Eduard
Masferrer-Rius
a,
Fanshi
Li
a,
Martin
Lutz
b and
Robertus J. M.
Klein Gebbink
*a
aOrganic Chemistry and Catalysis, Debye Institute for Nanomaterials Science, Utrecht University, Universiteitsweg 99, 3584 CG Utrecht, The Netherlands. E-mail: r.j.m.kleingebbink@uu.nl
bStructural Biochemistry, Bijvoet Centre for Biomolecular Research, Utrecht University, Universiteitsweg 99, 3584 CG Utrecht, The Netherlands
First published on 20th October 2021
The direct enantioselective hydroxylation of benzylic C–H bonds to form chiral benzylic alcohols represents a challenging transformation. Herein, we report on the exploration of new biologically inspired manganese and iron complexes bearing highly electron-rich aminopyridine ligands containing 4-pyrrolidinopyridine moieties ((S,S)-1, (R,R)-1, 2 and 5) in combination with chiral bis-pyrrolidine and N,N-cyclohexanediamine backbones in enantioselective oxidation catalysis with aqueous H2O2. The current manganese complexes outperform the analogous manganese complexes containing 4-dimethylaminopyridine moieties (3 and 4) in benzylic oxidation reactions in terms of alcohol yield while keeping similar ee values (∼60% ee), which is attributed to the higher basicity of the 4-pyrrolidinopyridine group. A detailed investigation of different carboxylic acid additives in enantioselective benzylic oxidation provides new insights into how to rationally enhance enantioselectivities by means of proper tuning of the environment around the catalytic active site, and has resulted in the selection of Boc-L-tert-leucine as the preferred additive. Using these optimized conditions, manganese complex 2 was shown to be effective in the enantioselective benzylic oxidation of a series of arylalkane substrates with up to 50% alcohol yield and 62% product ee. A final set of experiments also highlights the use of the new 4-pyrrolidinopyridine-based complexes in the asymmetric epoxidation of olefins (up to 98% epoxide yield and >99% ee).
Several examples of enantioselective aliphatic C(sp3)–H oxidation catalysts exist, generally making use of manganese, iron or ruthenium as the transition-metal, often in combination with porphyrin, salen or aminopyridine ligands.5–36 Among these systems, the ones based on manganese and iron supported by aminopyridine ligands have emerged as a very powerful family of catalysts.37 It has been well established that such complexes activate H2O2, in most cases with the help of a carboxylic acid as an additive, to generate powerful and selective electrophilic metal-oxo species.22,23,38 On the one hand, the use of both non-heme iron and manganese complexes with aminopyridine ligands in asymmetric epoxidation reactions has been described extensively with high yields and enantioselectivities.39–62 On the other hand, the enantioselective oxidation of aliphatic C(sp3)–H bonds remains amongst the most challenging reactions, in which aminopyridine-based manganese complexes have particularly appeared as effective catalysts in the last years.63
The general mechanism of this latter oxidation reaction entails an initial HAT step from the substrate to the high valent manganese-oxo species, followed by a hydroxyl rebound step to generate the final alcohol product. The main challenges are based on the stereoselection of the C–H bond oxidation, which usually originates from enantio-discriminating HAT (Fig. 1A) or OH rebound steps (Fig. 1B),36 and the undesired overoxidation of the initial alcohol product producing ketone products. Regarding the latter drawback, a useful approach has been developed that makes use of hydrogen bond donor solvents, such as fluorinated alcohol solvents, to strongly deactivate electron rich C–H bonds that are in alpha-position to a hydroxyl group toward reaction with electrophilic reagents, thereby disfavoring alcohol overoxidation and preventing the loss of chirality. Accordingly, the use of fluorinated alcohol solvents has been widely applied in different kinds of oxidation reactions.26,31,32,64–71
![]() | ||
Fig. 1 Enantioselective aliphatic (benzylic) C–H oxidation of propylbenzene catalyzed by aminopyridine-based manganese complexes. A) Mechanism in which the origin of the enantioselectivity is determined at the HAT step.36 B) Mechanism in which the origin of the enantioselectivity is determined at the rebound step (OH transfer step).36 |
Examples of enantioselective aliphatic C(sp3)–H oxidation reactions catalyzed by biologically inspired manganese complexes bearing aminopyridine ligands include a system to generate enantiomerically enriched products, that was reported by Bietti, Costas and co-workers. These authors developed a catalytic approach for the oxidative desymmetrization of mono-substituted cyclohexanes using bulky bioinspired manganese catalysts in combination with H2O2, generating γ-ketones with up to 85% yield and 96% ee.21,25 More recently, the enantioselective C–H lactonization of unactivated methylenes directed by carboxylic acids has also been described using the same kind of catalysts to afford chiral γ-lactones with up to 88% yield and >99% ee.30 Focusing on enantioselective benzylic oxidation, Sun and co-workers developed an oxidative desymmetrization approach for the enantioselective oxidation of benzylic methylene groups of spirocyclic hydrocarbons by the bioinspired manganese catalyst Mn(S-PEB) and H2O2, affording up to 94% yield and 98% ee of the resulting ketone products (Fig. 2A).27,29 Later on, the same authors used their manganese-catalyzed system also for the oxidation of oxindoles and dihydroquinolinones, with up to 67% yield and 99% ee; as well as for the oxidation of benzylic methylene C–H bonds of indane-based substrates using fluorinated alcohol solvents, with up to 78% yield and 95% ee for the alcohol product (Fig. 2A).28,31 Company, Bietti, Costas and co-workers developed a manganese-catalyzed benzylic hydroxylation of simple aromatic substrates to the corresponding alcohol products using fluorinated alcohol solvents, which prevented the overoxidation of the alcohol to the ketone (vide supra). Particularly, they showed that the use of the electron-rich manganese catalysts Mn(dMMbpbp) and Mn(Me2Nbpbp) (4) affords the alcohol products with up to 61% yield with respect to H2O2 and in 66% ee (Fig. 2B).68 Bryliakov and co-workers have employed the Mn(dpf) catalyst for the undirected enantioselective benzylic oxidation of simple arylalkanes with H2O2, first in acetonitrile as solvent. They could get up to 86% ee for the alcohol product in the oxidation of substituted-ethylbenzenes, despite the fact that the alcohol/ketone ratio, conversions and yields for the alcohol products were low (5–6% alcohol yield; Fig. 2C).24,72 In a follow-up study, they could enhance oxidation towards the alcohol product by switching to fluorinated alcohol solvents, providing up to 34% yield and up to 76% ee for alcohol products in the oxidation of substituted-ethylbenzene substrates. To the best of our knowledge these are the highest ee values reported for the direct oxidation of simple arylalkanes towards alcohol products.26 Much more recently, the same authors used a one-pot sequential oxidation and oxidative kinetic resolution approach to obtain 40% yield and 97% ee for the 1-phenylethanol product, as well as up to 60% yield and 93% ee for the oxidation of 3,4-dihydroquinolinone derivatives.32 Remarkably, the Mn(dpf) catalyst in these latter examples is used in combination with an enantiopure amino acid additive, namely Boc-L-proline (Fig. 2C).
Previously, we have shown that bulky manganese complexes supported by aminopyridine ligands are highly selective in aromatic oxidation reactions towards phenol products, whereas the use of electron-rich manganese complexes switches the chemoselectivity of the reaction, favoring benzylic C(sp3)–H bond oxidation in substituted arene substrates. Based on these observations, we concluded that the electronic nature of the ligand is a key factor in controlling the chemoselectivity of these Mn-catalyzed C–H oxidations.73 In addition, Bryliakov and co-workers have recently developed several electron-rich aminopyridine ligands based on para-substituted aminopyridines (NEt2, NMeiPr, N(CH2)4 substituents). Particularly, they have shown that the corresponding diferric complexes supported by these ligands can generate high-spin oxoiron(V) intermediates upon reaction with H2O2, which are active in asymmetric epoxidation and aliphatic hydroxylation reactions.74,75 This strategy of including electron donating groups in the para-position of the pyridine ligand was first reported by Di Stefano et al.76 and Costas et al.43,57
Inspired by these findings, we have explored electron-rich N2Py2-type ligands bearing para-pyrrolidine substituted pyridine donors in enantioselective oxidation catalysis with manganese as the metal center. Since 4-pyrrolidinopyridine is a stronger N-heteroaromatic electron donor ligand compared to 4-dimethylaminopyridine, DMAP (pKa = 18.33 and 17.95 for 4-pyrrolidinopyridine and DMAP, respectively),77,78 we hypothesize that the greater electron-donating capacity of the 4-pyrrolidinopyridine donors might lead to a better stabilization of the active manganese-oxo species formed upon reaction with H2O2, and accordingly to a better catalytic performance. Herein, we report on the rational development of the new manganese complexes (S,S)-1, (R,R)-1 and 2 bearing 4-pyrrolidinopyridine moieties (Fig. 3B) and their use in the undirected enantioselective catalytic oxidation of benzylic C–H groups using aqueous H2O2 as benign oxidant, carboxylic acids as co-ligands, and fluorinated alcohol solvents to provide good alcohol yields and ee's. Interestingly, the current complexes outperform the analogous manganese complexes 3 and 4 containing 4-dimethylaminopyridine moieties in terms of benzylic alcohol product formation.68 Furthermore, we have also explored manganese and iron complexes derived from the 4-pyrrolidinopyridine-modified ligands in the asymmetric epoxidation of olefin substrates.
![]() | ||
Fig. 3 A) Synthesis of tetradentate aminopyridine ligands (S,S)-L1, (R,R)-L1 and L2. B) Metal complexes used in this study. |
Complexation reactions were then performed by the reaction of equimolar amounts of the corresponding ligand with [MnII(CF3SO3)2] in dry THF under an inert atmosphere to afford manganese complexes (S,S)-[MnII(L1)(CF3SO3)2] and (R,R)-[MnII(L1)(CF3SO3)2] ((S,S)-1 and (R,R)-1), and (S,S)-[MnII(L2)(CF3SO3)2] (2) as microcrystalline solids (for further details see ESI†). Complexes (S,S)-1 and 2 were characterized by high resolution mass spectrometry (HRMS). HRMS analysis of (S,S)-1 showed a prominent mass peak at m/z 666.2378 corresponding to the [MnII(L1)(CF3SO3)]+ ion (calc. 666.2372). For complex 2 a prominent mass peak at m/z 664.2212 corresponding to the [MnII(L2)(CF3SO3)]+ ion (calc. 664.2215) was found. Manganese complexes (S,S)-[MnII(L3)(CF3SO3)2] (3) and (S,S)-[MnII(L4)(CF3SO3)2] (4) were also synthesized in order to compare the catalytic properties of the new complexes.43,49
Synthesis of non-heme iron complex 5 was performed using equimolar amounts of L2 and [FeII(CF3SO3)2(CH3CN)2] in dry THF under an inert atmosphere. HRMS analysis of the complex showed a prominent mass peak at m/z 665.2189 corresponding to the [FeII(L2)(CF3SO3)]+ ion (calc. 665.2184).
![]() | ||
Fig. 4 From left to right: ORTEP diagrams of the molecular structure of (S,S)-[MnII(CF3SO3)2(Pyrbpbp)] (2), (R,R)-[MnII(CF3SO3)2(Me2Nbpbp)] ((R,R)-4), (S,S)-[FeII(CF3SO3)2(Pyrbpbp)] (5), and (S,S)-[MnII(CF3SO3)2(bpbp)] (Mn(bpbp)), showing the atom numbering scheme. Triflate anions are omitted except for the oxygen atoms directly bound to the metal center, and hydrogen atoms are omitted for clarity. The structure of complex (R,R)-4 was reported by Costas and co-workers,43 the structure of Mn(bpbp) by Bryliakov and co-workers.40 |
2 | (R,R)-4 (ref. 43) | 5 | Mn(bpbp) 40 | ||||
---|---|---|---|---|---|---|---|
Mn1–N1 | 2.222(3) | Mn1–N1 | 2.210(4) | Fe1–N1 | 2.157(3) | Mn2–N8 | 2.250(2) |
Mn1–N4 | 2.215(3) | Mn1–N5 | 2.210(3) | Fe1–N4 | 2.162(3) | Mn2–N5 | 2.239(2) |
Mn1–N2 | 2.294(3) | Mn1–N3 | 2.301(4) | Fe1–N2 | 2.226(3) | Mn2–N7 | 2.273(2) |
Mn1–N3 | 2.300(3) | Mn1–N4 | 2.315(3) | Fe1–N3 | 2.227(3) | Mn2–N6 | 2.300(2) |
Mn1–O1 | 2.152(3) | Mn1–O1 | 2.177(4) | Fe1–O4 | 2.153(3) | Mn2–O7 | 2.145(2) |
Mn1–O4 | 2.184(2) | Mn1–O4 | 2.195(3) | Fe1–O1 | 2.153(3) | Mn2–O10 | 2.152(2) |
N6–C25 | 1.474(5) | N2–C6 | 1.450(6) | N6–C28 | 1.472(5) | — | — |
N6–C28 | 1.462(5) | N2–C7 | 1.457(8) | N6–C25 | 1.463(5) | — | — |
N6–C18 | 1.346(4) | N2–C3 | 1.344(7) | N6–C18 | 1.346(5) | — | — |
N5–C24 | 1.466(4) | N6–C24 | 1.434(6) | N5–C24 | 1.462(5) | — | — |
N5–C21 | 1.466(5) | N6–C23 | 1.469(6) | N5–C21 | 1.470(5) | — | — |
N5–C3 | 1.346(4) | N6–C20 | 1.352(5) | N5–C3 | 1.348(5) | — | — |
N1–Mn1–N2 | 76.26(10) | N4–Mn1–N5 | 75.45(11) | N1–Fe1–N2 | 77.76(11) | N8–Mn2–N7 | 74.80(8) |
N3–Mn1–N4 | 75.87(10) | N3–Mn1–N1 | 74.80(13) | N3–Fe1–N4 | 77.37(11) | N6–Mn2–N5 | 76.03(9) |
N2–Mn1–N3 | 77.11(10) | N3–Mn1–N4 | 77.25(12) | N2–Fe1–N3 | 79.40(10) | N7–Mn2–N6 | 78.68(9) |
N2–Mn1–N4 | 95.53(10) | N1–Mn1–N4 | 91.89(12) | N2–Fe1–N4 | 96.48(11) | N7–Mn2–N5 | 101.02(9) |
O1–Mn1–O4 | 101.19(12) | O1–Mn1–O4 | 104.08(14) | O4–Fe1–O1 | 94.01(11) | O7–Mn2–O10 | 103.52(8) |
C25–N6–C28 | 112.3(3) | C6–N2–C7 | 117.2(5) | C28–N6–C25 | 111.7(3) | — | — |
C18–N6–C28 | 124.0(3) | C3–N2–C7 | 120.7(4) | C18–N6–C25 | 124.5(3) | — | — |
C21–N5–C24 | 112.3(3) | C23–N6–C24 | 117.7(4) | C21–N5–C24 | 112.2(3) | — | — |
C3–N5–C21 | 124.1(3) | C20–N6–C23 | 120.4(4) | C3–N5–C21 | 124.5(3) | — | — |
The Mn–N bond distances in complex 2 range from 2.222(3) to 2.300(3) Å and the Mn–O bond distances from 2.152(3) to 2.184(2) Å. These values compare quite well with the Mn–N and Mn–O bond distances of complex (R,R)-4 (from 2.210(4) to 2.315(3) Å and 2.177(4) to 2.195(3) Å, respectively).43 On the other hand, complex 2 displays a slightly smaller O–Mn–O angle (101.19(12)°) relative to the corresponding angle in complex (R,R)-4 (104.08(14)°), which means that the 4-pyrrolidinopyridine moieties introduce some steric strain in the complex. The Fe–N bond distances in complex 5 range from 2.157(3) to 2.227(3) Å and the Fe–O bond distances are 2.153(3) Å, which are indicative of a high-spin iron complex.57,81 Comparing complexes 2 and 5, we find that their structures are very much alike, with slightly longer Mn–N distances and similar Mn/Fe–O distances. The O–Fe–O angle of complex 5 is much smaller (94.01(11)°) than the O–Mn–O angle in 2 and (R,R)-4 though, which we attribute to the difference in the ionic radius of the two metal ions (the ionic radius of Mn(II) being larger than that of Fe(II)). This observation is a general trend that has been observed for other iron and manganese complexes bearing the same ligand, such as for Mn and Fe complexes with the parent bpbp ligand (O–M–O angle of 107.45(9) and 85.81(5), respectively)40,82 and with the (S-PEB) ligand (see Fig. 2A for structure of the (S-PEB) ligand, O–M–O angle of 105.1(1) and 101.5(2), respectively).83,84
Overall, the molecular structure of complexes 2 and 5 do not differ significantly from the structure of 4-dimethylamino-substituted complex (R,R)-4, nor from the non-substituted bis-pyrrolidine manganese complex (Mn(bpbp)).40 This shows that the pyrrolidine and dimethylamino substituents provide similar structural properties to the complexes. Accordingly, the introduction of a pyrrolidine substituent in the para-position of each pyridine ring of the bpbp ligand does not produce significant changes in the structural geometry of the complex.
Catalyst | r.s.ma | Alcoholb | Ketoneb | p-Phenolb | o-Phenolb | Quinoneb | eec | MBd |
---|---|---|---|---|---|---|---|---|
a Remaining starting material (r.s.m) in %. b Yields in % with respect to substrate determined by GC against an internal standard. c Enantiomeric excess determined by HPLC on a chiral stationary phase. (S)-(−)-1-Phenyl-1-propanol (6a) was obtained as the major enantiomer. d Mass balance (MB) was calculated considering remaining starting material and all products formed: MB = (r.s.m%) + (Product Yields%). n.d. = non-detected. AA = acetic acid. | ||||||||
(S,S)-1 | 26 | 35 | 4 | 1 | <1 | 1 | 32 | 68 |
2 | 34 | 34 | 5 | <1 | n.d. | 1 | 33 | 74 |
3 | 43 | 23 | 1 | 2 | 1 | 1 | 39 | 71 |
4 | 37 | 27 | 4 | 1 | <1 | 1 | 36 | 70 |
5 | 25 | <1 | n.d. | <1 | <1 | <1 | — | 25 |
Under these conditions, complexes (S,S)-1 and 2 generate the alcohol product 6a in 34 to 35% yield, together with the ketone product 6b in 4 to 5% yield and trace amounts of the aromatic oxidation products. For both complexes the (S)-alcohol product formed in 32 and 33% ee, respectively, showing that a change in the amine backbone does not induce a significant change in enantioselectivity. Complexes 3 and 4 yield the benzylic alcohol product 6a in 23 and 27%, respectively, in this reaction, whereas the ketone yield is low (1% and 4%, respectively). The ee value for the (S)-alcohol obtained for these latter manganese complexes ranges between 36% and 39%, showing a slight increase compared to complexes (S,S)-1 and 2. Overall, when acetic acid is used as additive, complexes (S,S)-1 and 2 afford a higher catalytic activity than complexes 3 and 4 based on substrate conversion and combined alcohol and ketone yield, whereas ee values are slightly lower. Mass balances of these reactions are not excellent, which could indicate that overoxidation to non-detected products may occur.
Worthy of note is that use of non-heme iron complex 5 in this reaction resulted in the formation of only trace amounts of benzylic alcohol product 6a, while substrate conversion was considerable (75%), indicating a poor mass balance for this reaction. This observation indicates that the use of iron as the metal is not optimal for this aliphatic (benzylic) hydroxylation reaction. Moreover, para-phenol, ortho-phenol and quinone products were also detected in trace amounts, indicating that complex 5 shows aromatic oxidation to some extent, which was also noted by Bryliakov and co-workers for related non-heme iron complexes supported by tetradentate aminopyridine ligands.88–91 Accordingly, we focused our study on benzylic oxidation exclusively on the use of manganese complexes as catalysts.
Because of the importance of carboxylic acids in H2O2-mediated oxidation catalysis, we decided to investigate different carboxylic acid additives. One of the carboxylic acids that has shown promising results in aliphatic C–H hydroxylation, as well as in epoxidation reactions, is racemic 2-ethylhexanoic acid (2-eha). When using this acid, an increase in ee values of the alcohol product has been documented for several manganese-catalyzed and iron-catalyzed oxidations.42,43,45,57,59,92 Therefore, here we have studied the oxidation of substrate 6 using 2-eha following the previously mentioned conditions (Table 3).
Catalyst | r.s.ma | Alcoholb | Ketoneb | p-Phenolb | o-Phenolb | Quinoneb | eec | MBd |
---|---|---|---|---|---|---|---|---|
a Remaining starting material (r.s.m) in %. b Yields in % with respect to substrate determined by GC against an internal standard. c Enantiomeric excess determined by HPLC on a chiral stationary phase. (S)-(−)-1-Phenyl-1-propanol (6a) was obtained as the main enantiomer. d Mass balance (MB) was calculated considering remaining starting material and all products formed: MB = (r.s.m%) + (Product Yields%). n.d. = non-detected. 2-eha = 2-ethylhexanoic acid. | ||||||||
(S,S)-1 | 31 | 34 | 7 | <1 | 1 | n.d. | 59 | 73 |
2 | 39 | 34 | 8 | n.d. | n.d. | n.d. | 58 | 81 |
3 | 44 | 26 | 3 | <1 | <1 | n.d. | 50 | 73 |
4 | 38 | 30 | 6 | n.d. | n.d. | n.d. | 58 | 74 |
With 2-eha, complexes (S,S)-1 and 2 showed similar benzylic alcohol yields as with the use of acetic acid (34% yield), whereas the formation of overoxidized ketone product slightly increased to 7–8% yield. Interestingly, ee values for the alcohol product increase for all complexes when 2-eha is employed. Complexes (S,S)-1 and 2 showed ee values for the (S)-alcohol product up to 59%, which means a two-fold increase in comparison with the use of acetic acid as additive (compare Tables 2 and 3). For the manganese complexes bearing 4-dimethylaminopyridine groups (3 and 4), the ee value increased in a similar way only in the case of complex 3 containing the bis-pyrrolidine backbone (58% ee), whereas complex 4 based on the N,N-cyclohexanediamine backbone showed a smaller increase (50% ee). Also under these conditions, complexes 3 and 4 showed lower conversions and alcohol and ketone yields compared to complexes (S,S)-1 and 2.
From these results, we concluded that manganese complexes (S,S)-1 and 2 with either a bis-pyrrolidine or a N,N-cyclohexanediamine backbone are promising catalysts for benzylic oxidations, since they show high ee values for the alcohol product. Comparing our results to the systems previously described by Costas68 and Bryliakov,24,26,32,72 we can conclude that the current complexes perform the benzylic hydroxylation of an alkylbenzene with ee's commensurate to state-of-the-art homogeneous catalysts (see Fig. 2B and C). In addition, complexes (S,S)-1 and 2 show higher conversions and benzylic alcohol yields (34% yield) than complexes 3 and 4 bearing 4-dimethylaminopyridine moieties. We believe that the reason for the (slight) increase in alcohol yield is caused by the higher basicity of the ligands resulting from the pyrrolidine substituents, which provide the complex with a more electron-donating ligand and therefore might provide a better stabilization of the active oxidant.
Complex 2 was then chosen for further reaction optimization, since a better mass balance was observed compared to the use of complex (S,S)-1. Initially the use of another fluorinated solvent, i.e. 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP), was explored using acetic acid and 2-eha as additives (see Table S1†). In these experiments, a high alcohol/ketone product ratio (A/K of 31 and 18, using AA and 2-eha, respectively) was observed with only trace amounts of overoxidized ketone product being formed. This observation agrees with the stronger hydrogen bond donor ability of HFIP compared with TFE (A/K = 4.25 for complex 2 in combination with 2-eha in TFE), providing an enhanced polarity reversal to alcohol groups and favoring the deactivation of proximal C–H bonds toward oxidation by high valent metal-oxo species.68 However, due to the higher melting point of HFIP compared to TFE (−3.3 and −43.5 °C, respectively), the reaction in HFIP was performed at a higher reaction temperature of 0 °C, resulting in lower ee values of the alcohol product when 2-eha was employed (48% and 58% for HFIP and TFE, respectively). Using acetonitrile as the solvent in the current oxidation reaction provided a low alcohol/ketone ratio (A/K of 0.2 and 0.5, using AA and 2-eha, respectively), indicating that overoxidation of the primary alcohol product is highly favored in this solvent (see Table S1†). Thus, TFE was chosen as the solvent for further reaction optimization, because it provides good A/K product ratios and allows the reaction to be performed at a lower temperature (−35 °C), which has been shown to be crucial to obtain good enantioselectivities.
First, we decided to test a series of carboxylic acids with different types of alpha-carbons. We have considered an acid with a primary alpha-carbon (acetic acid, AA), a secondary (propionic acid, PA), a tertiary (2-ethylhexanoic acid, 2-eha), a quaternary (2,2-dimethylbutanoic acid, 2,2-DMBA), as well as an sp2-hybridized alpha-carbon (benzoic acid, BZA). Table 4 summarizes the catalytic data for the use of this set of additives in the oxidation of propylbenzene (6). Increasing the length of the alkyl chain of the carboxylic acid, by using propionic acid, did provide a slight increase in alcohol and ketone yield (38% and 7% yield, respectively) compared to the use of AA. However, the ee value for the alcohol product did not increase (Table 4, entry 1 and 2). Interestingly, when acids with tertiary and quaternary alpha-carbons were used (2-eha and 2,2-DMBA), ee values for the benzylic alcohol significantly increased (58% ee), without deterioration of the alcohol yield (Table 4, entry 3 and 4). Worthy of note is the use of 2,2-DMBA, which provided a significant increase in ketone formation (14% yield), clearly favoring oxidation of the initial alcohol product compared to the other carboxylic acids tested. The use of this acid does also provide the alcohol with an increased ee. The use of a carboxylic acid with an sp2-hybridized alpha-carbon, such as benzoic acid, resulted in a decrease in alcohol ee (25% ee) compared to the use of AA, as well as in a lower alcohol yield (Table 4, entry 5). Accordingly, this first data set indicated that the use of a carboxylic acid additive with a tertiary sp3-hybridized alpha-carbon provides the best results in terms of alcohol yield and ee value. In all these cases, aromatic oxidation is basically suppressed to a minimum.
Entry | CAa | r.s.mb | Alcoholc | Ketonec | p-Phenolc | o-Phenolc | Quinonec | eed | MBe |
---|---|---|---|---|---|---|---|---|---|
a Carboxylic acid: AA = acetic acid, PA = propionic acid, 2-eha = 2-ethylhexanoic acid, 2,2-DMBA = 2,2-dimethylbutanoic acid, BZA = benzoic acid. b Remaining starting material (r.s.m) in %. c Yields in % with respect to substrate determined by GC against an internal standard. d Enantiomeric excess determined by HPLC on a chiral stationary phase. (S)-(−)-1-Phenyl-1-propanol (6a) was obtained as the main enantiomer. e Mass balance (MB) was calculated considering remaining starting material and all products formed: MB = (r.s.m%) + (Product Yields%). n.d. = non-detected. | |||||||||
1 | AA | 34 | 34 | 5 | <1 | n.d. | 1 | 33 | 74 |
2 | PA | 29 | 38 | 7 | 1 | n.d. | n.d. | 34 | 75 |
3 | 2-eha | 39 | 34 | 8 | n.d. | n.d. | n.d. | 58 | 81 |
4 | 2,2-DMBA | 31 | 27 | 14 | n.d. | n.d. | n.d. | 58 | 72 |
5 | BZA | 37 | 23 | 7 | n.d. | n.d. | n.d. | 25 | 67 |
Next, we decided to screen a set of carboxylic acids with tertiary alpha-carbons in which the substitution on one of the beta-carbons varies (see ESI,† Table S2). On basis of this analysis, we concluded that an acid additive containing a tertiary alpha-carbon and a secondary beta-carbon (such as 2-eha) provides the best performance in the H2O2-mediated benzylic oxidation of propylbenzene with manganese catalyst 2. Interestingly, using this set of carboxylic acid additives led to a complete suppression of aromatic oxidation activity.
Finally, we considered the use of chiral amino acids as carboxylic acid additives. Recent studies have shown the advantageous use of these additives in other H2O2-mediated oxidations using bioinspired manganese complexes as catalysts.24,26,32,72 Amino acids comprise a tertiary alpha-carbon, which seems optimal for enantioselective benzylic oxidation with catalyst 2 on the basis of our screening of carboxylic acids with different types of alpha-carbons (Table 4). Accordingly, we have tested N-protected prolines, leucines and phenylalanines containing different protecting groups (Boc, Cbz and Phth) and chiralities (L and D) (Table 5).
Entry | CAa | r.s.mb | Alcoholc | Ketonec | p-Phenolc | o-Phenolc | Quinonec | eed | MBe |
---|---|---|---|---|---|---|---|---|---|
a Carboxylic acid: Boc-L-Pro = N-tert-butylcarboxy-L-proline, Cbz-L-Pro = N-carbobenzyloxy-L-proline, Cbz-D-Pro = N-carbobenzyloxy-D-proline, Boc-L-tert-Leu = N-tert-butylcarboxy-L-leucine, Boc-D-tert-Leu = N-tert-butylcarboxy-D-leucine, Phth-L-Phe = phthalimido-L-phenylalanine. b Remaining starting material (r.s.m) in %. c Yields in % with respect to substrate determined by GC against an internal standard. d Enantiomeric excess determined by HPLC on a chiral stationary phase. (S)-(−)-1-Phenyl-1-propanol (6a) was obtained as the main enantiomer. e Mass balance (MB) was calculated considering remaining starting material and all products formed: MB = (r.s.m%) + (Product Yields%). n.d. = non-detected. | |||||||||
1 | Boc-L-Pro | 46 | 31 | 7 | <1 | n.d. | n.d. | 47 | 84 |
2 | Cbz-L-Pro | 27 | 38 | 6 | 1 | n.d. | n.d. | 52 | 72 |
3 | Cbz-D-Pro | 39 | 27 | 4 | 1 | n.d. | n.d. | 61 | 71 |
4 | Boc-L-tert-Leu | 29 | 42 | 9 | <1 | n.d. | n.d. | 58 | 80 |
5 | Boc-D-tert-Leu | 37 | 30 | 16 | n.d. | n.d. | n.d. | 52 | 83 |
6 | Phth-L-Phe | 43 | 24 | 5 | <1 | n.d. | n.d. | 46 | 72 |
Regarding the prolines employed, we have considered both Boc-L-proline and a Cbz-L-proline. For Boc-L-proline we obtained up to 31% alcohol yield in 47% ee and 7% ketone yield, (Table 6, entry 1). For Cbz-L-proline we obtained a higher conversion and yields, with the alcohol product being formed in 38% yield, and the ketone in 6% yield. The ee value for the alcohol product in this case was 52%. Since the Cbz protecting group provided better conversion and yields, we decided to also test the Cbz-proline additive with opposite stereochemistry D (Table 6, compare entries 2 and 3). Interestingly, we found that the overall activity of the current system changes by switching the chirality of the amino acid additive, indicating that a proper engineering of the chiral environment around the catalytic active site is crucial, and that subtle modifications may translate into different performances. The experiment using Cbz-D-Pro provided a lower conversion and alcohol yield, whereas the alcohol ee increased to 61% (Table 6, entry 3). Next, we have also considered leucines with different chiralities (Boc-L-tert-leucine and Boc-D-tert-leucine) as amino acid additives. Boc-L-tert-leucine provides the benzylic alcohol product in up to 42% yield, together with the ketone product in 9% yield. The ee value in this case was as high as 58%. The use of the opposite enantiomer, Boc-D-tert-leucine, again led to a different catalytic performance. The yield for the alcohol product decreased to 30%, whereas the ketone product was formed in a much higher amount (16%). The ee value for the alcohol product slightly decreased to 52%. Finally, we have also considered phthalamido-protected L-phenylalanine as additive. However, product yields were much lower with only 24% alcohol yield and 5% ketone yield, and an ee value of 46% was observed for the alcohol product (Table 6, entry 6).
Entry | H2O2a | r.s.mb | Alcoholc | Ketonec | p-Phenolc | o-Phenolc | Quinonec | eed | MBe |
---|---|---|---|---|---|---|---|---|---|
a Total equivalents of H2O2 used in the oxidation reaction (added in two portions). b Remaining starting material (r.s.m) in %. c Yields in % with respect to substrate determined by GC against an internal standard. d Enantiomeric excess determined by HPLC on a chiral stationary phase. (S)-(−)-1-Phenyl-1-propanol (6a) was obtained as the main enantiomer. e Mass balance (MB) was calculated considering remaining starting material and all products formed: MB = (r.s.m%) + (Product Yields%). n.d. = non-detected. | |||||||||
1 | 2 × 0.5 | 23 | 46 | 9 | n.d. | n.d. | n.d. | 60 | 78 |
2 | 2 × 1 | <1 | 23 | 55 | <1 | n.d. | n.d. | 72 | 78 |
An important observation from these experiments is that by switching the chirality of the amino acid additive, the enantioselectivity in the alcohol product does not change for this catalytic system, i.e. the (S)-alcohol is observed as the major product in all cases. Therefore, we can conclude that the enantioselectivity of the reaction is dictated by the chirality of the starting Mn-complex and is not perturbed by (chiral) additives, as was previously described in other studies using similar manganese complexes for the oxidation of aliphatic C–H bonds.26,32 Indeed, by using complex (S,S)-1 in combination with 2-eha, the (S)-(−)-1-phenyl-1-propanol product was generated as the main enantiomer in 59% ee, while the (R)-(+)-1-phenyl-1-propanol product formed as the main enantiomer in 57% ee at a similar conversion and yield when using complex (R,R)-1 of opposite chirality (see ESI,† Table S3).
From the data compiled in Table 6, we concluded that the best carboxylic acid additive is Boc-L-tert-leucine, providing the highest benzylic alcohol yield (42%) and a good ee value (58%). These characteristics could be slightly increased by making use of an iterative addition protocol (see ESI,† for further details). This methodology consists of adding a first portion of manganese complex (0.5 mol%) and carboxylic acid additive (0.2 equiv.) and adding H2O2 (0.5 equiv.) over a period of 1 h. Then, a new portion of complex (0.5 mol%) and carboxylic acid (0.2 equiv.) is added, and a second portion of H2O2 (0.5 equiv.) is added again over a period of 1 h. This procedure provided us with a slight increase in conversion and alcohol yield (up to 46%), keeping a similar ee value of 60% (Table 7, entry 1). This observation could indicate that catalyst lifetime is an issue in these H2O2-mediated C–H oxidations, as it has been previously described that oxidative degradation of the ligand occurs for related non-heme iron and manganese complexes.95–97 When the same iterative addition protocol was used but employing a total of 2 equiv. of H2O2, the overoxidized ketone product was obtained as the main product in 55% yield, together with the alcohol product in 23% yield, indicating that overoxidation is highly favored when a large excess of oxidant is used (Table 7, entry 2). Interestingly, the ee value for the alcohol product increased to 72%, which can be explained by a kinetic resolution effect in the secondary oxidation step. This kinetic resolution methodology has been previously used to reach high alcohol ee values by Bryliakov et al.24,32,98 The iterative addition protocol using an overall 2.0 equiv. of oxidant also led to full substrate conversion. Overall, the catalytic system 2/H2O2/Boc-L-tert-leucine performs the oxidation of monoalkylbenzene 6 with higher alcohol yields compared to the system developed by Bryliakov and co-workers.26 Similar conversions were obtained for both systems (∼75%), whereas alcohol yield was higher with the current complex (46% and 34% alcohol yield for the use of catalyst 2 and Mn(dpf), respectively). However, ee values for the alcohol product were lower when 2 was used instead of Mn(dpf) (60% and 76% ee for the use of catalyst 2 and Mn(dpf), respectively). Of note is that ee's have been considerably increased using a kinetic resolution approach with Mn(dpf), obtaining up to 97% alcohol ee (Fig. 2C).32
![]() | ||
Scheme 1 Asymmetric synthesis of benzylic alcohols by a manganese-catalyzed C–H oxidation. Reactions were performed on 0.2 mmol scale in 2.5 mL of TFE, 0.5 mol% of complex 2 and 0.2 equiv. of Boc-L-tert-Leu (complex 2, carboxylic acid, and oxidant were all added in portionwise twice; for details, see the ESI†). Yields were determined by GC, and ee was determined by HPLC on a chiral stationary phase. Yields for the overoxidized ketone products are shown in parenthesis, whereas remaining starting material (r.s.m%) is shown in brackets. |
Oxidation of ethylbenzene (7) leads to the benzylic alcohol product 7a in 41% yield and 52% ee, along with the corresponding ketone 7b in 16% yield. Interestingly, isobutylbenzene (8) was also considered, which bears a reactive tertiary aliphatic C–H bond. Benzylic alcohol product 8a was obtained in 50% yield and 62% ee, together with the corresponding ketone product 8b in 13% yield. The 2-methyl-1-phenyl-2-propanol product derived from oxidation at the tertiary position was only detected in trace amounts, which agrees with the favorable oxidation of a benzylic C–H bond (activated C–H bond, BDE = 85.4 kcal mol−1) compared to a tertiary aliphatic C–H bond (non-activated C–H bond, BDE = 96 kcal mol−1).99 Next, we extended our study to the oxidation of para-substituted ethylbenzenes as substrates, containing electron-withdrawing and electron-donating substituents. Oxidized products were obtained in good benzylic alcohol yields ranging from 21 to 50%, with overoxidized ketone products between 5 and 21% yield. Enantioselectivity values for the alcohol product range from 45 to 62%. Oxidation of 1-chloro-4-ethylbenzene (9), containing an electron-withdrawing substituent, provided the alcohol product 9a in 41% yield and 59% ee, together with ketone byproduct 9b in 21% yield. 1-Bromo-4-ethylbezene (10) was also considered, which yielded the desired benzylic alcohol product 10a in 44% yield and 45% ee, generating the ketone product 10b in 21% yield. Oxidation of 4-ethylanisole (11), which bears an electron-donating substituent, provided lower activities, with the benzylic alcohol product 11a in 21% yield and 50% ee, with the ketone 11b being formed in only 5% yield. A dialkylbenzene was also considered in the current work. Particularly, we explored the oxidation of 4-ethyltoluene (12), which contains two benzylic positions. Interestingly, reaction mainly occurred on the ethyl substituent, which bears the C–H bonds with the lower BDE (85.4 and 89.7 kcal mol−1 for the BDE of the benzylic C–H bond of ethylbenzene and toluene, respectively),99 affording the benzylic alcohol product 12a in 42% yield and 62% ee, with the ketone product 12b in 10% yield. Oxidation at the other benzylic position occurred to a much lower extent, generating 4-ethylbenzyl alcohol in 5% yield. Other by-products were also detected, which might be assigned to products in which oxidation takes place at both alkyl substituents.
Entry | Catalyst | CAa | r.s.mb | Epoxidec | eed |
---|---|---|---|---|---|
a Carboxylic acid: AA = acetic acid, 2-eha = 2-ethylhexanoic acid. b Remaining starting material (r.s.m) in %. c Yields in % with respect to substrate determined by GC against an internal standard. d Enantiomeric excess determined by chiral GC. | |||||
1 | (S,S)-1 | AA | <1 | 98 | 78 |
2 | (S,S)-1 | 2-eha | <1 | 98 | 93 |
3 | 2 | AA | <1 | 98 | 84 |
4 | 2 | 2-eha | <1 | 98 | 97 |
Entry | Catalyst | CAa | r.s.mb | Epoxidec | eed |
---|---|---|---|---|---|
a Carboxylic acid: AA = acetic acid, 2-eha = 2-ethylhexanoic acid. b Remaining starting material (r.s.m) in %. c Yields in % with respect to substrate determined by GC against an internal standard. d Enantiomeric excess determined by HPLC on a chiral stationary phase. Ee values for Mn-catalyzed oxidations were not possible to determine due to low epoxide formation. | |||||
1 | (S,S)-1 | AA | 24 | 28 | — |
2 | (S,S)-1 | 2-eha | 57 | 4 | — |
3 | 2 | AA | 35 | 19 | — |
4 | 2 | 2-eha | 48 | 10 | — |
5 | 5 | AA | 13 | 67 | 53 |
6 | 5 | 2-eha | 20 | 56 | >99 |
For these reactions we have not been able to determine the enantioselectivity due to low concentration of the epoxide product. Next, we tested the non-heme iron complex 5, which provided much higher efficiencies for the epoxidation of 14, with yields up to 67 and 56% when acetic acid and 2-eha were employed, respectively. This observation contrasts with the previous study on enantioselective benzylic oxidations, where catalyst 5 was not capable of performing the aliphatic C–H hydroxylation of propylbenzene towards the benzylic alcohol product, and it shows that a highly electron-rich iron complex performs better for the epoxidation of aliphatic olefins compared to the analogous manganese complexes. Remarkably, the enantioselectivity obtained when 5 is used in the presence of 2-eha was excellent (>99%), which represents an increase compared with the related non-heme iron complex supported with the aminopyridine ligand containing dimethylamino substituents.59
Overall, the current complexes containing 4-pyrrolidinopyridine moieties provide enhanced enantioselectivities for the epoxidation of olefins. This finding is in accordance with the increase in enantioselectivity reported in previous studies by the introduction of dimethylamino or other similar amine substituents into several manganese and iron complexes, compared to the use of complexes with non-substituted pyridines.37,43,45,49,52,53,61
Future efforts in our laboratory will focus on the further development of highly electron-rich manganese and iron complexes that make use of strong electron-donating ligands for a better stabilization of the metal-oxo active species. Improvement of enantioselectivities for benzylic alcohol products, as well as the understanding of the factors that govern product chemoselectivity (aliphatic vs. aromatic oxidation) are currently being explored.
Footnote |
† Electronic supplementary information (ESI) available: Experimental details, synthesis and characterization of ligands and complexes and selected GC and HPLC chromatograms. CCDC 2100950 and 2100951. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/d1cy01642c |
This journal is © The Royal Society of Chemistry 2021 |