Development of N-F fluorinating agents and their fluorinations: Historical perspective

This review deals with the historical development of all N-F fluorinating agents developed so far. The unique properties of fluorine make fluorinated organic compounds attractive in many research areas and therefore fluorinating agents are important. N-F agents have proven useful by virtue of their easy handling. This reagent class includes many types of N-F compounds: perfluoro-N-fluoropiperidine, N-fluoro-2-pyridone, N-fluoro-N-alkylarenesulfonamides, N-fluoropyridinium salts and derivatives, N-fluoroquinuclidium salts, N-fluoro-trifluoromethanesulfonimide, N-fluoro-sultams, N-fluoro-benzothiazole dioxides, N-fluoro-lactams, N-fluoro-o-benzenedisulfonimide, N-fluoro-benzenesulfonimide, 1-alkyl-4-fluoro-1,4-diazoniabicyclo[2.2.2]octane salts, N-fluoropyridinium-2-sulfonate derivatives, 1-fluoro-4-hydroxy-1,4-diazoniabicyclo[2.2.2]octane salts, N-fluorodinitroimidazole, N-fluoro-trichloro-1,3,5-triazinium salt, N-F ethano-Tröger’s base derivatives, N-fluoro-methanesulfonimide, N-fluoro-N-arylarenesulfonamides, bisN-F salts such as N,N’-difluorobipyridinium salts and N,N’-difluoro-1,4-diazoniabicyclo[2.2.2]octane salts, and their many derivatives and analogs, including chiral N-F reagents such as optically active N-fluoro-sultam derivatives, N-fluoro-alkaloid derivatives, DABCO-based N-F derivatives, and N-F binaphthyldisulfonimides. The synthesis and reactions of these reagents are described chronologically and the review also discusses the relative fluorination power of each reagent and their mechanisms chronicling developments from a historical perspective.


Introduction
Fluorinated organic compounds occupy an important position in pharmaceuticals [1], agrochemicals [2], and materials [3]. Especially, in the first two areas, the presence of fluorine has attracted attention during the last decades. Nowadays, a considerable number of medicines [4,5] and agrochemicals [6] contain at least one fluorine atom in their structures. The fluorine atom has unique properties such as the highest electronegativity, extremely low polarization, strong C-F bonds, and the smallest size after a hydrogen atom [7]. Thus, introduction of fluorine into selective positions of a bioactive compound can produce remarkable changes in efficacy. Fluorine-scan/fluorine editing of a lead molecule is now a routine step in drug discovery [8]. Organofluorine compounds are very rare in nature [9] and therefore without natural compounds, chemical processes are required to generate building blocks. Molecular fluorine (F 2 ) is a useful fluorinating reagent, however, unlike Cl 2 and Br 2 , F 2 is extremely reactive, toxic, and corrosive and its handling requires specialist skills and equipment. Therefore, easy-tohandle and selective fluorinating agents are essential for the wide-spread advancement of organofluorine chemistry to nonspecialist chemists. Alternatives to F 2 , such as perchloryl fluoride (FClO 3 ) [10] and the O-F reagents such as CF 3 OF [11], CF 2 (OF) 2 [11], CsOSO 2 OF [12], CF 3 COOF [13], and CH 3 COOF [14] have been used as fluorinating agents for many years. However, these reagents have significant risks for safe handling. Although XeF 2 [15] was considered as a safer alternative, it is expensive because of the scarcity of Xe in nature. The appearance of the safe and easy-to-handle N-F fluorinating agents described in this review have brought about a breakthrough in synthetic fluorine chemistry enabling an increasing number of researchers to engage in organofluorine chemistry. Their development has significantly contributed towards the current 'golden age' of fluorine chemistry. The N-F fluorinating agents now stand out as particularly useful electrophilic or radical fluorinating agents by virtue of their easy handling, efficiency, and selectivity. These non-hygroscopic nature and stability make them easier to handle than nucleophilic fluoride reagents. Potassium fluoride (KF) and naked fluoride anion salts are extremely sensitive to moisture, while HF seriously attacks human skin. The N-F fluorinating agents can be classified into two categories: these are neutral and cationic. This review covers the chronological advancement of these reagents regardless of their classification, as they advanced side by side. The history of the N-F compounds acting as fluorine atomtransfer reagents can be traced back to 1964 when Banks and co-worker [16] first reported that perfluoro-N-fluoropiperidine (1-1) could fluorinate the sodium salt of 2-nitropropane to form 2-fluoro-2-nitropropane in a 40% yield (Scheme 1). The reaction with sodium diethyl malonate was also reported to produce the difluoromalonate, but in a very low yield of ca. 5%. N-F amine 1-1 is very volatile (bp 49.5 °C), and could only be prepared in 7.5% or 13% yield by electrochemical fluorination of pyridine or 2-fluoropyridine in anhydrous hydrogen fluoride [17,18] (Scheme 2). Not surprisingly, 1-1 did not become a popular reagent.
In 1986, Banks and co-worker reported the preparation of polymeric analogues of perfluoro-N-fluoropiperidine (1-1) [21] and then in 1991, Banks et al. reported the improved yields of the reactions of 1-1 with sodium salts of 2-nitropropane, malonate esters, and a keto ester, and phenylmagnesium bromide [22]. However, the fluorinated products were still accompanied by considerable amounts of byproducts resulting from the reaction of the substrates with 1-5.

1-2. N-Fluoroperfluorosuccinimide and N-fluoroperfluoroglutarimide
In 1981, Yagupols'kii and co-worker reported the synthesis of N-fluoroperfluorosuccinimide  and N-fluoroperfluoroglutarimide (2-2) by reaction of precursor imides with XeF 2 (Scheme 4) [23]. However, there were no reports on the fluorination capability of these N-F compounds. The purpose of this research was to establish if the presence of perfluoroacyl groups was sufficient to stabilize the Xe-N bond. Their experiment revealed that the intermediate F-Xe-N compounds were not detected, but N-fluoroimides 2-1 and 2-2 were formed. The stability of these N-F compounds was low, since they decomposed to H 2 NCO(CF 2 ) n COOH upon standing with air.
Notably, attempts to prepare N-fluorosuccinimide from succinimide, or one of its salts by reaction with fluorine (F 2 ), trifluoromethyl hypofluorite, or perchloryl fluoride in a variety of solvents, and at temperatures ranging from −78 °C to room temperature, all but failed, as reported in the margin of the paper cited in [24].

1-4. N-Fluoro-N-alkylarenesulfonamides
In 1984, a series of stable N-fluoro-N-alkylsulfonamides 4-1a-g was reported by Barnette [26]. The treatment of N-alkylsulfonamides with very dilute F 2 (1% or 5%) in N 2 at −78 °C afforded the fluorinated products 4-1a-g. As detailed in Figure 1, various kinds of N-fluoro-N-alkylsulfonamides were synthesized by this method. However, the yields were low except for the case of compound 4-1d. In the cases of secondary and tertiary alkyl groups on the amine side, low yields were obtained and these were attributed to concomitant N-S-bond cleavage reactions with F 2 . This method, using the dilute F 2 , was inefficient for their production due to long reaction times. N-Fluoro-N-alkyl-p-toluenesulfonamides 4-1b,c,f proved to be efficient fluorinating agents in the fluorination of carbanions (Scheme 7). The yields of reactions with sodium malonates and Grignard reagents were largely improved to up to 81% and 50%, respectively. The carbanions of aromatics, ketones, nitroalkanes, amides, etc. could also be reasonably well fluorinated and this study showed great progress. However, although these fluorinating agents were stable and easy-to-handle, their fluorinating power was low. They could fluorinate only reactive carbanions, but not aromatics, olefins, vinyl acetates, trimethylsilyl or alkyl enol ethers, and so on.
Soon after (1986), Schwartz and co-worker reported the stereospecific synthesis of alkenyl fluorides with N-fluoro-N-tertbutylbenzenesulfonamide (4-1h), a compound which is soluble at low temperature [27]. Alkenyllithium reagents, generated in situ, reacted at −120 °C with 4-1h in THF/Et 2 O/pentane to give the desired alkenyl fluorides 4-2 in good yields (Scheme 8). However, reactions that were run above −120 °C, or in pure ether or THF gave higher yields of the protonated products 4-3.

1-5. N-Fluoropyridinium salts and their derivatives
In 1986, Umemoto et al. reported N-fluoropyridinium triflate and its derivatives 5-4 as new stable cationic fluorinating agents. These possessed either electron-donating or -withdrawing substituents on the pyridinium nuclei, and were the first reactive, easy-to-handle fluorinating agents with wide application [28,29]. The work continued with additional disclosures until 1991 [30][31][32][33][34]. Before that, reactive fluorinating reagents were difficult to handle because of toxicity, a tendency to explode, instability, and/or hygroscopicity, while easy-tohandle reagents had limited application because of their low reactivity.
Umemoto et al. found that the hygroscopic pyridine·F 2 complex 5-2a decomposing vigorously at temperatures above −2 °C [35], which was formed by the fluorination of pyridine (10% F 2 /N 2 ) at low temperature in a freon solvent, could undergo straightforward counteranion replacement with a non-nucleophilic anion. Therefore, exchange with salts such as sodium triflate in acetonitrile generated non-hygroscopic N-fluoropyridinium triflate salts as highly thermally stable reagents as illustrated in Scheme 9 [28]. Moreover, they found that the fluorination power (reactivity) of these N-fluoropyridinium salts could be tuned by the substituents on the pyridinium nuclei.
The transformation of the unstable pyridine·F 2 complex to stable N-fluoropyridinium salts could be conducted by direct fluorination (10% F 2 /N 2 ) of the pyridine in acetonitrile and in the presence of a suitable salt at low temperature (method B in Scheme 10). This one-step process was successfully applied to many other pyridine derivatives (method B in Figure 2).
It was finally shown that the stable N-fluoropyridinium salts 5-4 could be synthesized in four different ways using 10% F 2 /N 2 [28,31,33] (Scheme 10): method A (stepwise method) involved the fluorination of a pyridine derivative with F 2 /N 2 , to form a pyridine·F 2 complex 5-2, followed by treatment with a nonnucleophilic anion salt, acid, or silyl derivative. Method B (onestep method) involved the fluorination of a pyridine derivative with F 2 /N 2 in the presence of a non-nucleophilic anion salt. Method C involved mixing a pyridine derivative with an acid or its silyl derivative, forming a salt 5-5, before fluorination. Method D involved mixing a pyridine derivative with a Lewis acid, forming complex 5-6, and then the fluorination with F 2 /N 2 was carried out.   Note: a this yield was the one by the improved method reported in [37]; b this improved method was reported in [38].
In total, sixty-two stable N-fluoropyridinium salts possessing different non-nucleophilic counteranions and electron-withdrawing or -donating groups were efficiently synthesized [33]. Figure 2 shows 31 examples and their methods of preparation. Umemoto and co-worker also reported the synthesis of a polymer version, poly(vinyl-N-fluoropyridinium salts) of these reagents [36].
The reactivities of many N-fluoropyridinium salts were examined [32] and mainly five kinds of N-fluoropyridinium salts, shown in Scheme 11, emerged as useful fluorinating agents due to their availability. The fluorination power greatly changed depending on the electron density at the nitrogen, an aspect con-trolled by the electronic nature of the substituents. The fluorinating power increased in the order of 2,4,6-triMe 5-4j < unsubstituted 4a < 3,5-diCl 4t < 2,6-diCl 4r < pentachloro 4v, in good agreement with the decreasing order of the pK a values of the pyridines. For example, in order to fluorinate phenol, triMe 5-4j needed heating at 100 °C in a haloalkane solvent for 24 h, whereas pentachloro 5-4v required only room temperature within 0.1 h for a successful reaction. The salt 5-4v was so powerful that it fluorinated an equimolar amount of benzene in dichloromethane in 2 h at 40 °C. In general, triflate salts were more effective than BF 4 salts because of the higher solubility of the triflate salts in a haloalkane solvent. As outlined in Scheme 12, fluorinations of many kinds of substrates with these N-fluoropyridinium salts were performed. It was shown that less reactive substrates can be fluorinated well with the more powerful reagents, and reactive substrates can be fluorinated with less powerful reagents, a match process which minimizes side reactions. Thus, these N-fluoropyridinium salts made possible the fluorination of a diversity of nucleophilic organic compounds with different reactivities ranging across aromatics, carbanions (Grignard reagents, enolate anions), active methylene compounds, olefins, silyl enol ethers, vinyl acetates, sulfides and so on, under mild conditions with high selectivity and yields [29][30][31][32]. All these reactions could be carried out routinely using standard glassware in normal laboratory environments and without any specialist training.
In 1991, N-fluoropyridinium pyridine heptafluorodiborate (NFPy), C 5 H 5 NF(C 5 H 5 N)B 2 F 7 , was introduced as a fluorinating agent [41]. NFPy was prepared by the reaction of fluorine with pyridine·BF 3 complex and its fluorination ability is shown in Scheme 14. However, it was subsequently reported that the correct structure was N-fluoropyridinium pyridinium tetrafluoroborate trifluorohydroxyborate, C 5 H 5 NF(C 5 H 5 NH)BF 4 (BF 3 OH), which was a 1:1 mixture of the N-fluoropyridinium salt and N-hydropyridinium salt (anion parts; BF 4 and BF 3 OH), based on an X-ray diffraction study of a commercial sample [42].

1-8. N-Fluoroquinuclidinium triflate
In 1988, Banks and co-worker developed the stable and nonhygroscopic N-fluoroquinuclidinium triflate (8-1) [50], which was an alternative to N-fluoroquinuclidinium fluoride (6-1) (Scheme 20). The triflate 8-1 was prepared in high yield by the counteranion replacement reaction developed by Umemoto and co-worker [28]. The fluorinating power of triflate 8-1 was the same as that of the fluoride 6-1, but its nonhygroscopic nature made it a useful fluorinating agent in terms of handling and storage. Details on the synthesis and reactivities of triflate 8-1
The enantioselectivities of products were examined after the fluorination of different metal enolates. In the best case a 63% yield and 70% enantiomeric excess (ee) was obtained. Even though other products gave less satisfactory outcomes, the potential of the N-F fluorinating agents for an enantioselective fluorination was demonstrated (Scheme 22).
The reagent 10-2 having no α-proton to the N-F site proved to be a good choice for fluorinating enolate anions (Scheme 24). The side reaction, involving HF elimination, and which was a problem in reactions with the Barnette's reagents 4-1 having the α-proton(s) except for 4-1b [26], was avoided here. The HF elimination is a decomposition process that is observed with N-F reagents that have an α-proton and occurs under strong base conditions.
The fluorination of benzene and anisole under excess substrate conditions gave fluorobenzene and fluoroanisoles in 88% and 98% yield, respectively. The reaction with sodium diethyl phenylmalonate gave the fluorinated product in 93% yield (Scheme 26).

1-12. N-Fluorolactams
A new class of N-F fluorinating agents, N-fluorolactams 12-1, was synthesized by Sathyamurthy et al. in 1990 [55]. These compounds had already been prepared by Grakauskas and co-worker in 1970 [56], but in low yields and the N-fluorolactams were not recognized as fluorinating agents at that time. For positron emission tomography (PET), Sathyamurthy et al. allowed these lactams to react with 0.05% 18 F 2 /Ne in a freon and obtained the N-[ 18 F]fluorolactams 12-1 in good yields (Scheme 27, entry 1). The 18 F-transfer ability was demonstrated by fluorination reactions with various Grignard reagents in up to 51% yield (Scheme 27, entry 2). In the event the β-elimination of HF proved to be an obstacle for the fluorination of strong bases such as phenyllithium.
NFSI was shown to fluorinate a variety of nucleophiles. As seen in Scheme 31, trimethylsilyl enol ethers, enolate anions of ketones and esters, and aryl-and vinyllithiums were fluorinated with NFSI in moderate to high yields. Although aromatics such as anisole, toluene, and acetanilide could also be fluorinated by NFSI, these reactions required neat conditions and high temperatures, indicating that NFSI was not so powerful.

1-15. N-Fluorosaccharin and N-fluorophthalimide
In 1991, Gakh et al. reported the synthesis of N-fluorosaccharin  and N-fluorophthalimide  in their studies on the re-  The fluorinating power of reagents 16-3 strengthened as the electronegativity of the R group increased in the order of CH 3 < CH 2 Cl < CH 2 CF 3 . The reagents 16-3 were able to fluorinate enol and conjugated enol acetates of steroids, sodium malonates, enamines, Grignard reagents, and aromatic compounds under mild conditions. In each case Selectfluor gave the corresponding fluorinated products in good to high yields (Scheme 34).
In 1995, the same group reported that Selectfluor reacted with quinuclidine to form N-fluoroquinuclidinium tetrafluoroborate in quantitative yield [62] (Scheme 35). They described this as a "transfer fluorination" since there was an intermolecular transfer of the fluorine atom of Selectfluor to the nitrogen of quinuclidine. In 1996, full details were published on the reactivities of all 16-3 reagents and the syntheses of 16-3 and intermediates 16-2 including additionally C 2 H 5 and C 8 H 17 as R group and PF 6 − and FSO 3 − as anion X − [63,64].
The generous provision of free samples of Selectfluor after its commercialization resulted in many publications concerning its applications from a diversity of research groups and in a short time period (1992)(1993)(1994)(1995)

1-18. Zwitterionic N-fluoropyridinium salts
In 1995, Umemoto and co-worker disclosed a zwitterionic N-fluoropyridinium salt system 18-2 which had a broad fluorinating power and high selectivity [76]. A series of N-fluoropyridinium-2-sulfonates with electron-withdrawing or -donating substituents were synthesized in high yields by fluorination of the corresponding pyridinium-2-sulfonates 18-1 with 10% F 2 /N 2 in acetonitrile or a mixture of acetonitrile/water at −40 to −10 °C ( Figure 5). In a few cases, a catalytic amount of triethylamine was used to improve the yields. The starting pyridinium-2-sulfonates possessing an electron-withdrawing group(s) were prepared in high yields from the reaction of the corresponding 2-chloropyridines with sodium sulfite. All of these N-F reagents are easy-to-handle and stable crystalline solids.  The nature of the lipophilic alkyl or trifluoromethyl substituents had a significant effect on the reactivity. Previously, N-fluoropyridinium-2-sulfonate and its 6-chloro derivative had been synthesized and shown to have excellent selectivity in fluorination reactions, but these reagents exhibited a low reactivity due to their low solubility in organic solvents [32].  Their fluorinating power increased in the order of 18-2a < 2b ≈ 2c ≈ 2d ≈ 2e < 2f < 2g < 2h, consistent with the order of the pK a values of the pyridines (Scheme 39). The least powerful 18-2a was suitable for the fluorination of reactive carbanions and easily oxidizable sulfides, whereas the most powerful 18-2h was suitable for less-reactive substrates such as olefins, aromatics, and neutral active methylene compounds. N-Fluoro-6-(trifluoromethyl)pyridinium-2-sulfonate (18-2f') was prepared later [77]. As shown in Scheme 40, the 18-2 pyridinium series proved to be effective fluorinating agents for a wide range of substrates. Moreover, an extremely high ortho-selectivity was observed in the fluorination of phenol. This could be attributed to a hydrogen-bonding interaction between the sulfonate anion and the phenol hydroxy group in the transition state. This assumption could also explain the high o-selectivity obtained in the fluorinations of naphthol, phenylurethane, and trimethylsilyl ether of phenol, and the exclusive 6-selectivity observed in the fluorination of conjugated enol triisopropylsilyl ethers of steroids. An additional advantage of the 18-2 reagents is that, after the fluorinations, the resulting pyridine-sulfonic acids are easily removed in an aqueous work-up.
As illustrated in Scheme 41, triflic acid was shown to promote the fluorination of anisole with 18-2h. While the reaction time was ≈30 h without the additive (entry 1, Scheme 41), 1 equiv of TfOH shortened this to less than 1 h (entry 2). Therefore, it would appear that this additive acted as a catalyst. With 0.1 equiv of TfOH, the reaction time was 5.5 h (entry 3) which can be attributed to the strong electron-withdrawing effect of the 2-SO 3 H substituent formed by the protonation of the sulfonate anion in 18-2h. The fluorination scope of NFTh is shown in Scheme 43 [79,80]. NFTh reacted with aromatics, enols, ketones, activated olefins, and substrates with active methylene groups to produce the corresponding fluorinated products in good to high yields. NFTh is a powerful fluorinating agent comparable to Selectfluor. Since NFTh has an acidic proton, its effectiveness is curtailed in the case of anionic substrates.
As a reagent N-F carboxamide 21-2 fluorinated electron-rich substrates such as sodium diethyl (phenyl)malonate, 1-morpholinocyclohexene, phenol, and anisole (Scheme 47). The fluorination power of the carboxamide 21-2 was less than that of its N-F sulfonamide analog 11-2. tane salts 22-1a-f in a pure form and in good to excellent yields (Scheme 48) [83]. All the salts were fully identified by elemental analysis and spectral analysis. Previously, in 1992, Banks and co-workers described how attempts to synthesize 22-1 proved to be unsatisfactory [42] and in 1995, they reported their synthesis characterized only by 19 F NMR and described that the salts were moisture-sensitive [84].

1-29. N-Fluorinated cinchona alkaloid derivatives by combination with Selectfluor
In 2000, Shibata and Takeuchi reported a far more practical enantioselective fluorination method. They discovered that the fluorination of carbanions with Selectfluor occurred in a highly enantioselective manner when carried out in the presence of cinchona alkaloid derivatives [95]. This method consisted of two simple steps. Firstly, the cinchona alkaloid is reacted with Selectfluor in acetonitrile at room temperature for 1 h, and then this is followed by addition of the substrate. The resulting mixture is stirred at a suitable temperature. They proposed that Selectfluor transfers fluorine to the alkaloid to give a chiral N-F alkaloid species, in a manner that followed the fluorine transfer reported by Banks when  fluor combination was effective for the fluorination of silyl enol ethers of indanones and tetralones, forming the fluorinated products in up to 91% ee. The DHQDA/Selectfluor combination was effective also for acyclic esters, with outcomes up to 87% ee, and for cyclic keto esters, up to 80% ee. For oxindoles, the (DHQD) 2 PYR/Selectfluor combination was effective too, generating the products with up to 82% ee.
Salt 32-2 was prepared in high yield in a small-scale batch reactor (Scheme 72). Accordingly, 100% F 2 (1 equiv) and BF 3 (1 equiv) gas were condensed into a stainless steel autoclave cooled at −196 °C, which contained cyanuric chloride (32-1, 1 equiv) in CFCl 3 . This was followed by gradual warming to room temperature and the reaction mixture was stored for 5 days. The resultant salt 32-2 is a moisture-sensitive white solid that decomposes at 153-155 °C. Reagent 32-2 reacted with deactivated benzenes such as chlorobenzene and nitrobenzene at ambient temperature (Scheme 73). This outcome indicated that 32-2 was a more powerful fluorinating reagent than the N-fluoropentachloropyridinium salts 5-4v,w. However, reagent 32-2 is not easy-tohandle because of its moisture-sensitivity.
This feature improved the enantioselectivity in cinchona alkaloid-catalyzed enantioselective fluorinations. The enantiomeric excesses of the products obtained with NFBSI 33-3 increased by 18% compared to that of (PhSO 2 ) 2 NF (NFSI, 14-2), while the chemical yields decreased (Scheme 75). Since the active fluorination agent in these reactions is considered to be the in situ generated N-fluorocinchona alkaloid salt, after fluorine transfer from NFBSI or NFSI to generate a counter anion, the bulky in situ generated counter anion may help enhance the % ee of the products.   Figure 11) [104]. In these reactions, the actual enantioselective fluorinating agents should be the N-F cinchona alkaloid salts formed in situ after the transfer from 34-3. Electron-donating groups (R) afforded higher enantioselectivities compared to R = H (NFSI, 14-2), while chemical yields were decreased. For example, while NFSI afforded 56% ee and 66% chemical yield (entry 1, Figure 11), R = OMe (34-3a) and tert-butyl (34-3b) gave 94% ee and 96% ee (entries 2 and 3 in Figure 11), but with only 49% and 17% chemical yields, respectively. Electronwithdrawing groups, R = CF 3 34-3c and OCF 3 34-3d, failed to furnish the fluorinated product under the same conditions (entries 8 and 9, Figure 11).  ciency effect, this could also be attributed to the strong lipophilic effect of the bis(CF 3 )phenyl group in 35-5a to the organic layer (PhCF 3 ).

1-36. Chiral dicationic DABCO-based N-F reagents
In June 2013, Gouverneur et al. reported the synthesis of chiral dicationic DABCO-based N-F reagents 36-5 that made possible the asymmetric electrophilic fluorocyclization of monoolefins with carbon nucleophiles [107]. This can be contrasted with Toste's method, described in section 1-35 above, for asymmetric fluorocyclization of dienes with nitrogen nucleophiles using Selectfluor and an optically active phase-transfer catalyst, a reaction which did not work for monoolefins with carbon nucleophiles.
The solubility, reactivity, and enantioselectivity of these types of reagents could be tuned by varying the substituents on the aryl rings and the p-trifluoromethyl-derivative 36-5b proved to be the most efficient for this type of reaction (Scheme 80). Up to 99% chemical yield and an average of 71% ee were obtained for a series of indene derivatives. However, the enantioselectivity was low (19% ee) for the case of a dihydronaphthalene derivative (the bottom in Scheme 80). Racemates of these fluoro products were prepared in high yields with N-fluoro-2,6dichloropyridinium triflate (5-1r).
The fluorination ability of 37-2a and -2b was evaluated with β-keto ester derivatives (Scheme 82). Moderate yields and good enantioselectivities of the fluorinated products were obtained.

1-38. N-Fluoromethanesulfonimide (Me-NFSI)
In 2016, Shibata et al. reported the synthesis and reactivity of N-fluoromethanesulfonimide (Me-NFSI, 38-2) [111]. Me-NFSI was first reported in a patent in 1994 [112], however, the reported fluorinations were vague. Prior to Shibata's report, Me-NFSI had not appeared in the literature in over 20 years. Although Me-NFSI is vulnerable with acidic protons on the methyl groups, Shibata realized the high atom economy of Me-NFSI and claimed that Me-NFSI was more effective for the fluorination of active methines than (PhSO 2 ) 2 NF (NFSI, 14-2) under Lewis acid-catalysis and non-catalysis. Me-NFSI was prepared in good yield by the fluorination of methanesulfonimide 38-1 with 10% F 2 /N 2 in acetonitrile at −40 °C in the presence of NaF (Scheme 85). Many active methine compounds were fluorinated in high yields using Me-NFSI in the presence of 10 mol % Ti(OiPr) 4 in methylene chloride as a solvent at room temperature (Scheme 86). The reactions with Me-NFSI were faster than those with NFSI. In particular, the fluorination of sterically demanding tert-butyl keto ester 38-3b with Me-NFSI gave the product 38-4b in 90% yield within only 5 min (92% in 3 h), while NFSI gave the product in 20% yield after 10 min and the yield gradually increased to 80% over 24 h. The faster reaction of Me-NFSI was explained by the activation of its N-F moiety via the stronger complexation with the Lewis acid Ti(OiPr) 4 than NFSI.
For the fluorination of malonates, different conditions were needed. Treatment with 2 equiv of Me-NFSI in the presence of 20 mol % Ti(OiPr) 4 in toluene at reflux temperature gave the fluorinated malonates in satisfactory yields (Scheme 87).
Keto esters were fluorinated in high yields with Me-NFSI employing methanol or water as solvent under catalysis-free conditions (entries 1-3, Scheme 88).

1-39. An N-F reagent derived from the ethano-Tröger's base
In 2016, Gouverneur and Cvengroš reported the synthesis of a novel N-F reagent 39-3 derived from the ethylene-bridged Tröger's base 39-1 [113]. The fluorination of the precursor 39-2, obtained from 39-1, was achieved only after reaction with N-fluoropentachloropyridinium triflate (5-1v) in acetonitrile at −35 °C, giving the N-F reagent 39-3 in higher than 95% conversion (Scheme 89). All other attempts to fluorinate, either with F 2 , XeF 2 , Selectfluor, or N-fluoro-2,6-dichloropyridinium triflate (5-1r) failed. This suggested that reagent 39-3 was more reactive than the other known reagents such as Selectfluor, but less active than the N-fluoropentachloropyridinium salts. The salt 39-3 is not stable and decomposition occurred when a solution of 39-3 in acetonitrile was left standing at room temperature for 8 hours or more. Therefore, 39-3 was best prepared with 5-1v immediately before use. The characterization of 39-3 was made by high-resolution mass spectrometry and NMR spectroscopic analyses containing 2D 19 F-15 N heteronuclear correlation experiments. As shown in Scheme 93, Selectfluor CN was used as an oxidant to improve the Pd-catalyzed bistrifluoromethoxylation of alkenes with AgOCF 3 . It was shown that the yield of the bis(CF 3 O) product 40-4 increased as the fluorinating power of the N-F reagent increased in the order of NFSI < 16-3 (R = Me) < Selectfluor (R = CH 2 Cl) < Selectfluor CN (R = CH 2 CN). A considerable number of olefin derivatives were bistrifluoromethoxylated using this method with Selectfluor CN [114].  The hydrofluorinated products were obtained in moderate yields along with good stereoselectivities in some cases (Scheme 95).

Evaluation of fluorinating power of N-F fluorinating agents
It emerges from the overview above that many N-F fluorination reagent variants have been developed, each with a different fluorination power (reactivity). Therefore, the power ordering of N-F reagents has become a matter of interest. Early on, Umemoto et al. revealed that the fluorinating power of N-fluoropyridinium salt systems could be tuned by the electron densi-ty at the nitrogen site, a property which is controlled by the electronic nature of the substituents. In order to quantitatively explain the variability of the N-F reagents, many attempts have been made: Theoretical calculations were performed by Hachisuka and Umemoto in 1991 [116,117], Woolf in 1994 [118], and Fainzil'berg in 1994 and 2001 [119,120]; N-F 19-fluorine NMR chemical shift analysis was accessed by Umemoto in 1991 [34]; measurements of peak reduction potentials were carried out by Gilicinski in 1992 [121], Differding in 1992 [122], Evans in 1999 [123], Zhang in 2013 [104], and reported by Umemoto in 2016 [124]; relative fluorination rates (kinetic data) were conducted by Togni in 2004 [125], and electromotive forces with Li, Mg, and Zn metals were reported by Umemoto in 2016 [124].
In a significant advancement in this regard, in 2016 Xue and Cheng published a comprehensive energetic scale for the quantitative estimation of the fluorinating power of N-F reagents [126,127]. They applied the Fluorine Plus Detachment (FPD) energy parameter, introduced by Christe and Dixon in 1992 [128], to electrophilic N-F reagents and calculated the FPD energies of 130 N-F reagents (Scheme 98).
As seen in Figure 14, a large number of N-F reagents were clearly arranged in their power order; the smaller the FPD value In addition, since some N-F reagents have been proven to be useful radical fluorinating agents [129][130][131][132][133][134][135], Xue and Cheng introduced the N-F homolytic bond-dissociation energy (BDE) parameter for radical fluorination abilities of N-F reagents in 2017 (Scheme 99) [127,136]. They calculated the BDE strength of 88 N-F reagents and clearly arranged them in terms of their radical fluorination power; the smaller the BDE value the stronger the radical fluorinating power ( Figure 15). The BDE is useful for identifying and designing new atomic fluorine sources.

Scheme 99: N-F homolytic bond dissociation energy (BDE).
In 2018, another significant advancement was reported. Since FPD and BDE are thermodynamic parameters, they cannot be directly used for predicting reaction rates. Mayr et al. [137] and Hodgson et al. [138] independently reported the quantitative reactivity scales for electrophilic fluorinations of popular N-F reagents. They provided detailed kinetic fluorination studies of a series of popular N-F fluorinating agents such as N-fluoropyridinium salt derivatives, NFSI, and Selectfluor TM with enamines, carbanions, or 1,3-diketones. Figure 16 shows the relative reactivity scale of the popular N-F reagents, which covers eight orders of magnitude. These reactivity scales can be used for predicting the actual fluorination rates of N-F fluorinating agents. Hodgson et al. further advanced their kinetic studies with N-F fluorinating agents in 2019 [139] and 2020 [140].  suggested a 2,6-dinitro-substituted N-fluoropyridinium salt as the most promising fluorinating reagent [141].

Reaction mechanisms of N-F fluorinating agents
Another noteworthy issue is the reaction mechanism for the fluorination with N-F reagents. There have been long standing arguments regarding which mechanism is operating in these fluorinations: a single-electron transfer (SET) or a nucleophilic substitution (S N 2) mechanism (Scheme 100) [22,142]. In the SET process, a single-electron transfer from a substrate (Nu − ) to a N-F reagent [N-F] occurs to form a radical anion [N-F] ·− and a Nu · radical, and the former undertakes the homolytic N-F bond cleavage to give a F · radical and a N: − anion, and then the F · radical and the Nu · radical combine to generate the final product Nu-F. In the S N 2 mechanism, a nucleophile directly attacks the F atom of the N-F reagent to give the fluorinated product via the heterolytic cleavage of the N-F bond. The key point is 8-1 produced only the non-cyclized products II and III. They also showed that the observed reaction rates were much faster than the calculated electron-transfer rates and therefore they concluded that a S N 2 mechanism occurred to give the fluorinated products [146].
In 1995, based on the study on the fluorination reactions of the N-F reagent NFOBS 13-2, Davis et al. described that all of their available data supported the S N 2 reaction process, but the SET mechanism could not be rigorously excluded [58]. In 1998, Zupan et al. studied the reactions of norbornene with Selectfluor, NFTh 19-2, and N-fluoro-2,6-dichloropyridinium triflate [147]. Although their results ended toward S N 2, they too could not completely exclude the SET. Soon after, in 1999, Wong et al. reported their studies using compound VI having a phenylcyclopropyl moiety which was the hypersensitive radical probe (rate: 10 11 s −1 ) (Scheme 103) [148].
They found that Selectfluor (OTf salt) produced product VII in 45% yield with no rearrangement products despite the complete consumption of VI (entry 1, Scheme 103). This result is aligned with the S N 2 mechanism, however, the reaction of VI with Selecfluor was completely blocked by the radical scavenger, 2,2,6,6-tetramethylpiperidinoxyl (TEMPO), which suggested a SET mechanism. The reaction of VI with NFSI gave VII (40%) and a product IX (5%) derived from the rearrangement fluoro product VIII (entry 2 in Scheme 103). This result is consistent with the SET mechanism. In their review published in 2005 [149], Wong discussed these results in more detail.
Since the reactivity of the radical (F · ) is extremely fast and diffusion-controlled, the combination of F · with any alkyl radical R · should proceed at an equal or greater rate than the cyclization of the alkyl radical as observed in the case of Differding [145] and the ring-opening of the cyclopropyl radical in the case of Wong [148]. Despite extensive research, the mechanism remains unclear as a result of the difficulty of assessing reaction rates. The key may be the lifetime of the intermediate N-F radical or radical anion, [N-F] · or [N-F] ·− , which releases the radical F · . Finally, Wong mentioned that, if these radicals react faster than the rates of diffusion, anything beyond that would probably remain unproven [149].
In the meantime, in 2002, Togni and Rothlisberger provided evidence which strongly supported the SET mechanism from their studies on the fluorination mechanism of β-ketoesters with Selectfluor and a chiral TiCl 2 (TADDOLato) catalyst [150]. Their computational study suggested a one-electron transfer followed by simultaneous N-F bond cleavage and the accompanying experimental outcomes demonstrated that a chlorination side reaction was quenched by a radical scavenger, while the fluorination reaction was unaffected. Since Selectfluor is inert towards chloride, they proposed that any [N-F] · radical formed by the SET process would react with chloride ions (Cl − ) present in the reaction solution to form the relatively stable Cl · radical, which caused the side chlorination.
On the other hand, in 2002, Laali and co-worker reported that competitive fluorination studies of mesitylene and durene with Selecfluor in CH 3 CN and an ionic liquid supported the conventional polar mechanism (S N 2) [151]. However, in 2006, Stavber et al. made similar competition studies and proposed a SET process as the dominant mechanism that explained the higher reactivity of durene viz-à-viz mesitylene towards Selectfluor in aq CH 3 CN [152]. In 2009, Shubin et al. performed competitive studies of mesitylene and durene with NFSI under solvent-free reaction conditions and reported that the reaction rate ratio (k Mes /k Dur ) followed a polar mechanism [153].
Since 2015, many papers supporting either SET or S N 2 mechanisms have appeared, including those reported by C. TEMPO has been often used as a radical scavenger to explore the reaction mechanisms of N-F fluorinating agents but, in Nelson's paper [163] in 2019, it was revealed that Selectfluor reacted with TEMPO (X) to form the oxoammonium salt XI (Scheme 104). This author concluded that the use of TEMPO as a radical trap in the reactions with Selectfluor was unreliable. This conclusion was subsequently confirmed by Borodkin et al. [164]. The latest review [165] by Hodgson and co-worker in 2021 discussed the reactivities and reaction mechanisms of electrophilic N-F fluorinating agents in detail.

Conclusion
A historical timeline on the progress of N-F fluorinating reagents and their fluorinations has been described in this review. Table 1 shows a historical list of all the N-F reagents. The fluorination power and reaction mechanism of the N-F fluorinating agents have also been summarized briefly and from a historical perspective. The fluorination reactions described here for each N-F fluorinating agent are those reported at the time. The synthetic applications of each N-F reagent since then are too numerous to be covered in this review. As many reviews on the reactions and applications of the N-F fluorinating agents have been published [39,133,142,[165][166][167][168][169][170][171][172][173][174][175][176][177][178][179][180][181], readers are recommended to consult these reviews.
The seminal work of perfluoro-N-fluoropiperidine by Banks was reported in 1964 [16]. However, this compound could not be contemplated for two decades because of the very low yields in its production and fluorination reactions. At that time, FClO 3 [10] and O-F reagents such as CF 3 OF and CF 2 (OF) 2 [11] were used as electrophilic fluorinating agents despite their inherent safety issues. The real attention on N-F compounds started in the 1980's when the first report on a N-F fluorinating agent, N-fluoro-2-pyridone, was published by Purrington in 1983 [24].
A very large number of N-F fluorinating agents have been developed so far and a considerable number of easy-to-handle N-F reagents have been commercialized. In particular, Selectfluor, NFSI, and N-fluoropyridinium salt derivatives are produced in factories on a large scale. This has greatly contributed to the remarkable advancement of organofluorine chemistry. Before the appearance of N-F reagents, electrophilic fluorinating agents were thought dangerous and difficult to handle and special techniques were required. Chemists avoided them. Now, practicing chemists use them without any problem. The N-F reagents have brought such a dramatic change that many chemists nowadays may not realize the magnitude of this change and the contributions of the pioneers.
For practitioners in the field, high thermal stability and nonhygroscopicity of reagents are key factors for easy handling and a long shelf life. Many N-F reagents satisfy this requirement, however, there are limitations as the fluorinating power increases. N-Fluoropentachloropyridinium triflate (5-4v) and tetrafluoroborate (5-4w) rank as the most powerful of the nonhygroscopic and thermally stable N-F agents. However, the more powerful N-fluoro-2,4,6-trichloro-1,3,5-triazinium tetrafluoroborate (31-2) is moisture-sensitive and not easy-tohandle. It is now rare for researchers and laboratories to use molecular fluorine (F 2 ) in universities and institutes, because of strict safety regulations, although 20% or 10% F 2 diluted with N 2 can be handled without significant problems. As a result, currently reported fluorination reactions have been limited to the use of a few commercially available and inexpensive N-F reagents such as Selectfluor and NFSI. As mentioned in this review, there are many other useful N-F fluorinating agents and the plentiful supply will satisfy the strong demand for fluorinations. That said, there is still room for improvement or refinement. The authors think that further research and development on N-F fluorinating reagents is needed for the advancement of synthetic fluorine chemistry but, for this to happen, it would be highly desirable to develop a safe and convenient F 2 generator.

Funding
We are grateful to the National Institutes of Health (R01GM121660) for financial support.