3.3.1. Multi-Metallic Catalysts

Similar to what was observed for the azidolytic ARO of *meso*-epoxides catalyzed by Cr–salen complexes, preliminary kinetic studies indicated a second-order dependence on catalyst **63** in the HKR of terminal epoxides [109]. Consequently, a similar cooperative bimetallic pathway was proposed as a possible mechanism for the HKR of epoxides (Figure 1). Since then, numerous studies have reported multi-metallic catalysts where several salen moieties are linked together in order to facilitate this bimetallic pathway. Larger systems also have the advantage of often being less soluble and therefore more easily recyclable.

Kureshy and coworkers [111,112] synthesized dimeric Co-complexes **65** and **66** (Figure 27). Two Co(III)salen moieties were covalently linked by a methylene or propane-2,2-diyl group, with the expectation that the increased molecular weight would result in an easier catalyst recovery compared to the monosalen complexes. This very straight-forward way of linking two salen moieties together was also employed by Jacobs and coworkers [44] in their design of dimeric Cr–salen complex **23** (Figure 7), which was impregnated on silica and used as a heterogeneous catalyst for the azidolytic ring-opening of *meso*-epoxides. Complex **65** was employed in the HKR of epichlorohydrin, propylene oxide and styrene oxide under solvent-free conditions. Complex **66** was investigated in the HKR of epichlorohydrin and a variety of alkyl terminal epoxides. For all substrates, the ring-opened diols and the unreacted epoxides were obtained with excellent yields (up to 50% for the diols and up to 52% for the epoxides) and enantiomeric excesses (up to 99% for both the diols and the epoxides). In addition, the dimeric complexes could be recycled and reused for up to four cycles without significant loss of reactivity or enantioselectivity [111,112].

**Figure 27.** Dimeric Co(III)salen complexes **65** and **66**.

In an e ffort to design versatile and easily recyclable Co–salen complexes capable of cooperative catalysis, Jones and coworkers synthesized styryl-substituted bissalen complex **67**, which could be polymerized to give access to homogeneous catalysts **68** and **69** and heterogeneous catalyst **70** (Figure 28) [113]. All the complexes were investigated as catalysts for the HKR of epichlorohydrin under solvent-free conditions, with reaction conditions optimized for the enantioenrichment of the unreacted epoxide. For homogenous catalysts **67**–**69**, full conversion of one of the enantiomers (50%) was achieved in 7 h with only 0.02 mol% Co loading, a ffording highly enantioenriched unreacted epoxide (99% *ee*). As a comparison, the corresponding monosalen catalyst **64** (Figure 26) a fforded the unreacted epoxide in 13% *ee* under the same reaction conditions and reaction time. The increased enantioselectivity and reactivity was attributed to the facilitated cooperative pathway with the dimeric catalyst. For the heterogenous catalyst **70**, a slightly higher catalyst loading (0.04 mol% Co) was needed. At this catalyst loading, catalyst **70** achieved full conversion of one of the enantiomers

of the epoxide and 99% *ee* for the unreacted epoxide of the opposite absolute configuration in 5 h. Furthermore, the insoluble catalyst **70** could be recovered by simple filtration, then regenerated and reused. The enantioselectivity of catalyst **70** was retained over three cycles, although each cycle required longer reaction time to reach full conversion.

**Figure 28.** Dimeric Co–salen complex **67** and polymeric complexes **68**–**70**.

Calix[4]arene-based dimeric Co(III)salen complex **71** (Figure 29) was reported by Wezenberg and Kleij [114]. Two salen moieties were installed on the upper rim of a calix[4]arene, with the expectation of facilitating the cooperative pathway in the HKR of terminal epoxides. Catalyst **71** and monosalen analogues **72** and **73** (Figure 29) were separately employed in the HKR of 1,2-epoxyhexane, epichlorohydrin and styrene oxide in acetonitrile, with conditions optimized for the production of the enantioenriched epoxide. All catalysts gave comparable conversions (32–52%), while catalyst **71** gave slightly lower *ee* values of both the ring-opened product and the unreacted epoxide (83–91% and 83–97%, respectively) than monosalen analogues **72** and **73** (88–95% and 97–99%, respectively). Kinetic studies showed that the HKR reactions employing catalyst **71** predominantly followed an intramolecular cooperative pathway. Comparison of the intra- and intermolecular rate constants for **71** and **72** revealed that although **71** had a significant intramolecular rate constant, as expected, the intermolecular rate constant was lower than for monosalen analogue **72**. This is remarkable since many other bimetallic salen catalysts show both increased intra- and intermolecular rate constants in the HKR reaction compared to the corresponding monosalen catalysts [21]. For catalyst **71**, the lower intermolecular rate constant meant that at higher catalysts loadings, the overall initial rate was lower than for the monosalen analogue **72**, while at lower catalyst loadings the intermolecular pathway was suppressed and catalyst **71** showed higher reaction rates. The authors sugges<sup>t</sup> that the decreased intermolecular reaction rate could be caused by the way the salen moieties were immobilized on the calixarene [114].

**Figure 29.** Calix[4]arene-based dimeric Co–salen complex **71** and monomeric analogues **72** and **73**.

Jacobsen and coworkers [58,69,115] developed three generations of oligomeric Co(III)salen complexes. Apart from catalyzing the ARO of *meso*-epoxides with water and carbamate as nucleophiles, the first generation (catalyst **40**, Figure 18) also catalyzed the HKR of terminal epoxides [69]. With enantioselectivity similar to monosalen complex **64** (Figure 26), catalyst **40** allowed a 10- to 50-fold decrease in catalyst loading (mol% Co) and up to 16-fold decrease in reaction time compared to catalyst **64** (Table 1). For the second generation of catalysts (complex **74** and **75** in Figure 30), the major difference was the linker. The use of a pimelate tether gave an oligomeric complex with a more predictable and reliable conformation [115]. Both complexes worked very well in the HKR of terminal epoxides, and the catalyst loading and reaction time were further decreased compared to the first generation (Table 1). The third generation (catalyst **35**, Scheme 10) exhibited better solubility and could be synthesized with high overall yield on a large scale [58]. The catalyst loading in the HKR of terminal epoxides could be reduced down to 0.003 mol% Co while retaining excellent enantioselectivity. Furthermore, its lipophilicity made catalyst **35** a good catalyst for solvent-free reactions (Table 1).


**Table 1.** Comparison of catalytic properties of oligosalen Co(III) complexes in the hydrolytic kinetic resolution (HKR) of terminal epoxides.

**Figure 30.** Jacobsen and coworkers' [115] second generation of oligosalen catalysts **74** and **75**.

Besides Jacobsen, Weck, and coworkers [48] also developed macrocyclic oligomers. They reported macrocyclic Co(III)salen oligomers **76** and **77** (Figure 31). Unlike other macrocyclic oligosalen complexes where the salen was part of the macrocycle, oligomers **76** and **77** consisted of a central macrocyclic backbone onto which the salen moieties were connected in a pendant-like fashion. This added flexibility afforded excellent reactivity and enantioselectivity in the HKR of terminal epoxides with very low catalyst loadings (down to 0.01 mol% Co). For a number of terminal alkyl epoxides and glycidyl ethers, the unreacted epoxides could be obtained in up to 48% yield and >99% *ee* under solvent-free conditions. More sterically hindered epoxides such as styrene oxide and *tert*-butyloxirane could also be resolved with similar reactivities and enantioselectivities, but higher catalyst loading (0.1–0.25 mol% Co) and longer reaction times were required. Interestingly, a linear polymeric analogue **78** (Figure 31) gave poorer results than macrocyclic catalysts **76** and **77** in terms of both reactivity and enantioselectivity, which reinforced the important role of the macrocyclic structure for this class of catalysts. A later study from the same group described the copolymerized cross-linked macrocyclic Co-oligosalen mixture **79** (Figure 32). With similar catalytic properties to **76** and **77**, oligomeric mixture **79** was more synthetically available and, therefore, more practical for larger scale catalysis [116].

**Figure 31.** Macrocyclic oligosalen catalysts **76** and **77**, and linear polymeric analogue **78**.

**Figure 32.** Cross-linked macrocyclic oligosalen catalyst **79**.

Khan and coworkers [117] designed a group of cyclic bissalen complexes where two Co–salen moieties were tethered by ethylene glycol chains (**80**–**84** in Figure 33). Co-complexes **80**, **81**, and **84** catalyzed the HKR of a number of terminal epoxides and glycidyl ethers with excellent yields (46–53% for the so formed 1,2-diols and 43–47% for the unreacted epoxides) and enantioselectivities (92–96% *ee* for the 1,2-diols and 96–99% *ee* for the unreacted epoxides) under solvent-free conditions and at low catalyst loadings (down to 0.016 mol% Co). Catalyst **84** could be recycled up to three times with maintained reactivity and enantioselectivity and could be used directly without catalyst regeneration between cycles. For the HKR of epichlorohydrin, catalyst **84** maintained its performance on a multigram scale, illustrating the scalability of the protocol. (*S*)-Propylene oxide obtained from this methodology was applied in a short synthesis of (*R*)-mexiletine with high overall yield (80%) and *ee* (98%).

**80**; 2\$F**81**; 27V**82**; %) **83**; 6E) **84**; 3)

**Figure 33.** Cyclic bissalen complexes **80**–**84** containing ethylene glycol chains. The counterions (X) are formally illustrated as bound to the metal.

Kureshy and Ganguly [118] reported a series of polymeric catalysts where the Co–salen moieties were tethered by different chiral and achiral linkers. Preliminary studies of the HKR of 1,2-epoxyhexane revealed that the highest enantioselectivity was obtained with polymeric Co–salen complexes with chiral BINOL linkers where the absolute configuration of the BINOL linker and the salen ligand were opposite (i.e., (*S*,*R*,*R*) in catalyst **85** and (*R*,*S*,*S*) in catalyst **86**, Figure 34). The preferred enantiomeric pairing of the two components was supported by DFT calculations comparing (*R*,*S*,*S*) catalyst with (*S*,*S*,*S*) catalyst. When employed in the HKR of a broad range of terminal alkyl and aryl epoxides and glycidyl ethers, catalyst **86** enabled excellent yields and enantioselectivities of both the ring-opened products and the unreacted epoxides (Scheme 42). Catalyst **86** could be recycled six times in the HKR of 1,2-epoxyhexane without loss of enantioselectivity, and only required one catalyst regeneration (between cycle 4 and 5). To show the scalability of the protocol, catalyst **86** was successfully employed in gram-scale syntheses of chiral β-blockers (*S*)-metoprolol, (*S*)-toliprolol and (*S*)-alprenol, where the final products were obtained in moderate overall yields (up to 44%) and excellent enantioselectivities (>99%).

**Figure 34.** Polymeric Co–salen catalysts with 1,1'-bi-2-naphthol (BINOL) linkers.

$$\begin{array}{ccccccccc} & \text{AO} & & & \text{O}\_{\text{C}} & & & \text{O}\_{\text{C}} & & & \text{O}\_{\text{SO-7}} & & \text{O}\_{\text{SO-7}} & & & & \text{O}\_{\text{O}} & & \\ & \text{AO} & & & & & & & & & & & & & \\ \text{(5.6.5-17)} & & & & & & & & & & & & & \\ \text{(6.6.5-0.5)} & & & & & & & & & & & & \\ \text{(6.6.5-0.5)} & & & & & & & & & & & \\ \text{(6.6.5-0.5)} & & & & & & & & & & & \\ \end{array}$$

**Scheme 42.** The hydrolytic kinetic resolution of terminal epoxides catalyzed by catalyst **86** (Figure 34).

Schulz and coworkers [119] reported the polymeric bithiophene-linked Co–salen catalyst **87** (Figure 35). The insoluble catalyst was employed in the dynamic HKR of epibromohydrin in THF. The catalyst was successfully reused in 11 subsequent catalytic runs, giving ring-opened products in good to excellent yield (60–99%) and with consistent enantioselectivity (around 84% *ee*). In a following study, the same group synthesized cyclic calixsalen catalyst **88**, which contained a phenyl linkers instead of the previously used thiophene linkers (Figure 35) [120]. Catalyst **88** could be used either as a pure trimer, a pure tetramer or as a mixture. The best results were obtained with the tetramer, which afforded the ring-opened product in full conversion (>99%) and with high enantioselectivity (92% *ee*) when used in the dynamic HKR of epibromohydrin in THF. The trimer exhibited significantly lower reactivities, both in terms of conversion and reaction rate. All the catalysts could be easily recovered from the reaction mixture by filtration and reused with maintained enantioselectivity, although the yield decreased after each run.

**Figure 35.** Polymeric Co–salen catalyst **87** and calixsalen Co catalyst **88**.

In addition to dimers and macrocyclic oligomers, there are also some examples of other types of multi-salen structures. Jacobsen and coworkers synthesized dendrimeric Co-complexes based on polyamidoamine (PAMAM) central core containing 4, 8, and 16 salen moieties (represented by 8-dendrimer **89** in Figure 36), expecting that the dendritic framework would enforce the bimetallic cooperative pathway in the HKR of 1,2-epoxyhexane in THF [121]. The dendrimeric catalysts exhibited higher reactivities than monomeric and dimeric analogues **90** and **91** (40–42% yield compared to 29–37% yield for the ring-opened product), as well as significantly higher rate constants. All catalysts (including **90** and **91**) a fforded the ring-opened product with excellent enantioselectivity (>98% *ee*). For the enantioenrichment of the unreacted epoxide, it was reported that catalyst **89** effected the HKR of 2-cyclohexyloxirane with 50% conversion and 98% *ee*.

Zheng and coworkers developed linear and cross-linked polymeric Co–salen complexes. They reported the linear polymeric Co–salen catalysts **92** and **93** (Figure 37), which catalyzed the HKR of propylene oxide, epichlorohydrin and phenyl glycidyl ether with excellent conversions (up to > 49%) and *ee* values (up to 98% for the unreacted epoxides and the 1,2-diols) in organic solvents (THF and CH2Cl2) or under solvent-free conditions [122]. No obvious di fferences in catalytic performances were observed between the two polymeric catalysts.

**Figure 36.** Polyamidoamine-based 8-dendrimer **89** and its monomeric (**90**) and dimeric (**91**) analogues.

**Figure 37.** Linear polymeric Co–salen catalysts **92** and **93**.

The same group also reported the synthesis and catalytic performance of a set of cross-linked copolymers [123]. Catalyst **94** was synthesized by the condensation of enantiopure *trans*-1,2-diaminocyclohexane and a mixture of dialdehyde **95** and trialdehyde **96** (Scheme 43). A series of catalysts, oligomers **94**, were thus synthesized with different ratios of the two aldehydes (ranging from 100% dialdehyde **95** to 100% trialdehyde **96**). The catalysts were applied in the HKR of epichlorohydrin, styrene oxide, and phenyl glycidyl ether under solvent-free conditions. Conversions of 43–53% were reported, with high enantioselectivities for both the ring-opened products and the unreacted epoxides (up to > 99% *ee* and up to 97% *ee*, respectively). The mixed cross-linked polymeric catalysts exhibited slightly higher reactivity than the catalysts based on either pure dialdehyde or pure trialdehyde. Recycling of the catalysts was unsuccessful due to the decomposition of the catalyst during the reaction. The authors attributed this to the hydrolysis of the ester groups under the reaction conditions.

**Scheme 43.** The synthesis of cross-linked polymeric Co–salen catalyst **94** from dialdehyde **95** and trialdehyde **96**. LPTS = lutidinium *p*-toluenesulfonate.

Weck and Jones [124] reported a set of homopolymerized and copolymerized Co(III)salen complexes, where the salen moieties were installed in a pendant-like fashion on the polystyrene chain (**97**–**100** in Figure 38). Catalysts with different ratios of salen moieties and styrene moieties were investigated in the HKR of epichlorohydrin in CH2Cl2. All the catalysts gave 49–55% conversion and afforded the unreacted epoxide with excellent enantioselectivities (>99% *ee*). Copolymeric catalysts **99** and **100** exhibited improved reactivity compared to homopolymeric catalyst **97** (1 h reaction time compared to 2 h). The authors hypothesized that the lower ratio of the salen moieties on the polystyrene chain could make the catalytic sites more accessible to the substrate. The same group also evaluated the optimal flexibility of polymeric Co–salen catalysts by synthesizing a series of homopolymeric catalysts with different lengths of the linker connecting the Co–salen complexes to the polymeric backbone (catalyst **101**–**104** in Figure 38) [125]. The catalysts were employed in the HKR of epichlorohydrin. Kinetic studies revealed that a 6-atom distance (corresponding to m = 1, catalyst **102**) between the polymer backbone and salen unit gave the highest reactivity and enantioselectivity. Catalyst **102** executed the successful HKR of a number of terminal epoxides under solvent-free conditions and with low catalyst loading (0.01–0.1 mol%), affording unreacted epoxides in 43–46% yield and ≥98% *ee*. These papers illustrate the importance of the flexibility and composition of polymer-supported Co–salen catalysts for the HKR of terminal epoxides.

**Figure 38.** Homo- and copolymerized Co–salen complexes **97**–**100** and homopolymerized Co–salen complexes **101**–**104**.

Weberskirch and coworkers [126] synthesized the first water-soluble amphiphilic block copolymeric Co–salen catalyst for the employment in HKR reactions (Figure 39) epoxides. The polymeric complex **105** could aggregate into micellar assemblies in aqueous environment, forming a hydrophobic core with a high local concentration of the active catalyst. The so-formed catalyst **105** enabled the efficient and enantioselective HKR of a number of terminal aryl epoxides and glycidyl ethers. The reactions proceeded with 50–54% conversion and with high-to-excellent enantioselectivities (up to 95% *ee* for the 1,2-diols and >98% *ee* for the unreacted epoxides). The catalyst could be recycled and reused in four successive cycles with consistent *ee* values for the unreacted epoxides, although the reaction time needed to be increased after each cycle to maintain the yield.

**Figure 39.** Amphiphilic copolymerized Co–salen catalyst **105**.

As described above, covalently linking metal–salen moieties together is one approach to obtaining multi-metallic catalysts. However, these larger motifs often require considerable synthetic efforts. In addition, many of the oligomeric complexes and polymers are prepared and used as mixtures in the catalysis, making it more complicated to elucidate the most active structure and understand how efficient catalysis is achieved. One alternative that could reduce the synthetic cost is the use of multi-metallic metal–salen assemblies based on non-covalent interactions. The most attractive case is one where the assembly can be completely controlled and only one well-defined catalyst is formed.

Following this line, Kim and coworkers [127] developed a series of dimeric catalysts where two identical Co–salen complexes were connected through the coordination of one oxygen of each salen unit to an Al(III)-containing Lewis acid (catalyst **54** in Scheme 34). The dimeric catalysts could induce high reactivities (41–46% yield for the ring-opened products and 40–46% yield for the unreacted epoxides) and enantioselectivities (up to 86% *ee* for the ring-opened products and up to 99% *ee* for the unreacted epoxides) in the HKR of epichlorohydrin, 1,2-epoxybutane, and glycidol, whereas monomeric analogues catalyzed the reactions with significantly lower reactivities and enantioselectivities. The catalysis could be performed either in THF or under solvent-free conditions. The same strategy was also extended to Lewis acids of other group 13 elements (catalyst **55** and **56** in Scheme 34) [128]. The dimeric Co–salen complex **55** (linked by GaCl3) was employed in the HKR of a wide range of 3-substituted propylene oxides and glycidyl ethers, a ffording unreacted epoxides in 40–49% yield and 97–99.8% *ee*. Ring-opened products were obtained in 42–50% yield and >85% *ee*. The reactions were performed under solvent-free conditions in THF or in a CH2Cl2/THF mixture. Kinetic studies showed the participation of both intramolecular and intermolecular pathways in the HKR reactions and significantly higher reaction rates for dimeric complexes than monomeric analogues. Other Lewis acids (ZnCl2, FeCl3, SnCl4, etc.) have also been attached to Co(III)salen complexes, and the so-formed monomeric catalysts exhibited improved reactivities in comparison with monosalen complex **64** (Figure 26) in the HKR of terminal epoxides [129–131].

Supramolecular interactions, such as aromatic donor–acceptor interactions and hydrogen bonding, have also been utilized to construct multi-metallic catalysts with controlled structures. The previously mentioned aromatic donor–acceptor complexes **42**–**44** (Figure 19) also worked very well as catalysts for the HKR of various terminal epoxides such as styrene oxide and sterically hindered *tert*-butyloxirane in CHCl3 or CCl4 [70]. The resolved epoxides were isolated in high yields (>40%) and enantioselectivities (up to 99% *ee*).

Hong and coworkers [132] designed and synthesized bis-urea-functionalized Co–salen complexes capable of forming self-assembled hydrogen-bonding structures in solution. The complex with *p*-CF3-phenyl substituents on the urea groups gave the highest rate constant in the HKR of epichlorohydrin (catalyst **106** in Figure 40). Catalyst **106** was investigated in the HKR of epichlorohydrin, allyl glycidyl ether, 1,2-epoxybutane and 1,2-epoxyhexane under solvent-free conditions, a ffording unreacted epoxides in 41–43% yield and 99% *ee*. In addition, complex **106** showed significantly higher reactivity than the monosalen complex **64** (Figure 26) at the same catalyst loading. Kinetic and self-association studies supported the hypothesis that the observed rate enhancement could be attributed to the enforcement of the bimetallic mechanism by the proximal self-association of Co–salen units through urea–urea hydrogen-bonding.

**Figure 40.** Bis-urea-functionalized Co–salen catalyst **106**, displayed as hydrogen-bonded dimer.

Another approach to achieving more e fficient catalysis is to utilize space constraints. Li and Yang [133] placed homogenous monosalen catalyst **64** (Figure 26) in a nanocage of mesoporous silica SBA-16 (one catalyst per cage) via the *ship in a bottle* strategy. A key step was tuning the size of the pores so that the reactants used for synthesizing the catalytically active complex (salen ligand and cobalt salt) were allowed to enter the nanocage but the formed complex was hampered from escaping. The so-immobilized catalyst was easy to recycle and was e ffective for up to 11 successive

runs, but the system gave lower enantioselectivity and reactivity than reactions performed with catalyst **64** under homogeneous conditions. To enable the bimetallic cooperative pathway, the same group described a new catalytic system where two Co–salen complexes were accommodated in one nanocage (Scheme 44) [134]. With nanoreactor **107**, the HKR of propylene oxide in CH2Cl2 could be completed with full conversion (50%) and excellent enantioselectivity (98% *ee* for both the unreacted epoxide and the 1,2-diol) at very low catalyst loading (1:12000 Co–salen/epoxide).

**Scheme 44.** The hydrolytic kinetic resolution of terminal epoxides performed in nanoreactor **107** in CH2Cl2.
