Microwave-Assisted 1,3-Dipolar Cycloaddition of Azomethine Ylides to [60]Fullerene: Thermodynamic Control of Bis-Addition with Ionic Liquids Additives

The cycloaddition of azomethine ylides to [60]fullerene (C 60 ) has been studied in ortho-dichlorobenzene ( o -DCB) by evaluating the impact of an ionic liquid (IL) additive. The solvent effect has been addressed by evaluating the activation parameters of the cycloaddition and the boosting effect of the microwave (MW) induced dielectric heating. The IL additive plays a twofold role of stabilizing the dipolar ylide intermediate and favoring the retro-cycloaddition at high temperature regime. Under the conditions explored, a combined kinetic and thermodynamic preference favors the selective formation of trans bis-full-eropyrrolidine regioisomers, in agreement with the DFT computational analysis.


Introduction
Fullerene functionalization by the azomethine ylide cycloaddition, yielding fulleropyrrolidines (FPs), [1] has significantly contributed to the synthesis of new molecular materials for transformative research in strategic fields, including solar energy conversion and nanomedicine. [2]FPs synthesis tolerates many functional groups and can be accelerated by microwave (MW) irradiation. [3,4]In this case, toluene, benzene or odichlorobenzene (o-DCB) have been used as solvents, whereby the latter displays a relatively higher boiling point and medium MW-absorbing properties (with a tangent loss, tan δ = 0.280, higher than that of water, tan δ = 0.123). [5]Moreover, we have previously described the preparation of FPs in a mixture of ionic liquids (ILs) and o-DCB, showing that higher reaction rates and conversions can be achieved in such conditions. [6,7]The use of ILs is aimed at addressing the replacement of hazardous volatile organic solvents, being instrumental for fast and selective MWheating by ionic conduction, with negligible vapor pressure and safer protocols. [8]n this work, we have investigated the effect of MW irradiation on the cycloaddition of azomethine ylides to [60]  fullerene, in o-DCB/IL mixtures (Scheme 1), in order to address the thermal effect of MW-irradiation on reaction rates and selectivity.When the 1,3-dipolar cycloaddition is carried out on the fullerene sphere, indeed, not only the mono-adduct (m-FP) is obtained, but a complex mixture of poly-adducts (p-FP) can also form at higher fullerene conversion.Thus, 8, 46 and 262 FP regioisomers can be expected for symmetrical addends respectively for the bis, tris and tetrakis-addition at the fullerene core.
FP bis-adducts, in particular, are of great interest in a wide number of applications in materials science. [9]However, their isolation and purification may be difficult and time-consuming.In this context, several approaches have been developed to increase the selectivity of bis-addition to the fullerene core. [10]hile fullerene cyclopropanation has been successfully controlled with tether-directed covalent methods, supramolecular masks or photo-triggers, [11] to the best of our knowledge, there are no precedent reports on how to direct and optimize the bisaddition selectivity during FPs synthesis.
Our results address: (i) the impact of MW-assisted heating on the cycloaddition kinetics and selectivity in o-DCB and IL containing media; (ii) optimization of the synthetic protocol towards the bis-substitution of the fullerene core; (iii) DFT calculations to interrogate mechanism and the thermodynamic drivers leading to stable FP bis-adducts.

Results and Discussion
The reactivity of [60]fullerene with sarcosine and heptaldehyde was initially used as a benchmark reaction to screen the influence of the synthetic conditions.In particular, we focused on the use of MW-activated IL phases to accelerate the azomethine ylide cycloaddition and control the formation of fullerene poly-adducts, p-FP 1 (Scheme 1).
IL nature and content, as well as [60]fullerene concentration were shown to affect both the yield and the selectivity of the reaction. [7]In particular, long chain ILs are effective to improve fullerene dispersion.Therefore, a 1-methyl-3-n-octyl imidazolium tetrafluoroborate (OmimBF 4 ): o-DCB = 1 : 3 mixture was used to prevent clustering/aggregation phenomena, and the [60]fullerene concentration was set to ensure its complete solubilization, while yield optimization was explored as a function of the relative ratio between azomethine ylide precursors, sarcosine and heptaldehyde.A faster conversion of pristine [60]fullerene to m-FP 1 was thus observed at a fullerene concentration of 7 mM, in the presence of two and four equivalents of sarcosine and heptaldehyde, respectively.
ILs are known to play a role in the stabilization of the incipient 1,3-dipole intermediate, however favoring cyclo-reversion [6,[12][13][14] under MW-induced dielectric heating.Indeed, the cyclo-reversion of FPs can be readily achieved in pure ILs under MW irradiation by unlocking the pyrrolidine ring to backrelease pristine [60]fullerene. [6]Noteworthy, when o-DCB is introduced as a co-solvent, retro-cycloaddition is remarkably slowed down, which provides a facile control over the process reversibility giving access to a thermodynamic equilibration of the reaction.
The effect of the temperature on the reaction was initially investigated under conventional heating, in a temperature range between 343 and 443 K, in the absence of MW irradiation, using o-DCB, either alone or with the IL additive, as the solvent phase.
Table 1 collects the first order rate constants (k obs ) obtained for [60]fullerene conversion under the explored reaction conditions (Scheme 1).
The experimental kinetics show the progress of [60] fullerene cycloaddition, which generally occurs with consecutive formation of m-FP 1 and p-FP 1 products (Figures S1-S12).
A typical kinetic trace is reported in Figure 1a for the reaction carried out at 423 K (entry 5 in Table 1) in o-DCB, showing a [60]fullerene conversion up to 58 % in 2 h, with formation of m-FP 1 (35 %) and a mixture p-FP 1 (23 %).When IL is added to the solvent phase, the cycloaddition kinetics is remarkably faster so that formation of p-FP 1 becomes prevalent above 400 K (Figures S10-S12).In Figure 1b is reported the corresponding kinetic trace for the reaction with IL additive, at the same temperature (423 K), showing the prevalent formation p-FP 1 (52 %) vs. m-FP 1 (39 %) in 60 min.Noteworthy, the IL additive leads to a 4-fold increase of the reaction rate with respect to the control reaction in pure o-DBC at 443 K (R obs = 17.3 and 3.77 μmol L À 1 s À 1 , respectively), with [60]fullerene conversion rising up to 64 % in 5 minutes (Table 1, entry 12 vs.entry 6).
The Eyring plot in Figure 2 reports the dependence of the reaction kinetic constant (k obs ) on the temperature conditions (Table 1), according to Equation (1): where ΔH � and ΔS � are the enthalpy and entropy of activation, respectively, T is the reaction absolute temperature, R is the gas constant (1.986 Kcalmol À 1 ), K is the transmission coefficient (assumed to be 1), k B = Boltzmann constant, h = Planck constant.
In all cases, k obs values are obtained considering first order kinetics for the [60]fullerene reactant, in agreement with the experimental concentration-time profiles and consistent with the generally accepted mechanistic scheme (Table 1 and Scheme S1 in Supporting Information).
Inspection of graphs in Figure 2 shows the line slopes (ca.4200 K À 1 ) yielding similar values of activation enthalpy, ΔH � = 8.4 � 0.2 and 8.0 � 0.8 Kcalmol À 1 ascribed to o-DCB and o-DCB/ IL solvent conditions, respectively.Similarly, the activation entropies are calculated from the graph intercepts, showing large negative values, that turn out to be slightly mitigated by the addition of the IL additive (ΔS � = À 55.2 � 0.5 and À 52.3 � 1.7 calK À 1 mol À 1 for o-DCB and o-DCB/IL, respectively).The almost identical enthalpy activation barriers together with the negative activation entropy are consistent with a concerted 1,3dipolar cycloaddition mechanism.The rate enhancement, observed in the presence of ILs, is therefore explained by a less negative activation entropy in the presence of ILs, due a favorable desolvation of the cyclic transition state with respect to the zwitterionic azomethine ylide reagent (Scheme S1). [15]ith the aim to exploit the ILs ionic phase as excellent MWabsorbers, the reaction was performed under MW irradiation (50 W), flowing compressed air to avoid possible overheating effects (Table 2).
MW-assisted kinetics were instrumental to address the temperature conditions operating under MW induced dielectric heating.The operating temperature can be determined indirectly by extrapolation of the Eyring plot in Figure 2 (T calc in Table 2). [16,17]he combined use of IL and MW is responsible for a remarkable initial rate acceleration, reaching up to 40.1 μM s À 1 (Figure S13).This kinetic effect (k obs = 573 × 10 À 5 s À 1 ) corresponds to a temperature read-out T calc = 471 K, as provided by the   m-FP 1 (%) [c] Sel [d] R obs (μM s À 1 ) (k obs × 10 5 /s À 1 ) T calc (K) [e] 1 o-DCB 9 6.6 Eyring linear relationship in Figure 2 (light red circle).With respect to the control reaction, performed in pure o-DCB and under analogous MW irradiation power, the IL additive leads to a ca.20-fold rate acceleration, which is explained considering a poor dielectric heating of the pure o-DCB solvent (T calc = 419 K, light grey circle in Figure 2).Despite the MW-induced boosting effect, [60]fullerene conversion levels off at 65 % (Figure S13 and Table 2) with formation of both m-FP 1 and p-FP 1 in ca.
1.1 ratio (Table 2).While the evolution to poly-substituted products is expected to increase at higher [60]fullerene conversion (Figure 1 and Table 2), the observation of a limiting plateau yield is likely ascribed to the equilibrium regime involving the thermally activated retro-cycloaddition.This latter is effective in the presence of the IL additive and likely occurs at all the fulleropyrrolidine sites.
To further explore the effect of the thermodynamic control on the cycloaddition products, we turned our attention to the regioselectivity of the 1,3-dipolar cycloaddition obtained with sarcosine and formaldehyde as a model reaction, yielding FPs 2 (Scheme 1). [18]In particular, the N-methyl unsubstituted azomethine ylide generated from formaldehyde rules out the formation of additional bis-FP 2 stereoisomers and minimizes the steric hindrance of a second pyrrolidine ring installed on the C 60 scaffold. [19],3 dipolar cycloaddition on [60]fullerene can involve 30 reactive double bonds, shared by condensed [6,6] cycles, so that after installation of the first pyrrolidine ring, 8 isomeric bisadducts can be formed, either located on the same C 60 hemisphere with respect to the first functionalization (three cis isomers), in the opposite one (four trans isomers), or at the equatorial positions (1 eq isomer) (Figure 3).[19] With the aim to tune the selectivity of the bis-cycloaddition, MW-irradiation was optimized at an applied power of 12 W, and the product profile was monitored via HPLC analysis, by identification of the single analytes specifically attributed to m-FP 2 and to trans-, cis-and equatorial bis-fulleropyrrolidines (Figures S14 and attributions in Table S1), while considering the remaining peaks as an overall signature of further [60]fullerene poly-functionalization.Using pure o-DCB as solvent, a complex mixture of fulleropyrrolidines readily forms in solution after ca. 10 min irradiation (Figure S14a).Analysis of the product profile reveals that [60]fullerene conversion (> 80 %) leads to the formation of m-FP 2 (13 %) together with seven bis-FP 2 regioisomers (31 %) where trans-bis-fulleropyrrlidines are the prevalent isomers (23 %).In this reaction, a mixture of ill-defined p-FP 2 turns out to be dominant (56 %), hampering the isolation of desired bis-functionalized products (Figure 4, trace a).A similar product distribution is also obtained under conventional heating conditions.
Noteworthy, the introduction of the ionic liquid additive combined with MW-induced dielectric heating was crucial for the simplification of the product profile.At shorter irradiation time (< 5 min), [60]fullerene conversion remains below 60 %, which limits the formation of p-FP 2, while enhancing the production of m-FP 2 (46 %) and of bis-FP 2 where the preferential selectivity towards trans and equatorial regioisomers is confirmed (54 %) (Figure S14b and Figure 4, trace b).The selective formation of trans and equatorial regioisomers at short reaction time, is indicative of a marked kinetic preference, regulated by steric hindrance that favors the bis-functionalization at the equatorial or sub-hemisphere distal positions. [19]o verify a converging thermodynamic control over the equilibration of bis-fulleropyrrolidine regioisomers, we designed a two-stage/follow-up experiment.In the first stage, a crude reaction mixture was obtained under MW irradiation (12 W, < 10 min) in pure o-DCB and purified from the unreacted ylide precursors.The "purified" crude was then irradiated again, in a second stage (12 W for 5 minutes), using a o-DCB:IL = 3 : 1 solvent phase.HPLC analysis confirmed that the initial mixture (Figure 5a), containing the unreacted [60]fullerene and fulleropyrrolidine products (m-FP 2, and seven bis-FP 2 regioisomers), [20] evolved in the second stage, under thermodynamic re-equilibration.The resulting product profile shows an increased conversion of [60]fullerene (up to 95 %) that occurs by consuming the trans-4, cis 1-3 and eq isomers while leading  to the preferential formation of trans-1, trans-2, trans-3 bis-FP 2 regioisomers (Figure 4, trace c and Figure 5b).The further progress of the 1,3 dipolar cycloaddition, without recharging the ylide precursors (Scheme 1), is indicative of a reversible reaction pathway under thermodynamic control.
To address reaction path and product distribution, we performed DFT calculation at the b3lyp/6-311G level (see Figure 3, Figure 6, Figure S15, Figure S16).
The formation of bis-fulleropyrrolidines results from the favorable interaction between empty orbitals of the m-FP 2 (in particular, the LUMO, LUMO + 1 and LUMO + 2) and the HOMO of the ylide reagent, with a calculated energy gap in the range 0.625-0.975eV (14.4-22.5 Kcal mol À 1 , Figure 6 and Table 3).
Geometry optimization of all the expected bis-FP 2 has been performed with the aim to address their relative thermodynamic stability.In all cases the calculated free energies (ΔG) of the diverse regioisomers are referred to the most stable equatorial bis-FP 2 (eq, Figure 3).
Indeed, the LUMO-LUMO + 2 molecular orbitals of m-FP 2 display an electron density distribution involving diverse [6,6]  bonds of the fullerropyrrolidine scaffold, which provides a rational for a regioselective 1,3-dipolar cycloaddition controlled by the frontier orbital coefficients of the interacting species (Figure 6 and Table 3). [21]ccording to the computational analysis, the higher energy bis-FP 2 isomers display a cis-orientation (cis-1, cis-2 and cis-3) with relative energy gap in the range ΔG = + 2.5-8.0Kcal mol À 1 .
Because of the lower distortion of the fullerene scaffold, bis-FP 2 with a trans-orientation are more stable regioisomers, with similar energy content (ΔG below 1.5 Kcal mol À 1 , thus at the limit of the significance value for this level of calculation).
These results are in good agreement with the regioselectivity outcome observed under diverse heating conditions, indicating that the trans-bis-FP 2 isomers are the dominant products, formed under thermodynamic control, by unlocking the cyclo-reversion of the high energy cis-isomers (Figure 4 and Figure 5).

Conclusion
The functionalization of [60]fullerene by azomethine ylide 1,3dipolar cycloaddition to fulleropyrrolidines has been performed by the combined use of IL phases and MW dielectric heating.The reaction of [60]fullerene with heptaldehyde and sarcosine as ylide precursors has been considered as a benchmark system    (cis-3, trans-2, trans-3)  to address the cycloaddition performance under conventional heating and under MW irradiation.Here, the use of the IL additive allows a flash thermal activation induced by the efficient MW absorption, which is responsible for a 20-fold rate acceleration of [60]fullerene conversion.Inspection of the activation parameters by the Eyring relationship points to a favourable stabilization of the zwitterionic azomethine ylide reagent in the presence of ILs, which mitigates the negative activation entropy expected for the evolution of the cyclic transition state.Optimized conditions were then applied for the regioselective formation of bis-FP 2 trans-regioisomers, as observed for the reaction of [60]fullerene with a symmetrical ylide generated in situ from formaldehyde and sarcosine precursors.In this case, the combined use of ILs additive and MW irradiation is effective to unlock the reverse regime of retro-cycloaddition, thus allowing a desirable thermodynamic control of the product distribution.Controlling [60]fullerene functionalization by a proper choice of the solvent composition and of the temperature conditions offers a straightforward tool to exploit a convergent kinetic and thermodynamic control on the cycloaddition selectivity.

Experimental Section
General: Continuous microwave irradiation was carried out in a CEM-Discover-Coolmate mono-modal microwave apparatus, with simultaneous monitoring of irradiation power, pressure and temperature.Compressed air was applied to improve the temperature control of the reaction mixtures.All reactions were monitored by HPLC, using a Shimadzu LC-10AT VP pump system, equipped with a UV detector SPD-10A VP set at 340 nm.Phenomenex Luna and Cosmosil Buckyprep columns (250 × 4.6 mm 5 μm particles) were used, with toluene as eluent flowing at 1 mL/min.
Procedure for the cycloadditions: Unless otherwise stated, 5 mg of [60]fullerene (7 μmol) were dispersed upon sonication in a mixture of o-dichlorobenzene (and ionic liquid), together with sarcosine (1.2 mg, 14 μmol) and aldehyde (3.2 mg, 28 μmol), in a closed glass test tube containing 1 mL of solvent.Samples were diluted using toluene and analyzed by HPLC.
DFT calculations have been performed with Gaussian 09 software.
Geometries have been optimized at the b3lyp/6-311G level in the gas phase.Inclusion of a continuous pcm model for the solvent (water) did not lead to significant changes in the relative energies of the bis-adducts discussed in Figure 3.

Figure 2 .
Figure 2. Eyring plot for the reaction performed under conventional heating conditions, in o-DCB (black circles) and in the presence of 25 % OmimBF 4 (red circles).MW-induced T calc (Table2) were extrapolated from this graph (light grey and red circles).

Figure 4 .
Figure 4. Histogram plot illustrating the % composition of the reaction mixtures obtained with 14 μmol of [60]fullerene, 56 μmol of formaldehyde and 28 μmol of sarcosine, irradiated at 12 W in (a) 2 mL of o-DCB, > 10 min (1 step o-DCB, blue columns) or (b) in 1.5 mL of o-DCB and 0.5 mL of OmimBF 4 , 5 min (1 step o-DCB/IL, orange columns).Yellow columns (c) are referred to an ylide precursors-free mixture, obtained from the reaction in o-DCB, after thermodynamic equilibration in 1.5 mL of o-DCB and 0.5 mL of OmimBF 4 , upon 5 min irradiation (2 steps o-DCB/IL).In all cases, % are based on chromatographic peak areas by assuming a similar response factor for all compounds.

Figure 5 .
Figure 5. HPLC profile of (a) a mixture of [60] fullerene monoadduct (m-FP 1) and bis-adducts obtained from a reaction of 14 μmol of fullerene, 56 μmol of formaldehyde and 28 μmol of sarcosine irradiated at 12 W in 2 mL of o-DCB (< 10 min) and (b) of the mixture in (a) after irradiation (12 W for 5 min) in 1.5 mL of o-DCB and 0.5 mL of OmimBF 4 .

Table 1 .
Reaction rate and conversion, observed during the reaction of [60]fullerene with sarcosine and heptaldehyde in o-DCB and o-DCB