Mechanochemical solid state synthesis of copper(I)/NHC complexes with K3PO4

We therefore examined different approaches to avoid the transmetallation step (45) and to establish a protocol for the direct synthesis of 5 in solution from imidazolium salt 3 (Table 1, liquid state approaches) [48]. The use of strong bases such as n-BuLi or NaOt-Bu (in various equivalents, Table 1, entries 1 and 2) or weak bases (Et3N, K2CO3 or K3PO4,Table 1, entries 3–5) either did give no conversion to 5 at all or delivered a catalytically inactive complex, which we assign to a CO2 adduct of 5 (Table 1, entry 4) [51,52]. We hypothesize that in this CO2 adduct, the guanidine moiety is unavailable to perform its assisting part in catalysis through hydrogen-bonding interaction [48]. As additional evidence to support the formation of the CO2 adduct of 5, we can show that bubbling of CO2 through a solution of 5 leads to catalytically inactive complexes (see Supporting Information File 1 for details). This also supports the notion that during catalytic ester hydrogenation, the guanidinium moiety acts as a hydrogen bond donor to the esters [48]. The formation of a CO2 adduct hinders the ability to form hydrogen bonds. Furthermore, utilizing Cu2O for a “built-in base” approach did not give complex 5 (Table 1, entry 6).

Table 1: Attempted direct synthesis of bifunctional catalyst 5 from imidazolinium salt 3: liquid and solid state approaches.

[Graphic 1] Entry Reagents Conditions Results Liquid state aproaches
[20] 1 strong bases 1.00 equiv 3, 1.10 equiv CuCl, 3.00 equiv n-BuLi THF, 0 °C → rt, 16 h no formation of 5 2   1.00 equiv 3, 1.10 equiv CuCl, 1.05/2.00/3.00/5.00 equiv NaOt-Bu THF, rt, 16 h no formation of 5 3 weak bases 1.00 equiv 3, 1.10 equiv CuCl, 3.00 equiv NEt3 THF, 0 °C → rt, 16 h no formation of 5 4   1.00 equiv 3, 1.00 equiv CuCl, 2.00 equiv K2CO3 acetone, 60 °C, 16 h formation of catalytically inactive 5∙CO2 5   1.00 equiv 3, 1.00 equiv CuCl, 2.00 equiv K3PO4 acetone, 60 °C, 16 h no formation of 5 6 “built-in base” approach 1.00 equiv 3, 2.0 equiv Cu2O CH2Cl2, 4 Å MS, 60 °C, 16 h no formation of 5 Solid state approaches (steel vessel (12 mL), 6 steel balls (1 cm diameter) if not noted otherwise) 7 strong bases 1.0 equiv 3, 1.0 equiv CuCl, 1.5 equiv NaOt-Bu 450 rpm, 4 h no formation of 5 8   1.0 equiv 3, 1.0 equiv CuCl, 1.5 equiv NaOH 450 rpm, 4 h no formation of 5 9   1.0 equiv 3, 1.0 equiv CuCl, 1.5 equiv KHMDS 450 rpm, 4 h no formation of 5 10   1.0 equiv 3, 1.0 equiv CuCl, 1.5 equiv NaH gastight zirconia vessel (45 mL), 6 zirconia balls (1.5 cm diameter), 450 rpm, 4 h formation of 5 observed, inseparable mixture of products 11 no added base [46] 1.0 equiv 3, 1.0 equiv CuCl 450 rpm, 4 h formation of [3][CuClBr]– observed by HRMS 12 “built-in base” approach 1.0 equiv 3, 0.5 equiv Cu2O 450 rpm, 4 h no formation of 5 13 weak base 1.0 equiv 3, 1.0 equiv CuCl, 1.5 equiv K2CO3 450 rpm, 4 h 44% of 5 14   1.0 equiv 3, 1.0 equiv CuCl, 1.5 equiv K3PO4 450 rpm, 4 h 91% of 5

Since our attempts to establish direct synthetic routes to 5 from 3 in liquid state were not fruitful, we turned our attention to the mechanochemical synthesis of bifunctional catalyst 5, based on two recent reports on preparation of copper(I)/NHC complexes [45,47].

All mechanochemical syntheses were carried out in a planetary ball mill and the vessel was loaded in an argon-filled glovebox. Copper(I) chloride, imidazolium salt 3 and the appropriate base were mixed (in a molar ratio of 1.0:1.0:1.5, respectively) and ground for 4 hours. Afterwards purification included dissolving the crude product in CH2Cl2, filtration over a PTFE syringe filter and concentrating the filtrate under reduced pressure. Employing strong bases such as KHMDS, NaOt-Bu or NaOH did not lead to the desired product (Table 1, entries 7–9). All three approaches have in common that the conjugated acid of the added base is a liquid. In the literature, the improvement of mechanochemical syntheses by addition of small amounts of a liquid have been reported (LAG, liquid-assisted grinding) [43]. However, in our case, the formation of small amounts of liquid during the milling process lead to agglutination of the remaining solids and therefore insufficient homogenization of the reaction mixture. This gave a mixture of compounds, in which the envisaged complex 5 could not be identified.

A different approach was made using sodium hydride as a base (Table 1, entry 10). Instead of small amounts of liquid, here, deprotonation leads to the formation of dihydrogen. Hence, another gastight mill was utilized for this approach. Unfortunately, a successful synthesis of 5 directly from 3 was not possible under these conditions: NMR analysis of the resulting mixture indicated the presence of 5, but also of unwanted side-products that could not be identified. Purification of 5 from this complex mixture turned out not to be feasible. Further modifications of the milling conditions did not lead to the elimination of these side-products, therefore the experiments with NaH as a base were discontinued. As a side comment, the addition of no base at all led to the formation of the imidazolinium cuprate ([3][CuClBr]−, Table 1, entry 11) [46]. The direct transition of the “built-in base” approach conditions to mechanochemical synthesis (copper(I) oxide and imidazolium salt 3 as starting materials), lead to no formation of 5 (Table 1, entry 12).

The use of K2CO3 for the mechanochemical synthesis of copper(I)/NHC complexes [46,47] was feasible for the preparation of 5 (Table 1, entry 13). Importantly, we found that extending on this concept also K3PO4 could be employed equally well while giving significantly higher yields than the previous protocol [46,47] (Table 1, entry 14). All of the approaches discussed here are attractive due to the use of copper(I) chloride as the copper source. Interestingly, the use of K2CO3, which led to the formation of a catalytically inactive postulated CO2 adduct of 5 in the liquid state synthesis, did lead to catalytically active 5 in the mechanochemical approach. In a similar vein, the different outcome with K3PO4 as a base (which led to no catalytically active complexes in the liquid state synthesis) was surprising, as in the ball mill, clean and catalytically active copper(I) complex 5 was obtained. To avoid the possible formation of the catalytically inactive CO2 adduct when employing K2CO3 for the synthesis of 5 we decided to use K3PO4 for subsequent investigations (see also below, Table 2). Even though the imidazolium bromide salt 3 was employed in combination with CuCl as copper(I) precursor, elemental analysis of 5 clearly supported the formation of 5 as a chloride salt (see Supporting Information File 1).

Table 2: Synthesis of standard Cu(I)/NHC-complexes using K3PO4 as a weak base (standard procedure: steel vessel (12 mL), 6 steel balls (1 cm diameter), 450 rpm, 4 h). (dimer = [(NHC)2Cu+]Cl−).

[Graphic 2] Complex Yield Cu(0)/O2
[45] CuCl/K2CO3[46] [Cu(IMes)Cl] (7a) 71% (8% homoleptic cationic Cu(I) complex) 85% 65% [Cu(SIMes)Cl] (7b) 73% 76% 53% [Cu(IPr)Cl] (7c) 63% 82% 78% [Cu(SIPr)Cl] (7d) 64% (12% homoleptic cationic Cu(I) complex) 65% 66%

For the optimized protocol, the starting materials were mixed in a steel vessel and ground at 450 rpm for a total time of four hours. After ball milling, an off-white powder was obtained which gave complex 5 in very good yield of 91% after extraction with CH2Cl2 and filtration. NMR analysis of 5 matched previously reported data [48,49] and showed no side products. It has to be mentioned that complex 5 synthesized via the mechanochemical route is isolated as a CH2Cl2 adduct (5/CH2Cl2 = 1:1) as confirmed by NMR spectroscopy and elemental analysis. If complex 5 is formed via the liquid state synthesis [48,49], also a CH2Cl2 adduct is isolated, albeit with a 5/CH2Cl2 ratio of 2:1.

In order to demonstrate the general applicability of the K3PO4-based protocol for the mechanochemical synthesis of copper(I)/NHC complexes, we decided to prepare the most common copper(I)/NHC complexes 7ad [5,6] employing our method (Table 2). When the corresponding imidazoli(ni)um salts 6ad were submitted to the standard protocol, complexes 7ad were obtained with acceptable yields, with similar yields compared to previous methods. In some cases, the homoleptic cationic copper(I) complexes [(NHC)2Cu]+CuCl2− were observed as side products [48,53].

We decided to directly compare complex 5 from mechanochemical synthesis (5bm) with its counterpart from the liquid state transmetallation route (5ls) in catalysis. We found that 5bm was catalytically active, however displaying slightly diminished activity in general most likely due to different adduct ratio inhibiting the catalytic activity (Scheme 3). This was established using the standard reactions for catalytic hydrogenations with copper(I)/NHC complexes [4]. In this vein, we tested complex 5 from solid and liquid phase synthesis in the catalytic hydrogenation of esters, carbonyl compounds and in the semihydrogenation of alkynes. In the catalytic hydrogenation of ethyl benzoate (8) lower overall conversion to benzyl alcohol (9) and lower yield was found with 5bm (65% conv. and 53% yield with 5bm, in comparison to 100% conv. and 80% yield with 5ls; Scheme 3a). We hypothesize that the higher amount of CH2Cl2 as part of the prepared complex, which is not a suitable solvent for catalytic ester reduction with H2[48], led to lower catalyst activity. Possible coordination of residual phosphate to the guanidine moiety was excluded as analysis by 31P NMR experiments. The copper(I)-catalyzed 1,2-reduction of functionalized ester 10 was also successfully achieved using the ball mill synthesized bifunctional catalyst 5bm, again with slightly diminished yields and conversions.

[1860-5397-19-34-i3]

Scheme 3: Application of bifunctional catalyst 5 in copper(I)-catalyzed hydrogenations: comparison of 5 prepared by solid state/ball milling (5bm) and liquid state (5ls) synthesis. Standard conditions: Substrate (0.40 mmol), 10 mol % 5, 1.1 equiv NaOt-Bu, 1.3 equiv 15-crown-5, 100 bar H2, 1,4-dioxane (3 mL), 70 °C, 24 h.

Application of the ball mill-synthesized complex 5bm in the alkyne semihydrogenation of tolane (12) gave (Z)-stilbene (13) with full stereoselectivity in good yield (86%, Scheme 3b). Noteworthy, the complex 5 was never evaluated in this reported reaction. Therefore, 5bm behaves similarly to other copper(I)/NHC complexes in this transformation [54-60]. The catalytic 1,2-reduction of carbonyl compounds is mainstay for copper(I)/NHC complexes [61-67], which is why we also tested 5bm in these transformations: The 1,2-reduction of benzaldehyde (14) and acetophenone (15) proceeded with good yields (Scheme 3c). No aldol addition for the acetophenone substrate has been observed although working under strongly basic conditions [68,69].

留言 (0)

沒有登入
gif