Key words organic synthesis - electrochemistry - fluorination - C–H functionalization - radicals
Within the realm of synthetic organic electrochemistry,[1 ] few applicable options exist for the C–H functionalization of unactivated aliphatic
centers. In 2016 our lab reported a simple and inexpensive method for the oxidation
of allylic C–H bonds featuring N -hydroxytetrachlorophthalimide as a mediator for hydrogen atom transfer.[2 ] Shortly thereafter, it was found that the use of quinuclidine as mediator allowed
for the oxidation of stronger C–H bonds such as unactivated methylenes.[3 ] Based on requests from industrial collaborators in medicinal chemistry, we were
compelled to extend this precedent to the problem of C–H fluorination. Although both
photochemical[4 ] and purely chemical means[5 ] exist for accomplishing such a transformation (Scheme [1 ]A), an electrochemical alternative was pursued to determine if there was any specific
advantage in terms of scalability and/or selectivity. Disclosed herein is a practical
and scalable approach to C(sp3 )–H fluorination that utilizes Selectfluor in a unique way when coupled to anodic
oxidation in the presence of a nitrate additive.
Scheme 1 (A) Electrochemical C(sp3 )–H fluorination. (B) Reaction development
A truncated optimization table is depicted in Scheme [1 ]B, wherein Selectfluor was chosen as a fluorine atom donor based on its wide availability.
From a reactivity standpoint, one could also envisage three distinct roles for Selectfluor:
as (1) its own electrolyte due to its ionic nature; (2) an electrophilic fluorine
source; and (3) itself a mediator similar to quinuclidine.[3 ] Using 1 as a model substrate, the impact of various reaction parameters was investigated
(Scheme [1 ]B). The fully optimized conditions called for the use of Selectfluor (3.0 equiv)
and sodium nitrate (0.2 equiv) in acetonitrile with a pair of reticulated vitreous
carbon (RVC) electrodes at 3 mA to deliver the desired fluorinated product 2 in 62% NMR yield (54% isolated, entry 1). Not surprisingly, the reaction was found
to be sensitive to oxygen, as is rationalized from the proposed radical chain mechanism
(vide infra , entry 2). The reaction was confirmed to be electrochemically driven and required
a constant supply of current (entry 3). Extensive screening revealed that the sodium
nitrate was essential for the initiation as well as for improving the reproducibility
of this reaction (entries 4 and 5). The peculiar use of nitrate has precedent in the
electrochemical literature; nitrate is known to be oxidized anodically to generate
reactive radical species capable of abstracting hydrogen from substrates.[6 ] The alternative fluorinating agent Selectfluor II (B ) was also evaluated and did not improve the yield (entry 6). The likely role of Selectfluor
as a mediator was supported by the fact that other electrophilic fluorinating sources
(some known to capture nucleophilic radicals) failed to effect this transformation
(entries 7 and 8).
The use of a Ni foam cathode instead of RVC had a deleterious effect upon this reaction
(entry 9, for a detailed summary of electrodes screened, see the Supporting Information).
Finally, increasing the current to 6 mA also resulted in lower yield (entry 10). In
this case, the reaction resulted in incomplete conversion of the starting material
with unidentified byproducts, likely due to the higher voltage.
With an optimized set of conditions in hand, the utility of electrochemical C–H fluorination
was explored. The generality of this transformation is shown in Scheme [2 ]
[7 ] demonstrating efficient C–H fluorination on various classes of molecules including
terpenes, amino acids and pharmaceutically relevant structures. Comparisons with chemical
conditions available in the literature are also included in the Scheme [2 ]. In general, fluorination occurs at the unactivated secondary or tertiary C–H bonds
that are the most distal from the electron-withdrawing group. This regioselectivity
tracks with innate reactivity towards an electrophilic oxidant[8 ] and is reminiscent of the selectivity of electrochemical unactivated C–H oxidation,[3 ] which strongly suggests that the C–H abstraction step proceeds via a mediated radical
mechanism. Within the realm of secondary systems (Scheme [2 ]A), simple acyclic and cyclic alkanes were fluorinated, including those bearing esters
(3 and 4 ), ketones (5 and 6 ), and even alkyl bromides (8 ). In the case of sclareolide, the reaction gave a mixture of the corresponding fluorinated
compounds 9 in 58% yield. Notably, even in the absence of sodium nitrate, 9 was obtained in 67% yield. Fluorination of tertiary systems generally proceeded in
higher yield (Scheme [2 ]B). Thus, acyclic, amino acid derivatives, and adamantanes were all fluorinated in
synthetically useful yields. Fluorination of unsubstituted adamantanes generally led
to a mixture of di-/tri-fluorinated products in reasonable yields, whereas high yields
of mono-fluorination were observed when only one tertiary C–H bond was available (19 and 20 ). Access to fluorinated amino acids (12 , 13 , 17 , and 18 ) is a promising application with known uses in drug discovery contexts.[9 ] This methodology was also field-tested at Eisai where numerous building blocks were
subjected to electrochemical fluorination and synthetically useful yields of valuable
fluorinated products emerged (24 , 25 , and 26 in Scheme [2 ]D). In order to demonstrate the simplicity with which this chemistry could be conducted
on scale, a 100-gram scale fluorination of protected l -valine 27 provided the corresponding fluorinated adduct 12 in 78% yield (Scheme [2 ]E) without significant erosion of enantiopurity (96% ee , see the Supporting Information). This was accomplished using a simple batch reactor
(see the Supporting Information for details), but in principle could also be easily
adopted to a flow setup.
Regarding the limitations of this method, several substrates delivered mixtures of
isomers, showed no reaction, or decomposed under standard conditions (Scheme [2 ]C). For example, C–H bonds adjacent to a heteroatom are prone to oxidation, resulting
in degradation. No product formation but partial decomposition was observed in the
case of strained systems such as the bicyclo[2.1.0]pentane framework 22 after passing a total charge of 1.0 F/mol at 1.5 mA under constant current conditions.
Certain nucleophilic groups were not tolerated under the current conditions. Regardless
of these limitations, the operational simplicity, reproducibility, and short reaction
times are useful attributes of the present method. To further simplify the reaction
conditions, it is worth noting that in the cases of sclareolide, adamantane, and the
protected valine sodium nitrate was not required for efficient initiation of the reaction.
A proposed mechanism for electrochemical C–H fluorination is described in Scheme [3 ]. Considering the fact that this reaction is a net redox-neutral transformation and
that the regioselectivity is analogous to that of other radical-based C–H functionalizations,
a radical chain mechanism is proposed. Initially, a small amount of carbon radical
is generated by either direct or nitrate-mediated electrochemical oxidation of a substrate.
Subsequent fluorination by Selectfluor delivers radical cation 28 , which can then abstract hydrogen from the substrate. Oxidative initiation was confirmed
by conducting a reaction in a divided cell (without nitrate), where reaction progress
was observed only in the anodic chamber. Evidence for a radical chain reaction includes
the observation of >100% current efficiency in the reaction shown in Scheme [4 ]. The necessity of constant electricity supply for this radical chain process is
not unusual because 28 can undergo nonproductive reaction pathways such as fragmentation and hydrogen abstraction
from solvent or decomposed material.
Scheme 2 Electrochemical fluorination: Scope and applications
Scheme 3 Proposed mechanism
We assume that the role of nitrate is to improve the efficiency of initiation, rather
than as a chain carrier. This is supported by the fact that C–H fluorination of several
substrates do not require a nitrate additive (vide supra ), which clearly indicates that nitrate radical is not crucial for the reaction progress.
In contrast, DABCO-containing fluorine sources are essential for successful reactions
(see Scheme [1 ]). Regarding the capability of nitrate as initiator, it is known that nitrate can
be electrochemically oxidized.[10 ] Cyclic voltammetry indicated that oxidation of nitrate indeed occurred at +2.2 V
in acetonitrile with respect to Ag/AgCl reference electrode (see the Supporting Information).
Scheme 4 Evidence for a radical chain mechanism: 0.25 equiv of electrons delivers >50% yield
Thus, nitrate oxidation likely occurs before any direct anodic C–H abstraction.[11 ] Taken together, it appears that (1) the nitrate additive is considered to be helpful
in the initiation step; (2) Selectfluor-derived species (such as 28 ) could participate in the propagation step as a chain carrier, though the efficiency
of the propagation strongly depends on the substrate used. The triple utility of Selectfluor
as a fluorine donor, mediator, and electrolyte is a rather memorable aspect of this
chemistry.
Finally, as the mechanisms for electrochemical and photochemical C–H fluorinations
are radical based, a small study was pursued to see if reactivity differences could
be observed with different medicinally relevant scaffolds. As illustrated in Figure
[1 ], substrates 24 , 25 , and 29 reacted quite differently and in all cases showed superior reactivity through the
electrochemical mode. The origin of these differences is not clear but points to the
need for complementary new approaches to achieve this valuable transformation.
Figure 1 Comparison of electrochemical and photochemical C–H fluorination using actual med-chem
intermediates
In summary, a simple and scalable protocol for the electrochemical fluorination of
unactivated C(sp3 )–H bonds has been developed. The scope has been explored across a range of substrates
bearing numerous types of functional groups and the ease of scale-up is evidenced
by the 100-gram scale fluorination of a valine derivative. As electrochemical functionalization
processes become more mainstream, it is likely that this method will find use alongside
analogous C–H oxidation processes for both building-block diversification and metabolic
prediction.
