1
Introduction
Over the last decade, photoredox catalysis has become a powerful method in modern
synthetic organic chemistry. Since 2018, between 900 and 1,000 publications appeared
each year on the topic ‘photoredox catalysis’. Light, preferably in the UV-A or visible
range, provides enough energy to overcome the activation barriers of reactions that
cannot be passed under thermal conditions by employing excited state reactivity.[1] Photoredox catalysis complements the current synthetic methodology of so far unknown
transformations and allows the limits of available methods to be overcome.[2]
[3] Thus, photoredox catalysis is an important addition to the repertoire of transformations
involved in the total synthesis of natural products[4] and late-stage functionalization of pharmaceutically active compounds.[5] Currently, the majority of methods developed use transition-metal catalysts, mainly
based on ruthenium and iridium, due to their advantageous photophysical properties
and their (photo)chemical robustness.[6] The concept of photoredox catalysis implies a certain degree of inherent sustainability
using sunlight or energy-saving LEDs. Furthermore, generating redox active species
in low concentrations and under spatiotemporal control allows the reactivity to be
precisely controlled by suppression of competing side reactions. In order to enhance
sustainability by combining photodriven reactions with organic (in the sense of non-metalated)
photoredox catalysts, dyes have become important alternatives to metal catalysts,
for instance, eosin y,[7] rhodamine 6G,[8] mesityl-[9] and aminoacridinium,[10] naphthochromenones,[11] phenones[12] and 4,6-dicyanobenzenes.[13] It is also worth mentioning here that both photoredox catalytic as well as stochiometric
reactions can be performed even in the absence of any dye by exciting in situ formed donor–acceptor complexes.[3] Organic photoredox catalysts span a broad variety of molecular scaffolds that can
easily be tuned by core modifications, a crucial prerequisite to adjust a particular
catalyst system to be most effective in a distinct transformation.[14]
While divergent photochemical synthesis has become an important tool in modern photocatalysis,[15] the photoactivation of notoriously inert sulfur hexafluoride (SF6) and switching between its fluorination and pentafluorosulfanylation reactivity illustrate
recent challenges in photochemistry. In contrast to its smaller fluorinated analogues
bearing a dipole moment along the elongated C–X bond, namely a simple fluorine substituent
and the CF3 group, the late-stage introduction of the SF5 group is still a major challenge and its chemistry is highly underdeveloped.[16]
[17] While a variety of nucleophilic and electrophilic fluorination or trifluoromethylation
reagents today allow the routine introduction of fluorine in standard organic chemical
laboratories under conventional inert conditions, the introduction of the SF5 group is still restricted to the use of highly reactive and highly toxic gaseous
mixed sulfur fluorides, requiring special equipment and allowances to handle these
reagents, depending on the location of the laboratories.[18] Interestingly, the SF5 group represents a special substituent out of a collection of chemically stable S(VI)-based
functional groups, the chemistry of which has only been scarcely explored to date.
Sulfoximines, sulfonediimines, sulfurimidoylfluorides and sulfonimidamides have been
seriously neglected for quite some time in drug discovery programs. Only recently
have these ‘forgotten’ S(IV) motifs experienced a tremendous increase in research
interest, having been shown to uniquely contribute to modern medicinal chemistry,
whilst offering novel modes of catalysis or being applicable in inverse drug discovery
approaches.[19]
[20]
2
Sulfur Hexafluoride (SF6)
In contrast to the high reactivity of sulfur fluorides in lower oxidation states or
partly fluorinated sulfur fluorides, sulfur hexafluoride (SF6) is a highly inert gas (bp –63.9 °C). It is non-flammable, odorless, colorless, tasteless
and non-toxic.[21] Its inertness towards almost any chemical agent is mainly due to its fully symmetric
octahedral fluorine shield that causes the lack of availability of a low-lying unoccupied
orbital at the S(VI) center or the fluorine atoms for interaction with nucleophiles.[22] Furthermore, it is the strongest greenhouse gas known to mankind today; it displays
a 22,800- to 23,500-fold higher greenhouse potential than carbon dioxide and has a
mean lifetime in the atmosphere of about 3,200 years.[23] However, SF6 remains indispensable in many applications, especially in the context of high-voltage
switchgears, plasma-etching in semiconductor manufacturing and metallurgy, and needs
to be destroyed after use.[24] Due to its widespread technical applications, SF6 is produced on large scale (~10,000 tons/a), is quite cheap (44–54 €/kg) and easily
available.[19] In contrast to lower sulfur fluorides such as SF4 (including SF3-NR2 and F2S=N), SF5Cl and SF5Br, which have found widespread use as pentafluorosulfanylation and fluorination reagents,[16]
[25] SF6 was virtually discounted as a reagent in organic synthesis due to its intrinsic inertness;
its ability to serve as a pentafluorosulfanylation agent[26] for routine applications has even been excluded.[17]
3
The Pentafluorosulfanyl (SF5) Group
The incorporation of fluorinated substituents into organic molecules significantly
affects their physical, chemical, biological and pharmaceutical properties. Fluorinated
compounds not only play an important role in pharmaceutical chemistry,[27]
[28] but also in agrochemistry,[29] dye chemistry,[30] and materials chemistry including optoelectronics.[28]
[31] Fluorination of a lead structure often increases metabolic stability and bioavailability
and reduces the pK
a of acidic groups in the surroundings.[32] The most routinely used fluorinated motif is the trifluoromethyl (CF3) group.[33] Each individual fluorinated motif has a unique set of properties, including metabolic
stability, steric demand, acidity, lipophilicity and polarity, which need to be matched
with the requirements of the target and the desired mechanism of action during the
optimization process to reach an optimized binding situation with the target and optimal
pharmacological properties. Hence, it is an important task to search for more effective
and stable fluorinated groups. Among the arsenal of commercially available fluorinated
motifs like SCF3, OCF3 and CF3, the pentafluorosulfanyl (SF5) group is the most underexplored, often designated a ‘forgotten functional group’.
This arsenal is complemented by some more exotic emerging fluorinated motifs for which
there is a lack of almost any synthetic accessibility today, for example, SF4-bridged motifs, -NRCF3 or -OSF5.[34]
SF5 compounds behave as promising analogues of CF3 motifs in drugs and other functional organic compounds like agrochemicals and liquid
crystals, which connect high lipophilicity with low rotational barriers and steric
bulk. Highly beneficial properties have been proposed for the SF5 group in pharmaceutically active drugs,[35]
[36] functional materials,[37,38] metal complexes,[38,39] biologically active compounds,[40] or for 19F MRI by improving the signal-to-noise ratio (SNR) in combination with an ultrashort
echo-time (UTE).[41] For instance, trifluoralin has a 5-fold enhanced activity against quackgrass and
crabgrass if the CF3 group is replaced by SF5, whilst fenfluramine (an appetite suppressant) shows 10-fold strong binding to the
receptor.[40a]
[42]
[43] Moreover, the SF5 group is both thermally and widely chemically stable and not prone to hydrolysis
under physiological conditions. After initial metabolic processing, it has recently
been shown for some compounds that the SF5 substituent can finally be metabolized with formation of fluoride anions.[42]
[44] The SF5 substituent on phenyl rings shows a group electronegativity of 3.65 in contrast to
3.35 for CF3, as well as a Hammett parameter of σp = 0.68, between CF3 (σp = 0.54) and NO2 (σp = 0.78) (Figure [1]), and is highly lipophilic (Hansch parameter π = 1.23, between SCF3: π = 1.44 and OCF3: π = 1.04).[45] The SF5 group has been discussed as a bioisosteric replacement not only for the CF3 group but also for tBu, NO2 and halogen substituents.[35] The first organic pentafluorosulfanyl compound was described by Cady in 1950 who
prepared SF5CF3 by excessive fluorination of CS2.[46]
Figure 1 Steric size of the SF5 group in comparison to the CF3 group (top) and the Hammett/Hansch parameters (π and σp)[47] for a variety of fluorinated substituents and the NO2 group in comparison (bottom)
Triggered by the pronounced interest in the SF5 group during the last decade, major progress has been made in developing new protocols
to access SF5-containing small molecules.[26]
[48] In principle, two general strategies can be distinguished that rely either on the
formation of (i) S–F bonds (oxidative fluorination), or (ii) the C–S bond (direct
pentafluorosulfanylation). The first approach has contributed majorly to a routine
access to SF5-substituted arenes to be employed as building blocks in early synthetic steps or
divergent synthetic routes. In 1997, Ou et al. reported a chloride-supported, XeF2-based strategy to access ArSF3, ArSF4Cl and ArSF5 compounds.[49] Mechanistic studies strongly point to the relevance of the presence of chloride
as a halogen donor during an anion-radical transition during the course of the reaction.
A major breakthrough finally enabling the large-scale preparation of pentafluorosulfanylated
arenes was reported by Umemoto in 2012 by employing Cl2 as the oxidizing agent in the presence of KF to access SF4Cl-substituted arenes that could finally be converted into the SF5 compounds by treatment with HF, ZnF2 or AgF.[50] Ultimately, in 2018, Togni reported a modified protocol enabling a gas-free synthesis
of SF5 arenes by employing trichloroisocyanuric acid as the oxidizing agent.[51] However, these synthetically highly valuable approaches do not transfer the desired
functional group, but rather require prefunctionalized disulfides or thiophenols as
starting materials. Limitations are posed by the rather aggressive reaction conditions,
restricting the application of these methods mainly to the introduction of the SF5 group in early synthetic steps. Arylphosphorothiolates have been demonstrated as
convergent substrates for Ar-SF4Cl and Ar-SF5 synthesis.[52] Furthermore, today this strategy cannot be applied to aliphatic substrates.
The above-mentioned approach of forming the C–S bond is currently limited to aliphatic
substrates, and is by far more underdeveloped than the oxidative fluorination methodology.
It fully depends on the generation of the SF5 radical by application of the mixed sulfur fluorides SF5Cl, SF5Br and S2F10. Alternative pathways that employ the corresponding charged SF5 species suffer from the low nucleophilicity and lability of the SF5 anion or the very high energy of the SF5 cation. However, the extraordinary toxicity of the available pentafluorosulfanylation
reagents excludes their use in standard research laboratories and renders the broad
industrial use of these methodologies nearly impossible.[16]
[17] Taken together, the exploration and use of the SF5 group in organic compounds is still very limited because of a lack of synthetic accessibility.
However, ramping up from 2016, several reports have described the photochemical activation
of sulfur hexafluoride, finally harnessing it either as a fluorination or a pentafluorosulfanylation
reagent. In particular, the recent progress in pentafluorosulfanylation chemistry
will enable the significant future potential of this long time written-off molecule
in modern organofluorine chemistry.
4
Photoredox Catalytic Activation of SF6
The inertness of SF6 poses a significant challenge to chemists attempting to harness it as a reagent in
synthesis. Early examples of SF6 activation involve extreme reaction conditions employing very high temperatures,[53] or UV irradiation at wavelengths of <190 nm.[54] Modern SF6 chemistry in contrast allows the molecule to be activated under much milder reaction
conditions. These methods (Scheme [1]) comprise the fluorination of low-valent transition-metal complexes developed by
Ernst et al.[55] or Pt catalysts allowing deoxyfluorination reactions described by Braun et al.[56] Dielmann and co-workers reported the nucleophilic activation of SF6 by superbasic phosphines either resulting in complete degradation to phosphine sulfides
and difluorophosphoranes or conversion into a bench-stable SF5 anion.[57]
Rueping and co-workers used bipyridine compounds as two-electron donors for the metal-free
activation of SF6. The formed SF5 anion dissociates into SF4 and allows deoxyfluorinations of benzylic alcohols, aldehydes and carboxylic acids.[58] The activation of SF6 by non-coordinated phenolate anions was reported in 2021 by Hoge and co-workers,
resulting in the formation of phosphazenium pentafluorosulfanylide salts.[59] SF6 can be completely deconstructed by electrochemical reduction (Magnier et al.),[60] or by an aluminum(I) compound according to Crimmins et al.[61] (Scheme [1]).
Scheme 1 Overview of methods for the thermal/electrochemical activation of SF6
While these thermal or electrochemical methods contribute significantly to the efficient
destruction of SF6, such approaches suffer from an inherent disadvantage. One strategy of activation
is the reduction of the high barrier of activation by increasing the thermodynamic
driving force of the reactions. However, these highly exergonic reactions tend to
harm the selectivity and suffer from the alternating bond dissociation enthalpies
of SF6 and its partly defluorinated reaction products.[62] Thermal reaction approaches therefore tend to end up either in complex reaction
mixtures or a thermodynamic sink generating either strong M–F bonds or inorganic fluoride
as reaction products. These products cannot easily be applied in downstream fluorination
reactions, and thus limit the versatility of these approaches. In general, photochemistry
allows this problem to be circumvented by permitting a significant part of the reaction
to progress on the potential energy surface of a particular excited state and only
ultimately to cross-over to the ground state potential energy surface of the product.
This product does not necessarily need to be the thermodynamic product of the ground
state reaction, nor is it controlled by thermal reaction barriers connecting potential
energy surfaces.
The reduction potential of SF6 was determined to be –2.17 V vs Fc+/Fc (Magnier et al.),[60] being –1.8 V vs SCE, and –1.9 V vs SCE (Nargony et al.).[63] In the case of SF6, one-electron reduction forms the radical anion SF6
•–, which can exist in at least four negative ion states. At least two of them can serve
as synthetically relevant fragmentation pathways (Scheme [2]) to form reactive species showing fundamentally different chemical reactivity.[64] The ground state of the radical anion SF6
•– was proposed best to be described as an ‘association complex (SF5
•F)–’, which is kept together by a very weak interaction of 1.35 ± 0.1 eV (~130 kJ/mol).[65] The first excited state leads to decomposition into the fluorine radical F• and the anion SF5
– (pathway I), a formal Lewis acid–base adduct of SF4 and the fluoride anion, which parallels exactly the reactivity of SF4.[64] This state was successfully employed in fluorination- and deoxyfluorination-type
reactions. However, it cannot serve as a pentafluorosulfanylation pathway due to its
low stability and very weak nucleophilicity. To establish the ladder process for the
fragmentation of the radical anion SF6
•– into the pentafluorosulfanyl radical SF5
• and a fluoride anion (pathway II), the second excited state of the radical anion
SF6
•– needs to be populated.[64] These considerations are aligned with mass spectrometric analysis as well as high-level
theoretical calculations that have shown that different decomposition pathways of
the radical anion SF6
•– (in the gas phase) depend on the kinetic energy of the transferred electron for the
preceding reduction from SF6.[66] This also suggests internal conversion to effectively compete with the rates of
dissociative relaxation of the initially populated negative ion state. Below approximately
2 eV excess electron energy, the radical anion SF6
•– decomposes into the fluorine radical F• and the anion SF5
–.[67]
[68] In order to fragment into the radical SF5
• this energy should be higher than a threshold of approximately 2 eV. Accordingly,
Beier et al. showed that the one-electron reduction of SF6 by TEMPOLi yielded aliphatic SF5 compounds, and a mechanism employing the SF5 radical was discussed.[69] These considerations are fully aligned with conventional pentafluorosulfanylation
protocols relying on the use of SF5Cl, SF5Br or S2F10, being the only identified sources of the SF5 radical.[16]
[26]
[70]
Scheme 2 The decomposition pathways of the SF6
•– radical anion depend on the kinetic energy of the electron for the preceding reduction
from SF6.[41] Below approximately 2 eV, SF6
•– decomposes into the fluorine radical F•, which gives access to fluorinations. Above 2 eV, SF6
•– decomposes into the radical SF5
•, which allows pentafluorosulfanylations.
Based on these physical–chemical studies, a suitable photoredox catalyst for the activation
of SF6 by one-electron reduction firstly should be able to generate the SF6 radical anion as a primary condition. To switch between its modes of reactivity,
namely fluorination or pentafluorosulfanylation, the data suggests the requirement
to gain control over the target negative ion state to induce the desired fragmentation
of SF6
•–. To realize fragmentation to a SF5 radical, higher reduction potentials (resulting in a higher contribution to electron
excess energy) are suggested to be required compared to induce the formation of the
SF5 anion for fluorination purposes. This idea will be elaborated in the following parts
of this short review.
Scheme 3 Photoredox catalytic activation of SF6 by Ir(ppy)2(dtbppy)PF6 for the deoxyfluorination of allylic alcohols, e.g., 1 to allylic fluorides 2 and 3 (top), the proposed photoredox catalytic cycle (middle) and examples of the product
scope 4–12 (bottom)
In 2016, Jamison and co-workers developed the first method to activate SF6 by photoredox catalysis employing the widely used and commercially available catalyst
Ir(ppy)2(dtbppy)PF6 (excitation at 470 nm). This approach employs an oxidative quenching cycle that relies
on a back-electron transfer by (iPr)2NEt as a sacrificial reductant (Scheme [3]).[70] The oxidation potential of the Ir(III) center is Eox(IrIII/IrII) = –1.61 V (vs SCE).[71] The fragmentation of the formed radical anion SF6
•– yields reactive fluoride radicals that under the strongly reducing reaction conditions
most likely will be instantaneously reduced to fluoride and SF6
– anions to undergo deoxyfluorination-type reactions.[58] The suggested mechanism comprises activation of the alcohol by O–S bond formation
to give a putative R–O–SFx intermediate. Interestingly, the observed retention of the configuration when trans- and cis-(–)-carveol are subjected to the reaction conditions hints at an SNi-type reaction mechanism, as has been observed in the case of deoxychlorination reactions
employing SOCl2.[72] The reaction tolerates a variety of functional groups, including the acid-labile
Boc protecting group (8), vinylic sites (9) as well as aldehyde functions (10). The application of continuous-flow reactors gave products 4, 6 and 12 on gram scale.
Later, Nagorny and co-workers applied a similar photocatalytic strategy to access
glycosyl fluorides using SF6 (Scheme [4]) that more recently has also been translated into an electrochemical approach.[63]
[73] Here, 4,4′-dimethoxybenzophenone (13) served as an organic photoredox catalyst together with an amine as a sacrificial
donor. This ketone was chosen as a photoredox catalyst due to its increased extinction
coefficient at 365 nm, the long lifetime of its triplet state and its appropriate
redox potential of Ered(13/13
•–) = –2.20 V (vs SCE). It was hypothesized that the strong reducing agent causes SF6 to mainly fragment into the radical SF5
•. It was further postulated that this radical undergoes a reaction with the previously
oxidized aminyl radical cation to give SF4, which was suggested to be the active fluorinating agent. The reaction conditions
tolerate different protecting groups on the carbohydrate (Bn, 15 and 17; Ac, 16; benzylidene, 18; PMB, 19) and is applicable for a broad variety of monosaccharides (e.g., 20 and 21). Even disaccharides were converted into the corresponding fluorides. Remarkably,
flow chemistry allowed glycosyl fluoride 15 to be produced on gram scale.
Scheme 4 Photoredox catalytic activation of SF6 using ketone 13 for the deoxyfluorination of glycosides, e.g., 14, to glycosyl fluoride 15 (top), the proposed photoredox catalytic cycle (middle) and examples of the product
scope 16–21 (bottom)
Kemnitz and co-workers followed the idea of nucleophilic activation of SF6 by envisioning N-heterocyclic carbenes being suitable nucleophiles. However, the
incubation of various NHCs, e.g., 22, with SF6 showed only very weak activation under thermal reaction conditions (Scheme [5]). Exploiting the excited state properties of NHC* under irradiation at 311 nm improved
the efficacy of the desired activation remarkably.[74] The excited state reduction potential of NHC 22 has been determined as Ered(22*/22
•–) = –2.2 V (vs SCE), being sufficient to photoreduce SF6 and induce a fragmentation of the radical anion SF6
•– into the SF5
• radical. After a second single-electron transfer, the SF5
– anion is formed which initiates the formation of SF4 as a fluorinating agent able to deoxyfluorinate a variety of substrates including
1-octanol (23), allylic alcohol 24 and benzoic acid (25) to the corresponding products 26–28. Recently, Huang et al. provided more detailed insights into the mechanism of the
nucleophilic activation of SF6 by NHCs through in silico experiments. This work proved comparably high barriers for the nucleophilic activation
of SF6 (43.4 and 33 kcal/mol) by a variety of NHCs. Furthermore, a linear correlation between
the Gibbs free energies of activation and the HOMO energies of the NHC were observed.
Following the same path, they predicted the thermodynamic and kinetic feasibility
of the nucleophilic activation of SF6 by NHCs.[75] Due to the requirement of stoichiometric amounts of carbene 22 that are converted into the difluorinated urea derivative, this method does not represent
a photoredox catalytic approach.
Scheme 5 Photochemical activation of SF6 by N-heterocyclic carbene 22 for the deoxyfluorination of substrates 23–25 to give the products 26–28
The hitherto discussed literature starting from 2016 represent groundbreaking and
pioneering work in overcoming the inertness of SF6 for fluorination reactions. However, the work covered so far only represents one
aspect of the Janus-faced reactivity of the SF6
•– radical anion. In 2015, we became interested in the so far veiled backward oriented face of SF6, i.e., its ability to undergo pentafluorosulfanylation reactions. For such a process
neither thermal nor photochemical activation conditions had been reported previously.
In this context it is noteworthy that the SF5 radical, to a certain extent, resembles the F radical in its kinetic and its thermodynamic
properties comprising its global electrophilicity index (ω = 3.7 eV vs 3.94 eV for
F•), as well as its electron affinity.[76]
[77] The electron affinity of the SF5 radical has been determined to be 3.8 ± 0.15 eV, which even exceeds that of a fluoride
radical (3.42 eV) by about 0.4 eV.[64,73,78] Stabilization of the radical in solution in the presence of bulk reducing agents
is therefore highly unlikely.
In 2002, Kirsch et al. filed a patent on the reaction of SF6 with tetrakisdimethylamino(ethylene) (TDAE) (30) forming the mixed bisamidinium fluoride pentafluorosulfanylide 31. The efficacy of the reaction could be enhanced by irradiating the reaction mixture
with visible or UV light. Furthermore, the reagent was described as being a useful
fluorinating and pentafluorosulfanylation reagent; however, only a somewhat general
procedure was described.[79]
Recently, the group of Tlili resumed the exploration of this system, studying the
metal-free activation of SF6 by 30 to form reagent 31 under irradiation with blue light in pentane.[80] The versatile utility of 31 was demonstrated by the development of deoxy- and dethiofluorination reactions by
subjecting CO2 or CS2 to the pentafluorosulfanylide species (Scheme [6]). The deoxyfluorination of the primarily formed carbamic or thiocarbamic acids grants
access to (thio)carbamoyl fluorides that can further be converted into precious N-trifluoromethylamines.[34d]
[e] SF4 was proposed to be the reactive intermediate for fluorination reactivity. Interestingly,
in the presence of trichloroisocyanuric acid (32), pentafluorosulfanylide 31 could be converted into SF5Cl in situ, which can add to alkenes or alkynes (Scheme [6]). Furthermore, the mechanism of activation has been investigated. TEMPO trapping
experiments revealed the transient occurrence of TDAE•+ as well as SF5 radicals. However, a simple single-electron reduction of SF6 by TDAE was ruled out based on electrochemical data; therefore the authors suggested
an unknown intermediate to render the single-electron reduction of SF6 thermodynamically possible.
Scheme 6 Photochemical and metal-free activation of SF6 into the SF5-based reagent 31 for both deoxyfluorinations and pentafluorosulfanylations
We envisioned to gain control over the mode of reactivity, namely the in situ formation of a highly oxidizing SF5 radical, that is frightened away by even weak electron donors by application of a photoredox catalytic approach. The
rationale behind this was manifold. Firstly, the desired electron-transfer process
could be tuned by excited state lifetimes, irradiation power, the emission spectrum
of the light source as well as the concentrations of the reactants. Secondly, the
amount of excess energy of the transferred electron needed to switch the mode of reaction
of SF6 should be able to be controlled by the energy difference between the acceptor state
and the corresponding excited state energy of the photoredox catalyst. Ultimately,
reducing the excited state density in the medium finally allowed destructive overreduction
to be avoided, which is another stumbling stone of SF6-based pentafluorosulfanylation chemistry as described above (see Scheme [1]). A net neutral photoredox catalytic cycle in the absence of any sacrificial reductant
could warrant for generating the SF5 radical in a ‘redox shelter’, opening up a timeframe for transfer to an organic substrate
before either colliding with another excited state of the catalyst or any reducing
reaction intermediate.
N-Phenylphenothiazines based on 33 are important photoredox catalysts because (i) they are synthetically well accessible,
(ii) their modular structure allows the introduction of electron-donating or electron-withdrawing
groups at the core or at the phenyl group to tune the optoelectronic properties, (iii)
they are strongly reducing photoredox catalysts, and (iv) they are photochemically
stable.[81]
N,N-Diisobutylaminophenyl-phenothiazine is currently the most strongly reducing catalyst
in this series that allows, for the first time, the photoredox catalytic alkoxylation
of alkyl olefins, as non-activated substrates, to give products with Markovnikov selectivity.
Such photocatalytic reductions do not require any additional reagent, tolerate other
functional groups, including allyl, alkynyl, cyanide and even acid-labile Boc groups
within the substrate scope, and allow exo-trig cyclizations.[82] Furthermore, N-phenylphenothiazines have been shown to form deeply colored stable radical cations,
e.g., 33
•+, which can be excited in the near-infrared (NIR) or visible region to access a variety
of strongly oxidizing doublet states Eox(33
•+
*/33) ≥ 2.1 V (vs SCE).[83] Its participation in chemical reactions was reported earlier by Moutet and Reverdy
in 1979.[84] This class of catalysts therefore has the general prerequisites to be employed in
oxidative ‘conPET’ (consecutive photoinduced electron transfer) processes spanning
a potential range of operation of ca. 5 V.
The first method to use SF6 as a pentafluorosulfanylation reagent to yield valuable SF5-containing organic compounds was therefore realized by a photoredox catalytic approach
that precisely activates SF6 using LED light at 365 nm and transfers the SF5 group onto the organic substrates (see Scheme [5]). In contrast to the mentioned photoredox catalytic activation of SF6 for deoxyfluorination, our approach precisely controls the local reductivity by N-phenylphenothiazine (33) as a strong photoredox catalyst with an excited state potential of Eox(33*/33
•+) = –2.5 V (vs SCE). It is able to transfer the SF5 group from SF6 to α-methyl- (34) and α-phenylstyrene (35) to yield compounds 36 and 37 (Scheme [7]).[77] Furthermore, the vicinal fluoride anion can be abstracted to give the pentafluorosulfanylated
vinyl and allyl compounds 38 and 39, respectively. Additionally, the low loading of photocatalyst 33 in these experiments prohibits a potential overreduction that would yield the unpreferred
SF5 anion. Surprisingly, initial mechanistic investigations hinted at a reaction mechanism
that is not based, as initially proposed, on a simple Giese-type addition to the styrene.[85] Instead, more detailed mechanistic studies revealed the participation of the radical
cation in a twofold excitation process, mirroring the anionic ‘conPET’ process reported
by König and co-workers.[8] Quenching of the excited state of 33 by SF6 generates the correct negative ion state of the SF6 radical anion, which fragments into the desired SF5 radical. Radical cation 33
•+ is not able to oxidize 34 or 35, because back electron transfer, which would be required to close the photoredox
cycle, is endergonic by about 100 kJ/mol according to electrochemical data and theoretical
analysis. However, re-excitation of the radical cation 33
•+ at 365 nm or 530 nm, allowing it to reach its highly oxidizing excited doublet states,
allows for a second photoelectron transfer and activates the substrates by formation
of their radical cations 34
•+ and 35
•+, respectively. This process is suggested to be critical in establishing an efficient
pentafluorosulfanylation protocol due to its dual function. Firstly, it closes the
photoredox catalytic cycle, and secondly it prepares the substrate by turning it into
a strongly electron-deficient open-shell state that cannot be oxidized during the
approach of the strongly oxidizing SF5 radical before reaching the C–S bond-forming transition state. Such a process was
detrimental to the reaction since it annihilates the reactive species by turning the
‘Janus-faced’ coin to its fluorination side. This mechanistic proposal goes along with the suggested
mechanism of fluorination by Selectfluor, including the formation of a radical cation and subsequent fluorine radical transfer.[86] The preproduct cations 36+
and 37+
, respectively, can be trapped by in situ generated anhydrous fluoride anions to products
36 and 37. Unfortunately, the substrate scope was limited to the styrenes 34 and 35.
Scheme 7 Photoredox catalytic activation of SF6 by N-phenylphenothiazine (33) provides enough excitation energy into SF6 to yield the pentafluorosulfanylated products 38/39, and the photoredox catalytic mechanism
The interceptability of critical intermediates that we investigated in follow-up work
corroborates the suggested reaction mechanism. Trapping of the intermediate cations
36+
and 37+
by alcohols as external nucleophiles not only allowed the scope of the reaction to
be significantly enlarged, but also gave more precious insights into the operating
reaction mechanism excluding a concerted addition of sulfur hexafluoride and installing
both the SF5 group and the vicinal fluoride substituent via a concerted reaction mode. In this
variation of the reaction, the competing nucleophilic attack by in situ generated fluoride anions could be reduced by addition of 10–20 mol% of BEt3. This almost completely suppressed the formation of the vicinal fluorides 38 and 39 by trapping available fluoride anions in the solution (Scheme [8]). While we chose MeOH as our model compound to generate products 40 and 41 in moderate to good yields, the method tolerated a broad variety of alcohols and
functional groups, including vinyl, allyl, ethynyl and cyanide (43–50) on the side chain.[87] Furthermore, the isolated products could be subjected to follow-up transformations
allowing Au(III)-catalyzed deoxyazidation to access the corresponding vinyl-, allyl-
or azidopentafluorosulfanyl compounds 40–50.
Scheme 8 Proposed photoredox catalytic mechanism to pentafluorosulfanylated products 38 and 39 by trapping with alcohols as external nucleophiles instead of the internal fluoride
anion, and examples of the product scope 40–50
Scheme 9 Trapping of the intermediate substrate radical cations 34
•+ and 35
•+ by alkynols and the formation of oxaheterocyclic products 51–54, 56 and 57, and the acyclic product 55
Further information regarding the proposed reaction mechanism could be acquired by
another advance of this SF6-based pentafluorosulfanylation protocol reported by our group in 2021. The discovery
of a divergent reaction outcome when bifunctional alkynols were subjected to the reaction
conditions led us to develop a domino-type reaction sequence to access the oxygen-containing
heterocycles 51–54, 56 and 57 in a single step, specifically tetrahydrofurans, tetrahydropyrans and oxepanes, starting
from substrates 34 and 35 (Scheme [9]).[88] This process is based on a cascade of C–O, C–C and terminal C–S bond-forming reactions
with formation of one stereocenter starting from a styrene, an alkynol and SF6 to generate a significantly high degree of complexity in a one-pot reaction. The
proposed mechanism is in full agreement with Baldwin’s cyclization rules and is supported
by thermodynamic considerations and in silico studies. The key step is a radical type 5-, 6- or 7-exo-dig cyclization. Our mechanistic studies suggest a competitive trapping of 34
•+ and 35
•+ by the SF5 radical or the in situ generated alkynol anion. The preformed equilibrium between the alkynol and the in situ formed anhydrous fluoride turned out to be critical to control the outcome of the
reaction based on theoretical calculations. An alternative mechanism comprising an
initial attack of the SF5 radical on the alkyne was ruled out by a control experiment employing an alkyne lacking
the hyxdroxy function and thermodynamic considerations. At high alkynol concentration
the enriched alkoxide outcompetes the direct attack of the SF5 radical forming the unusual terminal radical 51
•. This quickly cyclizes to the corresponding vinyl radical that finally traps the
SF5 radical. However, the corresponding eight-membered oxocane product is not formed
from substrate 34 due to significant ring strain; instead the ring-opened product 55 was formed in a comparable yield. With these heterocycles, the current SF5 product scope shows a high level of structural complexity. The yields are rather
low (20–32%), but it is important to keep in mind that these compounds cannot be synthesized
by any other methods. Furthermore, the formation of remote SF5-substituted reaction products 51–57 once more corroborates the existence of radical cations 34
•+ and 35
•+ as the key reaction intermediates of the reaction.
