Amino acid-derived N-heterocyclic carbene palladium complexes for aqueous phase Suzuki – Miyaura couplings †

In this work, three ligands produced from amino acids were synthesized and used to produce five bisand PEPPSI-type palladium–NHC complexes using a novel synthesis route from sustainable starting materials. Three of these complexes were used as precatalysts in the aqueous-phase Suzuki–Miyaura coupling of various substrates displaying high activity. TEM and mercury poisoning experiments provide evidence for Pd-nanoparticle formation stabilized in water.


Introduction
The current dependence of the chemical industry on petrochemically-derived solvents has sparked a recent movement in developing sustainable alternatives. 1 Water is one of the leading contenders, 2 alongside ionic liquids 3,4 and biomassderived solvents. 5,6 The drive to utilize water as a solvent is due to its large heat capacity, non-toxicity, thermal stability, and large abundance on the planet. 2,7 The issue of designing catalytic systems suitable for ''in-water'' and ''on-water'' organic synthesis still remains and has gained startling momentum within research. 8,9 The most popular approach of adapting metalcatalyzed processes for use in water is currently the coordination of water-soluble ligands to the metal centre appropriate for the reaction. 9,10 This approach facilitates simple separation of insoluble organic products from the water medium, allowing for the possible recycling of the catalytic water layer. 11 Notable examples of homogeneous catalysis in water include Ir in asymmetric hydrogenation, 12,13 Ru in olefin metathesis, 14,15 and the work on Pd catalyzed C-C coupling reactions. [16][17][18] N-Heterocyclic carbenes (NHCs) have played a dominant role in the water-soluble ligand approach. 10,19,20 Their strong sand p-back donation facilitate their utilization as stable ligands for many catalytic systems. [21][22][23] In addition, the simple functionalization of the N-bound groups allows for easy incorporation of water soluble moieties, thus making them more suitable for water soluble catalyst design as opposed to other competitive ligands such as phosphines, 15,16 which can be susceptible to oxidation in aqueous systems. 24 The majority of work in this field has been dedicated towards Pd-NHC complexes for Suzuki-Miyaura couplings. 9,19 Complexes featuring monodentate and multi-dentate chelating ligands have been utilized for homogeneous couplings. Pd-NHC complexes of-bis, 25,26 pincertype, [27][28][29] and PEPPSI-type 17,30 (pyridine-enhanced precatalyst, preparation, stabilization and initiation) are among the most active for aqueous cross coupling catalysis.
The most common route towards producing novel NHC precursors involves the functionalization of imidazoles via quaternization with alkyl tails featuring a sulfonate or carboxylate group. 31 In this work, we utilize the greener approach of introducing water-soluble R-groups into the imidazolium precursor via a modified one pot Debus-Radziszewski heterocycle synthesis, using amino acids as precursors. This approach was initially developed in our group to obtain building blocks for the preparation of ionic liquids. 4,32 Here, we further expand on the use of amino-acid derived imidazoliums using these precursors for the synthesis of novel Palladium precatalysts, which were tested for their activity in the aqueous Suzuki-Miyaura cross-coupling reaction.

Synthesis and characterization
As established through previous research, imidazolium salts are useful moieties, with applications in various fields of chemistry such as ionic liquids, 4 poly-ionic liquids (PILs) 33 and NHC precursors. 31 The current strategies for synthesizing imidazolium salts fall short in the green and sustainable aspects, as a result of the use of haloalkanes and arenes to introduce novel R-groups. In this research we used the strategy of reacting glyoxal, formaldehyde and two equivalents of amino acid to synthesize imidazolium building blocks as previously described within our group. 32 Glyoxal and amino acids can be in principle obtained from sustainable sources, and the heterocycle synthesis is performed rapidly in water as illustrated in Scheme 1. Amino acid-derived NHC precursors have been documented for the synthesis of Au and Ru compounds, 34,35 but no examples of similar Pd-NHC complexes from amino-acid precursors have yet been developed to the best of our knowledge.
The building-blocks are isolated as internal carboxylate zwitterions 1a-1c in excellent yields (62-99%). Since initial attempts to directly complex these zwitterions to palladium were unsuccessful, conversion of these building blocks into suitable NHC precursors was performed via a simple esterification of the carboxylate arms in ethanol, using acetyl chloride as the source of both the acidic catalyst and the chloride counterion. In this way, compounds 2a-2c were isolated in very good yields as viscous ionic liquids. Compound 2a was previously synthesized by the Santini group using a different synthesis route. 36 Palladium-NHC complexes of bis-and mono-type were then successfully synthesized using two different synthesis methods (Scheme 2). Bis-NHC complexes were prepared using Ag 2 O to produce the dimeric silver-NHC complexes 3a and 3b, followed by transmetallation with 0.5 equivalents of PdCl 2 . 3a could be isolated as a white solid, while 3b was unable to be isolated from a mixture of the adduct and unreacted ionic liquid precursor. Both adducts are light and moisture sensitive, however stable under argon atmosphere for up to 5 days. Complexes 4a and 4b were produced as pale yellow solids. Mono-NHC PEPPSI-type complexes are generally more active than bis-NHC complexes due to the presence of the labile pyridine ligand, which can be exchanged allowing the coordination of the reagents to the metal center. Therefore in order to compare the catalytic activity, PEPPSI complexes were also targeted. These were synthesized using the procedure developed by Organ et al. 37 from the precursors 2a-2c, using K 2 CO 3 as a base in neat pyridine; producing yellow solids 5a-5c.
NHC Precursors 2a-2c incorporate glycine, b-alanine and phenylalanine R-groups into their structure, in order to introduce a variety of steric characteristics into the complexes. These amino acids were selected based solely on the success of their complexation. Attempts were made to synthesize complexes based on NHC precursors featuring alanine and leucine as well as pyruvaldehyde in the imidazolium backbone; however the resulting complexes either gave poor yields or were unable to be isolated. This is likely due to the varying pK a of the proton in the a-position to the N atom, causing non-selective deprotonation of this position over the imidazolium 2-H proton. The attempt to synthesize a bis-complex featuring ligand 2c was also unsuccessful. The presence of a small amount of a silver-2c carbene adduct could be confirmed by MS; however the lightsensitive carbene adduct was unable to be isolated from the unreacted ionic liquid imidazolium precursor.
The structure of the isolated complexes 4a-5b could be confirmed using X-ray structure analysis [38][39][40][41] with single-crystals grown from slow-evaporation solvent techniques. The molecular structures are presented in Fig. 1-4, bond lengths and angles in Tables 1 and 2. Compound 4a features two complexes within the asymmetric unit, and compound 4b displays a half-molecule in the asymmetric unit; a C 2 symmetry point group is revealed upon generating the full molecule. All complexes feature a squareplanar Pd(II) center, typical of 16 electron 4-coordinate palladium complexes. Distortions within the Cl-Cl-C-C coordination plane are minimal, the maximum deviations calculated for 4a are 0.027(3) and 0.025(3) Å for the coordinated carbon atoms (C1 and C23) of molecules 1 and 2 respectively. 4b displays no distortion as a result of the symmetry element. Afforded crystal structures confirm the trans-NHC arrangement of the bis-NHC complexes. Dihedral angles between the coordination plane and the imidazolium rings vary between 61-781 (see ESI †), similar torsions to those observed in bis-NHC complexes featuring more bulky ligands such as IPr or IMes. 42 The Pd-C bond lengths from 2.018(3) to 2.033(3) Å are in the expected range.
Structures of the PEPPSI complexes confirm the typical transarrangement of the NHC ligand to the coordinated pyridine. Once again very little distortion from the Pd(II) square-planar center is observed; the C1 atom of compounds 5a and 5b show plane deviations of 0.100(3) and 0.064(2) Å respectively. Dihedral torsions between the coordination plane and the imidazolium ring are notably larger than those of the bis-NHC compounds; however torsions between the coordination plane and the pyridine ring are smaller, at 47.79(7)1 and 39.31(6)1 for compounds 5a and 5b respectively. Another notable feature is that the carbonyl oxygen atoms from Glycine-derived complexes 4a and 5a are orientated away from the plane of coordination; whereas in b-alanine-derived complexes 4b and 5b they are aligned in parallel with the coordination plane. This is most likely a result of packing in the unit cell; the distance of the oxygen atoms from the Pd center rules out any coordination influence. The Pd-C bonds (5a: 1.969(2); 5b: 1.958(2) Å) are shorter as those in the bis-NHC complexes. The packing of 5a is completed by a chloroform molecule of crystallization.  Table 1 and ESI. † [89.34 (8)] - Table 2 Selected bond lengths and angles of complexes 5a and 5b 5a 5b 178.00 (9) 177.50 (6)

Water solubility
All five complexes are not initially soluble in water at pH 7, however PEPPSI-complexes 5a-5c were found to be readily soluble in both acidic and basic aqueous solutions with gentle heating. Within 1 hour at 40 1C, a yellow solution was clearly observed for pH ranges 1-5 and 9-14, suggesting successfully induced water solubility of at least 15 mg mL À1 for all complexes 5a-5c, higher than the required catalyst concentration. However the bis-complexes 4a and 4b remain insoluble even after heating at high temperatures under acidic and basic conditions. As the ester-functionalized NHC complexes are not initially water soluble, we hypothesize that water solubility was induced via hydrolysis of the ester moieties in the presence of base to introduce carboxylate functionalities.

Catalytic screening and optimization of conditions
Although complexes 4a and 4b were observed to produce moderate conversion in the coupling of bromoanisole with phenylboronic acid in ethanolic medium, they were proven to be unsuitable for aqueous-based systems (data not shown). Catalytic activity of the three PEPPSI complexes in pure water was therefore tested. The coupling of bromoanisole with phenylboronic acid using K 2 CO 3 as a base was used as an initial standard for catalyst evaluation (Table 3). Aside from its traditional role in the catalytic cycle, the base is also required to produce the water-soluble complex. The b-alanine-derived precatalyst 5b produced the marginally highest yield over 24 hours, although all precatalysts showed very similar results, despite the large differences in steric bulk, particularly in the phenylalanine-derived 5c (entry 3). A comparison was also made between the PEPPSI precatalysts and palladium acetate, a Pd(II) source typically used in Suzuki couplings (entry 4). Precatalysts 5a-5c all perform substantially better than Pd(OAc) 2 , highlighting the positive influence of the NHC ligands. The effect of in situ complex dissolution over catalyst pre-dissolution was also tested, comparing the immediate addition of the complex at the start of the reaction, or the predissolution of complexes for 2 hours before addition of substrates.
In each case we observed largely no difference in conversion between pre-activated catalysts and in situ dissolution over 24 hours (Table 4, entry 6).
Since 5b exhibited the best activity, it was selected for further condition optimization studies, varying base and temperature (Table 4). After base screenings, K 2 CO 3 still produced the best yield and was selected for further screenings. 60 1C was chosen as the best reaction temperature. Lower temperature screenings were proven to be unsuccessful in producing competitive yields (entry 5), and higher temperatures accelerated the destabilization of catalyst with increased formation of palladium black (entry 4). We also discovered under the optimized conditions that the catalyst loading could be reduced to 1 mol% without loss of activity (entry 6).
Water-soluble catalytic processes are well known to benefit from the addition of a phase-transfer catalyst, which aids the shuttling of base between the catalytic aqueous layer and the immiscible bromoarene. 17,18 Addition of two equivalents of tetran-butylammonium bromide (TBAB) was seen to dramatically improve yield up to 92% under optimized conditions (entries 6-8).

Further substrate screening
Having determined the optimum conditions for the reaction, screenings of various different substrates with complex 5b were performed in order to comment on the catalytic activity towards  different directing groups and aryl ring positions ( Table 5). As a general trend, the more electron-withdrawing substituents produce higher yields, as observed in the 4-fluoro and 4-acetyl moieties (entries 4 and 8). The effect of lowering precatalyst concentration was also tested, revealing that catalyst loadings of just 0.01 mol% could activate some of the more active aryl precursors (entries 4 and 8). In a number of cases, lowering the precatalyst loading increases conversion, particularly in the case of bromobenzene (entry 5). 1-Ethylbenzene was also tested, showing the catalyst's ability to also activate benzyl reagents; a moderate 78% yield was observed in this case (entry 7). Good yields were also observed when the boronic acid is exchanged with a pinacol ester (entries 13-15). Entry 13 particularly highlights the increased yield with decreased precatalyst loading. Chloroaryls were also tested but performance was poor. All substrate screenings were performed over 24 hours to ensure full conversion, while a detailed kinetic study was not included in this work. However, preliminary kinetic investigations of the bromoacetophenone coupling indicate that the reaction is finished within 50 minutes for both the predissolved system, and within one hour for the in situ dissolved system. The kinetic plot is available in the ESI. † Finally, it is worth mentioning that the possibility of catalyst recycling is one of the prominent advantages of using aqueousphase systems. In a proof-of-principle experiment, the separated aqueous layer was able to be recycled for an additional run. 100% yield was reached in the second cycle within two hours for the bromoacetophenone coupling with 2 mol% catalyst loading, using toluene to extract the product. Since recycling was not of primary interest within this report, detailed investigation into recycling with this family of complexes will be the subject of further study.

Formation of nanoparticles
A number of observations during the initial screenings lead us to believe that Pd(0) nanoparticles were formed during the reaction, a common factor in NHC-complex catalysis in aqueous media. 17,18,28,45 In particular, the presence of a black precipitate, the discovery that higher temperatures and catalyst loadings can hinder higher yields, and the very similar yields despite large differences in ligand sterics pointed to the possible formation of Pd(0) nanoparticles. In order to confirm this, TEM analysis was performed on the isolated precipitate from a coupling of bromoanisole with phenylboronic acid using precatalyst 5b (Fig. 5), which clearly shows the formation of nanoparticles taking place. Interestingly, during the screenings using TBAB, we noticed a significantly smaller amount of precipitate being produced in comparison to the initial TBAB-free screenings, suggesting a more effective stabilization of the nanoparticle in solution. To investigate this we performed TEM imaging on the complex 5b after dissolution in alkaline water with TBAB. Nanoparticles were once again observed, gathered inside darker aggregates, with an average size of 8.7 AE 0.1 nm (Fig. 6). This suggests that the nanoparticles are stabilized by TBAB doubling as a surfactant, in line with what was observed in other nanoparticle-based systems. 46,47 In order to determine the activity of nanoparticles in the coupling, mercury poisoning experiments were performed for complexes 5a-5c using a molar Hg : Pd ratio of 300 : 1 in the presence of TBAB (Table 6). Catalytic activity was significantly reduced, 5b being poisoned entirely, and activity for 5a and 5c reduced to 16% and 14% respectively. This experiment confirms the active role of the nanoparticles as the catalyst, indicating that the reaction is mostly heterogeneously catalyzed. Although the nanoparticles are determined to be responsible for the majority of the catalytic activity, however, the residual activity after Hg addition in the cases of 5a and 5c suggests that a number of catalytic species are likely to be present. This may include different nanoparticle species, metal clusters and molecular complex species with partially-hydrolyzed ester groups.

Conclusions
A simple, green method of synthesizing N-heterocyclic carbene (NHC) precursors with various different functionalities was applied to produce five NHC-palladium complexes. The three pyridine-coordinated complexes were suitable as precatalysts for aqueous phase Suzuki-Miyaura cross-coupling reactions, achieving good catalytic performance with a precatalyst loading as low as 0.01 mol%. It was ascertained that the catalysis proceeds according to a ''catalyst cocktail'' model, with Pd-nanoparticles responsible for most of the catalytic activity, as evidenced by transmission electron microscopy and mercury poisoning. This research builds on work involving amino-acid derived NHCs and addresses a number of sustainability issues within homogeneous catalysis, which further builds bridges between green chemistry and organometallic catalysis.

Syntheses
All purchased chemicals were used as supplied without further purification. Complex synthesis was carried out under an argon atmosphere using standard Schlenk techniques. The solvents used for complexations were purchased dry from acros organics and stored over 3 Å molecular sieves. Water used as a catalysis medium was purified using an SG Integra UV Plus millipore water purification system. 1

Ligand precursors 2a-2c
A flask with a Soxhlet apparatus containing molecular sieves (3 Å) was charged with the corresponding imidazolium zwitterion (0.01 mol) and a stir bar under argon. Ethanol (150 mL) was then added as a reactant and solvent, followed by acetyl chloride (0.012 mol, 1.2 equiv.). The mixture was refluxed at 150 1C for 4 hours. After cooling, the solution was filtered through a PTFE membrane and the solvent was removed under reduced pressure. The obtained residue was washed with diethyl ether and dried affording the desired ligand precursors, which were stored under an argon atmosphere.

Silver adducts 3a and 3b
A Schlenk tube was charged with 2a or 2b (0.003 mol) and a stir bar under argon. Dry acetonitrile was added (8 mL) and the mixture was gently heated (45 1C) with stirring for 10 minutes. Silver oxide (0.32 g, 0.0015 mol) was then added under flow of argon and the mixture stirred at room temperature for 12 hours. The mixture was then gently heated again (45 1C) for 5 minutes to aid dissolution of the adduct, and filtered under argon by use of a cannula. The solvent was removed from the filtrate in vacuo to give the desired product.
3a: white solid, yield 83%. 1  Bis-NHC Pd complexes 4a and 4b 4a: a Schlenk tube was charged with 3a (0.79 g, 0.002 mol) and a stir bar under argon. Dry acetonitrile was added (8 mL) and the mixture was stirred until dissolution. Palladium(II) chloride (0.18 g, 0.001 mol) was then added under flow of argon and the mixture was further stirred at 50 1C for 12 hours. The suspension was filtered and the filtrate dried in vacuo. The obtained crude solid was recrystallized from CHCl 3 to afford the desired product. Crystals suitable for X-ray analysis were prepared from a slow evaporation chamber of CH 2 Cl 2 /THF. Yellow solid, yield 36%. 1 (8 mL) and the mixture was stirred until dissolution. Palladium(II) chloride (0.2 g, 0.0011 mol) was then added under flow of argon and the mixture was stirred at room temperature for 12 hours. The mixture was filtered and the filtrate was transferred to a freezer at À20 1C overnight to afford the desired product. Crystals suitable for X-ray analysis were prepared from a slow evaporation chamber of CH 2 Cl 2 /THF. Yellow solid, yield 33%. 1 6.51 (dd, J = 8.1, 6.9 Hz, 2H), 4.14 (q, J = 7.1 Hz, 4H), 3.52 (dd, J = 7.5, 2.7 Hz, 4H), 1.13 (t, J = 7.1 Hz, 6H). 13

Catalytic screening
General procedure for the Suzuki-Miyaura catalytic screenings. A reaction vial was charged with base (1.0 mmol), phenylboronic acid/pinacol ester (0.75 mmol), complex (0.01 mmol) and a stir bar under air. Water (2 mL) was added, followed by aryl halide (0.5 mmol). The mixture was stirred at 60 1C for the specified time. Upon completion, the mixture was cooled to 5 1C. The products were then extracted with ethyl acetate (2 Â 5 mL), filtered, and evaporated to dryness. The residue was directly dissolved in CDCl 3 for 1 H NMR analysis. Yields were calculated using trimethoxybenzene (0.3 mmol) as an internal standard.
General procedure for the mercury poisoning experiments. A reaction vial was charged with base (1.0 mmol), TBAB (1.0 mmol), complex (0.01 mmol) and a stir bar under air. Water (2 mL) was added, and the mixture was stirred at 60 1C for 2 hours to pre-dissolve the complex. Mercury was added (3 mmol), followed by phenylboronic acid (0.75 mmol) and bromoanisole (0.5 mmol). The reaction was stirred for 24 hours under 60 1C heating. Upon completion, the mixture was cooled to 5 1C. The products were then extracted with ethyl acetate (2 Â 5 mL), filtered, and evaporated to dryness. The residue was directly dissolved in CDCl 3 for 1 H NMR analysis. Yields were calculated using trimethoxybenzene (0.3 mmol) as an internal standard.

Microscopy
General procedure for TEM analysis. The samples were prepared by pipetting a drop of solution or dispersion onto copper plates coated in a carbon film. The precipitate sample was prepared by dispersing the centrifuged and washed solid in chloroform. The nanoparticle colloidal solution in water was prepared by dissolving 5b in water with K 2 CO 3 and TBAB.