Dioxygen Activation by Internally Aromatic Metallacycle: Crystallographic Structure and Mechanistic Investigations.

Mononuclear metal-peroxo species are invoked as the key intermediates in metalloenzymatic or synthetic catalysis. However, either transience or sluggishness reactivity of synthetic analogs of metal-peroxo species impedes our understanding of oxygen activation mechanism. Herein, we designed and characterized a dioxygen-derived mononuclear osmium-peroxo complex, in which the peroxo ligand is stabilized by internally aromatic metallacycle. We demonstrate that the osmium-peroxo species shows catalytic activity toward promoterless alcohol dehydrogenations. Furthermore, computational studies provide a new mechanism for the osmium-peroxo-mediated alcohol oxidation, starting with the concerted double-hydrogen transfer and followed by the generation of osmium-oxo species. Interestingly, the internally aromatic metallacycle also plays a vital role in catalysis, which mediates the hydrogen transfer from osmium center to the distal oxygen atom of Os-OOH moiety, thus facilitating the Os-OOH→Os=O conversion. We expect that these insights will advance the development of aromatic metallacycle toward aerobic oxidation catalysis.

Introduction

Capturing highly active species and illustrating their reactivities are central issues in dioxygen activation chemistry (Elwell et al., 2017; Karlin, 2010; Lewis and Tolman, 2004; Hong et al., 2017; Dong at al., 2013; Solomon et al., 2014; Cho et al., 2016). Mononuclear metal-peroxo complexes are commonly implicated as reactive intermediates in oxidative catalysis of enzymatic or abiological systems, and considerable efforts have been dedicated to the synthesis and the mechanistic studies of them (Meunier, 2000; Karlsson, 2003; Yee and Tolman, 2015; Nam, 2015; Quist et al., 2017; Costas et al., 2004). Despite the fact that a number of mononuclear metal-peroxides have been elegantly synthesized and characterized, either the transient nature or the sluggish reactivity of these intermediates hampers our insight into the detailed dioxygen activation mechanisms during catalytic turnover (Shook et al., 2008; Cho et al., 2011; Ohta et al., 2013; Barman et al., 2019; Tronic et al., 2011). As the relative reactivity and stability of the metal-oxygen species are highly dependent on the supporting ligands or coordination environments (Cook and Borovik, 2015; Kovacs, 2015; Neu et al., 2015; Yan Poon et al., 2019), the rational ligand design constitutes a promising way to tune the relative stability and reactivity of metal-peroxo complexes (Mirica et al., 2004; Baglia et al., 2017). Internally aromatic metallacycles, featured by a rigid CˆCˆC-tridentate framework with electronically delocalized metallacyclic cores (Zhuo et al., 2019), were demonstrated to be efficient in stabilizing various metallacycle derivatives (Zhu et al., 2015). Structurally, these derivatives are termed as “carbolong complexes,” due to their unique ring systems consisting of different number of carbon atoms and one metal center in the equatorial plane (Zhu and Xia, 2018). Inspired by these findings, we speculated that internally aromatic metallacycle could be an ideal framework for oxygen binding and activation, which ultimately enable us to characterize a stable metal-dioxygen intermediate and further study its catalytic transformations.

Oxidation of alcohols toward corresponding carbonyl compounds is a fundamentally important reaction in both laboratory and industry (Kopylovich et al., 2015). However, the conventional alcohol oxidations usually require stoichiometric inorganic oxidants, in which a large amount of toxic waste was generated (Tojo and Marcos, 2006). As a result, developing environmentally benign catalytic systems using molecular oxygen as terminal oxidant is one of the long-term pursuit of chemists (Arends and Sheldon, 2004; Parmeggiani and Cardona, 2012; Shi et al., 2012; Punniyamurthy et al., 2005). Although great progress has been achieved in transition-metal-catalyzed aerobic oxidations of alcohols in the past decades (Brink et al., 2000; Stahl, 2004, 2005; Ryland and Stahl, 2014; Chung et al., 2013; Konnick, and Stahl, 2008; Sigman and Jensen, 2006), examples of alcohol oxidation mediated by mononuclear metal-peroxo species are rare (Bianchini et al., 1990; Esteruelas et al., 2011). Interestingly, osmium-dioxygen adduct was found to be an active catalyst in alcohol oxidations (Esteruelas et al., 2011). Nevertheless, understanding on the structural and mechanistic features in osmium-dioxygen-mediated alcohol oxidation is still limited.

Herein, we present the synthesis, characterization, and reactivity of a mononuclear osmium-peroxo complex. Theoretical investigations, including aromatic stabilization energy (ASE), nucleus-independent chemical shift (NICS) and anisotropy of the current-induced density (ACID), as well as experimental evidences indicated that the active species was stabilized by the internally aromatic pincer backbone. In particular, we demonstrate that the osmium-peroxo species could act as a pre-catalyst in alcohol oxidations without using additives. In contrast to the commonly proposed β-hydride elimination mechanisms for alcohol oxidation, our calculations show a new mechanism that the osmium-peroxo-catalyzed aerobic oxidation of alcohol proceeds via the concerted double-hydrogen transfer. As far as we know, the result outlined above is the first experimental evidence of a structurally characterized mononuclear metal-peroxide that is able to perform the alcohol dehydrogenation.

Results and Discussion

Synthesis and Characterization of an Osmium-Peroxo Complex

Osmapentalyne 1 (Zhu et al., 2013) was used as the starting material to study the binding of dioxygen (Scheme 1). The nucleophilic attack at carbyne carbon atom by sodium methoxide led to the reduction of metal center and dissociation of chloride, which facilitated the coordination and the following reduction of dioxygen. The yellow solution of 1 and CH3ONa in methanol under dioxygen atmosphere at room temperature (RT) underwent gradual color change to generate a brown solution (Figure 1A, left insets), which was followed by ultraviolet-visible spectroscopic changes (Figure 1A). As also indicated by in situ NMR, the mixture of 1 and excess CH3ONa in the presence of dioxygen at RT for 3 h resulted in almost quantitative conversion to 2 (Figure S1, and see also Transparent Methods for details), which was characterized by high-resolution electrospray ionization-mass spectroscopy (HRMS). The HRMS of 2 shows a prominent peak at m/z = 1,187.2762 (Figure 1B), in which the mass and isotope distribution patterns correspond to the cation of 2 (calculated for cation of 2, m/z 1.187.2793). Upon introduction of 18O into 2 using 18O2, a shift of four mass units was observed, from m/z 1,187.2762 to 1,191.2884 (Figure 1B, insets), indicating that dioxygen bound to the metal center of 2.

Scheme 1

Synthesis of 2 via the Reaction of Osmapentalyne 1 with O2 in the Presence of CH3ONa

See also Figures S1 and S14–S16.

Figure 1

Spectroscopic Characterization of 2

(A) UV-visible spectral changes observed upon addition of CH3ONa (15 equiv; 1.5 mM in methanol) to a methanol solution of 1 (0.1 mM; yellow line) under O2 atmosphere at RT. The left insets show the color changes of the solution during the conversion of 1 (yellow) to 2′ (brown). The right inset shows the change over time of the absorbance at 382 nm.

(B) High-resolution electrospray ionization mass spectrometry spectra (positive ion mode) of the reaction mixture of 1 and CH3ONa (5 equiv, CH3OH, 3 h) in air. The insets show the observed isotope distribution patterns for 2-16O2 (green panel) and labeled 2-18O2 (purple panel).

(C) Fourier transform infrared spectra of 2-16O2 (green line) and 2-18O2 (purple dashed line) as recorded from KBr disks. Yellow line is the difference spectrum of 2-16O2 and 2-18O2.

See also Figure S17.

The good thermal stability of 2 (Figure S2) in the absence of substrates allowed us to obtain single crystals suitable for spectroscopic and structural analysis. The solid infrared spectrum (KBr) of 2 exhibits a vibrational band at 913 cm−1, which shifts to 860 cm−1 upon replacement of 16O by 18O (Figure 1C). The observed isotopic shift value of Δv = 53 cm−1 is consistent with the value of 52 cm−1, calculated using Hooke's law of harmonic oscillators. These values are also comparable to those recorded for metal-peroxo complexes (Cho et al., 2011; Tronic et al., 2011; Hu et al., 2004; Lee et al., 2012; Waidmann et al., 2010), and lead us to conclude that the O2 unit in 2 possesses a peroxo character.

Direct evidence for the geometry of osmium-dioxygen moiety is provided by the X-ray crystal structure of 2. As shown in Figure 2, the crystal structure revealed a mononuclear osmium complex with the triangular Os−O2 unit having a small bite angle (40.18(14)º for O1−Os−O2). The five coordination sites in the equatorial plane are occupied by the CˆCˆC-type tridentate ligand and the peroxo ligand, and the axial positions are occupied by two phosphine ligands. The peroxo ligand within the Os−O2 unit is bound almost symmetrically to the metal center with Os−O bond lengths of 2.034(4) and 2.051(4) Å. The O1−O2 bond length in 2 (1.403(5) Å) is within the range of O−O bond distances found in other osmium-peroxo complexes (ca. 1.396–1.452 Å) (Mezzetti et al., 1994; Maddock et al., 1996). In accordance with the experimental values, we concluded that 2 is a side-on osmium-peroxo complex. This structure contains three metal-carbon bonds and two metal-oxygen bonds in an equatorial plane, which has not been reported previously in coordination chemistry. To the best of our knowledge, this crystal structure is the first example of a metal-peroxo complex lacking heteroatomic binding sites of a polydentate ligand scaffold.

Figure 2

Molecular Structure of the Cation of 2

Molecular structure of 2 with thermal ellipsoids drawn at 50% probability level. Hydrogen atoms in the PPh3 moieties are omitted for clarity. Selected bond lengths (Å) and angles (deg): O1−O2 1.403(5), Os1−O1 2.034(4), Os1−O2 2.051(4), O1−Os1−O2 40.18(14), Os1−O1−O2 70.6(2), Os1−O2−O1 69.2(2).

See also Table S1.

Aromaticity Evaluation of Osmium-Peroxo Complex

Aromaticity plays vital role in stabilizing organic-aromatic or metalla-aromatic compounds (Krygowski and Cyrański, 2001; Chen and Jia, 2013; Schleyer, 2001). In this study, the stability of the osmium-peroxo complex 2 could be attributed to the aromaticity of the metallacycle composed of the CˆCˆC skeleton and a metal center, which is also thought to have strong influence on the oxygen transfer process. The good stability, ring planarity, bond length equalization, and low-field exocyclic proton chemical shifts of metallacycle indicate that complex 2 is aromatic in nature. To shed light on the aromaticity, density functional theory (DFT) calculations were performed on the simplified model complexes 2-1 and 2-2. The PPh3 and the ester in 2-1 are replaced by PH3 and methyl, respectively, whereas in 2-2, the ester group is reserved (Figure 3). The computed positive ASE values (Schleyer and Pühlhofer, 2002; Wannere et al., 2003) and negative NICS values (Schleyer et al., 1996; Chen et al., 2005; Fallah-Bagher-Shaidaei et al., 2006) generally act as the indicator for aromaticity. As depicted in Figure 3A the ASE values of 2-1 (28.7 and 25.8 kcal mol−1) are comparable to the corresponding ASE values of previously reported metalla-aromatics (Chen and Jia, 2013; Landorf and Haley, 2006; Fernández et al., 2015), indicating the aromaticity of 2. The calculated NICS(1)zz values for the two metallacyclic rings of 2-2 are −10.3 (ring a) and −12.7 (ring b) ppm (Figure 3B), which are similar to those reported for other metalla-aromatics (Wei et al., 2018; Frogley and Wright, 2018). The aromaticity of 2 is further supported by the ACID (Herges and Geuenich, 2001; Geuenich et al., 2005), which can simulate the density and direction of the induced ring current in a molecular system under an external magnetic field. As shown in Figure 3C, an obvious clockwise diatropic ring current can be observed in 2-2, indicating the aromaticity of the two fused five-membered rings.

Figure 3

Aromaticity Evaluation of 2

(A) Aromatic stabilization energy (ASE, kcal mol−1) evaluations of the aromaticity of model complex 2-1.

(B) NICS(1)zz (nucleus-independent chemical shift along the z axis at 1 Å above the ring critical point) evaluations of the aromaticity of model complex 2-2.

(C) Anisotropy of the current-induced density (ACID) plot of 2-2 from π contribution with an isosurface value of 0.035. The magnetic field vector is orthogonal to the ring plane and is directed upward (aromatic species exhibit clockwise diatropic circulations).

The Reactivity of Osmium-Peroxo Complex in the Catalytic Oxidation of Alcohol

Metal-oxygen intermediate was extensively invoked as the active species for the hydrogen atom abstraction from C−H or O−H bonds (Sacramento and Goldberg, 2018; Gordon et al., 2019). As such, we first investigated the catalytic activity of 2 with 4-methylbenzyl alcohol using pure oxygen as the terminal oxidant. It is seen from Figure 4 that the catalytic oxidation completes within 12 h, affording the desired aldehyde in 92% isolated yield. With the preliminarily optimal reaction conditions in hand, we explored the substrate generality of this protocol (Figure S3). We found that alcohols with weak C-H bonds and acidic O-H bonds, such as 4-nitrobenzyl alcohol, 1-pyrenemethanol, and benzhydrol, were converted to the corresponding carbonyl compounds in high yield. In contrast, cyclohexanol with stronger C-H bond and slightly acidic O-H bond had significantly low yield relative to benzylic alcohols, suggesting that the alcohol oxidation rate is dependent on the C-H strength of substrate (Rajabimoghadam et al., 2018).

Figure 4

Aerobic Catalytic Oxidation of p-methylbenzyl Alcohol by 2 and the Kinetic Isotope Effects

(A) Parallel KIE studies conducted for the 2-catalyzed oxidation of 4-MeC6H4CH2OH (3) and isotopically labeled 4-MeC6H4CH2OD (3-OD). (B) Parallel KIE studies conducted for the 2-catalyzed oxidation of 4-MeC6H4CH2OH (3) and 4-MeC6H4CD2OH (3-CD).

See also Figures S3–S6.

The parallel kinetic isotope effect (KIE) studies conducted with the 2-catalyzed oxidation of 4-MeC6H4CH2OH (3) and 4-MeC6H4CH2OD (3-OD), respectively, revealed an overall KIE value of 1.16 (Figures 4A and S4). In addition, with 3 and 4-MeC6H4CD2OH (3-CD), a relatively larger overall KIE value of 2.50 was observed (Figures 4B and S5). To ensure the reliability of the later KIE value, intermolecular competition experiment by mixing equal molar amounts of 3 and 3-CD in the same reaction vessel was carried out to give an overall KIE value of 2.86 (Figure S6). The aforementioned kinetic results imply that the bond cleavage is C−H rather than O−H occurring in the turnover-limiting step (Simmons and Hartwig, 2012; Jones, 2003).

Mechanistic Studies of Catalytic Oxidation

Despite extensive experimental studies, the mechanisms of O2 activation and the metal-peroxo-mediated oxidation of alcohols remain elusive. Herein, computational studies in combination with experiments were conducted to gain insight into the detailed mechanism. We first carried out the catalytic oxidation reaction with excess of 1,1-diphenylethylene and 2,4,6-tri-tert-butylphenol, respectively, and found that free radical scavenger did not obviously suppress the reaction rates and the yield of aldehyde. This finding suggests that the catalytic oxidation probably proceeds via a non-radical mechanism. In the control experiments, the amounts of PPh3 and Ph3P=O in the initial stage of the reaction mixture with and without substrate were monitored by gas chromatography-mass spectrometry (GC-MS) (Figure S7). It was found that the amount of free PPh3 in both control experiments were negligible, but the amount of Ph3P=O in the initial stage of the reaction mixture in the presence of substrate was about 11 mg according to the GC-MS analysis. As such, we think that the first step of the catalytic cycle involves the dissociation of one of the phosphine ligand, which was then oxidized to phosphine oxide. In addition, the addition of PPh3 would significantly suppress the catalytic reaction, probably due to the inhibition of first step in catalytic cycle or competitive oxidation between PPh3 and substrate. With these results, we proposed that the first step of catalysis involves the dissociation of one PPh3 ligand from 2 and the following alcohol coordinate to the metal center (Figure S8). Figure 5 presents the calculated free energy profiles for the osmium-peroxo-catalyzed aerobic oxidation of alcohols. It is seen from Figure 5 that two distinct binding conformations of benzyl alcohol can be located in this study. In reactant complex RC0, the hydroxyl oxygen atom is coordinated to osmium center, whereas in the alternative reactant complex RC1, the hydrogen of hydroxyl is H-bonded to one of the peroxo oxygen atom. Although RC0 is calculated to be 8.6 kcal/mol lower in energy than RC1, the aldehyde formation via the initial proton transfer from the hydroxyl group of substrate to the peroxo moiety in RC0, followed by another hydride dissociation from the benzylic C–H bond, is demonstrated to be highly unfavorable (the calculated barriers are over 30 kcal/mol in all three possible pathways, see Figure S9). By contrast, we located a new pathway from RC1 (Figure 5), in which the proton transfer from the hydroxyl group to the peroxo moiety is coupled with the hydride transfer from the benzylic CH2 group to metal center (TS1). This new reaction channel involves a Gibbs energy barrier of 22.1 kcal/mol relative to RC0, leading to the formation of aldehyde products and OsH(OOH) species in IC1. By investigating all the possible reaction pathways for the OsH(OOH) intermediate in IC2 (Figure S10), we found that the most favorable pathway is initiated by the hydride shift from osmium center to the neighboring C1 (TS2), with a Gibbs energy barrier of 16.1 kcal/mol relative to IC2, whereas the hydride transfer to the another neighboring carbon (C7) requires an even higher barrier of 20 kcal/mol (Figure S11). Notably, conversion of IC2 to IC3 involves dearomatization of internally aromatic metallacycle, which is similar to the previously reported metal-ligand cooperation via dearomatization-aromatization process of ligand (Gunanathan and Milstein, 2011; Zeng et al., 2014; Khusnutdinova and Milstein, 2015; Li et al., 2019) The subsequent proton transfer from C1–H to the distal oxygen atom of Os–OOH (TS3) leads to the heterolytic cleavage of O–O bond and the formation of osmium-oxo species (Os=O) in IC4. Meanwhile, the internal aromaticity for metallacycle was recovered. This step experiences an overall Gibbs energy barrier of 24.2 kcal/mol (TS3) and constitutes the rate-determining step for the overall transformations. Such computational finding is in accordance with KIE experiments that the hydrogen originally from substrate C-H bond participates in the rate-determining transformations.

Figure 5

DFT-Computed Gibbs Energy Profile (in kcal/mol) for the Osmium-Peroxo-Catalyzed Aerobic Oxidation of Alcohol

Key distances are given in angstroms.

See also Figures S7–S11 and S13.

To complete the catalytic cycle, we proceed to consider the Os=O-mediated oxidation of alcohols. The calculated free energy profile is depicted in Figure 6. It is seen that the formation of aldehyde occurs via a double-hydrogen shift mechanism (TS4), in which the proton from the hydroxyl group shifts to oxo of Os=O species, whereas the hydride from the benzylic CH2 group shifts to the metal center. Such mechanism is analogous to the osmium-peroxo-catalyzed oxidation of alcohol via TS1. This step experiences a Gibbs energy barrier of 16.5 kcal/mol relative to IC5, leading to the formation of aldehyde product and osmium-hydrido-hydroxo species (OsH(OH)) in IC6. All these indicate that the reactivity of osmium-oxo is much higher than that of osmium-peroxo species. The final hydride shift to the neighboring C1, followed by the H2O formation, recovers the initial state of osmium-peroxo species RC0 (Figure S12, and see also Figure S13 for full catalytic mechanism). Our calculations demonstrate that hydrogen transfer in TS3, where the hydrogen is initially derived from the benzylic CH2 group of alcohol, is involved in the rate-determining step in overall transformations. This is in line with our experimental findings. As such, our computational studies reveal that both osmium-peroxo and oxo species act as the reactive intermediate toward alcohol oxidations. Interestingly, the internally aromatic metallacycle plays an unusual role in catalysis by facile metal-ligand cooperative catalysis through reversible dearomatization-aromatization. Such reversible transformation would facilitate the Os–OOH→Os=O conversion via the neighboring carbon site-assisted proton transfer, and thus significantly enhance the catalytic reactivity of 2. This also explains why the osmium-peroxo species 2 could serve as an oxidant in the promoterless alcohol oxidations, whereas most of the previously catalytic processes require additives, such as nitroxyl radicals and base, to complete the catalytic cycle (Kopylovich et al., 2015).

Figure 6

DFT-Computed Reaction Gibbs Energy Profile (in kcal/mol) for the Osmium-Oxo-Catalyzed Aerobic Oxidation of Alcohol

Key distances are given in angstroms.

See also Figures S12 and S13.

Conclusion

In this work, we demonstrated that the relatively stable osmium-peroxo species can be achieved through the reaction of osmapentalyne with molecular oxygen in the presence of nucleophile. The peroxo ligand was almost symmetrically bound to the metal center in a side-on fashion determined from X-ray crystal structure. DFT calculations together with experiments showed that the aromaticity of metallacycle provided the stability for active osmium-peroxo species, which was also capable of promoterless alcohol dehydrogenation. Different from the commonly proposed β-hydride elimination mechanisms for alcohol oxidation, our calculations demonstrate that the osmium-peroxo-catalyzed aerobic oxidation of alcohol proceeds via the concerted double-hydrogen transfer mechanism, affording the osmium-hydrido-hydroperoxo species (OsH(OOH)) and aldehyde in one step. The subsequent ligand-assisted O–O heterolytic cleavage in OsH(OOH) generates the second active species of osmium-oxo complex (Os=O) for alcohol oxidation. Thus the catalysis involves an unusual dearomatization-aromatization of metallacycle. The kinetic results and theoretical calculations suggest that the turnover-limiting step involves the transfer of a hydrogen atom derived from the benzylic CH2 of the substrate. These observations represent the rare examples of structurally characterized metal-peroxo complex in the aerobic oxidation of alcohol, and mechanistic insights could help to develop the new and more efficient catalytic systems for green and sustainable chemistry. Moreover, it is expected that the osmium-peroxo complex synthesized here could be applied to the promoterless oxidations of other substrates, such as amines.

Limitations of the Study

Alkene oxidation was also studied under the present catalytic conditions, but no hydroxylation or epoxidation products were observed.

Resource Availability

Lead Contact

Further information and requests for resources should be directed to and will be fulfilled by the Lead Contact, Haiping Xia (hpxia@xmu.edu.cn).

Materials Availability

Any materials generated in this study are available from the Lead Contact without restriction.

Data and Code Availability

The accession number for the crystallographic data of complex 2 reported in this paper is Cambridge Crystallographic Data Center: 1892835.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.