Silver-Promoted Radical Cascade Aryldifluoromethylation/Cyclization of 2-Allyloxybenzaldehydes for the Synthesis of 3-Aryldifluoromethyl-Containing Chroman-4-one Derivatives

A convenient silver-promoted radical cascade aryldifluoromethylation/cyclization of 2-allyloxybenzaldehydes has been developed. Experimental studies disclosed that the addition of aryldifluoromethyl radicals in situ produced from easily accessible gem-difluoroarylacetic acids to unactivated double bonds in 2-allyloxybenzaldehyde was an effective route to access a series of 3-aryldifluoromethyl-containing chroman-4-one derivatives in moderate to good yields under mild reaction conditions.


Introduction
The chroman-4-one framework belongs to the privileged class of heterocycles and acts as a major building block in many natural products, pharmaceuticals, and biologically active compounds [1,2]. In particular, the compounds with substituents at the 3-position of the chroman-4-one exhibit significant physiological and biological activities, such as GPR119 agonist (Figure 1, a) [3], anti-HIV agent (Figure 1, b) [4], and anti-tobacco mosaic virus agent (Figure 1, c) [5] etc. Thus, a number of approaches for the synthesis of 3substituted chroman-4-one derivatives have been developed, such as transition-metalcatalyzed cross-coupling reaction [6], Stetter reaction [7], and cascade radical cyclization [8][9][10][11][12]. The introduction of difluoromethylene group into organic molecules can not only improve chemical, physical properties and the physiological activity of the parent molecule, but also affect their binding affinity, metabolic stability, lipophilicity, and membrane permeability and bioactivity, as well as modulate pharmacokinetic behavior and penetration in biological tissues [13][14][15]. Hence, gem-difluoromethylene group (CF2) is regarded as a valuable fluorinated moiety that is extensively present in pharmaceuticals, biologically active compounds, and material molecules [16,17]. Therefore, the development of The introduction of difluoromethylene group into organic molecules can not only improve chemical, physical properties and the physiological activity of the parent molecule, but also affect their binding affinity, metabolic stability, lipophilicity, and membrane permeability and bioactivity, as well as modulate pharmacokinetic behavior and penetration in biological tissues [13][14][15]. Hence, gem-difluoromethylene group (CF 2 ) is regarded as a valuable fluorinated moiety that is extensively present in pharmaceuticals, biologically active compounds, and material molecules [16,17]. Therefore, the development of effective and general methodologies for the selective incorporation of a difluoromethylene group has become one of the hotspots in the field of organic chemistry. Traditional methods for the synthesis of the difluoromethylene-containing compounds include the direct deoxyfluorination of the carbonyl group in corresponding aldehydes or ketones by using bis(2-methoxyethyl)aminosulfur trifluoride or DAST ((C 2 H 5 ) 2 NSF 3 ) [18][19][20][21] and the conversion of building blocks containing difluoromethylene under specific conditions. For example, visible light-promoted the difluoromethylenation of C-H bonds under xanthone catalyzed conditions [22][23][24], and transition-metal-catalyzed (Pd-or Ni-) difluoromethylenation can be achieved applying chlorodifluoromethane [25], and so on. However, some existing methods suffer from either functional group incompatibility or the use of toxic and explosive reagents [26][27][28]. Therefore, it remains highly desired to develop practical and mild synthetic methods by using stability and easily available substrates containingdifluoromethylene building blocks, such as α,α-difluoroarylacetic acids [29][30][31][32].
Radical cascade fluoroalkylation/cyclization of alkenes has been proved to be one of the effective methods to obtain fluoroalkyl-containing heterocycle products due to the advantages that it avoids the separation and purification of intermediates [33] and realizes the introduction a fluoroalkyl group while constructing the diversity and complexity of heterocyclic skeletons. However, some fluoroalkyl radicals, such as CF 3 radical, CF 2 H, and C n F 2n+1 radical, are electron-deficient σ-radicals and thus prone to add to electron-rich alkenes. Due to electronically unfavorable conditions, it is difficult to realize the free radical cascade fluoroalkylation/cyclization of unactivated or electrondeficient alkenes. Therefore, there are only few reports on the cascade difluoromethylenation/cyclization regarding the synthesis of 3-difluoromethylene-containing chroman-4-one derivatives through the addition difluoromethyl radicals to unactivated alkenes in 2-allyloxybenzaldehydes. The reported methods include that of Zhu's group, which developed a visible light photocatalytic aryldifluoromethylation/cyclization of alkenes using α,α-difluoroarylacetic acids (Scheme 1a) [34], and Zhou's group and Ma's group, which described radical cascade cyclization of alkene aldehydes using a 1-bromo-2-difluorocontaining building block (Scheme 1b) [35,36] under the visible-light photoredox catalysis conditions. Therefore, it is highly desirable to develop new, convenient methods for the synthesis of 3-difluoromethylene-containing chroman-4-one derivatives, to avoid using a relatively expensive photocatalyst in the mild reaction. Our previous research on the decarboxylation of gem-difluoroarylacetic acids under silver-catalyzed condition revealed that aryldifluoromethyl radicals have a slightly electron-rich property relative to the CF 3 radical and could react with the electron-deficient activated double bond in allyl sulfones [37]. Herein, we further tried to use aryldifluoromethyl radicals from the decarboxylation of gem-difluoroarylacetic acids under the promotion of the catalytic amount of silver to explore the cascade aryldifluoromethylation/cyclization reaction with unactivated double bonds in 2-allyloxybenzaldehydes to synthesize 3-aryldifluoromethyl-containing chroman-4-one derivatives (Scheme 1c).

Results and Discussion
Our previous research found that the system of silver salts and persulfate salts was effective in the decarboxylation of difluoroarylacetic acids in cross-coupling reactions Scheme 1. Synthesis of 3-difluoromethylene-containing chroman-4-one derivatives.

Results and Discussion
Our previous research found that the system of silver salts and persulfate salts was effective in the decarboxylation of difluoroarylacetic acids in cross-coupling reactions with allyl sulfones [37]. Therefore, 4-methyl difluorophenylacetic acid 2a and 2-(allyloxy)benzaldehyde 1a which included unactivated double bonds were chosen as model substrates to participate in the reaction in the presence of AgNO 3 (20 mol%) and K 2 S 2 O 8 (2 equiv.) in CH 3 CN/H 2 O (v/v = 1:1) at 80 • C under a N 2 atmosphere, replacing visible light redox catalytic conditions [34]. To our delight, 3-(4-CH 3 )phenyldifluoromethylcontaining chroman-4-one product 3aa was observed in 56% 19 F NMR yield ( Table 1, entry 1). Encouraged by the result, we try to optimize the reaction conditions. When solvent was shifted from CH 3 CN/H 2 O to EtOH/H 2 O or DMSO/H 2 O, the reaction gave the desired product in lower yields (Entries 2 and 3). The product 3aa was also produced when the solvent was pure H 2 O, but the reaction efficiency was reduced (Entry 4). When the solvent was DMSO, the reaction could also produce the product in 32% yield (Entry 22). However, the desired product was obtained if the reaction was carried out in CH 3 CN (Entry 5). Upon further screening of solvents, it was found that the reaction provided the highest yield (73%) in the solvent CH 3 CN/H 2 O (v/v = 1:3) when 2 equivalents of K 2 S 2 O 8 were used as oxidant (Entry 6). Other oxidants, including Na 2 S 2 O 8 and (NH 4 ) 2 S 2 O 8 , were also effective (Entries 7 and 8). In contrast, selectfluor failed to carry out this transformation (Entry 9). Additionally, when changing the amount of oxidant K 2 S 2 O 8 from 2 into 1.75 or 2.25 equiv., the reaction efficiency was slightly reduced (Entry 6 vs 17 and 18). Moreover, oxidant K 2 S 2 O 8 was actually required to bring about the cascade aryldifluoromethylation/cyclization (Entry 11). We also further evaluated the effect of the amount of AgNO 3 on the reaction. AgNO 3 could promote the reaction, but it was not necessary (Entry 10). A higher catalyst loading of AgNO 3 did not lead to a higher reaction efficiency, and 10 mol% of AgNO 3 as promoter resulted in 63% yield of 3aa (Entries 12 and 13). The other Ag (I) sources, such as AgOAc, Ag 2 O, or Ag 2 CO 3 , could also be used as the promoters, but they were less effective (Entries [19][20][21]. Adjusting the ratio of 1a and 2a, it was found that 3aa was provided in the highest yield when the ratio of 1a and 2a was 1:1.5 (Entries 6, 14-16) at 80 • C for 4 h (Supplementary Materials).  Having established the optimal experimental conditions, we explored the scope and limitation of this cascade aryldifluoromethylation/cyclization by the reaction of α,αdifluoroarylacetic acids and 2-allyloxybenzaldehydes. We first explored the scope with respect to the α,α-difluoroarylacetic acids 2, as shown in Table 2. Difluoroarylacetic acids (2a-2h) bearing an electron-donating group, such as CH 3 , OCH 3 etc., at different positions of benzene ring produced the desired product in moderate to excellent yields (3aa-3ah). When 2,2-difluoro-2-phenylacetic acid 2i was used, the reaction gave the desired product 3ai in 50% yield. The sterically hindered 2,2-difluoro-2-mesitylacetic acid 2h was also a compatible substrate in the cascade aryldifluoromethylation/cyclization leading to 3ah in 55% yield. Additionally, the α,α-difluoroarylacetic acids with different halogen groups at position-4 of the benzene ring could offer the desired products in 32 to 60% yields (3akam and 3ao), and the halogen groups were well tolerated. However, no desired product 3ap was obtained when the para-substituent of benzene ring in the α,α-difluoroarylacetic acid is strong electron-withdrawing nitro group. Difluoroarylacetic acids bearing a strong electron-withdrawing group, such as CN or CF 3 at benzene ring, also made it difficult to provide the desired product. The results showed that the substituents on the benzene ring of α,α-difluoroarylacetic acid have a great influence on the reaction activity. This could be attributed to the fact that the electron-donating group substituted aryldifluoromethyl radicals produced in situ under silver-promoted condition demonstrated greater electron-rich property than that of non-substituted or strong electron-withdrawing group substituted phenyldifluoromethyl radical. Thus, the electron-richer aryldifluoromethyl radicals were capable of facilitating addition to unactivated double bonds in 2-allyloxybenzaldehyde. To further investigate the reaction scope, 2,2-difluoro-2-((4-methoxyphenyl)thio)acetic acid 2q was also employed as substrate, and an unexpected regio-selective product, 3aq , was achieved in 29% yield besides the desired product 3aq in 30% yield. The structure of 3aq was further characterized by X-ray single crystal diffraction analysis ( Figure 2) and the NMR and HR-MS etc. Additionally, when aliphatic 2,2-difluorobutanoic acid (2r) was employed in this transformation, the reaction also gave the desired product 3ar in 13% yield.
The generality of this cascade aryldifluoromethylation/cyclization toward different 2allyloxybenzaldehydes was also explored.
As shown in Table 3, a variety of 2-allyloxybenzaldehydes 1 were employed to react with 2,2-difluoro-2-(4-methoxyphenyl) acetic acid (2b). This method successfully gave the corresponding products 3 in moderate to good yields. The substrates 1b-d bearing electron-donating groups reacted smoothly, delivering the corresponding chroman-4-one derivatives 3bb-3db in moderate yields (35-50%). Unexpectedly, the reaction of the di-t-Bu-substituted 2-allyloxybenzaldehyde (1e) provided the corresponding product 3eb in trace amount in the aqueous CH 3 CN solution (CH 3 CN:H 2 O = 1:3) under standard conditions, while it delivered 3eb in 45% yield in the DMSO. The substrates monosubstituted by different halogens at different positions of the benzene ring in 2-allyloxybenzaldehydes (1f-1k) gave the desired product in moderate yields as well. Additionally, halogen functional groups were well tolerated in this protocol. The di-bromo-substituted substrate 1m was not a good substrate for this transformation. However, the cascade aryldifluoromethylation/cyclization of 2-((2-methylallyl)oxy) benzaldehyde 1m could afford the product 3mb in 41% in DMSO. Notably, the reaction of 2-allylbenzaldehyde 1n with 2,2-difluoro-2-(4-methylphenyl)acetic acid (2a) could also deliver the corresponding product 3na in 37% yield.  The generality of this cascade aryldifluoromethylation/ cyclization toward different 2-allyloxybenzaldehydes was also explored. As shown in Table 3, a variety of 2-allyloxybenzaldehydes 1 were employed to react with 2,2-difluoro-2-(4-methoxyphenyl)acetic acid (2b). This method successfully gave the corresponding products 3 in moderate to good yields. The substrates 1b-d bearing electron-donating groups reacted smoothly, delivering the corresponding chroman-4-one derivatives 3bb-3db in moderate yields (35-50%). Unexpectedly, the reaction of the di-t-Bu-substituted 2-allyloxybenzaldehyde (1e) provided the corresponding product 3eb in trace amount in the aqueous CH3CN solution (CH3CN:H2O = 1:3) under standard conditions, while it delivered 3eb in 45% yield in the DMSO. The substrates monosubstituted by different halogens at different positions of the benzene ring in 2-allyloxybenzaldehydes (1f-1k) gave the desired product in moderate yields as well. Additionally, halogen functional groups were well tolerated in this protocol. The di-bromo-substituted substrate 1m was not a good substrate for this transformation. However, the cascade aryldifluoromethylation/cyclization of 2-((2-methylallyl)oxy) benzaldehyde 1m could afford the product 3mb in 41% in DMSO. Notably, the reaction of 2-allylbenzaldehyde 1n with 2,2-difluoro-2-(4-methylphenyl)acetic acid (2a) could also deliver the corresponding product 3na in 37% yield.  The generality of this cascade aryldifluoromethylation/ cyclization toward different 2-allyloxybenzaldehydes was also explored. As shown in Table 3, a variety of 2-allyloxybenzaldehydes 1 were employed to react with 2,2-difluoro-2-(4-methoxyphenyl)acetic acid (2b). This method successfully gave the corresponding products 3 in moderate to good yields. The substrates 1b-d bearing electron-donating groups reacted smoothly, delivering the corresponding chroman-4-one derivatives 3bb-3db in moderate yields (35-50%). Unexpectedly, the reaction of the di-t-Bu-substituted 2-allyloxybenzaldehyde (1e) provided the corresponding product 3eb in trace amount in the aqueous CH3CN solution (CH3CN:H2O = 1:3) under standard conditions, while it delivered 3eb in 45% yield in the DMSO. The substrates monosubstituted by different halogens at different positions of the benzene ring in 2-allyloxybenzaldehydes (1f-1k) gave the desired product in moderate yields as well. Additionally, halogen functional groups were well tolerated in this protocol. The di-bromo-substituted substrate 1m was not a good substrate for this transformation. However, the cascade aryldifluoromethylation/cyclization of 2-((2-methylallyl)oxy) benzaldehyde 1m could afford the product 3mb in 41% in DMSO. Notably, the reaction of 2-allylbenzaldehyde 1n with 2,2-difluoro-2-(4-methylphenyl)acetic acid (2a) could also deliver the corresponding product 3na in 37% yield. be attributed to the fact that the electron-donating group substituted aryldifluoromethyl radicals produced in situ under silver-promoted condition demonstrated greater electronrich property than that of non-substituted or strong electron-withdrawing group substituted phenyldifluoromethyl radical. Thus, the electron-richer aryldifluoromethyl radicals were capable of facilitating addition to unactivated double bonds in 2-allyloxybenzaldehyde. To further investigate the reaction scope, 2,2-difluoro-2-((4-methoxyphenyl)thio)acetic acid 2q was also employed as substrate, and an unexpected regio-selective product, 3aq′, was achieved in 29% yield besides the desired product 3aq in 30% yield. The structure of 3aq′ was further characterized by X-ray single crystal diffraction analysis ( Figure 2) and the NMR and HR-MS etc. Additionally, when aliphatic 2,2difluorobutanoic acid (2r) was employed in this transformation, the reaction also gave the desired product 3ar in 13% yield.
. Figure 2. X-ray crystallography of 3aq′ (gray for carbon atoms, green for fluorine atom, yellow for sulfur atom and red for oxygen atom).    To gain more insight into the reaction mechanism and identify whether a radical pathway is involved in the silver promoted cascade aryldifluoromethylation/cyclization or not, some control experiments were carried out. When 1.0 or 1.5 equiv. of radical scavenger TEMPO were added into the reaction system of 4-methyl difluorophenylacetic acid (2a) and 2-allyloxybenzaldehyde 1a under the standard conditions, the reaction provided neither the desired product 3aa nor adduct of TEMPO with phenyldifluoromethyl radical based on 19 F NMR analysis. Additionally, the conversion of 2a was completely inhibited under the standard conditions because 2a almost totally remained in the above reaction system by 19 F NMR monitoring (Scheme 2a). We deduced from the results that the reaction may undergo a radical process. To further capture the phenyl difluoromethyl radical to verify this speculation, the reaction of only 4-methyl difluorophenylacetic acid (2a) used as substrate under the standard conditions in the absence of 2-allyloxybenzaldehyde 1a was performed (Scheme 2b). The phenyl difluoromethyl radical self-coupling product 4 was obtained in 16% isolated yield. Thus, it was indicated that the phenyl difluoromethyl radical as intermediate could be involved in the reaction process (Scheme 2).   To gain more insight into the reaction mechanism and identify whether a radical pathway is involved in the silver promoted cascade aryldifluoromethylation/cyclization or not, some control experiments were carried out. When 1.0 or 1.5 equiv. of radical scavenger TEMPO were added into the reaction system of 4-methyl difluorophenylacetic acid (2a) and 2-allyloxybenzaldehyde 1a under the standard conditions, the reaction provided neither the desired product 3aa nor adduct of TEMPO with phenyldifluoromethyl radical based on 19 F NMR analysis. Additionally, the conversion of 2a was completely inhibited under the standard conditions because 2a almost totally remained in the above reaction system by 19 F NMR monitoring (Scheme 2a). We deduced from the results that the reaction may undergo a radical process. To further capture the phenyl difluoromethyl radical to verify this speculation, the reaction of only 4-methyl difluorophenylacetic acid (2a) used as substrate under the standard conditions in the absence of 2-allyloxybenzaldehyde 1a was performed (Scheme 2b). The phenyl difluoromethyl radical self-coupling product 4 was obtained in 16% isolated yield. Thus, it was indicated that the phenyl difluoromethyl radical as intermediate could be involved in the reaction process (Scheme 2). To gain more insight into the reaction mechanism and identify whether a radical pathway is involved in the silver promoted cascade aryldifluoromethylation/cyclization or not, some control experiments were carried out. When 1.0 or 1.5 equiv. of radical scavenger TEMPO were added into the reaction system of 4-methyl difluorophenylacetic acid (2a) and 2-allyloxybenzaldehyde 1a under the standard conditions, the reaction provided neither the desired product 3aa nor adduct of TEMPO with phenyldifluoromethyl radical based on 19 F NMR analysis. Additionally, the conversion of 2a was completely inhibited under the standard conditions because 2a almost totally remained in the above reaction system by 19 F NMR monitoring (Scheme 2a). We deduced from the results that the reaction may undergo a radical process. To further capture the phenyl difluoromethyl radical to verify this speculation, the reaction of only 4-methyl difluorophenylacetic acid (2a) used as substrate under the standard conditions in the absence of 2-allyloxybenzaldehyde 1a was performed (Scheme 2b). The phenyl difluoromethyl radical self-coupling product 4 was obtained in 16% isolated yield. Thus, it was indicated that the phenyl difluoromethyl radical as intermediate could be involved in the reaction process (Scheme 2). On the basis of the above-mentioned results and previous reports [35,[37][38][39][40][41][42], a possible mechanism for the reaction is depicted in Scheme 3. First, α, α-difluorophenylacetic acid 2 undergoes an oxidative decarboxylation process under the promotion of Ag(I) to give the corresponding phenyldifluoromethyl radical I [43]. Then, the addition of I to the unactivated C=C double bond of 2-allyloxybenzaldehyde 1a affords an alkyl radical II.

Conclusions
In conclusion, we established a facile radical cascade aryldifluoromethylation/cyclization method to synthesize a series of 3-aryldifluoromethyl-containing chroman-4-one derivatives in moderate to good yields. These reactions involved the addition of aryldifluoromethyl radicals in situ produced from easily accessible gem-difluoroarylacetic acids to unactivated double bonds in 2-allyloxybenzaldehydes under silver-promoted conditions. This benign protocol replaces the use of expensive photocatalysts in visible light redox catalytic conditions and provides a new perspective to construct pharmaceutically valuable, 3-difluoromethylene-containing chroman-4-one derivatives. On the basis of the above-mentioned results and previous reports [35,[37][38][39][40][41][42], a possible mechanism for the reaction is depicted in Scheme 3. First, α,α-difluorophenylacetic acid 2 undergoes an oxidative decarboxylation process under the promotion of Ag(I) to give the corresponding phenyldifluoromethyl radical I [43]. Then, the addition of I to the unactivated C=C double bond of 2-allyloxybenzaldehyde 1a affords an alkyl radical II.  On the basis of the above-mentioned results and previous reports [35,[37][38][39][40][41][42], a possible mechanism for the reaction is depicted in Scheme 3. First, α, α-difluorophenylacetic acid 2 undergoes an oxidative decarboxylation process under the promotion of Ag(I) to give the corresponding phenyldifluoromethyl radical I [43]. Then, the addition of I to the unactivated C=C double bond of 2-allyloxybenzaldehyde 1a affords an alkyl radical II. The alkyl radical intermediate II attacking the C-O double bond led to an alkoxy radical intermediate IIl, realizing an intramolecular cyclization. Afterward, a formal 1,2-hydrogen atom transfer produces the α-hydroxy carbon-centered radical IV. Subsequently, radical IV is oxidized to give the carbocation intermediate V [43]

Conclusions
In conclusion, we established a facile radical cascade aryldifluoromethylation/cyclization method to synthesize a series of 3-aryldifluoromethyl-containing chroman-4-one derivatives in moderate to good yields. These reactions involved the addition of aryldifluoromethyl radicals in situ produced from easily accessible gem-difluoroarylacetic acids to unactivated double bonds in 2-allyloxybenzaldehydes under silver-promoted conditions. This benign protocol replaces the use of expensive photocatalysts in visible light redox catalytic conditions and provides a new perspective to construct pharmaceutically valuable, 3-difluoromethylene-containing chroman-4-one derivatives.

Conclusions
In conclusion, we established a facile radical cascade aryldifluoromethylation/cyclization method to synthesize a series of 3-aryldifluoromethyl-containing chroman-4-one derivatives in moderate to good yields. These reactions involved the addition of aryldifluoromethyl radicals in situ produced from easily accessible gem-difluoroarylacetic acids to unactivated double bonds in 2-allyloxybenzaldehydes under silver-promoted conditions. This benign protocol replaces the use of expensive photocatalysts in visible light redox catalytic conditions and provides a new perspective to construct pharmaceutically valuable, 3-difluoromethylene-containing chroman-4-one derivatives.