Download PDF
Synlett 2021; 32(15): 1473-1478
DOI: 10.1055/a-1503-7339
synpacts

Cu(I)–Bis(phosphine) Dioxides as Catalysts for the Enantioselective α-Arylation of Carbonyl Compounds

Margarita Escudero-Casao
,
Giulia Licini
,
Manuel Orlandi  ∗

We thankfully acknowledge the Department of Chemical Sciences of the Università degli Studi di Padova (P-DiSC#08BIRD2019) for financial support and a postdoctoral fellowship for E.-C. M.
› Further Information
 
 PDF Download Permissions and Reprints All articles of this category


Dedicated to Prof. Franco Cozzi, an ‘evergreen’ mentor, on the occasion of his 70th birthday

Abstract

The transition-metal-catalyzed α-arylation of carbonyl compounds was first reported by Buchwald and Hartwig in 1997. This transformation has been used and studied extensively over the last two decades. Enantioselective variants were also developed that allow for controlling the product stereochemistry. However, these suffer several limitations in the context of formation of tertiary stereocenters. Presented here is our group’s contribution to this research area. The chiral Cu-bis(phosphine) dioxides catalytic system that we reported allowed accessing the enantioselective α-arylation of ketones that were not suitable for this transformation before in good yields and er up to 97.5:2.5. Preliminary insight and speculation concerning the reaction mechanism involving the unusual pairing of bis(phosphine) dioxides with transition-metal catalysts is also given.

1 Introduction

2 State of the Art

3 Enantioselective α-Arylation of Acyclic Ketones

4 Summary and Conclusions


#

Key words

Cu catalysis - asymmetric catalysis - ligand design - α-arylation - bis(phosphine) dioxides
1

Introduction

Zoom Image
Margarita Escudero-Casao (left) graduated in industrial chemistry at the Universidad Veracruzana (Mexico). She obtained her PhD at the CINVESTAV (Mexico) under the supervision of Prof. E. Juaristi. In 2015 she joined the group of Prof. S. Castillón at URV in Spain as a postdoctoral researcher. After a period in industry, she moved to the University of Padova where she is now working on catalysis with emphasis on sustainable processes. Giulia Licini (middle) studied chemistry at the University of Padova where she received her PhD with Giorgio Modena. After a postdoc (Colorado State University with A. I. Meyers), she was appointed at the University of Padova, where she is now professor of organic chemistry. Her research field is mainly related to metal-catalyzed processes and molecular recognition, with a particular emphasis in small-molecule activation, green and sustainable processes, and reaction mechanisms. Manuel Orlandi (right) was born in Italy. He obtained a Master’s degree and a PhD in chemical sciences at the University of Milan under the supervision of M. Benaglia and F. Cozzi. In 2016 he joined the group of M. S. Sigman (University of Utah) where he worked on Pd catalysis and multidimensional correlation analysis. In 2018 Manuel moved back to Italy and after a short period at the University of Milan working for Jansen, he joined the faculty of the Department of Chemical Sciences (DISC) at the University of Padova. His research interests span from asymmetric catalysis to physical organic chemistry.

The transition-metal-catalyzed α-arylation of carbonyl compounds is a powerful strategy for the formation C(sp3)–C(sp2) bonds present in bioactive and natural compounds (Scheme [1]A).[1] In the past years, several research groups have engaged in the development of efficient catalytic systems for the construction of α-arylated quaternary stereocenters in enantioselective fashion.[2] However, fewer methods have been reported for the asymmetric formation of enolizable (e.g., tertiary) stereocenters.[3] Arylated products display increased acidity with respect to their corresponding starting materials (Scheme [1]B),[4] which can result in facile postreaction racemization via enolization. Such undesired event is pronounced especially in the presence of strong bases typically employed in direct α-arylation reactions. Elegant approaches to overcome this limitation have been devised. These rely on the use of preformed nucleophiles (e.g., silyl enol ethers, tin enolates) or enamine catalysis in order to activate the carbonyl α-position under mild conditions.[5] [6] [7] [8] [9] [10] [11] Here, we introduce the state of art and contextualize our recent contribution to this field.[12]

Zoom Image
Scheme 1 A) Examples of bioactive molecules featuring an α-aryl carbonyl moiety. B) General α-arylation reaction resulting in lowering of the α-proton pK a.

# 2

State of the Art

In 2011, Zhou and coworkers reported the first example of enantioselective α-arylation of esters catalyzed by a chiral Pd complex to provide tertiary stereocenters (Scheme [2]A).[5] The combination of silyl ketene acetals 1 and aryl triflates 2 in the presence of Pd, phosphine 3a, and LiOAc provided the monoarylated esters 4 in high yields and enantioselectivities. The use of LiOAc was found to facilitate the transmetallation step to form the Pd-enolate, thus increasing the reaction performances. Similarly, Yamamoto and coworkers found that TlOAc was effective to this purpose when Josiphos-type ligands were employed.[6] Even though the geminal report by Zhou was limited to reaction of tert-butyl ester derivatives (less sterically demanding esters provided lower er), in 2013 the same group showed that fine tuning of the ligand structure to 3b could result in efficient coupling of lactone-derived nucleophiles 5 (Scheme [2]B).[7] This work is a brilliant example of how weak noncovalent interactions (NCIs depicted in structure 7, Scheme [2]) can be tuned in order to access better ligands. The authors showed that 3b is suitable for the α-arylation of ketones as well (Scheme [2]C).[8] However, given the lower nucleophilicity of silyl enol ethers, more reactive tin-enolates 8 had to be used that could be generated in situ from alkenyl acetates 9 and nBu3SnOMe. This strategy could be used to forge tertiary stereocenters in high er with cyclohexanones and tetralones. However, acyclic ketones were not suitable substrates (for instance, propiophenone could be arylated in only 66:34 er).[8] Moreover, the use of stoichiometric organotin reagents rises cogent issues concerning safety and toxicity. Overall, Zhou and coworkers showed that Pd complexes of chiral monophosphines are excellent catalysts for the arylation of activated carbonyl compounds, while the arylation of ketones still showed room for improvement.

Zoom Image
Scheme 2 Pd-catalyzed α-arylation of carbonyl compounds with aryl triflates: A) tert-butyl esters, B) lactones, C) hexanones and tetralones.

Catalytic systems involving Cu instead of Pd have been developed as well. In this context, diaryliodonium salts 12 are arylating agents of choice. These compounds are hypervalent iodine reagents bearing two aryl rings. The electrophilic character of diaryliodonium salts govern their reactivity as electrophiles and strong oxidants. Due to their convenient synthesis and high reactivity, they have been extensively used in organic synthesis and metal catalysis.[13] In 2011, MacMillan and Gaunt developed a strategy for the enantioselective α-arylation of silyl ketene imides 11 with diaryliodonium salts 12 in the presence of Cu(BOX) catalysts 13 (Scheme [3]A).[9] [10] A tentative mechanism was proposed that involves the oxidative addition of 12 to 13 resulting in the formation of an electrophilic chiral Cu(III)–Ar species 14. This would then enable a mild arylation with a suitable nucleophile 11 to furnish products 15 (Scheme [3]A). These mild catalytic conditions gave access to valuable synthetic building blocks in excellent yields and enantioselectivities. The same year, MacMillan also demonstrated that the enantioselective α-arylation of aldehydes 16 with diaryliodonium salts 12 was possible through substrate activation via aminocatalysis with 17 in the presence of CuBr (Scheme [3]B).[11] The effectiveness of the methodology relied on electrophilic activation of 12 with Cu(I) to give the Cu(III)–Ar species 19, while nucleophile activation and stereocontrol were provided by the chiral enamine 18. The desired α-arylated aldehydes 20 were obtained in high yields with excellent enantioselectivities.

Zoom Image
Scheme 3 Cu-catalyzed enantioselective α-arylation of imides and aldehydes with diaryliodonium salts: (A) substrate activation via silyl ketene imide or (B) via enamine catalysis.

# 3

Enantioselective α-Arylation of Acyclic Ketones

Recently, our research group reported the use of a catalytic system based on Cu(I) and chiral bis(phosphine) dioxides to achieve enantioselective α-arylation of noncyclic ketones.[12] Our initial studies showed that the silyl enol ether of propiophenone (21a) could react with diaryliodonium salt 12a to give the corresponding α-arylated product 23a in 74:26 er in the presence of (CuOTf)2·Tol and BINAPO 22a (Scheme [4]A). Interestingly, bis(phosphine) dioxides provided reactivity while several other classes of ligands gave no reaction. Among these were 13, semicorrins, bipyridines, phenolates, phosphines, and bisphosphine monoxide (R)-BINAP(O) (monoxide of (R)-BINAP). Due to their polar P–O bond, chiral phosphine oxides are effective Lewis base catalysts.[14] Nevertheless, these have found little application in combination with transition-metal catalysis.[15] [16] [17] [18] Therefore, the enantioselectivity observed in our preliminary results was somehow surprising.

Zoom Image
Scheme 4 Enantioselective α-arylation of silyl enol ethers with Cu(I)–bis(phosphine) dioxides: selected examples of ligands screening. Xyl = 3,5-Me2C6H3; DMM = 3,5-Me2-4-MeOC6H2; BTMF = 3,5-(CF3)2C6H3; DTBM = 3,5-tBu2-4-MeOC6H2.

Aiming at the optimization of the stereochemical outcome of our benchmark reaction, we carried out a screening of several chiral bis(phosphine) dioxides 22 easily obtained from their corresponding commercial phosphines. The enantioselectivities of selected examples are shown in Scheme [4]. The GARPHOSO (22i–m) and SEGPHOSO (22n–p) family ligands provided the best performances (up to 86:14 er with 22l). Importantly, we also noticed that the chemical activity was related to the electronic properties of the phosphine oxide used. For instance, using electron-poor compounds such as 22m or 22e resulted in little or no conversion. Similarly, exceedingly bulky ligands showed diminished reactivity and poor selectivity (e.g., 22p).

It has been shown that establishing relationships between the ligand structure and a reaction stereochemical outcome can serve as a guide towards the synthesis of ideal candidates.[19] Therefore, we reasoned that a proper descriptor(s) providing a trend with the observed enantioselectivity could be found. Following previous examples set by the Sigman group,[20] we turned to computational chemistry in order to access a number of molecular parameters for compounds 22. Optimization of the geometries and their frequency analysis were performed at the M06-2X/6-31G(d) level of theory. From the optimized structures, we could collect a number of descriptors accounting for the steric, geometric, and electronic features of compounds 22. These included: Verloop sterimol parameters[21] and buried volumes[22] accounting for size and steric for specific portions of the ligands (e.g., aryl groups or aryl substituents); key bond lengths and angles of the ligand geometry (e.g., length of the P–O bond); vibrational frequencies and intensities,[23] and charges accounting for the electronics. With these parameters in hand, we could perform single-parameter correlations. Among all of the descriptors acquired, νPOas (the asymmetric vibrational stretching of the two P–O bonds) was the only one providing a trend with the observed enantioselectivity (expressed as ΔΔG‡ in kcal/mol according to the Curtin–Hammett principle). Vibrational frequencies are directly related to both the reduced mass and the strength of the bonds involved in the vibrational mode. As such, frequencies are linked to the electronic features of these bonds. Therefore, the trend between the enantioselectivity and νPOas in Figure [1] suggests that electronics of the phosphine oxide binding site (the P–O groups) is important for accessing high er values. The trend improved to a correlation when the data points accounting for biased ligands (either sterically 22p, or geometrically 22f–h) were removed from the data set (purple points in Figure [1]). The removed data points suggest that more than one parameter is required for a complete description of the system. In retrospect, we found that more inclusive models could be obtained when using multivariate correlations, which would include steric descriptors for the P-aryl substituents, and the dihedral angle of the chiral biaryl scaffold.[12] However, even though very simple, the correlation in Figure [1] points towards electron-rich phosphine oxides as ligands with better performances. Therefore, a number of candidates were evaluated computationally in order to identify ligands with low νPOas value. This in silico evaluation resulted in 22u as the best predicted candidate (Figure [1]). On the other hand, chemical intuition and direct comparison of oxides 22i–l suggest 22s as the best candidate (also based on additivity of the Hammett σ values). Therefore, we synthesized these phosphine oxides together with ent-22t as an additional candidate to test our model. Pleasingly, when we tested compounds 22s–u in our benchmark reaction, 22u provided product 23a in 94:6 er and 58% yield. Moreover, the three synthesized ligands were found to obey the correlation with νPOas (red squares in Figure [1]).

Zoom Image
Figure 1 Single-parameter correlation of the measured enantioselectivity with the phosphine oxide P–O bond stretching frequency ν POas. Data points of the biased substrates that have been removed from the initial correlation are purple triangles. Predicted data points are reported as red squares.

With our optimized ligand 22u, we next explored the reaction scope. Silyl enol ethers 21 of aryl ketones were investigated first. We found that electron-donating aryl groups provide better performances due to their higher nucleophilicity (23a–e, Scheme [5]A). These compounds were obtained in high er and good yields. On the other hand, less nucleophilic counterparts (e.g., 23f–g) afforded modest er and yields. These results were improved by changing the ligand to ent-22z, which afforded ent-23f and ent-23g in acceptable yield and er (Scheme [5]A). Over time, no erosion of enantioselectivity was observed for ent-23g despite its lower pK a due to the 4-CF3 group.

Zoom Image
Scheme 5 Reaction scope

The scope of the iodonium salts 12 was also investigated. Both electron-rich and electron-poor aryl groups reacted in typically good yields and high er (23h–q, Scheme [5]B). As a general trend, we observed that electron-deficient aryl groups are transferred more slowly. Ultimately, miscellaneous α-arylated ketones were explored where the ketone alkyl component was modified as well (Scheme [5]C). This position was found to be more sensitive. A longer alkyl chain was tolerated (23t–u) even though changing the Et group to an iPr resulted in complete loss of reactivity (not shown). Tetralone could be arylated to give 23v in 30% yield and 94:6 er, while isoflavanone 23z was obtained in 88% yield but low er (Scheme [5]C). In this sense, our method is synthetically complementary to previous work by Zhou,[8] which ensured efficient α-arylation for cyclic ketones (vide infra).

We found that linear dialkyl ketones were also suitable reaction partners even though with diminished efficiency (23aa–ad, Scheme [6]). This is remarkable as the enantioselective α-arylation of this class of carbonyl compounds was unprecedented. Moreover, methyl propionate ent-23ae was obtained in 51% yield and 90:10 er (with ligand ent-22t) demonstrating that also esters are amenable of arylation under our catalytic conditions. Further optimization of our catalytic system for these substrates is currently ongoing in our laboratory.

Zoom Image
Scheme 6 Arylation of silyl enol ethers of dialkyl ketones and methyl propionate

After evaluation of the reaction scope, we turned our attention to the role of 22. Phosphine oxides are generally believed to be labile ligands in transition-metal catalysis.[24] Moreover, they are also known as excellent Lewis bases.[14] Therefore, one mechanistic hypothesis was the involvement 22 as a Lewis base in the activation of the arylating agent 12 via formation of a hypervalent complex 24 (Scheme [7]A).[14a] The formation of the PO–I–Mes 3c4e bond in 24 would ensure delocalization of electron density in the bonding and antibonding orbitals (localized on the PO and Mes ligands). This would result in the polarization of the Ar–I bond with consequent higher electrophilic reactivity of 24 with respect to 12. In an alternative mechanistic scenario, 22 could act as a strong donor ancillary ligand at the Cu center promoting reactivity via stabilization of a Cu(III)–Ar intermediate 25 (Scheme [7]B). According with the tentative picture previously proposed by MacMillan and coworkers (vide infra),[9] [11] the mechanism in Scheme [7]B shows Cu engaging 12 first (to give 25) and 21 after to deliver product 23. However, at this stage the reverse order cannot be ruled out, as cannot be ruled out the involvement of a radical pathway.

Zoom Image
Scheme 7 Preliminary mechanistic considerations

In order to gain additional information concerning the reaction mechanism, we performed an initial rate kinetic analysis of our benchmark reaction. If 22 acted as a Lewis base by activating 12 at the reaction rate-determining step via 24, a positive kinetic order for 22 would be observed. On the contrary, if 22 acted as a ligand at the metal center, either a 0th or negative kinetic order would be expected. In this latter case, the negative order would be informative about the formation of an inactive [Cu(22)2] complex 26, from which dissociation of one molecule of 22 would be required for the reaction to proceed via a 1:1 complex [Cu(22)] (Scheme [7]B). Plotting the reciprocal of the observed initial reaction rate 1/k obs against the concentration of bis(phosphine) dioxide 22a resulted in a linear correlation with positive slope. Therefore, the observed kinetic order of –1 for 22a suggests the involvement of phosphine oxides as ligands.[25] This work adds to other recent reports showing how phosphine oxides bind effectively to first row transition-metal centers such as Fe,[15] [16] Ni,[26] and Cu[12,18] in asymmetric catalysis.


# 4

Summary and Conclusions

The α-arylation of carbonyl compounds is a well-established transformation that typically relies on Pd catalysis. Asymmetric variants of this transformation have been reported that allow for efficient setting of quaternary stereocenters. However, only a handful of methods have been reported that allow for accessing tertiary stereocenters in enantioselective fashion. This is due to facile postreaction racemization of α-arylated products. Based on previous reports by Gaunt and MacMillan, we posited that Cu catalysis could be employed to access the asymmetric α-arylation of simple ketones, a missing link of the Buchwald–Hartwig arylation reaction. We found that this was indeed the case when chiral bis(phosphine) dioxides were employed as ligand. Optimization of the ligand structure was performed via rational design though a correlation analysis. With the optimized ligand in hand, we could perform the enantioselective α-arylation of a variety of ketones. Notably, these include linear dialkyl ketones, which, although obtained in modest stereochemical outcome, represent uncharted territory in the domain of asymmetric α-arylation reactions. Improvement and extension of this reaction, its mechanistic analysis, and application of our novel catalytic system to other transformations are currently ongoing in our group.


#
#

Conflict of Interest

The authors declare no conflict of interest.


Corresponding Author

Manuel Orlandi
Department of Chemical Sciences, University of Padova, CIRCC – Consorzio Interuniversitario per le Reattività Chimiche e la Catalisi, Padova Unit
via Marzolo 1, 35131 Padova
Italy   

Publication History

Received: 27 April 2021

Accepted after revision: 09 May 2021

Accepted Manuscript online:
09 May 2021

Article published online:
08 June 2021

© 2021. Thieme. All rights reserved

Georg Thieme Verlag KG
Rüdigerstraße 14, 70469 Stuttgart, Germany


   Permissions and Reprints All articles of this category
Zoom Image
Margarita Escudero-Casao (left) graduated in industrial chemistry at the Universidad Veracruzana (Mexico). She obtained her PhD at the CINVESTAV (Mexico) under the supervision of Prof. E. Juaristi. In 2015 she joined the group of Prof. S. Castillón at URV in Spain as a postdoctoral researcher. After a period in industry, she moved to the University of Padova where she is now working on catalysis with emphasis on sustainable processes. Giulia Licini (middle) studied chemistry at the University of Padova where she received her PhD with Giorgio Modena. After a postdoc (Colorado State University with A. I. Meyers), she was appointed at the University of Padova, where she is now professor of organic chemistry. Her research field is mainly related to metal-catalyzed processes and molecular recognition, with a particular emphasis in small-molecule activation, green and sustainable processes, and reaction mechanisms. Manuel Orlandi (right) was born in Italy. He obtained a Master’s degree and a PhD in chemical sciences at the University of Milan under the supervision of M. Benaglia and F. Cozzi. In 2016 he joined the group of M. S. Sigman (University of Utah) where he worked on Pd catalysis and multidimensional correlation analysis. In 2018 Manuel moved back to Italy and after a short period at the University of Milan working for Jansen, he joined the faculty of the Department of Chemical Sciences (DISC) at the University of Padova. His research interests span from asymmetric catalysis to physical organic chemistry.
Zoom Image
Scheme 1 A) Examples of bioactive molecules featuring an α-aryl carbonyl moiety. B) General α-arylation reaction resulting in lowering of the α-proton pK a.
Zoom Image
Scheme 2 Pd-catalyzed α-arylation of carbonyl compounds with aryl triflates: A) tert-butyl esters, B) lactones, C) hexanones and tetralones.
Zoom Image
Scheme 3 Cu-catalyzed enantioselective α-arylation of imides and aldehydes with diaryliodonium salts: (A) substrate activation via silyl ketene imide or (B) via enamine catalysis.
Zoom Image
Scheme 4 Enantioselective α-arylation of silyl enol ethers with Cu(I)–bis(phosphine) dioxides: selected examples of ligands screening. Xyl = 3,5-Me2C6H3; DMM = 3,5-Me2-4-MeOC6H2; BTMF = 3,5-(CF3)2C6H3; DTBM = 3,5-tBu2-4-MeOC6H2.
Zoom Image
Figure 1 Single-parameter correlation of the measured enantioselectivity with the phosphine oxide P–O bond stretching frequency ν POas. Data points of the biased substrates that have been removed from the initial correlation are purple triangles. Predicted data points are reported as red squares.
Zoom Image
Scheme 5 Reaction scope
Zoom Image
Scheme 6 Arylation of silyl enol ethers of dialkyl ketones and methyl propionate
Zoom Image
Scheme 7 Preliminary mechanistic considerations