diffraction.

Table 1. Crystallization behavior of the diphosphines 29-34.

Ligand Crude product Crystallization Ligand properties 29 (tBu-tBu) Colorless oil EtOH, r.t. -> -20 °C Waxy colorless crystal 30 (tBu-Cy) Colorless solid Hot EtOH -> r.t -> -20 °C Colorless needles 31 (tBu-iPr) Waxy solid EtOH, r.t. -> -20 °C Waxy colorless crystal 32 (Ad-tBu) Yellow foam Hot EtOH, CH2Cl2 -> r.t. Colorless crystal 33 (Ad-Cy) Colorless solid Extraction with H2O Colorless solid 34 (Cy-Cy) Colorless foam Hot pentane, CH2Cl2 -> r.t. Colorless crystal

Alternatively, monophosphines 20, 21 and 22 can be lithiated with n-BuLi in THF solution (instead of pentane) without precipitation. This in-situ generation of the lithium aryls turned out to be advantageous as compared to the limited stability of the isolated lithium aryls. The formation of the lithium salts was checked on completeness and the “in process control” (IPC) was made by NMR measurements. NMR samples were taken after 15 minutes when the lithiation was initiated.

The lithium salt has been identified as hydrolyzed compound o-deuterium-benzyl-di(tert-butyl)phosphine after quenching with methanol-d4.

In case the lithiation turned out to be incomplete, the formation of undesired side products such as the alkylated phosphine n-butyl-di(tert-butyl)phosphine (36) is likely (Figure 6).

Figure 6. Example of a ¹H NMR spectrum of a crude product mixture after incomplete lithiation.

As is shown in Figure 6, incomplete lithiation gave the following compound mixture: diphosphine 29 (7.2 %), hydrolyzed monophosphine 35 (21.4 %) and alkylated phosphine 36 (71.4 %) derived from n-BuLi and Cl-PtBu2. Increasing temperature will lead to -lithiation and thus facilitate the formation of impurities. Elongation of the reaction time completes the o-lithiation and thus decreases the formation of impurities. The presence of sterically hindered residues lead to prolonged reaction times of the o-lithiation of 30-60 minutes (e.g. 29, 32). Alternatively to n-BuLi the more reactive tert-BuLi has been tested in the lithiation reaction. In this regard, the isolation of the lithium salts can be advantageous as the addition of equimolar reagents (next step) does not generate side products.

The target compounds 29-34 were obtained in yields of 34 up to 94 % by quenching the in-situ generated lithium aryls with chlorophosphines 26-28 followed by subsequent recrystallization. The

31P NMR shifts of all synthesized diphosphine ligands (Figure 7) nicely correlate with the nature of the alkyl substituent in the following order δ: tBu ≈ Ad > Cy> iPr.

Figure 7. ³¹P NMR data for ligands 29-34 in C6D6, 25 °C, for 33 in CDCl3.

The crystal and molecular structures of diphosphines 29, 32 and 34 were determined by X-ray diffraction (Figure 8). Selected bond lengths and angles are summarized in Table 2. Bond lengths are in the typically range of P-C(alkyl) and differ only marginally for the diphosphines 29, 32 and 34, respectively. As expected the P-C(aryl) bonds are slightly shorter than the P-C(alkyl) bonds.

The bond angle between the two cyclohexyl groups and P¹ (C(14)-P(1)-C(8)) shows a value of 102.91(11), which is smaller as compared to the corresponding angles in the diphosphines 29 and 32. This is due to less steric hindrance of the cyclohexyl rings.

Figure 8. Molecular structure of 29, 32 and 34 in the solid state. Ellipsoids are drawn at the 50%

probability level. Hydrogen atoms have been omitted for clarity.

Regarding the P2-C7-axis in 29 (see Figure 8), the substituents C12 and C6 form a torsion angle of almost 180° (C(6)-C(7)-P(2)-C(12): 173.0°). The tert-butyl groups on C12 are placed at the maximum distance to the benzene ring at P² (diphosphine 29, Figure 8).

Table 2. Selected bond lengths [Å] and angles [°] for 29, 32 and 34.

29 32 34 diadamantylphosphine groups on both phosphorus atoms was synthesized. To carry out the final step, the corresponding lithium salt 24 should be converted to the target ligand 38 using chloro-diadamantyl phosphine 37 as reagent for the nucleophilic attack (Scheme 21).[49,50,51]

Scheme 21. Synthesis of diphosphine 38 using Cl-PAd2.

Chlorophosphine 37 is commercially available, but expensive,^[52] and is described in the literature to be generated in high yield by mixing diadamantylphosphine 15 and DBU in the presence of either highly reactive phosgene or carbon tetrachloride (CCl4) (Scheme 22). Due to the high toxicity of the chlorinating reagents - phosgene, triphosgene^[51b] as well as carbon tetrachloride - alternative sources for the chlorination or other halogenations were considered.^[49,50,51]

Scheme 22. Synthesis of Cl-PAd2 37.

According to the literature attempts to halogenate the phosphine with gaseous chlorine, bromine and/or iodine were not successful and led to product mixtures as the diadamantylphosphine turned out to be extremely unreactive, probably because of the +I-effect of the adamantyl cage.

Unfortunately, the use of inorganic chlorination agents like PCl5[53a,c] or C2Cl6[53b,c] resulted in unidentified product mixtures and hydrochlorinated adamantyl phosphine.

The search for alternative halogenations of diadamantyl phosphine 15 led to the use of solid CBr4

as reagent, having good dissolution properties in methylene chloride (Scheme 23). Indeed, the desired bromo-derivative 39 could be synthesized in 95 % yield after the side product bromoform was removed by evaporation (b.p.: 149.5°C).

Scheme 23. Brominaton of 15.

As bromophosphine 39 suffers from slow decomposition,^[49] which was observed by NMR after 24 hours, it turned out to be obligatory to react 39 directly after its synthesis. It should be noted that phosphine oxide 40 (³¹P NMR (400 MHz, CDCl3):  68.8, s) and hydrobromide 41 (¹H NMR (400 MHz, CDCl3):  6.2, d ¹J(P,H) = 468.8 Hz) are the major impurities resulting from decomposition reaction of 39. For the next step isolated lithiated benzylphosphine 24 (Table 3, entry i) as well as o-Br-benzylphosphine 21 (Table 3, entry iii, iv and v) were used as starting materials for the substitution reaction with bromophosphine 39. In addition, a Cu⁺ catalyzed bromo-aryl activation (Table 3, entry vi and vii) was also conducted. ^[54] Corresponding approaches and detailed reaction conditions are summarized in Table 3.

Table 3. Several attempts synthesizing diphosphine 38.

In order to generate the target ligand 38 only two reaction conditions (Table 3, entry iii and vi) were successful, although only very poor yields in the range of 10 % after recrystallization could be achieved. All other attempts failed completely and/or came along with large amounts of the hydrolyzed monophosphine 42.

However, most importantly sufficient amounts of ligand 38 could be manufactured according to the reaction conditions where using tert-BuLi (Table 3, entry iii) for testing its catalytic activity in the isomerizing alkoxycarbonylation reaction. In total, the obtained results show that the P-aryl coupling is quite challenging for the highly sterically demanding - and thus unreactive - diadamantylphosphine 15. Typical ¹H and ³¹P NMR data of the starting material 21, the hydrolyzed phosphine 42 and the product 38 are given in Figure 9.

Figure 9. NMR data for the product mixture.

In parallel Pringle et al. investigated alternative Pd catalyzed P-C(aryl) couplings (conditions:

[Pd(PPh3)4], HPCg, DABCO, 140 °C, 24 hours) with the aim of synthesizing similar phosphine ligands with bulky substituents (Section 3.1, Figure 5).^[48] Pringle’s results are characterized by obtaining rather low yields in the range of 14 %, whenever particularly bulky substituents such as phospha-trioxa-adamantane (Cg) were involved. These findings support our preference for other reaction conditions as already described.^[48]

3.2.2 Synthesis and Molecular Structure of Diphosphine Pd(II) Complexes

Starting from the diphosphine ligands 29-34, 38 the desired Pd(II) complexes were synthesized in two steps, initially in the presence of [Pd(dba)2], followed by adding two equivalents of trifluoromethanesulfonic acid (TfOH) and p-benzoquinone (BQ) (Pd(0) -> Pd(II)) in the final step (Scheme 24).^[22,63a]

Scheme 24. Synthesis of diphosphine Pd(II) complexes 51-(OTf)2-57-(OTf)2.

The diphosphines 29-34, 38 and [Pd(dba)2] were dissolved in THF and stirred at room temperature for 12 hours. After filtration of small amounts of Pd black and removal of the solvent, the reddish residue obtained was washed with pentane to remove residual diphosphine. The generated diphosphine Pd(0) complexes were reacted further without additional characterization. Crystals of 47 (R = Ad, R’ = tBu) suitable for X-ray crystallography (Figure 10) could be isolated by adding pentane to a solution of 47 in methylene chloride. Dissolution of the [(P^P)Pd(0)(dba)] complexes

The diphosphines 29-34, 38 and [Pd(dba)2] were dissolved in THF and stirred at room temperature for 12 hours. After filtration of small amounts of Pd black and removal of the solvent, the reddish residue obtained was washed with pentane to remove residual diphosphine. The generated diphosphine Pd(0) complexes were reacted further without additional characterization. Crystals of 47 (R = Ad, R’ = tBu) suitable for X-ray crystallography (Figure 10) could be isolated by adding pentane to a solution of 47 in methylene chloride. Dissolution of the [(P^P)Pd(0)(dba)] complexes

Im Dokument Bulky Diphosphines in the Palladium Catalyzed Isomerizing Alkoxycarbonylation of Fatty Acids (Seite 33-0)