Synthesis of unsymmetrical NCN ’ and PCN pincer palladacycles and their catalytic evaluation compared with a related SCN pincer palladacycle † ‡

1-(3-(Pyridin-2-yl)phenyl)methanamine derivatives have been synthesized and underwent C–H bond activation to afford unsymmetrical NCN’ pincer palladacycles, which were characterised in the solid state. 2-Pyridinyl-phenol and -benzyl alcohols were then used as precursors to unsymmetrical PCN pincer palladacycles. Catalytic applications, where the palladacycle remains in the Pd(II) state, have been carried out and show good activity and selectivity.


Introduction
Palladacycles contain a covalent Pd-C bond intramolecularly stabilised by a coordinating group such as an amine, phosphine or thioether and have been extensively studied since their discovery in the mid-1960s. 1 Pincer palladacycles, where a tridentate ligand can coordinate to palladium were first synthesized by Shaw and Moulton in 1976. 2 However, the field of palladacycle-mediated catalysis only truly gathered momentum after the seminal discovery by Herrmann and Beller et al. 3,4 that their eponymous palladacycle ( Fig. 1) was highly active in catalytic C-C bond coupling reactions. Since, a number of reviews and books have been published covering the wide array of catalytic applications. [5][6][7][8][9][10][11][12][13][14] The majority of pincer palladacycles are symmetrical, such as I (Fig. 2). 15 However, unsymmetrical analogues such as II (Fig. 2), 16 have been explored, 17 and may provide a greater opportunity to fine tune catalysis due to the potential hemilability of the ligand 18 and the ability to influence catalytic activity by altering the steric and electronic properties of the donor atoms. 19,20 The synthesis of NCN pincers can be complicated compared with the analogous PCP and SCS pincers due to the hardligand (amine) soft-acid (metal) mismatch between palladium and nitrogen. This results in competition in forming the kinetic product III or the thermodynamic pincer product IV (Scheme 1). 21 Hence, incorporating functional groups in the mutual ortho position such as SiMe 3 or Br can be beneficial in their synthesis, compared with C-H activation routes. 8 However, this is less attractive synthetically, since an additional step is required in making the functionalised ligand.
Palladacycles normally act as a reservoir of catalytically active Pd(0) species in applications such as the Suzuki-Miyaura/Heck reaction. 25,26 Other catalytic applications of pincer palladacycles utilise the palladium in its Pd(II) oxidation state, retaining its ligand structure, and are more likely to take advantage of the tuning abilities of the unsymmetrical ligand design. The use of pincer palladacycles as a Lewis acid catalyst in an aldol condensation (Scheme 2) demonstrated the ability to change the stereochemical outcome of the reaction depending on the ligand donor atoms. 27 Another application utilising pincer palladacycles in Pd(II) catalysis, is in the coupling of vinyl epoxides and boronic acids using a symmetrical SeCSe palladacycle (Scheme 3). 28 Unsymmetrical pincer palladacycles are relatively rare and their synthesis and that of their respective ligands often poses a greater challenge than for their symmetrical counterparts. Given some distinct advantages of employing unsymmetrical pincer palladacycles in catalytic applications, 17 such as the tandem catalysis reported by Szabó et al. 27 recent work in our group has focused on a robust synthetic route to useful, modular unsymmetrical ligands and their pincer palladacycles. 29 The present work is focused on the synthesis of several new unsymmetrical nitrogen-based pincer ligands, denoted NCN′, their respective palladacycles, phosphinite PCN pincer palladacycles, and catalytic evaluation in aldol condensation and vinyl epoxide coupling reactions in comparison to a related SCN pincer palladacycle.
Reductive amination of 1 with HNMe 2 ·HCl was attempted in order to synthesise the requisite unsymmetrical NCN′ ligand with the dimethyl amine and pyridine groups ideally placed to promote C-H activation towards the corresponding palladacycle. Initially sodium triacetoxyborohydride 31 proved unsuccessful although the use of titanium isopropoxide/ sodium borohydride 32 furnished the product 2a in 81% yield (Scheme 4). Next, we attempted reductive aminations with NEt 2 H using both of the above conditions but to no avail. Given that the reductive amination procedure proved ineffectual, an alternative synthesis was devised. In previous work, 29 we have shown the synthesis of the benzylic bromide 4, via benzylic alcohol 3 (Scheme 5) can allow late stage diversification via nucleophilic substitution by sulphur nucleophiles. 29 Hence, the corresponding nucleophilic displacement of 4 was undertaken with nitrogen nucleophiles, yielding NCN′ ligands 2b and 2c in excellent yield (Scheme 5).

NCN′ palladacycle synthesis
Ligands 2a-c were then selected for C-H activation towards the palladacycle products. Refluxing the ligand in AcOH in the presence of Pd(OAc) 2 , followed by salt metathesis yielded the desired monomeric palladacycles 5a-c (Scheme 6).
The yields for forming these palladacycles are disappointing, due to significant formation of Pd black, and therefore an optimisation study for the synthesis of 5c was attempted ( Table 2). Changing the solvent to MeOH (entry 2), the palladium source to in situ generated Pd(MeCN) 4 (BF 4 ) 2 (entry 3), 33 and a transcyclopalladation (entry 4) 34 using Pd 2 (dmba) 2 Cl 2 ( Table 2) were attempted. It was found that indeed, the initial Pd(OAc) 2 in AcOH method proved to be the most effective for palladation (entry 1).
However, literature yields for NCN pincers are often low, with several examples provided using Pd(OAc) 2 as the palladium source (Table 3), showing the wide range of yields achieved for the key C-H bond activation step from the corresponding NCN pincer ligands. Shaw's earlier work, in contrast, showed yields for PCP pincers of around 75%. 2 In order to delve into such poor yields, we looked at the stability of 5c in refluxing d 4 acetic acid over time via 1 H NMR, which showed no degradation of the palladacycle over time, under the conditions used for C-H activation. We also contemplated an oxidative addition of a Pd(0) salt to a bromide, 38 although attempts to ortho-brominate 2c were unsuccessful.

PCN palladacycle synthesis
The synthetic route to unsymmetrical NCN′ pincer palladacycles has been modified to provide a route to unsymmetrical PCN pincer palladacycles. The synthesis of phosphinite palladacycles has been shown by Eberhard et al. 39 from a benzyl alcohol using ClPR 2 , and therefore the benzyl alcohol 3 was used as the starting point for synthesis of a PCN pincer palladacycle. Due to the air sensitivity of the PCN ligand, after  [Pd(dmba)(μ-Cl)] 2 Toluene reaction with ClPPh 2 , C-H bond activation using Pd(OAc) 2 in AcOH was performed in situ (Scheme 8). This synthesis yields a mixed 5-,6-membered palladacycle, and for direct comparison with our 5-,5-membered SCN, 29 and NCN′ pincer palladacycles, a phenolic alcohol was synthesised (Scheme 7). Using the phenolic (6) and benzylic alcohols (3), the synthesis of PCN pincer palladacycles was undertaken (Scheme 8). The PCN pincer palladacycle structures were also confirmed using X-ray crystallography (Fig. 5).
Growth of crystals of 5a-c and 7a and 7b by slow evaporation of CH 2 Cl 2 from a saturated solution allowed their structures to be determined by X-ray diffraction ( Fig. 4 and 5). All structures were as expected, with the metal in a distorted square planar arrangement.
Compound 5a crystallises in the monoclinic Pc space group with two independent molecules in the asymmetric unit. These are subject to π-π interactions (benzene centroid-pyridine centroid distances: 3.564 Å and 3.715 Å, offsets: 1.301 Å and 1.386 Å respectively). These pairs of molecules form layers with a 'herringbone' arrangement propagating along the crystallographic b and c axes.
The structure 5b crystallises in the monoclinic P2 1 /c space group with two independent molecules in the asymmetric unit. The structure is formed from corrugated layers of molecules propagating along the crystallographic a and b axes with each alternate layer consisting of the same crystallographicallyindependent molecules.
The structure 5c crystallises in the monoclinic P2 1 /n space group. The structure consists of π-π stacks (benzene centroidpyridine centroid distances: 3.616 Å and 3.705 Å, offsets:     Structure 7a crystallises in the monoclinic P21/n space group. The structure consists of columns of molecules closepacked along the crystallographic 21 screw axis parallel to the b axis with π-π interactions to neighbouring columns via the phenyl pyridine moieties (centroid-centroid distance: 3.894 Å, offset: 1.322 Å).

Catalytic investigation
Next, 5c, 7a and 7b were tested in the catalytic aldol condensation (Scheme 2) and compared with a previously published SCN pincer palladacycle (8) by our group (Fig. 6) 29 along with commercially available PdCl 2 (dtbpf ) and Pd(OAc) 2 . The results are presented in Table 4, alongside the published data by Szabó and co-workers for symmetrical PCP and SCS pincer palladacycles, as well an unsymmetrical PCS pincer palladacycle (Scheme 2). 27 The results presented in Table 4 show that the family of palladacycles synthesised by our group, with SCN, NCN′ and PCN examples, provides the opportunity to fine tune catalytic activity, with the choice of donor group influencing the stereochemical outcomes of the aldol condensation. The novel PCN pincer palladacycle 7a provides the highest trans selectivity of all catalysts tested, and slightly more than the literature PCP example, whereas the SCN and NCN′ provide a greater proportion of the cis product. 27 In addition, catalysis of the coupling of a vinylepoxide and phenylboronic acid was tested, using the NCN′, SCN and PCN pincer palladacycles, and compared to literature SeCSe results shown in Fig. 7, 28 and our new results using the symmetrical SCS palladacycle (Scheme 2). The coupling and results are presented in Table 5.
The results presented in Table 5 show that the symmetrical pincer palladacycles, SeCSe and SCS achieve the greatest linear selectivity (>78%), whereas the unsymmetrical pincer palladacycles achieve a higher proportion of the branched product (>34%). Clearly the presence of unsymmetrical pincer palladacycles is having an effect of the stereochemical outcome of the reaction, increasing the proportion of the branched product.

Conclusions
The simple synthesis of three NCN′ ligands has been shown. Their C-H activation yielded three new NCN′ pincer palladacycles. Similarly, two novel PCN pincer palladacycles were syn-    thesised, with 5-,5-, and 5-,6-membered rings and all palladacycles were characterised by X-ray crystallography. The catalytic applications of the NCN′, SCN and PCN pincer palladacycles, along with several other examples were considered in two Pd(II)-mediated reactions, an aldol condensation and vinyl epoxide coupling. It was found that in the aldol condensation, the ability to fine tune the stereochemical outcome of the reaction is possible, by varying the donor atoms in the palladacycle catalyst. It was also shown in the vinyl epoxide coupling that the unsymmetrical pincer palladacycles achieve different product ratios than the symmetrical palladacycles tested, demonstrating the potential benefits of unsymmetrical palladacycles in catalytic applications.
Further work is being undertaken towards varying the families of unsymmetrical palladacycles, supported by DFT investigations into their bonding properties and reactivity. Moreover, it is hoped that such simple synthetic routes to unsymmetrical pincer ligands may encourage others to employ these in other areas of transition metal chemistry and catalysis.

General details
Solvents and chemicals were purchased from commercial suppliers and used without further purification, with reactions taking place open to atmosphere and moisture.