Manganese powder promoted highly efficient and selective synthesis of fullerene mono- and biscycloadducts at room temperature

Discovery of an efficient, practical, and flexible synthetic method to produce various important electron acceptors for low-cost organic photovoltaics (OPVs) is highly desirable. Although the most commonly used acceptor materials, such as PC61BM, PC71BM, IC60BA, bisPC61BM have been proved to be promising for the OPVs, they are still very expensive mainly due to their low production yields and limited synthetic methods. Herein, we report an unprecedented and innovative synthetic method of a variety of fullerene mono- and biscycloadducts by using manganese powder as a promotor. The reaction of fullerenes with various dibromides proceeds efficiently and selectively under very mild conditions to give the corresponding cycloadducts in good to excellent yields. The combination of manganese power with DMSO additive is crucial for the successful implementation of the present cycloaddition. Notably, the standard OPV acceptors, such as PCBMs, have been obtained in extraordinarily high yields, which cannot be achieved under the previously reported methods.

Functional fullerenes have been used broadly as excellent n-type semiconductors in solution processable organic electronics [1][2][3][4][5] , especially as the unique electron acceptors for organic photovoltaics (OPVs) due to their significant increase in solubility while preserving certain electronic and optical properties of pristine fullerenes [6][7][8] . At present, the state-of-the-art of OPVs have an overall power conversion efficiency approaching 10% 9-11 based on newly developed low bandgap electron donors and [6,6]-phenyl-C 61 -butyric acid methyl ester (PC 61 BM) or its C 70 analogue PC 71 BM as an electron acceptor [12][13][14] which are the most well-demonstrated benchmark acceptors for testing new donor materials in terms of their miscibility, solubility, and high electron mobility 15,16 . Recently, many endeavors on development of new functional fullerenes have been made to improve the efficiencies of OPVs [17][18][19][20][21][22][23][24][25][26][27] and it was found that the bisfunctional fullerenes with up-shifted LUMOs, such as indene-C 60 -bisadduct (IC 60 BA) 19,20 , bisPC 61 BM 21 and bis-o-quinodimethane C 60 (o-QDMC 60 ) [22][23][24] showed higher open circuit voltages and hence improved OPV efficiencies. In this context, it is expected that the efficiency could be further improved that may expedite the practical application of the OPVs in the next few years. To achieve low-cost OPVs, synthesis of OPV materials in a simple, practical process with a high production yield is one of the important strategies 28 . However, the standard acceptors PCBM, ICBA, and their analogues are still very expensive due to low yields, low selectivities, and harsh synthetic conditions. For example, PC 61 BM was prepared by a one-pot reaction over two-steps in 58% yield through the reaction of C 60 with methyl 4-benzoylbutyrate p-tosylhydrazone at 70 °C followed by isomerization of the resulting [5,6]PC 61 BM to PC 61 BM at 180 °C ( Fig. 1a) 13 . IC 60 BA was synthesized in 34% yield along with the formation of the monoadduct IC 60 MA in 25% yield in the reaction of C 60 with indene at 180 °C (Fig. 1a) 19 . Therefore, development of an innovative, practical synthetic method for those important fullerene acceptors under mild conditions with high production yields is highly desirable.
Recently, we have been interested in development of new and efficient fullerene functionalizations under mild conditions toward application in OPVs [29][30][31][32][33][34][35][36] . Various fullerene functionalizations for synthesis of monosubstituted hydrofullerenes, monocycloadducts, single-bonded fullerene dimers, 1,4-disubstituted fullerenes have been developed in the presence of transition metal catalysts or oxidant via the formation of the fullerene monoradical as an active species 37 . These results led us to challenge the innovative, flexible synthetic method for those most common fullerene acceptors in very high yields. Herein, we report an unprecedented, highly efficient, and selective manganese powder-promoted bis-and monocycloaddition of C 60 (or C 70 ) with various alkyldibromides at room temperature for constructing carbocycle-fused fullerenes with various ring sizes (Fig. 1b). Notably, the present cycloaddition afforded PC 61 BM and PC 71 BM in excellent yields of > 90%, as well as IC 60 BA and bisPC 61 BM in 75% and 92% yields, respectively, which cannot be achievable under the previously reported methods.

Results
Optimization of the reaction conditions. As aforementioned, we have reported that the CoCl 2 dppf catalyst combined with a Mn reductant in o-dichlorobenzene (ODCB) promoted the monocycloaddition of C 60 with active dibromides efficiently at room temperature to form the fullerene monocycloadducts in a high selectivity 30 . It was noted that the efforts to synthesize the biscycloadducts under the Co-catalyzed standard conditions even using an excess amount of dibromides were failed. Further investigations of the reaction conditions for the selective biscycloaddition of C 60 with 1,2-bis(bromomethyl)benzene (1a, 4 equiv) are summarized in Table 1.
The reaction with Mn powder (9 equiv) in ODCB did not produce any desired cycloadducts (entry 1). We have previously demonstrated that the use of polar cosolvents with ODCB remarkably enhanced the fullerene monoradical generation and stability [33][34][35][36] , which led us to examine various polar cosolvents. To our delight, the reaction proceeded smoothly in a mixture of DMSO/ODCB with Mn powder at room temperature, affording the corresponding biscycloadduct 2a (o-QDMC 60 ) in 60% yield together with a 20% yield of the multicycloadduct 3a (entry 2). The reaction did not take place without using Mn powder in this solvent system (entry 3). Other cosolvents were also tested in the presence of Mn powder. It was found that the use of DMF as a cosolvent exhibited a much higher activity, yielding the multiadducts predominantly without formation of both bis-and monocycloadducts (entry 4), while other polar solvents such as CH 3 CN, THF, and EtOH were totally inactive (entries 5-7). Notably, the decrease in amounts of Mn powder (3 equiv) and dibromide 1a (2.2 equiv) in ODCB/DMSO solvent systems hampered the formation of multiadducts under a prolonged reaction time (47 h), resulting in 2a in an 85% isolated yield (entry 8). Other metals were also tested instead of Mn powder. In contrast to the high activity of Mn(0), other Mn salts such as MnO 2 , Mn(OAc) 3 •2H 2 O 38 , and MnCl 2 •4H 2 O having a higher oxidation state were totally ineffective (entries 9-11). Zinc and iron powders which have been employed as reductants successfully in our previous cobalt catalysis 29 showed a moderate activity, in which Zn powder could promote the present biscycloaddition to give 2a in 55% yield, while Fe powder showed a relatively lower reactivity to produce the monocycloadduct 3a in 54% yield without formation 2a (entries 12 and 13). The reactions using other zero-valent metals, such as Mg, Cu, Pd/C, and Raney Ni did not show any activity for promoting the present cycloaddition (entries [14][15][16][17]. Overall, it is concluded that the use of Mn(0) powder in DMSO/ODCB solvent systems is crucial for the selective formation of the fullerene biscycloadduct 2a in a high chemical yield.

Synthesis of biscycloadducts.
Under the optimized conditions, various alkyl dibromides have been examined to study the substrate scope and selectivity of the present biscycloaddition (Fig. 2). All the reactions were monitored by HPLC and the corresponding products were isolated by silica gel chromatography. It is noted that very small amounts of the monoadducts and the recovered C 60 were observed in every reaction, and the major by-products were multiadducts whose yields did not show due to some overlapping peaks with bisadducts in HPLC chromatogram. The new structures of biscycloadducts containing a mixture of isomers were determined by 1 H and 13 C NMR spectra as well as the high resolution mass.
Firstly, we tested various bis(bromomethyl)arenes to obtain the corresponding bisadducts fused with a 6-membered ring. 1,2-Bis(bromomethyl)benzenes 1b and 1c having a ester or a methoxy group on the phenyl ring were tolerated under the present conditions, furnishing the corresponding biscycloadducts 2b and 2c in 70% and 71% yields, respectively. The reaction is also compatible with the heteroaryl-incorporated dibromide, 3,4-bis(bromomethyl)-2,5-dimethylthiophene (1d), affording the corresponding bisadduct 2d in 56% yield. The reaction of C 60 with 1,3-dibromo-2,3-dihydro-1H-indene (1e) produced IC 60 BA (2e) in 75% yield which is much higher than that using the previously reported method (34%, Fig. 1a) 19 . It is noted that the scale-up reaction of C 60 (108 mg) with 1e did not show significant decrease in the efficiency, giving IC 60 BA (2e) in a 72% isolated yield after 26 h at room temperature ( Figure S3). The reaction is also applicable to the construction of the bisadducts fused with a 5-membered ring, which could not be prepared by the previously reported methods [22][23][24] . The reactions of C 60 with 1,3-dibromo-1,3-diphenylpropane (1f), and its derivatives 1g and 1h having Br and F substituents on the phenyl ring showed a high reactivity, producing the corresponding bisadducts 2f-h in good to high yields in short reaction times. Again, it was confirmed that the scale-up reaction of C 60 (108 mg) with 1f produced a 75% yield of 2f under a prolonged reaction time (45 h) ( Figure S4). The allylic dibromide such as 3-bromo-2-(bromomethyl)prop-1-ene (1i) is also a suitable substrate for the present cycloaddition, giving the corresponding bisadduct 2i in 61% yield. Interestingly, the reaction of C 60 with (dibromomethyl)benzene (1j) gave the 3-membered ring fused bisadduct 2j in 90% yield. This result encouraged us to synthesize bisPCBM which has also been applied as a promising acceptor in OPVs 21 . The corresponding dibromide reactant, methyl 5,5-dibromo-5-phenylpentanoate (1k) was prepared in 90% yield by the dibromination of methyl 5-phenylpentanoate with NBS in the presence of a catalytic amount of AIBN at reflux ( Figure S1). Remarkably, the biscycloaddition of C 60 with 1k under the standard conditions produced the desired bisPC 61 BM (2k) in 92% yield. It is worthy to note that bisPC 61 BM is generally obtained as a by-product during the preparation of PC 61 BM 21 .
LUMO energy levels of biscycloadducts. It was reported that 56π -biscycloadducts, such as IC 60 BA and bisPC 61 BM exhibited a significantly improved open circuit voltage (V oc ) owing to its much higher LUMO energy than that of PCBM [19][20][21] . We have measured the LUMO energy levels of the selected new biscycloadducts by cyclic voltammetry (CV). Biscycloadducts 2b and 2c having a ester or a methoxy group on the phenyl ring showed high LUMOs of − 3.48 and − 3.41 eV, respectively, which are higher than that of PC 61 BM (− 3.60 eV) and comparable with that of IC 60 BA (2e, − 3.43 eV). The 5-membered ring-fused bisadduct 2i shows a slightly higher LUMO energy level (− 3.40 eV) compared with IC 60 BA (2e). Since the V oc of bulk heterojunction solar cells has an association with the energy difference between the LUMO of acceptor and the HOMO of donor 39 , it is expected that the new biscycloadducts possessing higher LUMO energies should be potential acceptor candidates for OPVs.

Synthesis of monocycloadducts.
Inspired by the successful biscycloaddition of C 60 with Mn powder, we further extended the present method to the selective monocycloaddition. To our delight, the monocycloadducts could be obtained at room temperature in good to excellent yields with a wide substrate scope by simply decreasing the amounts of Mn powder (1 equiv) and dibromides (1 equiv) as shown in Fig. 3. The reaction of C 60 with bis(bromomethyl)benzenes 1a and 1b in the presence of Mn powder produced the 6-membered ring-fused monocycloadducts 3a and 3b in 72% and 82% yields, respectively, with a small amount of the recovered C 60 . The indene-monoadduct 3c can be prepared in 74% yield under the standard conditions using 1,3-dibromo-2,3-dihydro-1H-indene (1e) as a dibromide source. Monocycloaddition of 3-bromo-2-(bromomethyl)prop-1-ene (1i) with C 60 furnished the corresponding 5-membered ring fused monocycloadduct (3d) in 60% yield. (Dibromomethyl)benzenes 1j and 1l underwent the selective monocycloaddition with C 60 smoothly to afford the corresponding cyclopropyl-fused monoadducts 3e and 3f in 81% and 75% yields, respectively. Surprisingly, when 1k was used as a dibromide reactant, PC 61 BM (3g) was obtained in a very high yield of 93% using 2 equiv of Mn powder after 7 h at room temperature (Fig. 4a). The large-scale reaction of C 60 (200 mg) with 1k under the identical conditions produced PC 61 BM (3g) in a slightly lower yield of 88% (12 h, Figure S6), demonstrating that the efficiency of the present method stays high at large scale. It is noted that a 38% yield of PC 61 BM could be obtained under our previously reported Co-catalyzed reaction conditions 30 after 24 h along with the recovered C 60 in 54% yield ( Figure S7). However, under the present reaction conditions, the cycloaddition of C 70 with 1k produced a mixture of [6,6]-and [5,6]-isomers after 12 h. Subsequently, the mixture purified by silica gel chromatography was further heated at 180 °C in ODCB for 24 h to give the corresponding [6,6]-isomers PC 71 BM (3h) in 90% yield over two steps (Fig. 4b).
Noted that the signals for three methoxy groups in the 1 H NMR spectrum indicated that the ratio of three isomers in PC 71 BM was 38:46:16, which is different with that observed from the reported method 14 .

Discussion
In general, the biscycloadduct o-QDMC 60 (2a) and the monocycloadduct 3a can be prepared through the Diels-Alder reaction of C 60 with o-quinodimethane generated in situ from 1,2-bis(bromomethyl)benzene (1a) at high temperatures [22][23][24]40 . The question should be whether the present cycloaddition proceeds through the formation of o-quinodimethane from 1a by Mn powder, which was ruled out based on the good reactivity of other dibromides 1f-k which could not form the o-quinodimethane-like intermediates. The other possible mechanism is the formation of a fullerene radical anion by metal reductants 41 followed by the reaction with dibromides 42 . However, this pathway seems to be unlikely because we found that the reaction of C 60 and H 2 O in the absence of dibromides under otherwise the standard conditions did not produce any hydrogenated C 60 -adducts including the expected dihydrofullerene (C 60 H 2 ), which is different from the Zn-mediated monoalkylation of C 60 with alkylbromides as reported by Meier et al. 43 Moreover, we found that the reaction of the dibromide 1a and benzaldehyde with Mn powder in the absence of C 60 under otherwise the standard conditions did not afford any products and the starting substrates were recovered quantitatively, implying that the formation of the organomanganese reagents 44  cannot be accounted for the present cycloaddition. It is also noted that the existence of a small amount of air slows the reaction rate and much amounts of Mn powder and dibromides are required with prolonged reaction times. At present, although the present cycloaddition mechanism is yet to be determined, we assume that the reaction should be initiated by the electron transfer among C 60 , dibromides, and Mn powder. The detailed mechanistic studies will be reported in due course.
In conclusion, we have described a novel and highly efficient manganese powder-promoted fullerene cycloaddition with various alkyl dibromides. The reaction procedure is flexible, practical, and mild, which produces a variety of new and known fullerene cycloadducts with various carbocycle sizes in high chemical yields and high mono-and biscycloaddition selectivities. Notably, we have succeeded for the first time in synthesis of the most common OPV acceptors, such as PC 61 BM, PC 71 BM, IC 60 BA, and bisPC 61 BM in good to excellent yields. The combination of manganese powder with the DMSO cosolvent is vital for the implementation of the present cycloaddition sufficiently. Our method not only provides an efficient, low-cost, and general approach for the formation of the important functional fullerenes, but also may boost the realization of the practical application of OPVs.
Scale-up procedure for synthesis of PC 61 BM (3g). To a mixture of 1,2-dichlorobenzene (40 mL), DMSO (4 mL), C 60 (200 mg, 0.277 mmol), and Mn (30.4 mg, 0.554 mmol, 2 equiv) was added methyl 5,5-dibromo-5-phenylpentanoate (1k, 97 mg, 0.277 mmol, 1 equiv) under an argon atmosphere in glove box. The reaction mixture was stirred at room temperature for 12 h to give a dark brown solution. After monitoring with HPLC, the mixture was purified directly by silica gel column chromatography using toluene as an eluent. The isolated product was washed with methanol and dried to afford PC 61 BM (3g) in 88% yield (222 mg).