Stereodivergent synthesis of chiral fullerenes by [3 + 2] cycloadditions to C₆₀.

A wide range of new dipoles and catalysts have been used in 1,3-dipolar cycloadditions of N-metalated azomethine ylides onto C60 yielding a full stereodivergent synthesis of pyrrolidino[60]fullerenes with complete diastereoselectivities and very high enantioselectivities. The use of less-explored chiral α-iminoamides as starting 1,3-dipoles leads to an interesting double asymmetric induction resulting in a matching/mismatching effect depending upon the absolute configuration of the stereocenter in the starting α-iminoamide. An enantioselective process was also found in the retrocycloaddition reaction as revealed by mass spectrometry analysis on quasi-enantiomeric pyrrolidino[60]fullerenes. Theoretical DFT calculations are in very good agreement with the experimental data. On the basis of this agreement, a plausible reaction mechanism is proposed.

Introduction

During the last two decades, new carbon forms have been discovered by the scientific community with a sensational pace. First, fullerenes as carbon molecular allotropes, then carbon nanotubes and, more recently, graphene have emerged as suitable materials with promising and outstanding properties.[1] Simultaneously, the chemical modification of these new carbon nanoforms has been developed to process and make these materials available for different applications. However, despite the high level of knowledge reached specially in the molecular chemistry of fullerenes, a fundamental issue of chirality control has been scarcely addressed so far. Indeed, the unavailability of most of the known asymmetric induction methods for the activation of fullerene double bonds hampered the direct synthesis of optically active fullerene derivatives and has limited their preparation and their use to just a few examples.[2]

On the other hand, fullerene chirality, far from being something purely exotic, has proven to exert a profound influence on some electronic properties.[3] Thus, control of fullerene derivative chirality is critical to many fields including biomedical devices, molecular electronics and nanoscience where fullerenes find their main potential applications.[4]

The direct synthesis of optically active [60]fullerene derivatives by using chiral metal catalysis constituted a very important milestone.[5] Shortly after, optically pure [70]fullerene derivatives[6] and endofullerenes[7] were prepared with a complete stereocontrol by the use of copper or silver chiral complexes. Organocatalytic cycloaddition of allenoates triggered by chiral phosphines has been another important step toward the direct preparation of chiral carbon nanostructures.[8]

Particularly, chiral metal complexes described proved to be versatile catalysts affording optically pure pyrrolidinofullerenes by a fully stereodivergent cycloaddition of azomethine ylides.[9]

Herein, we have carried out a thorough and systematic study to expand the scope of this highly versatile stereoselective catalytic cycloaddition onto C60 to other catalysts and dipoles, using also complementary mass spectrometry (MS) methodologies. Theoretical calculations at the DFT level by using the two-layered ONIOM (ONIOM: our own n-layered integrated molecular orbital and molecular mechanics) have been used to propose a plausible mechanism by which these reactions occur with such a high level of stereocontrol.

Results and Discussion

New Catalysts for the Synthesis of Chiral Pyrrolidino[60]Fullerenes

The set of four chiral metal complexes, previously described by us, directs the cycloaddition of N-metalated azomethine ylides onto fullerenes with a completely stereodivergent control onto the two new formed stereocenters. Thus, the formed 2-alkoxycarbonyl-5-aryl disubstituted pyrrolidino[60] and [70]fullerenes were obtained in all the possible stereoisomeric forms: copper(II) acetate/3 and silver acetate/4 displayed a cis diastereoselectivity affording enantiomers 2S,5S and 2R,5R respectively,[5] whereas copper(II) triflate/TEA and (R) or (S)-DTBM Segphos (5) led, respectively, to the two trans 2R,5S and 2S,5R stereoisomers in high optical purity.[9] In the search for other even more efficient chiral complexes, we have found in phosphoramidite 6 a suitable ligand, along with copper(II) acetate, able to direct the cycloaddition of fullerenes—in the same way as copper(II) acetate/Fesulphos 3—onto the 2Si,5Re face of the N-metalated azomethine ylide affording (2S,5S) pyrrolidino[3,4:1,2][60]fullerenes 2a–e (2 and 5 are also used for numbering the two prochiral centers in the iminoesters 1, see also below). The enantioselectivity was found to be very similar to copper(II) acetate/3 (Table 1, entries 1–5), and the cis diastereoselectivity was also complete for the more hindered iminoester 1e derived from alanine (entry 5). Remarkably, the cycloaddition occurs also by using a lower catalyst loading (5% mol) while maintaining a high value of reactivity and enantioselectivity (entry 6).

Table 1

1,3-Dipolar Cycloaddition of Different N-Metalated Azomethine Ylides onto C60 with Different Catalytic Systems

entry	α-iminoester	R1	R2	R3	chiral complex	T (yield %)	cis:transa	ee trans(%)[b]	ee cis (%)b
1	1a	4-MeO-Ph	H	Me	Cu(OAc)2/ 6	–15 °C (55)	>99:<1	–	90 (2S,5S)
2	1b	2-thienyl	H	Me	Cu(OAc)2/ 6	–15 °C (60)	>99:<1	–	88 (2S,5R)d
3	1c	4-F-Ph	H	Me	Cu(OAc)2/ 6	–15 °C (51)	>99:<1	–	91 (2S,5S)
4	1d	4-CN-Ph	H	Me	Cu(OAc)2/ 6	–15 °C (54)	>99:<1	–	90 (2S,5S)
5	1e	4-MeO-Ph	Me	Et	Cu(OAc)2/ 6	–15 °C (33)	>99:<1		80 (2S,5S)
6c	1a	4-MeO-Ph	H	Me	Cu(OAc)2/ 6	–15 °C (51)	>99:<1	–	89 (2S,5S)
7	1a	4-MeO-Ph	H	Me	Cu(ACN)4PF6/Et3N/ 7	–20 °C (80)	>99:<1	–	61 (2S,5S)
8	1a	4-MeO-Ph	H	Me	Cu(OTf)2/Et3N/ 7	25 °C (91)	>99:<1	–	53 (2S,5S)
9	1e	4-MeO-Ph	Me	Et	Cu(ACN)4PF6/Et3N/ 7	–25 °C (70)	>99:<1	–	10 (2R,5R)
10	1a	4-MeO-Ph	H	Me	Cu(ACN)4PF6/Et3N/ 8	–30 °C (71)	>99:<1	–	96 (2S,5S)
11	1c	4-F-Ph	H	Me	Cu(ACN)4PF6/Et3N/ 8	–30 °C (69)	>99:<1	–	95 (2S,5S)
12	1d	4-CN-Ph	H	Me	Cu(ACN)4PF6/Et3N/ 8	–30 °C (70)	>99:<1	–	95 (2S,5S)
13	1e	4-MeO-Ph	Me	Et	Cu(ACN)4PF6/Et3N/ 8	–30 °C (65)	>99:<1	–	80 (2S,5S)
14	1j	Cy	H	Me	Cu(OAc)2/ 3	25 °C (39)	70:30	–	33 (2S,5S)
15	1j	Cy	H	Me	AgOAc/ 4	25 °C (25)	71:29	–	42 (2R,5R)
16	1k	CO2Et	H	Et	Cu(OAc)2/ 3	25 °C (0)	–	–	–
17	1f	4-PrO-Ph	H	Me	Cu(OAc)2/ 3	–15 °C (70)	>99:<1	–	90 (2S,5S)
18	1f	4-PrO-Ph	H	Me	AgOAc/ 4	–15 °C (56)	>99:<1	–	84 (2R,5R)
19	1f	4-PrO-Ph	H	Me	Cu(ACN)4PF6/Et3N/ 8	–30 °C (69)	>99:<1	–	94 (2S,5S)
20	1g	Ph	H	Me	Cu(OAc)2/ 3	–15 °C (63)	>99:<1	–	90 (2S,5S)
21	1g	Ph	H	Me	AgOAc/ 4	–15 °C (60)	>99:<1	–	85 (2R,5R)
22	1h	4-(C=C)TMS-Ph	H	Me	Cu(OAc)2/ 3	–15 °C (55)	>99:<1	–	93 (2S,5S)
23	1h	4-(C=C)TMS-Ph	H	Me	AgOAc/ 4	–15 °C (50)	>99:<1	–	85 (2R,5R)
24	1h	4-(C=C)TMS-Ph	H	Me	Cu(ACN)4PF6/Et3N/ 8	–30 °C (71)	>99:<1	–	97 (2S,5S)
25	1i	4-MeO-Ph	H	tBu	Cu(OAc)2/ 3	–15 °C (64)	>99:<1	–	93 (2S,5S)
26	1i	4-MeO-Ph	H	tBu	AgOAc/ 4	–15 °C (60)	>99:<1	–	86 (2R,5R)
27	1l	2-Cl-Ph	H	Me	Cu(OAc)2/ 3	–15 °C (90)	>99:<1	–	95 (2S,5R)d
28	1l	2-Cl-Ph	H	Me	AgOAc/ 4	–15 °C (66)	>99:<1	–	85 (2R,5S)d
29	1l	2-Cl-Ph	H	Me	Cu(OTf)2/Et3N/ 5	25 °C (86)	>99:<1	–	88 (2R,5S)d
30	1m	2-MeO-Ph	H	Me	Cu(OAc)2/ 3	–15 °C (63)	99:1	–	66 (2S,5R)d
31	1m	2-MeO-Ph	H	Me	AgOAc/ 4	–15 °C (66)	99:1	–	83 (2R,5S)d
32	1m	2-MeO-Ph	H	Me	Cu(OTf)2/Et3N/ 5	25 °C (71)	70:30	74 (2R,5R)d	70 (2R,5S)d
33	1m	2-MeO-Ph	H	Me	Cu(ACN)4PF6/Et3N/ 8	–30 °C (72)	>99:<1		87 (2S,5R)d
34	1n	3-Cl-Ph	H	Me	Cu(OAc)2/ 3	–15 °C (75)	98:2	-	11 (2S,5S)
35	1n	3-Cl-Ph	H	Me	AgOAc/ 4	–15 °C (47)	>99:<1	-	87 (2R,5R)
36	1n	3-Cl-Ph	H	Me	Cu(OTf)2/Et3N/ 5	25 °C (75)	50:50	86 (2R,5S)	77 (2R,5R)
37	1o	3-MeO-Ph	H	Me	Cu(OAc)2/ 3	–15 °C (50)	99:1	-	90 (2S,5S)
38	1o	3-MeO-Ph	H	Me	AgOAc/ 4	–15 °C (77)	99:1	–	86 (2R,5R)
39	1o	3-MeO-Ph	H	Me	Cu(OTf)2/Et3N/ 5	25 °C (55)	1:99	95 (2R,5S)	–

Determined by 1H NMR.

Determined by chiral HPLC. The absolute configuration is indicated in brackets.

5% mol load of catalyst.

For compounds 2b, 2l, and 2m the configurations shown represent the change in prority of the substituent at position 5 of the newly created pyrrolidine ring relative to the other entries of this table.

We have also tested noncommercially available chiral complexes, recently reported by some of us, featuring both the adjustable chirality of unnatural prolines and planar chirality of the ferrocenyl skeleton such as catalysts 7 and 8.[10] Thus, chiral ligand based on N-unsubstituted d-proline 7—that gave rise to an exo diastereoselectivity when used with olefins—neither induce high values of enantioselectivity (entries 7–9) nor showed a trans diastereoselectivity when used with copper triflate on fullerene at room temperature.[11] However, in sharp contrast, the configuration of ligand 8 afforded an enantioselectivity that was found to be more favorable. Indeed, the chiral complex formed from the salt Cu(ACN)4PF6 and 8 directs the cycloaddition toward the enantiomers 2S,5S with enantioselectivities over 95% and with good yields (entries 10–12 and see below). For the more substituted iminoester 1e, derived from alanine, the system Cu/8 allowed a complete cis diastereoselectivity and a significant yield improvement, although we failed to pass the 80% of enantioselectivity showed with other ligands (entry 13).

Influence of the α-Iminoester Nature and Substitution Pattern

In order to evaluate the effect of the nature and substitution pattern on the stereochemical outcome, a variety of α-iminoesters have been prepared from different aldehydes and aminoesters.

The reaction does not reach a good stereocontrol with aliphatic dipoles such as α-iminoesters 1j and 1k (aliphatic iminoesters are often unstable and typically they undergo hydrolysis) affording poor selectivities (entries 14–15) and reactivities (entry 16).

On the other hand, phenyl iminoesters or four-substituted phenyl iminoesters cycloadd to C60 mantaining very high values of cis diastereoselectivity. Both enantiomers have been obtained in high optical purity depending on the catalyst used (entries 17–26). Particularly, pyrrolidino[60]fullerenes 2h,i, featuring a tert-butyl ester or a 4-trimethylsilylphenyl group, were obtained with 93% of enantiomeric excess (ee) (2S,5S enantiomers) with the ligand 3 (entries 22 and 25) and 85–86% (2R,5R enantiomers) with the complex silver-BPE 4 (entries 23 and 26). Once more, the new chiral complex copper/8 was found to be the best performer with enantioselectivities as high as 97% (entry 24).

All the aforementioned aromatic α-iminoesters tested so far as dipoles precursors have different substituents in para position. In this context, we decided to synthesize other aromatic α-iminoesters bearing different electron-releasing and electron-withdrawing substituents in ortho and meta positions of the phenyl ring to evaluate their potential influence on the stereoselectivity of the reaction. As summarized in Table 1, the reaction worked with all these new α-iminoesters when applying the respective catalysts to obtain either the cis or trans adduct but with remarkable differences, depending upon the substituent nature and position.

Ortho-methoxy substituted aryl iminoester, 1m, gave rise to slightly lower diastereoselectivity and moderate enantioselectivity as a result of a higher steric demand, affording better ee values with the system Ag/4 (83% versus 66%, entries 30–31). Furthermore, in sharp contrast to the other iminoesters,[9] catalyst Cu/5 is not able to switch the diastereoselectivity toward the trans adduct, probably as a result of the metal coordination with three anchor groups in the dipole that yields a cis stereoisomer (entry 32, see Figure 1). Nevertheless, chiral ligand 8 along with Cu(ACN)4PF6 proved to be very effective also with this substrate affording cis pyrrolidine (2S,5R)-2m with 87% of ee (entry 33). Interestingly, the cycloaddition of 2-chlorophenyl α-iminoester 1l occurs with a complete cis selectivity and good enantioselectivity regardless the catalytic system used (entries 27–29). Thus, Cu/3 catalyzed cycloaddition of 1l onto C60 occurs with excellent selectivity and yield (entry 27). Finally, meta substituted aryl iminoesters 1n,o showed an intermediate behavior between ortho and para substituted analogues (entries 34–39). Along with nitrogen atom and ester moiety, chlorine leads to a stronger binding of the dipole to the metals with respect to the more hindered methoxy group (Figure 1). The same reason could explain the lack of selectivity when the chlorine atom is located in the meta position (dipole 1n) because it is too far to coordinate the metal but can hinder the stereodifferentiation (11% ee). The different electronic nature of the methoxy group does not have any significant impact on the selectivity when located in the meta position affording the same ee than that obtained in the para position (dipole 1a).

Figure 1

W conformation of the N-metalated azomethine ylide with α-iminoester 2-methoxy (X = OMe) or 2-chloro (X = Cl) aryl substituted.

Double Chiral Induction

The majority of 1,3-dipolar cycloaddition reactions imply the use of α-iminoesters, providing pyrrolidines with 2-carboxylate substitution. The use of α-iminoamides as azomethine ylides precursors has been considerably less explored.[12] Taking this into account and with the aim of exploring the interplay of the chirality of the catalytic system described and the chirality of a substrate, we have carried out some experiments of double asymmetric induction with chiral α-iminoamides as dipoles. Thus, iminoamides 9 and 10 were prepared from commercially available (R) and (S)-α-methylbenzylamine repectively by reaction of the primary amine with Boc-glycine, subsequent deprotection with HCl, and finally dehydration with the corresponding aldehyde. All the reactions are nearly quantitative and the overall yield is 88% (see SI).

Iminoamides 9 and 10 could be used as substrates for a metal catalyzed 1,3-dipolar cycloaddition over C60, and two diastereoisomers bearing three different chiral centers were synthesized in good yield and in different ratios, depending upon the catalytic system used (Table 2). In fact, when the reaction with 9 was catalyzed by copper(II) acetate and Fesulphos 3 as chiral ligand at −15 °C, pyrrolidine 12, featuring a 2S,5S configuration [and obviously (R) in the benzylic carbon atom of the amide], was obtained with a 92% of diastereomeric excess (de). This product is consistent with a coherent effect of the chirality of the catalyst with the chirality of the substrate that directs the cycloaddition on the 2Si,5Re face of the dipole derived from 9 (Table 2, entry 1). On the other hand, despite the sense of asymmetric induction of silver(I) acetate/[(R,R)-BPE] 4 as catalyst was found to be predominant (compound 11 with a configuration 2R,5R as main product). A mismatched effect of the chirality of 9 led to a mixture of both diastereoisomers 11 and 12 with only a 60:40 of diastereomeric ratio. Analogously, carrying out the reaction with the opposite enantiomer of the α-iminoamide (10) a mismatched effect was found with the complex Cu/3 (entry 3). These experimental findings clearly indicate that the asymmetric induction sense of the chiral N-metaled azomethine ylides is not strong enough to overcome the effect of substrate chirality and to give rise to high levels of enantioselectivity, regardless of the effect of the iminoamide stereocenter. Thus, the metal-chelated dipole (whether it be copper/3 or silver/4) in its predictable secondary amide conformation - with the chiral methine hydrogen eclipsing the carbonyl group - slightly prefers approaching of C60 from the side of the methyl group rather than the side of the phenyl group.

Table 2

Influence of an Additional Stereocenter in the Selectivity of the 1,3-Dipolar Cycloaddition of Chiral α-Iminoamides and C60

entrya	α-iminoamide	chiral catalyst	de (%)b	yield (%)
1	9	Cu(OAc)2/ 3	92(12)	70
2	9	AgOAc/ 4	20 (11)	66
3	10	Cu(OAc)2/ 3	17 (14)	63
4	10	AgOAc/ 4	90 (13)	64

Conditions (cycloaddition): metal salt (10% mol), chiral ligand (10% mol), anhydrous toluene, −15 °C, 3 h.

Determined by HPLC; indicating in brackets the major product.

Mass Spectrometry Studies

Along with its classical use for the characterization of organic compounds, electrospray ionization mass spectrometry (ESI/MS) has proven to be a very useful tool for mechanistic investigations due to its ability to detect ions and also transient intermediates involved in different reactions.[13] ESI/MS technique has also been successfully used for the analysis, detection, and investigation of reactions involving fullerenes highlighting its usefulness to develop a procedure to modulate the 1,3-dipolar retrocycloaddition of pyrrolidinofullerenes.[14−16] Furthermore, ESI/MS has also proven to be a very useful tool for fast screening of asymmetric reactions by using mass labeled enantiomers.[17]

In this context and to gain some insight into the reaction mechanism, MS experiments were first conducted on the reaction mixtures of the different catalytic systems employed. All of them showed the peaks corresponding to the respective catalytic systems (metal + ligand in different oxidation state) and the reactive chiral complex (metal + dipole + ligand in different oxidation state), but none of them showed the presence of acetate anion in the complex as previously postulated (the absence of anion acetate in the reactive complex was also confirmed by theoretical calculations, see below).[5]

Furthermore, we have also evaluated the intrinsic enantioselectivity of one of our systems (Ag/4) by ESI/MS analysis of the corresponding back reaction. The reversibility of the 1,3-dipolar cycloaddition of azomethine ylides (retro-Prato reaction)[18−20] and the usefulness of ESI/MS in the study of the retrocycloaddition reactions of fullerene derivatives enable the use of reverse reaction screening by following the recent methodology reported by Pfaltz for the related retro-Diels–Alder reactions.[21] This approach is based on the use of “an artificial racemate” formed by two mass labeled (pseudo)enantiomers with opposite absolute configuration and a slight mass difference in such a way that they behave as real enantiomers. Thus, the capacity to be transferred to the gas phase and the peak intensities for all the pseudo-enantiomeric species are assumed to be the same.[17,21]

Thus, we have prepared a pseudoracemic mixture using equimolar amounts of 2-methoxycarbonyl-5-(4-methoxyphenyl)pyrrolidino[60]fullerene 2a with a 2R,5R configuration (obtained with the complex (R,R)-BPE 4 and silver acetate) and its pseudoenantiomer 2-methoxycarbonyl-5-(4-propoxyphenyl)pyrrolidino[60]fullerene 2f with the opposite 2S,5S configuration (this compound was prepared using copper-Fesulphos 3 as catalyst). Since the different substituents of 2a and 2f (the difference in the mass is 28 units) are located in a far para position, the same behavior is expected for both pseudo-enantiomers.

Thus, the pseudoracemic mixture, dissolved in toluene/acetonitrile along with 20% of chiral ligand (R,R)-BPE 4 and silver acetate, was submitted to ESI-MS experiments at room temperature (Figure 2). This mass spectrum displayed peaks of low intensity both for the clusters at m/z 1556 and 1572 of different oxidized forms of 15a ([2a+ (R,R)-BPE 4 + Ag + nO]+ and at m/z 1584 and 1600 for 15f ([2f+ (R,R)-BPE 4 + Ag + nO]+), being these latter present in double amounts (see also SI).

Figure 2

Retrocycloaddition of the pseudoracemic mixture formed by equimolecular amount of (R,R)-2a and (S,S)-2f. ESI/MS analysis revealed a faster formation of species derived from (R,R)-2a (k–1 > k–2) as a result of the silver/(R,R)-BPE 4 catalysis, analogously to that found for the direct reaction.

The presence of two clusters, at m/z 817–819 and at m/z 845–847 corresponding to the loss of C60 from 15a and 15f, respectively, could be justified only through a silver catalyzed retrocycloaddition process (the isotopic pattern also confirms the presence of silver in the cluster, see Figure 3).

Figure 3

Products derived from catalyzed retrocycloaddition revealed by ESI/MS with their relative intensities. (a) Peaks derived from the loss of C60. Inset: enantiomeric ratio determined by HPLC for the direct addition of 1a (first peak refers to the (R,R) stereoisomer) catalyzed by Ag/4 at room temperature. (b) Clusters of the homodimer of 1a (red) and heterodimer of 1a and 1f (green) along with catalyst 4.

Taking into account that the transition states for the forward and backward reactions are identical (principle of microscopic reversibility), the peaks intensity ratio encountered during the retrocycloaddition reflects the enantioselectivity of the catalyst (indeed we assume the same chemical behavior for 1a,f and 2a,f that differ only in the methoxy or propoxy para substituent).

Thus, as expected, silver/(R,R)-BPE 4 chiral complex that affords (R,R)-2a in the direct reaction with a enantiomeric ratio of 83:17 at room temperature (Figure 3 inset), displays the same direction of enantioselectivity also for the back reaction. Thus, as appears in the different intensity of the peaks derived by loss of C60 (16a and 16f, Figure 3a), silver/(R,R)-BPE 4 promotes a faster retrocycloaddition of 15a (R,R) with respect to 15f (S,S), albeit with a lower enantioselectivity (ESI/MS detects a 67/33 ratio for the peaks at 817 (16a-H2) and 845 (16f-H2)).

The same sense of enantioselectivity of the chiral complex Ag+/4 in the retrocycloaddition reactions was confirmed also by the prevalent formation of 1a with respect to 1f in a ratio similar to that of the direct reaction. Indeed, the final products of the retrocycloadditions 1a and 1f have been found as homodimer and heterodimer along with catalyst 4 at m/z 935 (2 × 1a-H + (R,R)-4 + O) and at m/z 963 (1a + 1f-H + (R,R)-4 + O), while the corresponding 1f homodimer (m/z 991) is not observed. It is worthy to note that such retrocycloaddition products are not observed in the absence of silver, and therefore, they could only be formed through a metal catalyzed process.

Theoretical Calculations

In order to shed some light on the origins of the high stereocontrol obtained in the synthesis of pyrrolidino[60]fullerenes employing the aforementioned catalytic systems, DFT calculations have been performed on selected systems. Initially, geometry optimizations of the metalated azomethine ylides derived from imine 1g and (R,R)-BPE 4-Ag(I), (Rp)-Fesulphos 3-Cu(II), or 8-Cu(I) catalytic systems were carried out (Figure 4).

Figure 4

Main geometric features and relative energies (in kcal mol–1) of azomethine ylides derived from imine 1g and (A) (R,R)-BPE-Ag(I), (B) (Rp)-Fesulphos-Cu(II), or (C) 8-Cu(I) catalytic systems computed at (A,C) M06/LANL2DZ//B3LYP/LANL2DZ + ZPVE or (B) UM06/LANL2DZ//UB3LYP/LANL2DZ + ZPVE. White surfaces represent the corresponding solvent accessible surface areas.

Our calculations predict that only one conformation of the complexed azomethine ylide (R,R)-BPE-Ag(I)-1g is energetically available. In this 1,3-dipole, the preferred tetrahedral environment around the silver atom generates an effective blockage of the (2Si,5Re) prochiral face. On the other hand, when a less constrained chiral ligand is employed, two possible azomethine ylides conformations are accessible (Figure 4B,C). Using both ligands (Rp)-Fesulphos 3-Cu(II) or 8-Cu(I), the 1g conformation was found to be the most stable as a result of the stabilizing interaction between the carboxy group of the α-iminoester moiety and the Fe atom.[22] The energy differences with respect to the alternative 1g conformation are large enough (2.6 and 4.9 kcal mol–1, respectively) to ensure that only 1g conformations are relevant in the catalytic cycles associated with both catalytic systems. In the 1g conformation, the catalytic system promotes the effective blockage of the (2Re,5Si) prochiral face (Figure 4B,C). As a consequence, (R,R)-BPE 4 catalytic system induces an opposite stereochemical outcome with respect to those obtained when (Rp)-Fesulphos-Cu(II) or 8-Cu(I) were used.[23]

We have also located all the stationary points associated with the 1,3-dipolar cycloaddition of the previously described metalated ylides and C60 by means of two layer ONIOM calculations (See SI). It is known that the reaction of the metal catalyzed 1,3-dipolar cycloaddition presents a stepwise mechanism, in which the first step is the responsible for the observed stereocontrol (see also Figures S26–28).[24] The main geometrical features and relative energies of the least energetic transition structures associated with the first step in the formation of pyrrolidino[60]fullerenes 2g are gathered in Figure 5.

Figure 5

Main geometrical features and relative energies (in kcal mol–1) of the transition structures associated with the first step of the reaction between C60 and (A) (R,R)-BPE-Ag(I)-1g, (B) (Rp)-Fesulphos-Cu(II)-1g, or (C) 8-Cu(I) metallic azomethine ylides computed at (A,C) M06/LANL2DZ//ONIOM(B3LYP/LANL2DZ:PM6) + ZPVE or (B) UM06/LANL2DZ//ONIOM(UB3LYP/LANL2DZ:UFF) + ZPVE levels of theory. Low-level atoms in the ONIOM partitions are shown as transparent ball and sticks representations. Bond lengths are given in Å. Selected hydrogen atoms of the ligands are omitted for clarity.

Our results show that (R,R)-BPE-Ag(I), (Rp)-Fesulphos-Cu(II) or 8-Cu(I) catalysts induce a critical effect on the stereochemical outcome of the obtained pyrrolidino[60]fullerenes 2g. When (R,R)-BPE-Ag(I) is used as catalyst, (R,R)-BPEAg-TS is 1.3 kcal mol–1 more stable than (S,S)-BPEAg-TS. This energy difference stems from in the effective blockage of the (2Si,5Re) prochiral face of the reactive metalated azomethine (R,R)-BPE-Ag(I)-1g which results in a longer distortion of (S,S)-BPEAg-TS (Figure 4). On the other hand, the stereocontrol observed when (Rp)-Fesulphos-Cu(II) system is employed is the opposite one. In fact, (R,R)-FesulphosCu-TS-b was found to be 2.7 kcal mol–1 less stable than (S,S)-FesulphosCu-TS-a. The same stereochemical outcome was obtained when 8-Cu(I) catalytic system was considered. In this latter case (R,R)-8-Cu-TS-a was found to be 2.6 kcal mol–1 less stable than (S,S)-8-Cu-TS-a. Both results are consistent with the stereochemical outcome experimentally observed and predicted from the geometry of the respective starting azomethine ylide complexes.

It is interesting to note that in the case of densely functionalized ligand 8 the coordination patterns of Cu(I) are quite different on going from the azomethine ylides to the possible transition structures. Thus, in the case of 8-Cu(I)-1g the Cu(I) center interacts with the PPh2 and N-Me groups of the chiral ligand as well as with the nitrogen atom and the carboxy group of the azomethine ylide. This structure is distorted in (S,S)-8-Cu-TS-a, for which the Cu–N(Me) interaction is lost as a consequence of the approach of C60 along the less hindered (2Si,5Re) face. In addition, in this latter TS, the ferrocenyl moiety lies away from the methoxycarbonyl group in contrast with the geometry computed for 8-Cu(I)-1g (vide supra). These results indicate that the geometries of the transition structures of these catalyzed cycloadditions cannot be anticipated from the geometric features of the N-metalated azomethine ylides.

Conclusions

In summary, we report a fully stereodivergent methodology for the synthesis of chiral pyrrolidinofullerenes by the correct choice of a wide and easily available arsenal of chiral ligands, metals, and iminoesters or iminoamides. The complete control of the stereochemical outcome could be affected by the chirality of the dipole as revealed by experiments of double induction. ESI/MS experiments on a quasiracemic mixture showed a similar value of enantioselectivity and proved to be a suitable methodology for the screening of chiral catalysts for the azomethine cycloaddition onto fullerenes. Finally, DFT calculations shed light on the origin of the stereoselectivity displayed by the catalysts copper/3, silver/4, and copper/8. The experimental findings have been accounted for by the preferential attack of the fullerene as dienophile from the less hindered face of the previously formed metal complex.