Visible-Light-Induced Decarboxylation of Dioxazolones to Phosphinimidic Amides and Ureas

A visible-light-induced external catalyst-free decarboxylation of dioxazolones was realized for the bond formation of N=P and N–C bonds to access phosphinimidic amides and ureas. Various phosphinimidic amides and ureas (47 examples) were synthesized with high yields (up to 98%) by this practical strategy in the presence of the system’s ppm Fe.


Optimization of the Reaction Conditions
We commenced our study with the model reaction between 3-phenyl-1,4,2-dioxazol-Scheme 1. The construction of N=P and N-C bonds from dioxazolones.

Optimization of the Reaction Conditions
We commenced our study with the model reaction between 3-phenyl-1,4,2-dioxazol-5one (1a) and triphenylphosphine (2a) under visible light and N 2 atmosphere. The results are shown in Table 1. Initially, the reaction was carried out by employing DCE as the solvent under irradiation of 10 W 430 nm blue LED at room temperature, and the desired product N-(triphenyl-λ 5 -phosphinylidene)benzamide (3a) could be detected in 11% yield (entry 1). Afterwards, the solvent effect on the yield was investigated (entries [2][3][4][5][6][7][8]. Different solvents, such as 1,4-dioxane, CH 3 OH, acetone, CH 3 CN, DMF, THF, and CH 2 Cl 2 , were surveyed, and the reaction exhibited excellent reaction performance in CH 2 Cl 2 to provide the target product in 81% yield (entry 8). Further examination of the wavelengths of LED and substrate ratios showed no more positive results (entries [9][10][11][12][13][14]. Control reactions confirmed that nearly no amidation product 3a was detected at room temperature in the absence of visible light (entry 15). Moreover, when the reaction was carried out in the air, only a trace amount of the product 3a was detected (entry 16). Therefore, the optimized reaction conditions were illustrated as follows: 1a (0.1 mmol); 2a (0.1 mmol); and CH 2 Cl 2 (1 mL) in a N 2 atmosphere under the irradiation of 430 nm blue LED (10 W) for 24 h at room temperature (entry 8). Scheme 1. The construction of N=P and N-C bonds from dioxazolones.

Optimization of the Reaction Conditions
We commenced our study with the model reaction between 3-phenyl-1,4,2-dioxazol-5-one (1a) and triphenylphosphine (2a) under visible light and N2 atmosphere. The results are shown in Table 1. Initially, the reaction was carried out by employing DCE as the solvent under irradiation of 10 W 430 nm blue LED at room temperature, and the desired product N-(triphenyl-λ 5 -phosphinylidene)benzamide (3a) could be detected in 11% yield (entry 1). Afterwards, the solvent effect on the yield was investigated (entries 2-8). Different solvents, such as 1,4-dioxane, CH3OH, acetone, CH3CN, DMF, THF, and CH2Cl2, were surveyed, and the reaction exhibited excellent reaction performance in CH2Cl2 to provide the target product in 81% yield (entry 8). Further examination of the wavelengths of LED and substrate ratios showed no more positive results (entries [9][10][11][12][13][14]. Control reactions confirmed that nearly no amidation product 3a was detected at room temperature in the absence of visible light (entry 15). Moreover, when the reaction was carried out in the air, only a trace amount of the product 3a was detected (entry 16). Therefore, the optimized reaction conditions were illustrated as follows: 1a (0.1 mmol); 2a (0.1 mmol); and CH2Cl2 (1 mL) in a N2 atmosphere under the irradiation of 430 nm blue LED (10 W) for 24 h at room temperature (entry 8). With the optimized conditions in hand, the scope of the organophosphorus compounds and dioxazolones 1 was investigated (Scheme 2). To our delight, various 3-phenyl dioxazolones bearing different electron-donating groups (-CH 3 , -t Bu, and -OCH 3 ) or electronwithdrawing groups (-CF 3 , -F, -Cl, and -CN) on the phenyl ring at different positions could react smoothly with 2a to produce the desired products (3a-3n) in moderate to excellent yields (42-98%). Among these cases, a slight steric hindrance effect was observed, and parasubstituted 3-phenyl dioxazolones (3a-3h, 63-98%) showed higher reaction reactivities than those of ortho-substituted 3-phenyl dioxazolones (3m-3n, 42-47%). Moreover, the desired products 3o and 3p, which contain the skeletons of thiophene and furan, could also be successfully obtained in 43% and 50% of the yields, respectively. Additionally, electron-poor and electron-rich triphenylphosphine derivatives were all applicable to this transformation to access the desired products (3q-3v) in 51-91% yields. In addition, the phosphorus ligand, 1.1 -binaphthyl-2.2 -diphenylphosphine (BINAP), was also a suitable substrate to react with 1a, providing the corresponding product 3w in 51% yield.

Scheme 2.
Substrate scope for the synthesis of phosphinimidic amides. Scheme 2. Substrate scope for the synthesis of phosphinimidic amides.
Then, we expanded the photocatalytic decarboxylation reaction of dioxazolones to the synthesis of unsymmetrical urea compounds (Scheme 3). To our delight, a wide range of 3-phenyl dioxazolones all reacted efficiently with diisopropylamine 4a to furnish the corresponding aryl ureas (5a-5m) in moderate to excellent yields (37-98%). In these cases, 3phenyl dioxazolones bearing electron-donating groups (-CH 3 , -t Bu, and -OCH 3 ) showed a better reaction efficiency than those of 3-phenyl dioxazolones bearing electron-withdrawing groups (-CF 3 , -F, -Cl). Moreover, the broad scope of the commercially available secondary amines all reacted smoothly in this transformation, adding to the formation of desired ureas (5n-5v) in good to excellent yields (80-96%). In addition, the primary amine, such as aniline, was also a suitable substrate for reaction with 1a, providing the corresponding product 5w in 52% yield. However, cyclohexylamine (4l) and benzylamine (4m) were not suitable in this transformation to react with 1a to access the corresponding products 5x and 5y. Compared with the previous report [32], our method effectively avoids the harsh conditions of high temperature, showing good sustainability.
corresponding aryl ureas (5a-5m) in moderate to excellent yields (37-98%). In these cases, 3-phenyl dioxazolones bearing electron-donating groups (-CH3, -t Bu, and -OCH3) showed a better reaction efficiency than those of 3-phenyl dioxazolones bearing electron-withdrawing groups (-CF3, -F, -Cl). Moreover, the broad scope of the commercially available secondary amines all reacted smoothly in this transformation, adding to the formation of desired ureas (5n-5v) in good to excellent yields (80-96%). In addition, the primary amine, such as aniline, was also a suitable substrate for reaction with 1a, providing the corresponding product 5w in 52% yield. However, cyclohexylamine (4l) and benzylamine (4m) were not suitable in this transformation to react with 1a to access the corresponding products 5x and 5y. Compared with the previous report [32], our method effectively avoids the harsh conditions of high temperature, showing good sustainability.  To our satisfaction, this method is also suitable for the reaction between 3-(p-tolyl)-1,4,2-dioxazol-5-one 1b and 1,3-diphenylpropane-1,3-dione 6 to give the corresponding amide product in 58% yield (Scheme 4a), which was previously reported in the presence of additional FeCl 3 catalyst [29]. To verify the practicability of this synthetic protocol, the gramscale synthesis of 3a was carried out (for details, see the Supplementary Materials). When the reaction was performed at a 5 mmol scale, the desired product 3a was isolated in 80% yield, indicating that this approach has a good practicability and application prospect (Scheme 4b).
1,4,2-dioxazol-5-one 1b and 1,3-diphenylpropane-1,3-dione 6 to give the corresponding amide product in 58% yield (Scheme 4a), which was previously reported in the presence of additional FeCl3 catalyst [29]. To verify the practicability of this synthetic protocol, the gram-scale synthesis of 3a was carried out (for details, see the Supplementary Materials). When the reaction was performed at a 5 mmol scale, the desired product 3a was isolated in 80% yield, indicating that this approach has a good practicability and application prospect (Scheme 4b). Furthermore, we also evaluated the sensitivity of the reaction of 1a and 2a. Compared with the standard conditions, the changes in concentration, temperature, oxygen level, water level, light intensity, and scale were measured. The yields were measured by 31 P NMR and the yield deviation was calculated (for details, see the Supplementary Materials). Among them, light intensity and oxygen levels are important parameters for the reaction. Moreover, this transformation is moderately sensitive to water. Other parameters, such as concentration and temperature, can be regarded as random errors, which have a negligible impact on reaction efficiency ( Figure S4, Supplementary Materials).
Next, we calculated the E-factor [54,55] and EcoScale scores [56,57] of the chemical process to evaluate the safety, economic, and ecological properties of the method. The results are summarized in Tables S2-S5, Supplementary Materials. As can be seen, the Efactor is extremely low at 0.38 and 0.82, respectively, and the EcoScale penalty is also low, at 21.5 and 15.5. Both parameters reflect the excellent green chemistry metrics of the protocol.
To understand the mechanism of this transformation, a set of control experiments were performed (Scheme 5). The phosphorylation of 4-methylbenzamide 8 with triphenylphosphine 2a was performed to determine whether the N=P bond was formed through the amide intermediate. However, 4-methyl-N-(triphenyl-λ 5 -phosphaneylidene) benzamide 3b was not detected (Scheme 5a). Moreover, intermolecular competition experiments of 1a and 8 were conducted, and only product 3a was obtained with 48% yield (Scheme 5b). These results demonstrated that the phosphorylation of dioxazolones was not conducted through amide intermediates. Furthermore, various radical trapping experiments were conducted (Scheme 5c). When (2,2,6,6-tetramethylpiperidine-1-yl)oxidanyl (TEMPO) was added to the model reaction under standard conditions, the reaction was significantly inhibited. The TEMPO-trapped acyl nitrene adducts were detected by high-resolution mass spectrometry (HRMS), with peaks at 277.1922 m/z. Subsequently, when another radical scavenger, 2,6-di-tert-butyl-4-methylphenol (BHT), was subjected under standard conditions, the reaction was also severely suppressed, indicating a radical process in the phosphorylation of dioxazolone with triphenylphosphine. Then, the radical trapping experiments of 1a and 4a were conducted (Scheme 5c). The decreased yields of product 5a indicated that the transformation also involved a radical process. Furthermore, we also evaluated the sensitivity of the reaction of 1a and 2a. Compared with the standard conditions, the changes in concentration, temperature, oxygen level, water level, light intensity, and scale were measured. The yields were measured by 31 P NMR and the yield deviation was calculated (for details, see the Supplementary Materials). Among them, light intensity and oxygen levels are important parameters for the reaction. Moreover, this transformation is moderately sensitive to water. Other parameters, such as concentration and temperature, can be regarded as random errors, which have a negligible impact on reaction efficiency ( Figure S4, Supplementary Materials).
Next, we calculated the E-factor [54,55] and EcoScale scores [56,57] of the chemical process to evaluate the safety, economic, and ecological properties of the method. The results are summarized in Tables S2-S5, Supplementary Materials. As can be seen, the E-factor is extremely low at 0.38 and 0.82, respectively, and the EcoScale penalty is also low, at 21.5 and 15.5. Both parameters reflect the excellent green chemistry metrics of the protocol.
To understand the mechanism of this transformation, a set of control experiments were performed (Scheme 5). The phosphorylation of 4-methylbenzamide 8 with triphenylphosphine 2a was performed to determine whether the N=P bond was formed through the amide intermediate. However, 4-methyl-N-(triphenyl-λ 5 -phosphaneylidene) benzamide 3b was not detected (Scheme 5a). Moreover, intermolecular competition experiments of 1a and 8 were conducted, and only product 3a was obtained with 48% yield (Scheme 5b). These results demonstrated that the phosphorylation of dioxazolones was not conducted through amide intermediates. Furthermore, various radical trapping experiments were conducted (Scheme 5c). When (2,2,6,6-tetramethylpiperidine-1-yl)oxidanyl (TEMPO) was added to the model reaction under standard conditions, the reaction was significantly inhibited. The TEMPO-trapped acyl nitrene adducts were detected by high-resolution mass spectrometry (HRMS), with peaks at 277.1922 m/z. Subsequently, when another radical scavenger, 2,6-di-tert-butyl-4-methylphenol (BHT), was subjected under standard conditions, the reaction was also severely suppressed, indicating a radical process in the phosphorylation of dioxazolone with triphenylphosphine. Then, the radical trapping experiments of 1a and 4a were conducted (Scheme 5c). The decreased yields of product 5a indicated that the transformation also involved a radical process.
In 2021, Yu and Bao et al., disclosed that FeCl 3 (15 mol%) was required for the imidization of phosphines with dioxazolones under visible light irradiation [29]. While in our case, the transformations worked very well without any other additives. Considering the contamination issues in coupling reactions [58], we reasoned that some iron contamination might be possible in the manufacture of the starting materials. Therefore, the model reaction mixture was analyzed with inductively coupled plasma mass spectrometry (ICP-MS). Consequently, it is found that the Fe content of the reactions for the preparation of phosphinimidic amide (3a) and urea (5a) is approximately 27 ppm and 3 ppm, respectively (for details, see the Supplementary Materials). ICP-MS experiments were also performed on the starting materials of the model reactions (dioxazolone, PPh 3 , and amine), and the results showed that the iron contents of the dioxazolone, PPh 3 , and amine were 123 ppm, 420 ppm, and 0.9 ppm, respectively (for details, see the Supplementary Materials). It is reasoned that iron contamination issues in commercial chemicals are unavoidable during the production and transportation processes. When additional iron catalyst FeCl 3 (5 mol%) was added to the model reaction under standard conditions, the reaction time was shortened and the yield was increased. These results confirmed that this reaction could be facilitated by iron catalysis (for details, see the Supplementary Materials). These results suggest that, although it is not a real transition-meta-free system, it is still a synthetically useful procedure for the synthesis of phosphinimidic amides and ureas, especially from an industrial chemistry standpoint. In 2021, Yu and Bao et al., disclosed that FeCl3 (15 mol%) was required for t zation of phosphines with dioxazolones under visible light irradiation [29]. Wh case, the transformations worked very well without any other additives. Consid contamination issues in coupling reactions [58], we reasoned that some iron co tion might be possible in the manufacture of the starting materials. Therefore, t reaction mixture was analyzed with inductively coupled plasma mass spectrome MS). Consequently, it is found that the Fe content of the reactions for the prepa phosphinimidic amide (3a) and urea (5a) is approximately 27 ppm and 3 ppm tively (for details, see the Supplementary Materials). ICP-MS experiments were formed on the starting materials of the model reactions (dioxazolone, PPh3, and and the results showed that the iron contents of the dioxazolone, PPh3, and am 123 ppm, 420 ppm, and 0.9 ppm, respectively (for details, see the Supplementar als). It is reasoned that iron contamination issues in commercial chemicals are u ble during the production and transportation processes. When additional iron FeCl3 (5 mol%) was added to the model reaction under standard conditions, the Based on these control experiments and previous literature reports, a plausible reaction pathway is proposed in Scheme 6. Initially, the N atom of dioxazolones 1 coordinates with the Fe center to form complex B, which is excited by visible light to generate the highly active iron-aminyl radical C with the release of CO 2 . Subsequently, radical C reacts with triphenylphosphine 2a to form the complex D, followed by a reduction and elimination process to obtain product 3. On the other hand, intermediate C underwent Curtius rearrangement to form intermediate E, which further reacts with secondary amines 4 to obtain product 5.

General Information
All nuclear magnetic resonance (NMR) spectra were recorded on a Bruker Avance 400 MHz in CDCl3 at room temperature (20 ± 3 °C), by using tetramethylsilane as the internal standard. High-resolution mass spectra (HRMS) were conducted on a 3000-mass spectrometer, using Waters Q-Tof MS/MS system with the ESI technique.
Photochemical reactions were carried out under visible light irradiation by a blue LED at 25 °C. The RLH-18 8-position Photo Reaction System manufactured by Beijing Roger Tech Ltd. was used in this system ( Figure S1, Supplementary Materials). Eight 10 W blue LEDs were equipped in this photochemical reactor. The wavelength for blue LED is 430 nm, peak width at half-height is 18.4 nm ( Figure S2, Supplementary Materials). The distance from the light source to the irradiation vessel was approximately 15 mm.

General Experimental Procedures for the Synthesis of (3a-3w)
In a 25 mL reaction tube, dioxazolones 1 (0.2 mmol, 1.0 equiv), organic phosphine substrate 2 (0.2 mmol, 1.0 equiv) in 1 mL CH2Cl2 were allowed to stir with irradiation of 10 W blue LED under N2 atmosphere at room temperature for 24 h. After the reaction, the solvent was evaporated under vacuum, and the residue was purified by column chromatography on silica gel to afford the desired products 3a-3w.

General Information
All nuclear magnetic resonance (NMR) spectra were recorded on a Bruker Avance 400 MHz in CDCl 3 at room temperature (20 ± 3 • C), by using tetramethylsilane as the internal standard. High-resolution mass spectra (HRMS) were conducted on a 3000-mass spectrometer, using Waters Q-Tof MS/MS system with the ESI technique.
Photochemical reactions were carried out under visible light irradiation by a blue LED at 25 • C. The RLH-18 8-position Photo Reaction System manufactured by Beijing Roger Tech Ltd. was used in this system ( Figure S1, Supplementary Materials). Eight 10 W blue LEDs were equipped in this photochemical reactor. The wavelength for blue LED is 430 nm, peak width at half-height is 18.4 nm ( Figure S2, Supplementary Materials). The distance from the light source to the irradiation vessel was approximately 15 mm.

General Experimental Procedures for the Synthesis of (3a-3w)
In a 25 mL reaction tube, dioxazolones 1 (0.2 mmol, 1.0 equiv), organic phosphine substrate 2 (0.2 mmol, 1.0 equiv) in 1 mL CH 2 Cl 2 were allowed to stir with irradiation of 10 W blue LED under N 2 atmosphere at room temperature for 24 h. After the reaction, the solvent was evaporated under vacuum, and the residue was purified by column chromatography on silica gel to afford the desired products 3a-3w.