Photoreactions of Sc 3 N@C 80 with Disilirane, Silirane, and Digermirane: A Photochemical Method to Separate I h and D 5 h Isomers

: Under photoirradiation, Sc 3 N@ I h -C 80 reacted readily with disilirane 1 , silirane 4 , and digermirane 7 to afford the corresponding 1:1 adducts, whereas Sc 3 N@ D 5 h -C 80 was recovered without producing those adducts. Based on these results, we described a novel method for the exclusive separation of I h and D 5 h isomers of Sc 3 N@C 80 . The method includes three procedures: selective derivatization of Sc 3 N@ I h -C 80 using 1 , 4 , and 7 , facile HPLC separation of pristine Sc 3 N@ D 5 h -C 80 and Sc 3 N@ I h -C 80 derivatives, and thermolysis of Sc 3 N@ I h -C 80 derivatives to collect pristine Sc 3 N@ I h -C 80 . In addition, laser ﬂash photolysis experiments were conducted to elucidate the reaction mechanism. Decay of the transient absorption of 3 Sc 3 N@ I h -C 80* was observed to be enhanced in the presence of 1 , indicating the quenching process. When Sc 3 N@ D 5 h -C 80 was used, the transient absorption was much less intensive. Therefore, the quenching of 3 Sc 3 N@ D 5 h -C 80* by 1 could not be conﬁrmed. Furthermore, we applied time-dependent density functional theory (TD-DFT) calculations of the photoexcited states of Sc 3 N@C 80 to obtain insights into the reaction mechanism. a mixture of conformational isomers of 1,4-adducts with C 2 symmetries, as observed for 2 , 10 , 11 , and 12 . To examine the conformational exchange in 9 , variable temperature (VT) 1 H NMR experiments were performed between 303 and 363 K (Figure S10). As expected, the signals coalesced as the temperatures increased to show broad signals at 363 K. The spectrum at 303 K was reproduced when the NMR probe temperature was decreased to room temperature. The matrix-assisted laser desorption ionization time-of-ﬂight (MALDI-TOF) mass spectrum showed that no molecular ion peak expected for 9 − was observed, although 1,1,4,4-tetraphenyl-1,3-butadiene (TPB), 9-nitroanthracene (9-NA), and 2,5-dihydroxybenzoic acid (DHB) were used as matrices. This result is attributable to the low stability of the molecular ion 9 − , as reported for the MALDI-TOF spectrum of 12 [29].


Introduction
Endohedral metallofullerenes (EMFs) have been investigated extensively because of their fascinating structures based on electron transfer from encapsulated metal species to carbon cages [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17]. Among EMFs, trimetallic nitride template endohedral metallofullerenes (TNT-EMFs) constitute a major EMF family for which extensive studies have been conducted to ascertain and apply their remarkable properties [2,9,10]. In fact, Sc 3 N@I h -C 80 has been well studied among TNT-EMFs because of its high production yield [9]. A Photochem 2022, 2 123 few years after the discovery of Sc 3 N@I h -C 80 , the D 5h isomer of Sc 3 N@C 80 was isolated and characterized to demonstrate its higher reactivity than that of the I h isomer [18,19]. For the synthesis of Sc 3 N@C 80 , however, separation of the I h and D 5h isomers by high-performance liquid chromatography (HPLC) is not efficient because the retention times of these isomers are mutually similar using commercial HPLC columns.
To date, many chemical procedures without HPLC separation have been reported for separation of mixtures of fullerenes based on their chemical reactivity differences [20]. For example, Diels-Alder reactions using a cyclopentadiene-functionalized resin followed by retro-addition was used to facilitate the separation of the I h and D 5h isomers of Sc 3 N@C 80 and Lu 3 N@C 80 [21]. Selective complexation procedures were developed using aminosilica and Lewis acids that precipitate with some EMFs [22,23]. These methods enabled separation of Sc 3 N@I h -C 80 in gram quantities. More recently, a selective oxidation procedure using a ferrocenium salt was applied to separate both isomers based on differences in their oxidation potentials. This procedure involves sequential column chromatographic separation of the unreactive Sc 3 N@I h -C 80 and the oxidized Sc 3 N@D 5h -C 80 , which were subsequently recovered by reduction [24]. Although Sc 3 N@D 5h -C 80 and Sc 3 N@C 68 were obtained as the same fraction in this method, it was subsequently reported that Sc 3 N@D 5h -C 80 was separated from Sc 3 N@C 68 based on the predominant reactivity of the latter in methano-derivatization using a tosyl hydrazone reagent [25].
For our ongoing study of fullerene chemistry, disilirane 1 has been used as a versatile derivatizing reagent [15,16]. In general, 1 reacts with EMFs efficiently under visible light to afford the corresponding silylated EMFs. Furthermore, EMFs that exhibit less negative first reduction potentials are reactive toward 1 under both thermal and photochemical conditions [15,16]. These results led us to examine the reactivity of Sc 3 N@D 5h -C 80 for comparison with that of the corresponding I h isomer. This report describes differences in the photochemical reactivities of the I h and D 5h isomers of Sc 3 N@C 80 toward 1 as well as silirane 4 [26,27] and digermirane 7 [28,29]. Very interestingly, Sc 3 N@D 5h -C 80 was found to be photochemically inert toward 1-3, whereas Sc 3 N@I h -C 80 undergoes facile addition reactions under identical conditions. This result indicates a novel photochemical method for the exclusive separation of the I h and D 5h isomers of Sc 3 N@C 80 as follows: (i) selective derivatization of Sc 3 N@I h -C 80 in mixtures of the I h and D 5h isomers, (ii) facile HPLC separation of Sc 3 N@D 5h -C 80 from the derivatized Sc 3 N@I h -C 80 , and (iii) recovery of pristine Sc 3 N@I h -C 80 by thermolysis of its derivative.
In addition, the laser flash photolysis of Sc 3 N@C 80 was conducted to elucidate differences in the reactivities of I h and D 5h isomers. To date, few examples of comparative studies of the photoreactions of fullerene isomers with different cage symmetries have been reported. We reported earlier that C 2v -C 78 undergoes photoreaction with 1 to afford the corresponding silylated C 78 , whereas D 3 -C 78 was inert under the same conditions, indicating a procedure for the separation of C 2v -C 78 and D 3 -C 78 [30]. More recently, the photodynamics of three isomers of Sc 2 C 2 @C 82 have been reported as depending on the different fullerene cage symmetries although the intermolecular reactions of the photoexcited Sc 2 C 2 @C 82 with organic molecules have not been examined yet [31]. For this study, we investigate the mechanistic origins for the difference in the reactivities of the Sc 3 N@C 80 based on cage symmetries using spectroscopic and theoretical studies.

Results and Discussion
2.1. Separation of Sc 3 N@I h -C 80 and Sc 3 N@D 5h -C 80 Using Photochemical Functionalization As described above, earlier reports have shown that Sc 3 N@I h -C 80 reacts with 1 under photolytic conditions to give the corresponding 1,2-adduct 2 and 1,4-adduct 3 [32]. To examine the reactivity of Sc 3 N@D 5h -C 80 , a toluene solution of Sc 3 N@D 5h -C 80 and 1 was irradiated for 20 h using two 500 W halogen lamps (cut off < 400 nm) under an argon atmosphere (Scheme 1). However, Sc 3 N@D 5h -C 80 was found to be inert toward 1, as shown in the HPLC profiles of the photoreaction ( Figure S1 in Supplementary Materials).
This result led us to apply this photoreaction to the separation of the Ih and D5h isomers of Sc3N@C80. When a mixture of the Ih and D5h isomers and 1 in toluene was irradiated for 40 h, the Ih isomer was consumed with the concomitant formation of 2 and 3, whereas the D5h isomer remained intact ( Figure 1). The pristine Sc3N@D5h-C80 and the mixture of 2 and 3 were separated easily from the reaction mixture by preparative HPLC without recycling procedures. Finally, thermal desilylation of the mixture of 2 and 3 was performed at 160-170 °C in o-dichlorobenzene (ODCB) for 20 h. Subsequent HPLC separation afforded pristine Sc3N@Ih-C80 ( Figure S2). These procedures established a straightforward method to separate the Ih and D5h isomers of Sc3N@C80. Such a separation method employing silirane 4 as an alternative derivatizing reagent was also examined based on our earlier reported result obtained from the photochemical addition of 4 to Sc3N@Ih-C80 (Scheme 2) [26,27]. In fact, it was confirmed that 4 did not undergo an addition reaction with Sc3N@D5h-C80 under the photolytic condition used for This result led us to apply this photoreaction to the separation of the I h and D 5h isomers of Sc 3 N@C 80 . When a mixture of the I h and D 5h isomers and 1 in toluene was irradiated for 40 h, the I h isomer was consumed with the concomitant formation of 2 and 3, whereas the D 5h isomer remained intact ( Figure 1). The pristine Sc 3 N@D 5h -C 80 and the mixture of 2 and 3 were separated easily from the reaction mixture by preparative HPLC without recycling procedures. Finally, thermal desilylation of the mixture of 2 and 3 was performed at 160-170 • C in o-dichlorobenzene (ODCB) for 20 h. Subsequent HPLC separation afforded pristine Sc 3 N@I h -C 80 ( Figure S2). These procedures established a straightforward method to separate the I h and D 5h isomers of Sc 3 N@C 80 .
This result led us to apply this photoreaction to the separation of the Ih and D5h isomers of Sc3N@C80. When a mixture of the Ih and D5h isomers and 1 in toluene was irradiated for 40 h, the Ih isomer was consumed with the concomitant formation of 2 and 3, whereas the D5h isomer remained intact ( Figure 1). The pristine Sc3N@D5h-C80 and the mixture of 2 and 3 were separated easily from the reaction mixture by preparative HPLC without recycling procedures. Finally, thermal desilylation of the mixture of 2 and 3 was performed at 160-170 °C in o-dichlorobenzene (ODCB) for 20 h. Subsequent HPLC separation afforded pristine Sc3N@Ih-C80 ( Figure S2). These procedures established a straightforward method to separate the Ih and D5h isomers of Sc3N@C80. Such a separation method employing silirane 4 as an alternative derivatizing reagent was also examined based on our earlier reported result obtained from the photochemical addition of 4 to Sc3N@Ih-C80 (Scheme 2) [26,27]. In fact, it was confirmed that 4 did not undergo an addition reaction with Sc3N@D5h-C80 under the photolytic condition used for  Such a separation method employing silirane 4 as an alternative derivatizing reagent was also examined based on our earlier reported result obtained from the photochemical addition of 4 to Sc 3 N@I h -C 80 (Scheme 2) [26,27]. In fact, it was confirmed that 4 did not undergo an addition reaction with Sc 3 N@D 5h -C 80 under the photolytic condition used for Sc 3 N@I h -C 80 ( Figure S3). As expected, 4 also worked well as a selective carbosilylating reagent for Sc 3 N@I h -C 80 without reaction with Sc 3 N@D 5h -C 80 . Consequently, photoirradiation of a mixture of the I h and D 5h isomers and 4 in toluene for 60 h followed by HPLC separation gave pristine Sc 3 N@D 5h -C 80 and a mixture of 5a, 5b, and 6 [27], as shown in Figure 2. However, photochemical reactivity of 4 was somewhat lower than that of 1 considering the reaction time necessary to consume Sc 3 N@I h -C 80 . In addition, thermal decomposition of the mixture of 5a, 5b, and 6 was performed at 160-170 • C in ODCB for 40 h to give Sc 3 N@I h -C 80 along with a recovered mixture of 5a, 5b, and 6 ( Figure S4). This result indicates that thermal extrusion reactions of the addends in 5a, 5b, and 6 were less efficient than those of 2 and 3, which might reflect the relative stabilities of these adducts.
Photochem 2022, 2, FOR PEER REVIEW 4 Sc3N@Ih-C80 ( Figure S3). As expected, 4 also worked well as a selective carbosilylating reagent for Sc3N@Ih-C80 without reaction with Sc3N@D5h-C80. Consequently, photoirradiation of a mixture of the Ih and D5h isomers and 4 in toluene for 60 h followed by HPLC separation gave pristine Sc3N@D5h-C80 and a mixture of 5a, 5b, and 6 [27], as shown in Figure 2. However, photochemical reactivity of 4 was somewhat lower than that of 1 considering the reaction time necessary to consume Sc3N@Ih-C80. In addition, thermal decomposition of the mixture of 5a, 5b, and 6 was performed at 160-170 °C in ODCB for 40 h to give Sc3N@Ih-C80 along with a recovered mixture of 5a, 5b, and 6 ( Figure S4). This result indicates that thermal extrusion reactions of the addends in 5a, 5b, and 6 were less efficient than those of 2 and 3, which might reflect the relative stabilities of these adducts.
An earlier report described that digermirane 7 is more reactive than its silicon analog 8 toward Lu3N@Ih-C80 under visible irradiation because of the excellent electron-donor property of 7 [29]. This remarkable result led us to evaluate 7 as a third candidate for use as a selective derivatizing reagent for Sc3N@Ih-C80. First, the photoreaction of 7 with Sc3N@Ih-C80 was performed in a manner similar to that used for 1 and 4. During photolysis, HPLC analysis indicated that a product peak developed intensively as the peak of Sc3N@Ih-C80 decreased ( Figure S5). After consumption of Sc3N@Ih-C80, preparative HPLC separation of the reaction mixture afforded the 1,4-adduct 9 as the first example of germylated Sc3N@Ih-C80 derivative. The structure of 9 was established by X-ray crystallographic analysis as described below.
However, as expected, Sc3N@D5h-C80 did not react with 7 under identical conditions for prolonged photoirradiation ( Figure S6). This result led us to apply this germylation reaction to the separation of the Ih and D5h isomers of Sc3N@C80 (Scheme 3). When a mixture of the Ih and D5h isomers and 7 in toluene was irradiated for 20 h, the Ih isomer was consumed with the formation of 9, whereas the D5h isomer remained intact ( Figure 3). The adduct 9 and the pristine D5h isomer were separated easily using preparative HPLC.

Scheme 2. Selective carbosilylation of Sc 3 N@I h -C 80 using 4.
Photochem 2022, 2, FOR PEER REVIEW 4 Sc3N@Ih-C80 ( Figure S3). As expected, 4 also worked well as a selective carbosilylating reagent for Sc3N@Ih-C80 without reaction with Sc3N@D5h-C80. Consequently, photoirradiation of a mixture of the Ih and D5h isomers and 4 in toluene for 60 h followed by HPLC separation gave pristine Sc3N@D5h-C80 and a mixture of 5a, 5b, and 6 [27], as shown in Figure 2. However, photochemical reactivity of 4 was somewhat lower than that of 1 considering the reaction time necessary to consume Sc3N@Ih-C80. In addition, thermal decomposition of the mixture of 5a, 5b, and 6 was performed at 160-170 °C in ODCB for 40 h to give Sc3N@Ih-C80 along with a recovered mixture of 5a, 5b, and 6 ( Figure S4). This result indicates that thermal extrusion reactions of the addends in 5a, 5b, and 6 were less efficient than those of 2 and 3, which might reflect the relative stabilities of these adducts.  An earlier report described that digermirane 7 is more reactive than its silicon analog 8 toward Lu3N@Ih-C80 under visible irradiation because of the excellent electron-donor property of 7 [29]. This remarkable result led us to evaluate 7 as a third candidate for use as a selective derivatizing reagent for Sc3N@Ih-C80. First, the photoreaction of 7 with Sc3N@Ih-C80 was performed in a manner similar to that used for 1 and 4. During photolysis, HPLC analysis indicated that a product peak developed intensively as the peak of Sc3N@Ih-C80 decreased ( Figure S5). After consumption of Sc3N@Ih-C80, preparative HPLC separation of the reaction mixture afforded the 1,4-adduct 9 as the first example of germylated Sc3N@Ih-C80 derivative. The structure of 9 was established by X-ray crystallographic analysis as described below.
However, as expected, Sc3N@D5h-C80 did not react with 7 under identical conditions for prolonged photoirradiation ( Figure S6). This result led us to apply this germylation reaction to the separation of the Ih and D5h isomers of Sc3N@C80 (Scheme 3). When a mixture of the Ih and D5h isomers and 7 in toluene was irradiated for 20 h, the Ih isomer was consumed with the formation of 9, whereas the D5h isomer remained intact ( Figure 3). The adduct 9 and the pristine D5h isomer were separated easily using preparative HPLC. An earlier report described that digermirane 7 is more reactive than its silicon analog 8 toward Lu 3 N@I h -C 80 under visible irradiation because of the excellent electron-donor property of 7 [29]. This remarkable result led us to evaluate 7 as a third candidate for use as a selective derivatizing reagent for Sc 3 N@I h -C 80 . First, the photoreaction of 7 with Sc 3 N@I h -C 80 was performed in a manner similar to that used for 1 and 4. During photolysis, HPLC analysis indicated that a product peak developed intensively as the peak of Sc 3 N@I h -C 80 decreased ( Figure S5). After consumption of Sc 3 N@I h -C 80 , preparative HPLC separation of the reaction mixture afforded the 1,4-adduct 9 as the first example of germylated Sc 3 N@I h -C 80 derivative. The structure of 9 was established by X-ray crystallographic analysis as described below.
However, as expected, Sc 3 N@D 5h -C 80 did not react with 7 under identical conditions for prolonged photoirradiation ( Figure S6). This result led us to apply this germylation reaction to the separation of the I h and D 5h isomers of Sc 3 N@C 80 (Scheme 3). When a mixture of the I h and D 5h isomers and 7 in toluene was irradiated for 20 h, the I h isomer was consumed with the formation of 9, whereas the D 5h isomer remained intact ( Figure 3). The adduct 9 and the pristine D 5h isomer were separated easily using preparative HPLC. Finally, degermylation of 9 was accomplished by thermolysis at 130 • C in ODCB for 15 h ( Figure S7). This thermolysis is apparently more efficient even at lower temperatures than in the cases of 2, 3, 5a, 5b, and 6, probably because of the lower bond energies of C-Ge bonds (242 kcal/mol) than those of C-C and C-Si bonds (348 kcal/mol and 301 kcal/mol, respectively) [33].
Photochem 2022, 2, FOR PEER REVIEW 5 Finally, degermylation of 9 was accomplished by thermolysis at 130 °C in ODCB for 15 h ( Figure S7). This thermolysis is apparently more efficient even at lower temperatures than in the cases of 2, 3, 5a, 5b, and 6, probably because of the lower bond energies of C-Ge bonds (242 kcal/mol) than those of C-C and C-Si bonds (348 kcal/mol and 301 kcal/mol, respectively) [33].

Characterization of Germylated Sc3N@Ih-C80 9
Structural analysis of 9 was conducted based on our earlier studies of the related derivatives 2, 10, 11, and 12 ( Figure 4) [29,32]. The visible-near-IR (vis-NIR) spectrum of 9 closely resembles that of 2 ( Figure 5). In addition, the NMR spectral features of 9 are similar to those of 11 and 12. The existence of two isomeric molecules is inferred because the 1 H NMR spectrum shows two Ge-CH2-Ge methylene groups, respectively, as singlets at 2.33 and 2.57 ppm ( Figure S8). In the 13 C NMR spectrum, two methylene carbon signals Photochem 2022, 2, FOR PEER REVIEW 5 Finally, degermylation of 9 was accomplished by thermolysis at 130 °C in ODCB for 15 h ( Figure S7). This thermolysis is apparently more efficient even at lower temperatures than in the cases of 2, 3, 5a, 5b, and 6, probably because of the lower bond energies of C-Ge bonds (242 kcal/mol) than those of C-C and C-Si bonds (348 kcal/mol and 301 kcal/mol, respectively) [33].

Characterization of Germylated Sc3N@Ih-C80 9
Structural analysis of 9 was conducted based on our earlier studies of the related derivatives 2, 10, 11, and 12 ( Figure 4) [29,32]. The visible-near-IR (vis-NIR) spectrum of 9 closely resembles that of 2 ( Figure 5). In addition, the NMR spectral features of 9 are similar to those of 11 and 12. The existence of two isomeric molecules is inferred because the 1 H NMR spectrum shows two Ge-CH2-Ge methylene groups, respectively, as singlets at 2.33 and 2.57 ppm ( Figure S8). In the 13 C NMR spectrum, two methylene carbon signals

Characterization of Germylated Sc 3 N@I h -C 80 9
Structural analysis of 9 was conducted based on our earlier studies of the related derivatives 2, 10, 11, and 12 ( Figure 4) [29,32]. The visible-near-IR (vis-NIR) spectrum of 9 closely resembles that of 2 ( Figure 5). In addition, the NMR spectral features of 9 are similar to those of 11 and 12. The existence of two isomeric molecules is inferred because the 1 H NMR spectrum shows two Ge-CH 2 -Ge methylene groups, respectively, as singlets at 2.33 and 2.57 ppm ( Figure S8). In the 13 C NMR spectrum, two methylene carbon signals that are attributable to Ge-CH 2 -Ge were observed at 23.04 and 26.81 ppm. The 13 C NMR spectrum also shows two sets of four methyl groups, two sp 3 carbon signals of the I h -C 80 cages, and a total of 102 sp 2 carbon signals that are attributable to the I h -C 80 cages and the Dep ring carbons ( Figure S9). These results indicate the existence of a mixture of conformational isomers of 1,4-adducts with C 2 symmetries, as observed for 2, 10, 11, and 12. To examine the conformational exchange in 9, variable temperature (VT) 1 H NMR experiments were performed between 303 and 363 K ( Figure S10). As expected, the signals coalesced as the temperatures increased to show broad signals at 363 K. The spectrum at 303 K was reproduced when the NMR probe temperature was decreased to room temperature. The matrix-assisted laser desorption ionization time-offlight (MALDI-TOF) mass spectrum showed that no molecular ion peak expected for 9 − was observed, although 1,1,4,4-tetraphenyl-1,3-butadiene (TPB), 9-nitroanthracene (9-NA), and 2,5-dihydroxybenzoic acid (DHB) were used as matrices. This result is attributable to the low stability of the molecular ion 9 − , as reported for the MALDI-TOF spectrum of 12 [29].
cages, and a total of 102 sp 2 carbon signals that are attributable to Dep ring carbons ( Figure S9). These results indicate the existence mational isomers of 1,4-adducts with C2 symmetries, as observed examine the conformational exchange in 9, variable temperatur ments were performed between 303 and 363 K ( Figure S10). As ex lesced as the temperatures increased to show broad signals at 363 K was reproduced when the NMR probe temperature was decreas The matrix-assisted laser desorption ionization time-of-flight (M trum showed that no molecular ion peak expected for 9 -was obs tetraphenyl-1,3-butadiene (TPB), 9-nitroanthracene (9-NA), and acid (DHB) were used as matrices. This result is attributable to molecular ion 9 -, as reported for the MALDI-TOF spectrum of 12  Fortunately, the 1,4-addition structure of 9 was determined ing X-ray crystallographic analysis. Black block crystals of 9 suita were obtained using the liquid-liquid bilayer diffusion method 0 °C. The crystal structure of 9 shows two disordered positions occupancies of 0.72 and 0.28, whereas the digermirane addend This result suggests that the crystal structure of 9 includes a pair is consistent with the NMR observations. Disorder also exists in Sc3N clusters. They involve six locations of Sc atoms with a com cages, and a total of 102 sp carbon signals that are attributable to the Dep ring carbons ( Figure S9). These results indicate the existence of a mational isomers of 1,4-adducts with C2 symmetries, as observed for examine the conformational exchange in 9, variable temperature (V ments were performed between 303 and 363 K ( Figure S10). As expec lesced as the temperatures increased to show broad signals at 363 K. T K was reproduced when the NMR probe temperature was decreased t The matrix-assisted laser desorption ionization time-of-flight (MALD trum showed that no molecular ion peak expected for 9 -was observe tetraphenyl-1,3-butadiene (TPB), 9-nitroanthracene (9-NA), and 2,5 acid (DHB) were used as matrices. This result is attributable to the molecular ion 9 -, as reported for the MALDI-TOF spectrum of 12 [29].  Fortunately, the 1,4-addition structure of 9 was determined firm ing X-ray crystallographic analysis. Black block crystals of 9 suitable f were obtained using the liquid-liquid bilayer diffusion method with 0 °C. The crystal structure of 9 shows two disordered positions in t occupancies of 0.72 and 0.28, whereas the digermirane addend is or This result suggests that the crystal structure of 9 includes a pair of d is consistent with the NMR observations. Disorder also exists in the Sc3N clusters. They involve six locations of Sc atoms with a commo These Sc atom sites fall into two Sc3N sets with occupancies of 0.68 presents orientation of the cage and the Sc3N cluster in 9 with major o Fortunately, the 1,4-addition structure of 9 was determined firmly using the following X-ray crystallographic analysis. Black block crystals of 9 suitable for X-ray diffraction were obtained using the liquid-liquid bilayer diffusion method with CS 2 and hexane at 0 • C. The crystal structure of 9 shows two disordered positions in the I h -C 80 cage with occupancies of 0.72 and 0.28, whereas the digermirane addend is ordered ( Figure S11). This result suggests that the crystal structure of 9 includes a pair of diasteromers, which is consistent with the NMR observations. Disorder also exists in the orientations of the Sc 3 N clusters. They involve six locations of Sc atoms with a common N atom position. These Sc atom sites fall into two Sc 3 N sets with occupancies of 0.68 and 0.32. Figure 6 presents orientation of the cage and the Sc 3 N cluster in 9 with major occupancies. Photochem 2022, 2, FOR PEER REVIEW 7 Front View Top View Figure 6. ORTEP drawings of 9 showing thermal ellipsoids at the 50% probability level at 120 K. The CS2 molecule is omitted for clarity.
We have already reported that the redox properties of silylated and germylated EMFs are altered considerably compared to the corresponding pristine fullerenes because of electron-donating effects of the silyl and germyl groups [16]. The redox property of 9 was verified using cyclic voltammetry (CV) and differential pulse voltammetry (DPV), as shown in Figure 7. The first oxidation (E ox 1) and first reduction (E red 1) potentials of 9 are shifted cathodically by 510 and 250 mV, respectively, relative to those of Sc3N@Ih-C80 as presented in Table 1. In addition, both E ox 1 and E red 1 potentials of 9 are slightly more negative than those of the silylated derivative 2 [32]. Furthermore, the density functional theory calculations of 9 were conducted at the B3LYP/6-31G*~SDD level to obtain a basis for its electronic structure [34][35][36][37]. The optimized structure of 9 was calculated using an initial structure resembling that of the X-ray structure, as shown in Figure 6. The highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) levels of 9 are higher than those of pristine Sc3N@Ih-C80 by 0.68 and 0.45 eV, respectively (Table 1). These changes of the HOMO and LUMO levels are qualitatively consistent with those of the redox potentials of 2 and Sc3N@Ih-C80.  We have already reported that the redox properties of silylated and germylated EMFs are altered considerably compared to the corresponding pristine fullerenes because of electron-donating effects of the silyl and germyl groups [16]. The redox property of 9 was verified using cyclic voltammetry (CV) and differential pulse voltammetry (DPV), as shown in Figure 7. The first oxidation (E ox 1 ) and first reduction (E red 1 ) potentials of 9 are shifted cathodically by 510 and 250 mV, respectively, relative to those of Sc 3 N@I h -C 80 as presented in Table 1. In addition, both E ox 1 and E red 1 potentials of 9 are slightly more negative than those of the silylated derivative 2 [32]. Furthermore, the density functional theory calculations of 9 were conducted at the B3LYP/6-31G*~SDD level to obtain a basis for its electronic structure [34][35][36][37]. The optimized structure of 9 was calculated using an initial structure resembling that of the X-ray structure, as shown in Figure 6. The highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) levels of 9 are higher than those of pristine Sc 3 N@I h -C 80 by 0.68 and 0.45 eV, respectively (Table 1). These changes of the HOMO and LUMO levels are qualitatively consistent with those of the redox potentials of 2 and Sc 3 N@I h -C 80 .

Front View
Top View Figure 6. ORTEP drawings of 9 showing thermal ellipsoids at the 50% probability level at 120 K. The CS2 molecule is omitted for clarity.
We have already reported that the redox properties of silylated and germylated EMFs are altered considerably compared to the corresponding pristine fullerenes because of electron-donating effects of the silyl and germyl groups [16]. The redox property of 9 was verified using cyclic voltammetry (CV) and differential pulse voltammetry (DPV), as shown in Figure 7. The first oxidation (E ox 1) and first reduction (E red 1) potentials of 9 are shifted cathodically by 510 and 250 mV, respectively, relative to those of Sc3N@Ih-C80 as presented in Table 1. In addition, both E ox 1 and E red 1 potentials of 9 are slightly more negative than those of the silylated derivative 2 [32]. Furthermore, the density functional theory calculations of 9 were conducted at the B3LYP/6-31G*~SDD level to obtain a basis for its electronic structure [34][35][36][37]. The optimized structure of 9 was calculated using an initial structure resembling that of the X-ray structure, as shown in Figure 6. The highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) levels of 9 are higher than those of pristine Sc3N@Ih-C80 by 0.68 and 0.45 eV, respectively (Table 1). These changes of the HOMO and LUMO levels are qualitatively consistent with those of the redox potentials of 2 and Sc3N@Ih-C80.

Transient Absorption Spectroscopy of Photoreactions of Sc 3 N@C 80
Laser flash photolysis experiments were conducted to shed light on the differences in the reactivities of the I h and D 5h isomers. Transient absorption spectra were observed in visible (420-760 nm) and NIR (810-1028 nm, 1055-1400 nm) regions using laser excitation at 387 nm. The gaps in the probed spectral range stem from the detection limits of the visible and NIR detectors, gap between 760 and 810 nm, and from the fundamental wavelength of the white light laser source, gap between 1028 and 1055 nm. In toluene, the absorption band of the triplet excited state of Sc 3 N@I h -C 80 ( 3 Sc 3 N@I h -C 80 * ) was observed at λ max 520 nm, as shown in Figures 8 and S12. A report of an earlier study described how the singlet excited state of Sc 3 N@I h -C 80 ( 1 Sc 3 N@I h -C 80 * ) undergoes facile intersystem crossing (ISC) to give 3 Sc 3 N@I h -C 80 * with an absorption band around 500 nm [38]. Upon addition of 200 times equimolar amounts of 7, the decay of 3 Sc 3 N@I h -C 80 * was accelerated considerably, as shown in comparisons of decay plots both at 500 nm (Figure 9a,b) and at 1104 nm (Figure 9c,d). Figure 8. Evolution-associated spectra (EAS) obtained by the photolysis o E = 400 nJ; Figure S12) at their respective relative delay times. The transie excited state of Sc3N@Ih-C80 is shown in black and for toluene in red.   However, when Sc 3 N@D 5h -C 80 was photolyzed under the same conditions, the observed transient absorption peak was much weaker than in the case of Sc 3 N@I h -C 80 , as shown in Figures 10 and S14. Singlet and triplet excited states of Sc 3 N@D 5h -C 80 ( 1 Sc 3 N@D 5h -C 80 * and 3 Sc 3 N@D 5h -C 80 * , respectively) have not been hitherto characterized by spectroscopic studies. Therefore, several mechanistic possibilities should be examined for the weak transient absorption in Figure 10. For example, the photoexcitation of Sc 3 N@D 5h -C 80 might not be as efficient as that of Sc 3 N@I h -C 80 , although the former has a lower but comparable molar extinction coefficient at λ = 387 nm (excitation wavelength in laser flash photolysis) compared to that of the latter, as shown in Figure S16. Alternatively, if it is assumed that the molar extinction coefficient and the lifetime of 3 Sc 3 N@D 5h -C 80 * is not significantly different from those of 3 Sc 3 N@I h -C 80 * , then the weak absorption observed in Figure 10 suggests the low concentration of 3 Sc 3 N@D 5h -C 80 * under photolytic conditions. One possible explanation for this point is that the ISC process from 1 Sc 3 N@D 5h -C 80 * to 3 Sc 3 N@D 5h -C 80 * might be less effective than in the case of Sc 3 N@I h -C 80 . In the presence of 200 times equimolar amounts of 7, no appreciable difference in intensity of the transient absorption at 500 nm was observed because intense absorption of the triplet excited state of toluene hindered the evaluation (Figure 11a,b) [39]. Although slight differences of intensity at 1215 nm were noted upon addition of 7 (Figure 11c,d), they are mostly attributable to the poor signal-to-noise ratio caused by the low intensity in the NIR region. As such, a quantitative analysis of the photoactivity of Sc 3 N@D 5h -C 80 and, in turn, a comparison with Sc 3 N@I h -C 80 is rendered impossible.   Figure S15).

Theoretical Calculations of Photoreactions of Sc 3 N@C 80
To gain insight into the photochemical processes of Sc 3 N@C 80 , we applied timedependent density functional theory (TD-DFT) [40] calculations on the ten lowest excited states including the singlet states (S n ; n = 1, 2, . . . ) and triplet states (T n ; n = 1, 2, . . . ), respectively, for the I h and D 5h isomers of Sc 3 N@C 80 . The corresponding electronic excitation energies are schematized in Figure 12. Major orbital transition configurations are presented in Tables S1 and S2. and Sm) are small, internal conversion (IC) processes from energetically higher singlet states to lower singlet states are enhanced more effectively than the ISC processes. In the case of Sc3N@Ih-C80, there are the triplet excited states T2 and T3, with energies closely approximating those of the lowest singlet excited state S1. The energy differences between S1-T2 and S1-T3 are, respectively, 0.0387 and 0.0381 eV (Figure 12a). In the case of Sc3N@D5h-C80, T1 is energetically lower than S1, which is lower than other triplet states Tn (n > 1). S1 and T1 are different from each other in energy by 0.1239 eV (Figure 12b). However, these energy differences do not explain the efficiencies of the ISC processes in Sc3N@C80. Alternatively, the efficiency of the ISC processes might depend on the spin-orbit coupling (SOC) interaction between the singlet and triplet excited states. The behaviors of encapsulated metal clusters inside the carbon cages might therefore affect enhancement of the SOC interaction. Further understanding of the photoreactivities of Sc3N@C80 must await investigations of those photoexcited states of the corresponding Ih and D5h isomers. It was proposed in an earlier report of the relevant literature that the photoreaction of C60 and 1 proceeds via the electron donor-acceptor interaction between 3 C60* and 1 based on a quenching experiment of 3 C60* by 1 [42]. The Rehm-Weller equation [43,44] for estimating the free energy change ΔG of electron transfer (ET) between 1 and 3 C60* in nonpolar solvents such as toluene exhibited a positive value. Therefore, results suggest that the ET between 1 and 3 C60* is not efficient in non-polar solvents, but that process is regarded as possible because the ΔG value is not so large [42]. Additionally, it has been reported that photoirradiation of C60 and siliranes that possess benzylsilane structures, which are good electron donors, afforded the corresponding adducts [45] In contrast, when siliranes without benzylsilane structures were used as substrates, the photoaddition reaction proceeded very slowly.
Based on these results, the ΔG values for the ET processes from 1 to the excited triplet states of Sc3N@Ih-C80 and Sc3N@D5h-C80 were calculated using the oxidation potential (Eox) of 1 (+0.27 V vs. Fc/Fc + ) [42] and the first reduction potentials (Ered 1 ) of Sc3N@Ih-C80 (-1.26 V vs. Fc/Fc + ) [21] and Sc3N@D5h-C80 (-1.33 V vs. Fc/Fc + ) [21], respectively. Assuming that the IC process from higher excited triplet states Tn (n >1) to the lowest state T1 occurs  According to Kasha's rule [41], when the energy gaps between two singlet states (S n and S m ) are small, internal conversion (IC) processes from energetically higher singlet states to lower singlet states are enhanced more effectively than the ISC processes. In the case of Sc 3 N@I h -C 80 , there are the triplet excited states T 2 and T 3 , with energies closely approximating those of the lowest singlet excited state S 1 . The energy differences between S 1 -T 2 and S 1 -T 3 are, respectively, 0.0387 and 0.0381 eV (Figure 12a). In the case of Sc 3 N@D 5h -C 80 , T 1 is energetically lower than S 1 , which is lower than other triplet states T n (n > 1). S 1 and T 1 are different from each other in energy by 0.1239 eV (Figure 12b). However, these energy differences do not explain the efficiencies of the ISC processes in Sc 3 N@C 80 . Alternatively, the efficiency of the ISC processes might depend on the spin-orbit coupling (SOC) interaction between the singlet and triplet excited states. The behaviors of encapsulated metal clusters inside the carbon cages might therefore affect enhancement of the SOC interaction. Further understanding of the photoreactivities of Sc 3 N@C 80 must await investigations of those photoexcited states of the corresponding I h and D 5h isomers.
It was proposed in an earlier report of the relevant literature that the photoreaction of C 60 and 1 proceeds via the electron donor-acceptor interaction between 3 C 60 * and 1 based on a quenching experiment of 3 C 60 * by 1 [42]. The Rehm-Weller equation [43,44] for estimating the free energy change ∆G of electron transfer (ET) between 1 and 3 C 60 * in non-polar solvents such as toluene exhibited a positive value. Therefore, results suggest that the ET between 1 and 3 C 60 * is not efficient in non-polar solvents, but that process is regarded as possible because the ∆G value is not so large [42]. Additionally, it has been reported that photoirradiation of C 60 and siliranes that possess benzylsilane structures, which are good electron donors, afforded the corresponding adducts [45]. In contrast, when siliranes without benzylsilane structures were used as substrates, the photoaddition reaction proceeded very slowly.
Based on these results, the ∆G values for the ET processes from 1 to the excited triplet states of Sc 3 N@I h -C 80 and Sc 3 N@D 5h -C 80 were calculated using the oxidation potential (E ox ) of 1 (+0.27 V vs. Fc/Fc + ) [42] and the first reduction potentials (E red 1 ) of Sc 3 N@I h -C 80 (−1.26 V vs. Fc/Fc + ) [21] and Sc 3 N@D 5h -C 80 (−1.33 V vs. Fc/Fc + ) [21], respectively. Assuming that the IC process from higher excited triplet states T n (n >1) to the lowest state T 1 occurs rapidly, the energies of the T 1 states were evaluated using TD-DFT calculations as 1.82 eV for Sc 3 N@I h -C 80 and 1.49 eV for Sc 3 N@D 5h -C 80 . As a result, the ∆G values were estimated as +10.36 kcal/mol for Sc 3 N@I h -C 80 and +19.50 kcal/mol for Sc 3 N@D 5h -C 80 [43,46,47]. These values are positive, as in the case of C 60 , but the value of the I h isomer is small, whereas that of the D 5h isomer is nearly twice as large as that of the I h isomer. These results suggest that the photoinduced electron transfer process of 3 Sc 3 N@D 5h -C 80 * should be less efficient than that of 3 Sc 3 N@I h -C 80 * even if they take place. Based on these estimations, the poor electron acceptor property of 3 Sc 3 N@D 5h -C 80 * might decrease its photochemical reactivity toward 1, 4, and 7.
processes between the 3 Sc 3 N@D 5h -C 80 * and 1 are much less likely to occur than those of Sc 3 N@I h -C 80 , judging from the corresponding values of changes of free energies ∆G. Therefore, the photochemical inertness of Sc 3 N@D 5h -C 80 toward 1, 4, and 7 might be partly attributed to the lower electron-acceptor ability of 3 Sc 3 N@D 5h -C 80 * . Further investigations of the dependence of the photochemical reactivities of EMFs on their structures including the carbon cage symmetries and the encapsulated metals are in progress.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/photochem2010010/s1, Materials and General Methods and Complete list of authors for Ref 51, Figure S1: HPLC profiles of the photoreaction of Sc 3 N@D 5h -C 80 with 1, Figure S2: HPLC profiles of the thermolysis of the mixture of 2 and 3, Figure S3: HPLC profiles of the photoreaction of Sc 3 N@D 5h -C 80 with 4, Figure S4: HPLC profiles of the thermolysis of the mixture of 5a, 5b, and 6, Figure S5: HPLC profile of the reaction mixture of Sc 3 N@I h -C 80 and 7, Figure S6: HPLC profiles of the photolysis of Sc 3 N@D 5h -C 80 with 7, Figure S7: HPLC profiles of the thermolysis of 9, Figure S8: 500 MHz 1 H NMR spectrum of 9 recorded at 293 K in CS 2 /CDCl 3 (1:3), Figure S9: 125 MHz 13 C NMR spectra of 9 recorded at 293 K in CS 2 /CDCl 3 (1:3), Figure S10: 500 MHz VT 1 H NMR spectra of 9 recorded in toluene-d 8 , Figure S11: Disorder of C 80 cage and Sc 3 N cluster with occupancies in the crystal of 9, Figure S12: Raw data of all the transient absorption measurements of Sc 3 N@I h -C 80 , Figure S13: Residual measurements of Sc 3 N@I h -C 80 in the absence and the presence of 7, Figure S14: Raw data of all the transient absorption measurements of Sc 3 N@D 5h -C 80 , Figure S15: Residual measurements of Sc 3 N@D 5h -C 80 in the absence and the presence of 7, Figure S16: UV-Visible Spectra of Sc 3 N@I h -C 80 and Sc 3 N@D 5h -C 80 in toluene, Figure S17: Pure toluene reference measurements, Table S1: Ten lowest excited states of Sc 3 N@I h -C 80 calculated by TD-B3LYP/6-31G*~SDD, Figure S18: Selected molecular orbitals of Sc 3 N@I h -C 80 , Table S2: Ten lowest excited states of Sc 3 N@D 5h -C 80 calculated by TD-B3LYP/6-31G*~SDD, Figure S19: Selected molecular orbitals of Sc 3 N@D 5h -C 80 , Table S3: Cartesian coordinates of optimized structures.