Stirring or milling? First synthesis of Rh(I)-(di-N-heterocyclic carbene) complexes both in solution and in a ball mill Journal of Organometallic Chemistry

An environment-friendly, convenient, fast and solvent-free mechanochemical approach have been accomplished for the synthesis of several diimidazolium salts and the bridging dinuclear rhodium(I) e N-heterocyclic carbene complexes of the type [{RhCl(cod)} 2 ( m -di-NHC)] derived from them. The com- pounds were synthesized also by the classical solvent method and the results of the two approaches were compared. A systematic study of both the mechanochemical and the solvent syntheses has also been carried out to determine the effects of various factors in ﬂ uencing the reactions. This is the ﬁ rst report on the mechanochemical synthesis of poly-NHC metal complexes as well as NHC e Rh complexes in ball mill. © 2020 The Authors.


Introduction
N-heterocyclic carbenes or NHCs have proven themselves as most powerful tools in the domain of modern chemistry because of their outstanding potential and wide application in the field of organometallic/coordination chemistry, catalysis, photophysics, medicine and material science [1e5]. Hydroformylation and carbonylation were among the first processes realized with rhodium-NHC complexes as the catalysts [6e9]. Poly-NHCs, of which di-NHCs are the most common and abundant, have attained significant attractions in the last two decades due to their allowance to form various organometallic compounds with diversity in geometries. They are relatively easy to synthesize and their properties can be modified simply by swapping the linker or changing the length of the linker, or placing various substituents onto the linker [2,10e13]. Transition metal complexes bearing NHCs as ligands have obtained widespread application in several fields of chemistry. RheNHC complexes deserve special mention due to their extensive usage in different domains of catalysis chemistry such as hydrogenation, dehydrogenation, hydroamination, hydration, CeC cross-coupling etc. [1e5, 10,14e17]. Moreover, a recent study tells about the potential of RheNHC complexes as anticancer drugs [18].
A number of excellent articles on the synthesis of NHC-metal complexes are already available [3,14e17,19e26]. Still, the development of safer, more efficient and cleaner synthetic methodology is on high demand. From the viewpoint of green chemistry, synthetic methods under solvent-free or solvent-less conditions are of high desire as the energy-cost as well as the waste-production are reduced. The traditional solvent synthesis has a number of disadvantages such as long reaction time, high energy costs, furthermore it most often has a negative impact on the environment due to the prevalent usage of volatile organic solvents.
Mechanochemistry has a long history in the everyday life and in laboratory-scale chemical syntheses [27]. In the laboratory, mechanochemically assisted reactions are usually carried out either in planetary or in vibrating ball mills, where the reagents are loaded in a milling jar and movement of balls generates the mechanical force to the reagents [28,29]. The possibility to adjust several instrumental parameters makes the processes highly efficient and reproducible [30e32]. Nevertheless, in some cases simple grinding together of the solid reactants in a mortar with a pestle may also bring excellent results. Typically, such procedures require solvents only for the workup and purification of the products. In the last two decades a huge increase in the applications of such methods in chemical research could be seen, and the developments have been discussed in several recent reviews [28,33e37]. Today, mechanochemistry ranks amongst the top ten world changing technologies according to IUPAC [38]. In addition to introduction of grinding and milling into classical organic synthesis to allow the design of solvent-less procedures [39], mechanochemical syntheses have been more and more applied also in organometallic chemistry [34]. Procedures in controlled atmospheres (H 2 , CO, etc.) have been developed [40] and various techniques have been introduced to follow the reactions in situ for kinetic analysis of the syntheses [41]. With regard to the subject of our present study, the mechanochemical synthesis of Pd(II)-and Pt(II)eN-heterocyclic carbene complexes (with mortar and pestle, [42]) and the synthesis of Ag(I)-, Au(I)-, Cu(II)-and Pd(II)eNHC complexes (in ball mills, [43]) deserve special mention. In general, however, applications of mechanochemical methods for synthesis of important organometallic catalysts are still rare.
In the present work, azolium salts serving as precursors to several new (eCH 2 e) n bridged (n ¼ 1 or 4) diimidazole-2-ylidene ligands and their bridging dinuclear Rh(I)-complexes have been synthesized both via the classical solution synthesis and by solventfree one-pot mechanochemical method and the results are compared. To the best of our knowledge, this is the first article to report the mechanochemical synthesis of di-NHC metal complexes as well as the first RheNHC complexes synthesized in ball-mill.
In our case, the classical solution synthesis of 2a,b and 3a,b as well as that of 4a,b and 5a,b involved heating of solutions of the reaction partners (Schemes 1 and 2) in the appropriate solvent for a period of time during which the reactions completed. Conversely, the mechanochemical synthesis of the same compounds consisted of milling together the reaction partners in a planetary ball mill for the time sufficient to achieve high conversions; these reactions were carried out in air. The successful formation of RheNHC complexes was validated by the appearance of the characteristic RheC carbene doublet signal(s) at d 180e185 ppm with J x 50 Hz in the 13 C NMR spectra of the compounds. 4a,b and 5a,b are dinuclear Rh(I) complexes in which two RhCl(cod) units are bridged by the appropriate carbene ligand derived from 2a,b or 3a,b. The presence of two rhodium centers was confirmed by the two doublet signals each in the region of d 98e101 ppm and d 68e71 ppm which are assigned to RheC vinyl carbons of coordinated cod. Interestingly, in the mechanochemical syntheses described here, the dinuclear complexes 4a,b and 5a,b were exclusively obtained with no sign of chelate [RhL] or macrocyclic [Rh 2 L 2 ] species which are known from solution syntheses [53,54]. HR ESI-MS measurements also confirmed the composition of the complexes. It should be added here, that none of the above ligands and complexes was obtained previously by mechanochemical synthesis.

Structural characterization by single crystal X-ray diffraction
The solid state structures of the carbene ligand precursors 3a, and 3b, and metal complexes 4a, 4b and 5b have been determined by single crystal X-ray diffraction studies.
3a was dissolved in methanol in a small tube and saturated with KPF 6 . Then, in a closed container, the tube was half immersed into diethyl ether and stored at À18 C. In four weeks, colourless crystals, suitable for XRD measurements, appeared on the wall of the tube. Crystals of 3b were obtained from a methanolic solution layered with diethyl ether.
Both 3a and 3b crystallized as triclinic (P1 space group). The asymmetric unit of 3a contains a cationic NHC-precursor and two disordered PF À 6 ions and a methanol ( Fig. 1) while that of 3b contains half of the molecule and a chloride.
In the molecule of 3b, the imidazole centroids are long apart (7.070 Å, Fig. S1); capped sticks representation is shown on Fig. 2. In the crystal, the molecules are packed in a stair-like arrangement (Fig. S2), in which the parallel imidazole ring planes are in a distance of 3.877 Å while the distance of the arene planes is 3.649 Å. All of the bond lengths and bond angles in the imidazole rings are as expected [55]. The supramolecular architecture is further stabilized by p-p stacking interactions between the fiveand six- Compound 2b (as a dinitrate salt) in its co-crystal with [Cu(NO 3 ) 2 (H 2 O) 2 ] has been previously studied by X-ray diffraction by Doimeadios [56]. Although the solid state structures of 3a, and 3b, discussed above, are very similar to that of 2b published in Ref. [56], however, the bond lengths and angles cannot be quantitatively compared due to the large error of the structure of 2b (R 1 ¼ 12.89%).
Crystallization of 5b was attempted by several methods, unfortunately the best crystals (obtained by slow evaporation of its solution in CH 2 Cl 2 ) were still of rather bad quality. 5b crystallized in monoclinic P2 1 /n space group, the unit cell contains half of the molecule. After refinement of the best dataset, the error still remained large (R 1 ¼ 17.51%, wR 2 ¼ 41.25%) so while the molecular model ( Fig. 3) proved to be suitable, the bond distances and bond angles cannot be evaluated.
The Rh(I)-complexes 4a and 4b were crystallized from benzene solutions layered with diethyl ether and were isolated as yellow crystals not sensitive to air and moisture. Both complexes crystallize in the monoclinic crystal system, however the space groups are different (C2/c for 4a, and P2 1 /c for 4b). Capped sticks representations of the molecular structures are shown on Fig. 4, while the selected bond distances and angles can be found in Table 1.
There little back-donation [55]. In 4a, the two imidazole rings are at a dihedral angle of 87.47 , in contrast, in 4b, these planes are almost parallel with a dihedral angle of 7.84 .
In the crystal of 4a, there are no interactions between the phenyl groups of the neighboring molecules, and the centroids of the imidazole rings are 3.719 Å and 5.021 Å apart (Fig. S3). In the case of 4b, the flexibility of the butylene bridge allows p-p stacking interactions between the imidazole rings (their distance is 3.654 Å) and also between the phenyl rings with a distance of 4.534 Å (Figs. S4 and S5).
On the basis of the Rh1eRh2 distances (6.938 Å in 4a and 8.277 Å in 4b) no RheRh bonding interactions can be assumed in these complexes.
It has previously been observed, that the Rh(I)-complexes of di-NHC ligands with a methylene bridge and aliphatic N-substituents preferred chelate coordination [53,54]. In contrast, 4a contains the N-benzyl-substituted di-NHC ligand (2a) in bridging position between the two RhCl(cod) moieties. Only two examples of similar structure have been determined so far [57,58]. The main aim of the present work was the exploration of the usefulness of mechanochemical synthesis in the field of Rh(I)eNHC complexes and its comparison to the solution methods of synthesis.
For this purpose, the syntheses were also carried out by using a planetary ball mill. Initially, the influence of ball milling conditions like milling time, frequency of milling and size of bearing balls were investigated (Tables 2 and 3). To do so, a milling cycle of 4 min (2 min milling, followed by another 2 min pause) to avoid the overheating, was used. Ligand 2b ( Table 2) was taken as reference, and 10 pieces of bearing balls with 5 mm diameter (in the  following: 10 Ø 5 mm) were employed to check the effects of frequency and milling time. The yield was found to be increased with the rise of frequency (Table 2, entries 2, 5) and after an initial increase, it levelled off with longer milling time ( Table 2, entries 1e3).
In contrast, the amount and the size of balls influenced significantly the yield ( Table 2, entries 4e8). It was finally found that a mixture of 10 Ø 5 mm and 10 Ø 8 mm balls yielded the best result (71% isolated yield) after 90 cycles at 550 rpm frequency ( Table 2, entry 8).
In the case of the metal complexes, 4b was taken as a reference (Table 3) with 10 Ø 5 mm and 10 Ø 8 mm balls and the best yield (74%) was observed after 45 cycles at 550 rpm frequency (Table 3, entry 4). Under these optimized conditions, during the synthesis of 4b, the outside surface temperature of the milling jar was regularly checked with the use of a remote infrared thermometer. It was found that by the end of the 45th cycle (i.e. in 3 h reaction time), the surface temperature of the jar increased to 34.1 C from 23.6 C measured at the start of milling.
As the volume of the milling jar was fixed (12.5 mL), the influence of the amount of starting material for the preparation of diimidazolium salt 2b from 1b, and metal complex 4b from 2b was examined (Tables 4 and 5). It was observed that 100 mg of 1b (Table 4, entry 2) and 150 mg of ligand precursor 2b (Table 5, entry 3) as reactants resulted in the highest yields.
In addition, the influence of inert milling aids like quartz, silica, alumina and Celite Hyflo Supercel were also investigated. However, no positive influence was observed.
In the case of solution syntheses, the impact of reaction time and  temperature were initially checked for the preparation of carbene ligand precursor 2b and metal complex 4b taken as references (Tables 6 and 7). According to the results, overnight reactions at high temperature (80 C to reflux temperature; Table 6, entry 4; Table 7, entry 4) provided the best yields.
To compare the mechanochemical method with the classical solvent approach, the synthesis of 2b and 4b were carried out maintaining the same reaction time, i.e. 6 h for the preparation of diimidazolium salt 2b ( Table 2, entry 8 vs Table 6, entry 3) and 3 h for the synthesis of the Rh(I)-complex ( Table 2, entry 4 vs Table 6, entry 3). The comparison clearly shows that the ball mill synthesis results in higher yields than the solution method (71% vs 49% for the carbene precursor 2b; and 74% vs 18% for complex 4b). In addition, since the mechanochemical synthesis requires solvents only for extraction and purification, the overall solvent need is significantly reduced e up to 50% in the small scale syntheses of ligands and complexes described in this study (see Experimental part).

Conclusion
Convenient and efficient novel mechanochemical methods of synthesis for two (eCH 2 e) n -bridged diimidazolium salts (precursors to di-NHC ligands) and four bridging dinuclear rhodium(I)e N-heterocyclic carbene complexes of the type [{RhCl(cod)} 2 (m-di-NHC)] were developed which are characterized with shorter reaction times and substantially reduced need of organic solvents compared to the classical solution syntheses of the same compounds. Together with the simplicity of the procedure, zero-solvent condition, and possible multipurpose applications, directed us to the conclusion that the mechanochemical (ball mill) synthesis of metal complexes in several cases may prove superior in comparison to the classical solution synthesis approach. In addition, the solid state structures of two new diimidazolium salts (3a and 3b) and three of the new [{RhCl(cod)} 2 (m-di-NHC)] complexes, 4a, 4b, 5b) were determined by single crystal X-ray diffraction.

General information
The Rh-metal precursor [RhCl(cod)] 2 was prepared as described in Ref. [61]. Synthesis of 1,1ʹ-di(imidazole-1-yl)methane, 1a and 1,4di(imidazole-1-yl)butane, 1b were done according to the literature     [46] are known compounds, however, in addition to the traditional solvent synthesis, we obtained them also in reactions in a ball mill. The identity and purity of these four compounds were checked by correlating their respective 1 H, 13 C and ESI-MS spectra to those available in the literature. All other chemicals and solvents were purchased from Sigma-Aldrich, Alpha Aesar, Merck, Molar Chemicals Kft. and VWR International and employed as received without further purification. Analytical thin-layer chromatography (TLC) was carried out on Kieselgel 60 F254 plates from Merck and the plates were visualised under UV fluorescence light at 254 nm. The column chromatography was executed on silica gel from Sigma-Aldrich (70e230 mesh, 63e200 mm).
Reactions in ball mill were carried out with the use of a planetary milling instrument model 'RETSCH PM 100' with a stainless steel jar (12.5 mL) and G100 ball bearings (Ø 5 mm and Ø 8 mm) operated at room temperature. In the generally used protocol, 1 cycle consisted of 2 min milling followed by 2 min cooling at ambient temperature; after an initial warming period the stabilized temperature of the milling jar in a 45 cycles procedure was around 34 C, estimated by the use of a remote infrared thermometer. 1  Single crystals were examined on a Bruker D8 Venture diffractometer (SC-XRD) and data processing was managed by Olex 2 software [63] including SHELX programs [64]. The molecular images were prepared by the Mercury CSD-4.3.0 software [65]. The crystallographic data (excluding the structure factors) for 3a, 3b, 4a, 4b, 5b were deposited at Cambridge Crystallographic Data Centre, as CCDC-1990551 and 1981017e1981020.

Mechanochemical synthesis
A 12.5 mL ball milling jar was charged with 10 Ø 5 mm and 10 Ø 8 mm stainless steel balls, 1a (100 mg, 0.675 mmol) or 1b (100 mg, 0.526 mmol) and 2-(chloromethyl)-1,3,5-trimethylbenzene (for 1a: 228 mg, 1.35 mmol; for 1b: 177 mg, 1.052 mmol). The mixtures were milled over a period of 90 cycles (1 cycle ¼ 2 min milling þ 2 min pause) at 550 rpm. Afterwards, the jar and the Table 6 Effect of the reaction temperature and time for the formation of 2b in its solution synthesis from 1b and benzyl chloride. a   in one portion. The solutions were stirred overnight at 80 C; the final conversion was checked with TLC. The resulting solutions were then filtered, the residues were washed with 2 Â 5 mL toluene and the filtrates were collected. The combined filtrates were evaporated to dryness and the residues were purified using silica gel column chromatography with a CH 2 Cl 2 and EtOAc mixture (1:1) as eluent and dried under vacuum. Yellow powders. Yield 80% for 4a (245 mg), 77% for 4b (224 mg), 51% for 5a (143 mg) and 65% for 5b (174 mg).

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.