Dirk M. Guldi

Radiation Laboratory

University of Notre Dame

Notre Dame, IN 46556

guldi@marconi.rad.nd.edu

The redox properties of pristine fullerenes and monofunctionalized fullerene derivatives in their ground and excited states are described. These compounds have drawn much attention in the last few years for the design of devices able to perform complex functions, as in molecular switches, receptors, photoconductors and photoactive dyads. These applications are generally based on the excellent electron accepting properties of the fullerene moiety. This, together with the low reorganization energy, makes C60 and its derivatives good candidates for building blocks of more complex systems for the conversion of light into electricity or fuels. The concept of linking fullerenes to a number of interesting electro- or photoactive species and the subsequent study of intramolecular transfer dynamics between the two moieties will be elucidated as a function of excited state energy of the antenna molecule, donor-acceptor distance, and solvent polarity.

The pioneering work by Kroto et al. and Krätschmer et al. regarding the initial discovery of [60]fullerene and the development of a large-scale method for their preparation, respectively, have prompted a lively interest to probe these 3-dimensional carbon allotropes in solution and to relate their properties to conventional 2-dimensional p-systems. Furthermore, a number of solid state properties have been reported ranging from superconductivity to nanostructered devices. The spherical symmetry of C₆₀, arranging the 60 carbon atoms in pentagon and hexagon architectures, leads to a substantial curvature of the carbon network and, in turn, to a strong pyramidalization of the individual carbon atoms with an average hybridization of sp^2.278. The sum of these effects exerts a strong impact on the reactivity of the fullerene core relative to 2-dimensional carbon networks, leading, in essence, to an extremely reactive exterior but, in contrast, to a nearly inert interior.

Figure 1: Structure of [60]fullerene and illustration of the pp orbital energy levels in C₆₀⁸

A particular intriguing consequence, which stems from this unique structure, is the electronic configuration of pristine fullerenes, comprising a five-fold degenerate HOMO (h_u) and a triply degenerate LUMO level (t_1u), linked by a moderate energy gap of ca. 1.8 eV. In accord with the degeneration of the LUMO level, the redox chemistry, conducted primarily by cyclic voltammetry, demonstrated unequivocally the ability of fullerenes to accept up to six electrons in reversible one-electron reduction steps.

C₆₀ —> (C₆₀•-) —> (C₆₀2-) —> (C₆₀3-)

—> (C₆₀4-) —> (C₆₀5-) —> (C₆₀6-) (1)

This suggests using C₆₀ and, to the same extent, of functionalized derivatives as potent electron acceptor moieties. The remarkably low redox potential of -440 mV versus SCE for the first one-electron reduction step, further underlines this aspect. More importantly, the exceptionally low reorganization energy of the fullerene core for such reductions is directly associated with a high degree of charge delocalization within the 3-dimensional carbon framework. Thus, forward electron transfer to the fullerene moiety is strongly favored, while, at the same time, back electron transfer experiences a noticeable slow down relative to conventionally probed electron acceptors, such as 2-dimensional aromatic molecules.

A more practical aspect of fullerenes is concerned with the optical absorption spectrum of, for example, the (C₆₀•-) p-radical anion. This species shows a narrow band in the near IR region around 1080 nm which serves as a diagnostic probe in assisting the identification of (C₆₀•^_) and, furthermore, allows an accurate analysis of inter- and intramolecular ET dynamics in C₆₀ containing systems.

Figure 2: NIR spectrum of the (C₆₀•-) p-radical anion, obtained upon pulse radiolytic reduction.

Another advantage of these 3-dimensional p-systems implements the flexibility of substituting C₆₀ by a higher fullerene, e.g., C₇₀, C₇₆, C₇₈, C₈₄ etc. Specifically, increasing the number of carbons in the fullerene network, from C₆₀ to C₈₄, eases the abstraction of an electron from the corresponding HOMO level which parallels an incremental decrease of the ground state reduction potential. In other words, higher fullerenes are unequivocally even better electron acceptors and electron donors than C₆₀.

With the advent of various protocols, regarding the controlled functionalization of the fullerene core, an extensive number of functionalized fullerene derivatives became available. These chemical modifications, in fact, provide the opportunity to combine the spectacular properties of fullerenes with those of other interesting materials, including electro- and photoactive species. The most intriguing examples, that stem from these versatile methodologies, are multicomponent donor-bridge-acceptor arrays as promising new materials for energy conversion devices. The focus of the current review is to summarize the intriguing properties of fullerenes as a multifunctional electron storage moiety. The reader should, nevertheless, be directed to a series of excellent review articles which appeared during recent years.

Based on the symmetry prohibitions, solutions of [60]fullerene (e.g., in cyclohexane) strongly absorb in the UV region, with maxima at 260 and 330 nm that have extinction coefficients (e) on the order of 105 M-1cm-1. In contrast, the visible region is characterized by only relatively weak transitions. For example, the maximum, noticed at 536 nm, displays an extinction coefficient of only 710 M-1cm-1. Similar to the fullerene ground state the photophysical properties of the corresponding excited states are governed by the fullerene’s 3-dimensional symmetry. A very low emission quantum yield (F) of 1.0 x 10-4, associated with the singlet excited state, relates to the combination of a short life-time, a quantitative intersystem crossing and, finally, the symmetry forbidden nature of the lowest-energy transition. To the same extent, the phosphorescence quantum yield (F < 10-6) is strongly impacted by the spherical structure of C₆₀.

Figure 3: Triplet-triplet absorption spectrum of C₆₀ in toluene, obtained upon flash photolysis.

In terms of transient absorption, the singlet excited state gives rise to a characteristic singlet-singlet absorption around 920 nm, whose lowest vibronic state has an energy of 1.99 eV., Once generated, this state is subject to a rapid and quantitative intersystem crossing process to the energetically lower lying triplet excited state (1.57 eV). In the case of [60]fullerene, this intersystem crossing takes place with a life-time of 1.3 ns, governed by a large spin-orbit coupling, which makes this process much faster than those noticed in 2-dimensional aromatic hydrocarbons. However, not only the rate is remarkable, but also the overall efficiency of the process is impressive, affording a triplet quantum yield close to unity.

The triplet-triplet absorption spectrum reveals a maximum in the visible region (750 nm; e = 20 000 M-1cm-1), and another transition in the UV (330 nm; e = 40 000 M-1cm-1). These features are easily detectable probes for uni- and bimolecular transfer reactions. The triplet life-times are dominated by triplet-triplet annihilation and ground-state quenching, and are quite short (50 and 100 ms). This is, again, much shorter than the triplet life-time of planar aromatic hydrocarbons.

C₆₀ —hn—> (¹C₆₀) —ISC—> (³C₆₀) (2)

Lifting an electron from a low-energy bonding orbital to a high-energy anti-bonding orbital increases the electron affinity and, at the same time, reduces the ionization potential of the fullerene by precisely the amount of the excited state energy. Consequently, the fullerene singlet and triplet excited states are stronger oxidants and reductants than the singlet ground state. Important values, concerning the redox potentials, together with some ground state and excited state properties, of C₆₀, C₇₀, and a monofunctionalized C₆₀ derivative, are listed in Table 1. Both excited states, namely, singlet and triplet excited states, are subject to bi- and unimolecular charge transfer reactions with adequate electron acceptor and electron donor moieties. A bimolecular reaction with the initially formed singlet excited state stands, however, in direct competition with the rapidly occurring intersystem crossing and is, in essence, overshadowed by the latter. Thus, it should be noted up front that truly bimolecular transfer reactions, as outlined in the following chapters, occur with the much longer-lived triplet excited state. Alternatively, a strong interaction between the fullerene and a donor, in form of a ground state complex or excited state complex (exciplex), is also an appropriate manner to accelerate transfer dynamics evolving from the singlet excited state.

A bimolecular electron transfer reaction is thought to lead to the formation of a primary contact radical pair. In the current context, this is the one-electron reduced (C₆₀•-) p-radical anion and the oxidized form of the applied quencher (Q•+). Specifically, the diagnostic peak of the (C₆₀•-) p-radical anion in the near-IR region is employed as a sensitive marker whose detection points unmistakably to the generation of a charge separated radical pair, i.e., {(C₆₀•-) / Q•+}. The latter evolves from the following sequence: an initial photoexcitation process, succeeded by a bimolecular electron transfer and, finally, a diffusional break-up of the contact pair. This dissociation depends on the ability of the environment, e.g., solvent, to retard the exergonic back electron transfer and stabilize the radical pair, for example, via strong dipole-dipole interactions.

The low reduction potential of (³C₆₀) (+1.01 V) permits an efficient reductive quenching with a large number of amines, e.g., diazabicyclooctane (DABCO). It is noteworthy to point out that the latter does not add to ground state C₆₀ which constitutes a fundamental advantage over primary, secondary, and tertiary amines. Selection of a polar solvent, such as alcohols or benzonitrile, is crucial in view of promoting an efficient charge separation which is, under these circumstances, typically confirmed through formation of the characteristic (C₆₀•-) p-radical anion absorption at 1080 nm. In contrast, the low dielectric constant of toluene or chlorinated hydrocarbons prevents this charge separation process and, in turn, no appreciable life-time of the (C₆₀•-) and (DABCO•+) radical pair is observed.

(³C₆₀) + DABCO —> (C₆₀•-) + (DABCO•+) (3)

Similarly, ferrocene (FC), tetrathiafulvalene (TTF), bis(ethylenedithio)-tetrathiafulvalene (BEDT-TTF), 3,3’,5,5’-tetramethybenzidine (NTMB), stable nitroxide radicals, N,N-dimethyl-1-naphthylamine, polysilane, oligothiophene, and tetrathiophene moieties, to name a few representative cases, are all excellent electron donors. These compounds have therefore been probed them as sacrificial electron donors in reductive quenching reactions with triplet excited fullerenes. Again, polar solutions play a central role in creating conditions that supports an efficient charge separation and gives rise to bimolecular rate constants for the charge recombination on the order of 1010 M-1s-1.

Based on the mildly electrophilic character of C₆₀, formation of fullerene charge transfer complexes is not only limited to NTME. In fact, aromatic amines, such as N,N-dimethylaniline (DMA), N,N-diethylaniline (DEA), diphenylamines (DPA), triphenylamine (TPA), triethylamine (TEA), and a number of substituted naphthalenes give rise to similar behavior. Generally, the resulting complexes are quite weak, with formation constants on the order of 0.1. Accordingly, high amine concentrations are required to study the characteristics and properties of these fullerene containing charge-transfer complexes. At low amine concentrations the fullerene excited states are, however, effectively quenched through electron transfer interactions. In aliphatic hydrocarbon solvents such as hexane or methylcyclohexane, the quenching of the monomer excited states occurs via the transient formation of exciplexes. The associated exciplex emissions are extremely sensitive to the solvent environment which in polarizable benzene and toluene, for example, leads to a complete quenching of the exciplex.

Finally, a bimolecular electron transfer with a ground state zinc phthalocyanine (ZnPc) exhibits an interesting twist. The oxidation potential of this chromophore, generating the ZnPc p-radical cation, is +0.8 V versus SCE and, thus, should favor electron transfer to the fullerene triplet excited state. Indeed, excitation of pristine C₆₀ in benzonitrile gives rise to the rapid formation of the fullerene’s diagnostic p-radical anion band. In non-polar solvents, the energy levels for the charge-separated radical pair are, however, significantly raised. As a direct consequence, the electron transfer (reaction 5a) is shut off, while the intriguing energy transfer route (reaction 5b) is still activated. It remains to be shown in the future how the exclusive excitation of the ZnPC chromophore impacts the pathway and formation of the final product.

(³C₆₀) + ZnPc —> (C₆₀•-) + (ZnPc•+) (4a)

(³C₆₀) + ZnPc —> C₆₀ + (³ZnPc) (4b)

The oxidative quenching of the fullerene triplet excited state_, leading to the formation of the highly unstable (C₆₀•+) p-radical cation, by electron acceptors (ranging from tetracyanoquinondimethane (TCNQ) and tetracyanoethylene (TCNE) to chloraniline (ClA)) or by cosensitizing the photoinduced electron transfer from a singlet excited N-methylacridinium heaxfluorophosphate (MA) with biphenyl (BP) have been established in solvents of different polarity, e.g., toluene and benzonitrile.

Cycloadditions provide a valuable tool for C₆₀ functionalization. Usually, 1,2 additions across a [6-6]-ring junction, which possess a higher electron density than the [6-5]-ring junction, are observed. Covalent attachment of a large number of addends across one of the double bonds located at the junctions of two hexagons or across the single bonds at the hexagon-pentagon heterojunction generally affords four different isomeric products. Among these four, the [6-5]-open and the [6-6]-closed are the most interesting isomers. The structure of the [6-5]-open isomer is very appealing, since its fullerene core completely sustains the p-system of pristine C₆₀, but still enables attachment of virtually any functionalizing group. Unfortunately, most of the [6-5]-open derivatives undergo conversion, by thermal, electrochemical, and photochemical initiation, into the thermodynamically more stable [6-6]-closed isomers.

In principle, more than just a single addition is possible to the C₆₀ core since the latter has 30 double bonds, all sharing the same reactivity. Thus, the isolation and characterization of C₆₀ polyadducts renders the C₆₀ framework a useful template for synthesizing novel and structurally complex molecules. For example, when two identical and symmetrical addends are taken up by C₆₀, up to eight isomeric bisadducts can be generated. Their nomenclature refers to the 3-dimensional geometry of the fullerene. In this light, the second addition can take place within either the same hemisphere ("cis"-isomer), at the equator ("equatorial"-isomer) or within the opposite hemisphere ("trans"-isomer).

In the following chapters, studies are reviewed that focus on time-resolved and steady-state experiments with functionalized fullerene derivatives. Upfront, a classification should be made to distinguish between the different types of derivatives. The simplest case describes the covalent attachment of addends that lack any VIS absorption and are redox-inactive, in the sense of accepting or donating an electron to the synthetically bound fullerene moiety. It should be emphasized, that these derivatives have evolved as important reference compounds for understanding the intramolecular dynamics within the more complex donor-acceptor dyads. Of increasing complexity are donor-acceptor arrays carrying electroactive addends. For example, in dimethylaniline (DMA), ferrocene (Fc) or tetrathiofulvalene (TTF) based multicomponent supermolecules, such fullerenes are implemented as photosensitizers that are thought to undergo intramolecular electron transfer processes from the adjacent electroactive moiety. Accordingly, these are classified as electroactive but, due to their insignificant VIS absorption characteristics, photoinactive functionalities. Finally, in light of the moderate fullerene absorption features in the VIS region, functionalization of pristine C₆₀ with chromophoric addends, such as metalloporphyrins or MLCT transition metal complexes, have developed as important objectives. The primary scope of the latter is to promote the VIS absorption characteristics of the resulting dyads and, most importantly, to improve the light harvesting efficiency of the fullerene core. Specifically, suitable photosensitizers assist to extend the fullerene absorption, which, upon excitation, yield energetically high-lying excited states further into the red. As a direct consequence, the role of the fullerene is significantly changed. Under these circumstances, i.e., bearing a photoactive and electroactive moiety, fullerenes operate exclusively as either electron or energy acceptor moieties.

2.1 Electroinactive, photoinactive moieties Converting the fullerene core into a 58 p-electron system greatly alters the electron accepting properties of mono- and multiply functionalized fullerene derivatives. In general, the reduction potentials of most of the known derivatives reveal a shift to more negative potentials. Physico-chemical properties, such as singlet excited states, triplet excited states, reductive quenching by DABCO, transient absorption spectra, and electrochemical reduction, sensitively reflect the functionalization of C₆₀.

Figure 4: Semilogarithmic plot of the quenching rate constants versus reduction potentials of the triplet excited states of C₆₀ and various functionalized fullerene derivatives.

A representative example is the triplet-triplet absorption, which exhibits a strong blue shift for the e,e,e-(³C₆₀)[C(COOEt)₂]₃ of almost 100 nm, relative to pristine C₆₀. This is rationalized to stem from a gradual lifting of the fullerene p-system with increasing number of bis(ethoxycarbonyl)methylene groups. Furthermore, these electronic parameters parallel a slow-down of the triplet excited state quenching with DABCO, amounting to nearly three orders of magnitude for e,e,e-(³C₆₀)[C(COOEt)₂]₃ as compared to ³C₆₀. A correlation, linking the reductive quenching rates of excited triplet fullerenes with the electrochemically determined redox potentials, shows that functionalization of pristine C₆₀ lowers the ease of reduction in both the ground and in the excited state.

Iminofullerenes have recently attracted attention, since the [6-5]-open structure has gained some degree of stabilization over those of the respective methanofullerene and pyrrolidinofullerene series. Although the [6-6]-closed configuration is still the thermodynamically more stable isomer, the [6-5]-open iminofullerene isomer - in contrast to other [6-5]-open structures - lacks, upon illumination, any evidence of an isomerization reaction. Most importantly, in the [6-5]-open isomer the fullerene core is essentially similar to C₆₀ and consequently may serve as a promising building block for more complex 3-dimensional systems.

A series of spectroscopic properties, concerning the singlet ground state and both excited states, namely, singlet and triplet, further supported by kinetic data, suggests that the electron pair on the nitrogen interacts with the fullerene p-electrons. This leads, in turn, to a number of important consequences for the electronic structure of the two aziridinofullerene isomers. Considering the free electron pair localized at the nitrogen atom, the [6-6]-closed isomer is best represented by a 60 p-electron system, two of them deriving from the nitrogen and the remaining 58 located within the functionalized fullerene core. In contrast, the [6-5]-open isomer, which, at first glance, may be assumed to be an isoelectronic analogue of pristine C₆₀, carries actually 62 electrons participating in the p-system. In terms of the intriguing electron accepting properties, e.g., ground and excited state, the [6-6]-closed aziridinofullerene reveals, besides a high photostability, not only a better activity than the [6-5]-open isomer, but even outscores the [6-6]-closed methanofullerene and [6-6]-closed pyrrolidinofullerene.

Figure 5: Semilogarithmic plot of the quenching rate constants versus reduction potentials of the triplet excited states of C₆₀ and various mono-functionalized fullerene isomers.