Limits of Activity: Weakly Coordinating Ligands in Arylphosphinesulfonato Palladium(II) Polymerization

Catalysts

3.1 Introduction

The coordination strength of the monodentate ligand L introduced with the catalyst precursors [(P^O)PdMe(L)] (^MeO1-L; P^O = κ²-P,O-Ar₂PC₆H₄SO₂O with Ar = 2-MeOC₆H₄) has a major impact on the catalytic activity in homo- and copolymerizations due to the equilibrium (P^O)PdR(L) + monomer (P^O)PdR(monomer) + L, which accompanies chain-growth. Thus, stronger coordinating ligands shift the equilibrium towards the dormant species ^MeO1-L. So far monodentate ligands, e.g. PPh3, tmeda, pyridine, 2,6-lutidine, DMSO and derivatives thereof have been used (Figure 3-1).24,25,38,43,45,86

Alternatively, carbon based ligands occupying two coordination sites, e.g. η³-allyl or η¹,η²-2-methoxycyclooct-5-enyl are suitable precursors to initiate chain growth.^20,87

Figure 3-1. Reported complexes ^MeO1-L with neutral, monodentate ligands L.

By comparison to the aforementioned N- and P-based ligands dimethylsulfoxide (DMSO) binds less strongly to the metal center and is more readily displaced by olefinic substrates. This enabled homooligomerization of methyl acrylate (MA) and the isolation of ethylene-methyl acrylate copolymers with more than 50 mol% MA incorporation.⁴³ Here, the weak coordination strength of DMSO permitted polymerization at low ethylene pressures and thus high MA/ethylene ratios. For completeness it should be mentioned that entirely 'base-free' species of the molecular composition [(P^O)PdMe] have been isolated^25,44,49 or synthesized in situ by abstraction of L from (P^O)PdMe(L).21,45-47,87

However, so far no improved polymerization activities in comparison to ‘base-coordinated’ compounds have

been reported. For in situ activated catalysts this might be due to incomplete activation or side reactions by activation reagents or catalyst precursors.^21,45,87 For isolated material the reported low solubility likely renders part of the catalyst inactive.⁴⁴

The significantly higher activity observed with DMSO- vs. pyridine-coordinated catalyst precursors suggests studies of further weaker coordinating ligands. Phosphine oxides (O=PR₃) as a ligand class lend themselves for this purpose, as they are easily accessible from the corresponding phosphines and exhibit a defined coordination site at the oxygen atom.

Furthermore, a great variety of phosphines is commercially available and allows for electronic and steric fine-tuning. While chelating, hemilabile ligands (X^O; X = N,P,O; O = phosphine oxide), and especially the phosphine-phosphine oxide ligands have attracted much attention in homogeneous catalysis,^88-93 the application of monodentate tertiary phosphine oxides in homogenous catalysis is rare,^94-98 even though coordination towards metal centers is well studied.⁹⁹ With respect to ethylene polymerization, Starzewski et al. reported a remarkable effect with a catalyst system based on a hemilabile κ²-P,O ligand, Ni(cod)2 and an additional monodendate ligand. By changing the monodentate ligand from triphenylphosphine to the corresponding phosphine oxide molecular weight of the produced polymer could be significantly increased from low-molecular weight waxes to high molecular weight polyethylene (Mη > 10⁶ g/mol).^97,98

3.2 Results and Discussion

3.2.1 Coordination Strength of Phosphine Oxides

Since the coordination strength is influenced by steric as well as electronic properties, both parameters should be varied independently. Here, the cone angle θ and the electronic parameter χ of the corresponding phosphines enable an educated selection of phosphine oxides (χ is defined as the difference between the ν(CO) (A1) absorption in L-Ni(CO)3 and the ν(CO) (A₁) absorption in (t-Bu)₃P-Ni(CO)₃).¹⁰⁰ For this study OPBu₃, OPOct₃, OPPh₃, OP(o-Tol)₃, and OP(p-CF₃C₆H₄)₃,^100,101 for which θ– and χ–parameters have been reported, as well as the even more electron deficient OP(3,5-(CF₃)₂C₆H₃)₃ were selected (Figure 3-2). For OP(3,5-CF₃C₆H₃)₃ no parameter data was available, but coordination strength can be expected to be rather low due to increased electron deficiency in comparison to OP(p-CF₃C₆H₄).

Phosphine oxides not available commercially were synthesized by phosphine oxidation with H₂O₂ (c.f. Chapter 7.2.1).

Figure 3-2. Coordination strength of phosphine oxides based on parameters θ and χ of the corresponding phosphines.^33,100

Regarding electronic properties, the comparison of the literature derived electronic parameters with the observed ³¹P NMR shifts of the phosphine oxides show a good correlation. With increasing electron deficiency χ increases, while δ decreases. It is important to note that the ³¹P NMR shift of phosphine oxides is a reliable measure of the basicity as opposed to the ³¹P NMR shift of phosphines which is also influenced by sterics.¹⁰² As expected, the aryl phosphine oxides exhibit a weaker basicity than OPBu₃. In comparison to OPPh , steric bulk is increased by introduction of a methyl group in ortho-position in

OP(o-is enlarged in comparOP(o-ison to the phosphine. In contrast, the introduction of a CF₃-group in para-position in OP(p-CF₃C₆H₄)₃ does not change steric properties but increases electron deficiency. In OP(3,5-(CF₃)₂C₆H₃)₃ electron deficiency is further increased while the steric influence is believed to be similar to OP(p-CF₃C₆H₄)₃ (θ = 145° vs. ~151° for P(3,5-Me₂C₆H₃)₃).¹⁰³

The relative coordination strength of these phosphine oxides compared to DMSO (KOPR3) was determined by ¹H NMR spectroscopy. The ¹H resonance of DMSO in CD2Cl2 is downfield shifted from 2.54 ppm (free DMSO) to 2.95 ppm by complexation to the Pd center in ^MeO1-dmso. Here, the exact coordination mode of DMSO remains unclear, but for a related complex an S-coordination was observed by X-Ray analysis.⁴³ Partial replacement of DMSO by OPR3 leads to a high field shift due to a fast equilibrium between Pd-bound and uncoordinated DMSO. From the shift difference the ratio between ^MeO1-dmso and ^MeO1-L and consequently KL at 25 °C was calculated (cf. Chapter 7.1.10).

Table 3-1. KL for ^MeO1-dmso + L 1-L + DMSO.

a∆δ is low due to limited solubility of OPR3 consequently inaccuracy of KL is enhanced.

The results are summarized in Table 3-1. Whereas both alkyl phosphine oxides coordinate slightly stronger than DMSO (K_OPBu3 = 3.5, K_OPOct3 = 3.3), the more bulky and electron-deficient OPPh₃ exhibits K_OPPh3 = 0.2 for the equilibrium ^MeO1-dmso + OPR₃

MeO1-OPR3 + DMSO. An even weaker coordination is evident for the comparison with OP(o-Tol)₃ (K_OPTol3 = 0.03), and OP(p-CF₃C₆H₄)₃ (KOP(pCF3Ar)3 = 0.04, Table 3-1, compare K_L for MeOH and 2,6-lutidine). Hence, coordination strength can be controlled by either steric bulk or electron deficiency over a large range. The introduction of a second electron withdrawing group in OP(3,5-(CF3)2C6H3)3 further reduces coordination strength significantly (KOP(3,5CF3Ar)3 ~ 0.001). Note, that in a related study KL was determined vs. DMSO for ethyl acetate (KEA < 10^-2), methyl ethyl sulfone (KMES < 10^-2), propionic acid (KPrA < 0.1),

N,N-dimethylacetamide (K_DMAcA ≈ 0.4), N-methylacetamide (K_MAcA ≈ 0.6) and acrylonitrile (K_ACN ≈ 1).⁵⁶

In order to investigate the influence of the temperature on the coordination equilibrium, K_OPPh3 was determined in the temperature range between -25 °C and 25 °C. Only minor changes of K_OPPh3 in this temperature range have been observed (Table 3-2).

Table 3-2. KL at different Temperatures ([^MeO1-dmso] = 0.016 mol L^-1; equiv. L: 10.3).

Plotting the natural logarithm of the equilibrium constants against the temperature in a Van’t Hoff analysis allowed for the determination of ∆H°DMSO/OPPh3 = 8 kJ mol^-1,

Figure 3-3. Van’t Hoff plot for the equilibrium ^MeO1-dmso + OPPh3 MeO1-OPPh₃ + DMSO determined by variable temperature ¹H NMR spectroscopy from T = -25 °C to 25 °C.

It is assumed that a similar small temperature dependence applies to all KOPR3 and that

according to the equilibrium [(P^O)PdR(L)] + ethylene [(P^O)PdR(ethylene)] + L, complexes ^MeO1-L derived from phosphine oxides with a weaker coordination strength than DMSO, i.e. from OPPh₃, OP(o-Tol)₃, OP(p-CF₃C₆H₄)₃, and OP(3,5-(CF₃)₂C₆H₃)₃ are expected to exhibit higher turnover frequencies than ^MeO1-dmso as long as saturation kinetics are not reached. Consequently, such complexes ^MeO1-L represent valuable synthetic targets for highly active single component catalysts.

3.2.2 Complex Synthesis and Characterization

For the synthesis of phosphine oxide complexes ^MeO1-OPR3 standard procedures were not applicable since they are either based on introduction of the ligand with the Pd-precursor as with L = TMEDA from [(tmeda)PdMe2],20,24,38,87

or subsequent ligand substitution by a stronger coordinating ligand.25,38,44,45,62,86

Notably, the weaker coordinating DMSO could be introduced by substitution of TMEDA. This substitution occurs since TMEDA is removed from the equilibrium ½ (^MeO1)₂-tmeda + DMSO ^MeO1-dmso + ½ TMEDA under vacuum due to the considerably higher volatility of TMEDA vs. DMSO.⁴³ However, an analogous procedure, e.g. solvent evaporation from a mixture of (^MeO1)₂-tmeda and phosphine oxide in high boiling solvents, did not result in the isolation of clean products. In an alternative approach, multinuclear 'base-free' palladium alkyl complexes which are accessible e.g. by pyridine or lutidine abstraction with B(C₆F₅)₃^25,44 may be suitable precursors for the preparation of phosphine oxide complexes ^MeO1-OPR3. However, a more convenient synthesis starts from [{(^MeO1-Cl)-µ-Na}₂)]⁶¹ and chloride abstraction in the presence of phosphine oxides is expected to generate ^MeO1-OPR3 if the presence of stronger coordinating ligands is avoided (Scheme 3-1).

Scheme 3-1. Synthesis of ^MeO1-L.

The viability of this general route was demonstrated by the synthesis and isolation of

MeO1-dmso. As expected, also the complexes ^MeO1-OPBu3 and ^MeO1-OPOct3 with the