Nickel-catalyzed trifluoromethylthiolation of Csp2–O bonds

A computationally guided development of the first efficient protocol to trifluoromethylthiolate aryl and vinyl Csp2–O bonds is presented, showcasing important reactivity requirements for the introduction of potentially reactive functional groups under homogeneous Ni-catalysis.


Introduction
Owing to nickel's non-precious nature and its higher reactivity in the rst elementary step of cross coupling cycles, i.e. the oxidative addition, the eld of homogeneous Ni-catalysis has long been considered promising, yet also challenging. 1 This is because difficulties have frequently been encountered in taming nickel's reactive nature to achieve desired selectivities and scope. 2 In spite of that, in recent years there has been impressive progress in the activation of the least reactive bonds, such as aromatic ethers or aryl uorides. 3 However, these milestones typically featured the conversion of C-OMe (or C-F 4 ) to inert C-C or C-H bonds. 5,6 By contrast, less is known about the reactivity limits and molecular requirements for the installation of potentially reactive functional groups. We therefore envisioned that a computationally assisted development 7 of an unprecedented Nicatalyzed protocol for C-heteroatom bond formation presents an ideal challenge to (i) identify the general reactivity requirements for efficient Ni-catalysis and (ii) demonstrate the viability of applying computational tools to assess substrate scope.
As a suitable test case, we focused on the nickel-catalyzed triuoromethylthiolation of Csp 2 -O bonds. 8 The SCF 3 group makes molecules more lipophilic, increasing their membrane permeability and bioavailability. 9 These properties are of considerable interest in a pharmaceutical and agrochemical context. Consequently, numerous efforts have been undertaken to synthesize aryltriuoromethyl suldes. 10,11 In particular the direct catalytic introduction of SCF 3 is an attractive approach. While aryl halides 12 or boronic acids 13 have successfully been converted to C-SCF 3 via metal catalyzed crosscoupling strategies or oxidative protocols, 14 to date, there is no report of a direct and catalytic triuoromethylthiolation of Csp 2 -O bonds.

Results and discussion
Given the widespread abundance of phenols, the tri-uoromethylthiolation of phenol derivatives would be highly attractive for synthetic diversity. In this context, the scope could in principle range from more activated derivatives (e.g. aryl tri-ates) to the least reactive derivatives, i.e. aryl ethers which are present in biomass feedstocks (such as lignin 15 ). 6 However, while Ni-catalysis has recently been successfully utilized to activate aromatic ethers, 3 we hypothesized that there might be a fundamental reactivity conict in introducing SCF 3 : the created SCF 3 -product would be expected to be inherently more reactive towards oxidative addition 16 which may impede further transformation.
To test this, we subjected Ni(cod) 2 /dppf to PhSCF 3 1 (see Fig. 1). We recently showed that this system triggers the mild triuoromethylthiolation of aryl chlorides, proceeding via Ni (0) / Ni (II) catalysis with [(dppf)Ni(cod)] formed as the active catalyst. 12e In accordance with our hypothesis, the reaction of the [Ni (0) ] catalyst with PhSCF 3 is indeed seen, even under mild reaction conditions (45 C), as judged by 31 P-NMR spectroscopic analysis. A complete disappearance of the characteristic 31 P-NMR singlet signal of [(dppf)Ni (0) (cod)] (33.8 ppm) 12e occurred, and the formation of a new species was seen that appears as two triplets at 30.8 ppm (with J ¼ 23.0 Hz) and at 22.1 ppm (with J ¼ 37.6 Hz) by 31 P-NMR spectroscopic analysis (see Fig. 1). While our efforts to structurally characterize the latter by X-ray crystallography have so far been unsuccessful, the formed species clearly constitutes a catalyst deactivation product. The subjection of this species as a catalyst (or also stoichiometrically) in the triuoromethylthiolation of aryl chlorides did not yield ArSCF 3 . This indicates that oxidative addition by a [Ni (0) ] catalyst to the product is facile and eventually leads to catalytically inactive species. To achieve productive catalysis and high overall conversion, it is therefore of utmost importance to avoid this deactivation process.
Our computational assessment 17 of the oxidative addition of [(dppf)Ni(cod)] to Ph-SCF 3 1 suggests an activation free energy barrier of DG ‡ ¼ 19.2 kcal mol À1 , and it uses the M06L method with a CPCM solvation model to account for toluene and the mixed 6-311++G(d,p) and LANL2DZ (for Ni, Fe) basis set. 17,18 This value now sets the bar for the possible reaction scope. The 'to-be-transformed' bond must show a barrier lower than 19.2 kcal mol À1 to avoid catalyst loss via an unproductive reaction with the product (ArSCF 3 ).

Identication of suitable leaving groupscomputational assessment & experimental tests
We subsequently undertook computational studies to identify matching leaving groups 'OR' (Fig. 2) that would show the desired greater reactivity than the Csp 2 -SCF 3 bond. For the cleavage of the C-O bonds, mechanistic support for Ni (0) /Ni (II)5i, 6 and also Ni (I) -catalysis 19 has previously been reported. However, on the basis of our previous mechanistic study 12e and the observation that (dppf)Ni (I) Cl is ineffective as a catalyst in C-SCF 3 bond formation, 12e,20 as a rst approximation, we calculated the activation barrier of oxidative addition using [(dppf) Ni (0) (cod)] to a range of phenol derivatives (Ph-OR), with R ¼ alkyl (ether), R 0 C]O (pivalate), SO 2 R 00 (sulfonic esters). Fig. 2 presents the results. This computational assessment suggests that in the context of C-O to C-SCF 3 conversion, the inherently high reactivity of C-SCF 3 only allows for triate precursors as suitable starting materials. Alternative C-O leaving groups that have previously been employed in the Ni-catalyzed construction of inert C-C bonds, such as aryl ethers (OMe), mesylates (OMs), tosylates (OTs) or pivalates (OPiv) 3,6 are predicted to be incompatible with Ni (0) -catalyzed triuoromethylthiolation, as they would generally be less reactive than Ar-SCF 3 , hence favoring catalyst deactivation via reaction with the product. 21 To experimentally test this computationally predicted trend, we subjected Ni(cod) 2 /dppf along with the easily accessible SCF 3 -source (Me 4 N)SCF 3 to Ar-OR derivatives (in toluene at 45 C), ranging from the predicted low (aryl ether) to high (aryl triate) reactivity (Fig. 2). In accordance with expectations, at best, a low conversion was seen for phenyl mesylates (5%), tosylates (1%) or pivalates (0%). In stark contrast, phenyl triate showed excellent conversion to PhSCF 3 (83%).
We additionally followed the conversion ArOTf / ArSCF 3 with ReactIR®. This analysis showed that the transformation was rapid, being essentially complete in 1.5 h with only little increase in conversion over the subsequent hours (see ESI, Fig. S1 †). We also analyzed the reactions of those substrates that showed little conversion (#5%), i.e. ArOMs and ArOTs, by 31 P-NMR spectroscopic analyses. We observed that essentially all of the [Ni (0) ] catalyst had transformed to the catalytically inactive species described in Fig. 1 within 3 h reaction time. This clearly highlights that while [Ni (0) ] is in fact capable of reacting with Ph-OMs or -OTs, the catalyst is rapidly consumed as soon as some of the more reactive PhSCF 3 molecules are generated. This corroborates with the strict requirement of suitably matching functionality and tailored reactivity progression from a "more" to "less reactive" functionality.

Computational assessment of functional group tolerance
We subsequently set out to test for the generality of the iden-tied Ni-catalyzed triuoromethylthiolation of activated C-O bonds and computationally assess the functional group (FG) tolerance (see Fig. 3). As we determined a barrier of DG ‡ ¼ 14.4 kcal mol À1 for the oxidative addition of [(dppf)Ni (0) (cod)] to Ph-OTf, all additional functional groups (FG) in the substrates will only be compatible if the reactivity of the C-FG bond is lower than that of Ph-OTf.
The computational results depicted in Fig. 3 suggest a tolerance of the protocol to ketone functional groups, C-C or benzylic C-O bonds. In all cases, the requirement of DG ‡ C-FG > 14.4 kcal mol À1 is fullled. Even aromatic C-CN bonds that were previously shown to be reactive under Ni-catalysis conditions 22 are predicted to be compatible.

SCF 3 -coupling of aryl triates
On the basis of this computationally guided substrate scope, we subjected a range of aryl triates to standard catalysis conditions. Table 1 presents the results. A number of aryl-and heteroaryl triates were coupled in good to excellent yields. The transformation was compatible with ketone (6, 7 and 8, Table  1), ether (9) and cyano (5) functional groups. Two heterocyclic examples (10,11) were also triuoromethylthiolated in good yields (see Table 1).
We next searched for bioactive molecules of greater complexity that would full our reactivity requirements and show compatibility with the computationally predicted scope. Estrone (an estrogenic hormone), 6-hydroxy avanone (a plant secondary metabolite used inter alia as an antioxidant) and dtocopherol (vitamin E) show an excellent functional group match, containing predominantly ketone and benzylic C-O bonds that are predicted to be less reactive than C-OTf and C-SCF 3 . Triuoromethylthiolation was successfully accomplished in 62-96% yield, highlighting the potential of this method for pharmaceutical applications (see Scheme 1).

SCF 3 -coupling of vinyl triates
Vinyl SCF 3 -compounds are also of signicance, nding applications as herbicides for example. 23 However, the current methodological repertoire to access these compounds relies predominantly on indirect strategies 24 or requiring stoichiometric amounts of metal. 13b, 25 The direct construction of C vinyl -   SCF 3 in a catalytic manner would be a highly attractive approach. It has been accomplished via the Cu-catalyzed tri-uoromethylthiolation of vinyl boronic acids with electrophilic SCF 3 -sources. 13c-e In a nucleophilic context, the catalytic installation of C vinyl -SCF 3 is limited to vinyl iodides and requires harsh reaction conditions (110 C). 26 A mild Ni-catalyzed conversion of readily accessible C vinyl -OR derivatives to C vinyl -SCF 3 would thus substantially widen the synthetic repertoire.
Our calculation of the barrier for the oxidative addition of [Ni (0) ] to C vinyl -SCF 3 indicated DG ‡ ¼ 18.8 kcal mol À1 . This barrier constitutes the upper limit for the reactivity of a potential leaving group (OR). C vinyl -OPiv and C vinyl -OMs show higher or similarly high barriers for oxidative addition (DG ‡ ¼ 22.1 and 17.7 kcal mol À1 ) and are hence ruled out. C vinyl -OTf on the other hand is predicted to be highly reactive (DG ‡ ¼ 5.2 kcal mol À1 ) and should hence be a compatible match.
Aer applying standard catalysis conditions, 27 we successfully transformed a number of vinyl triates to the corresponding triuoromethylthiolated counterparts (see Table 2).
The protocol proved to be compatible with a heterocyclic moiety (20, Table 2), a benzyl protecting group (17), and was successful for fully aliphatic (15) as well as conjugated (18,19) vinyl triate derivatives. Compound 19 ( Table 2) was afforded in a slightly lower yield (44%). However, upon closer inspection, it became clear that this was related to the inherent instability of the vinyl triate starting material.

Assessment of aryl and vinyl nonaates
We therefore shied our attention to potentially more stable analogues and considered nonaates. 28 Both, aryl and vinyl nonaates are computationally predicted to be compatible with Ni-catalyzed triuoromethylthiolation, showing similarly low or even lower barriers for oxidative addition by [Ni (0) ] than the corresponding triates (DG ‡ ¼ 4.8 for addition to C vinyl -ONf and DG ‡ ¼ 10.6 kcal mol À1 for addition to Ph-ONf). In accordance with these computational predictions, excellent conversions to aryl-and C vinyl -SCF 3 were observed (see Table 3). Particularly notable is the synthesis of 19 0 (Table 3) which was now highyielding (as opposed to its preparation in Table 2), reecting the greater robustness of vinyl nonaates over vinyl triates. 29

Conclusions
The inherently high reactivities of Ni-catalysts may be fundamentally at conict with introducing a wide range of functional groups, as shown here for the introduction of the pharmaceutically and agrochemically valuable SCF 3 group. We identied that the reaction of the Ni-catalyst with the desired product, ArSCF 3 , triggers undesirable catalyst deactivation reactions that ultimately inhibit catalysis. The overall substrate scope is therefore dictated by the reactivity of the desired functionality towards the catalyst (here: C-SCF 3 ). The application of computational tools allowed for the identication of matching   functional groups in terms of suitable leaving groups and tolerated functional groups. As a result, the rst Ni-catalyzed C-SCF 3 coupling of aryl and vinyl C-O bonds has been developed. Given the highly reactive nature of C-SCF 3 , only those C-OR derivatives of even greater reactivity, i.e. triates and nonaates, allow for efficient C-SCF 3 coupling. The protocol is mild, general and operationally simple.
Given that computational methods, soware and hardware have evolved to a level, at which calculations can nowadays frequently be done faster than experiments, 30 we anticipate that the herein applied approach will nd applications in the development of, but not limited to, homogeneous Ni-catalysis.