Initiating Electron Transfer in Doubly Curved Nanographene Upon Supramolecular Complexation of C60.

The formation of supramolecular complexes between C60 and a molecular nanographene endowed with both positive and negative curvatures is described. The presence of a corannulene moiety and the saddle shape of the molecular nanographene allows the formation of complexes with 1:1, 1:2, and 2:1 stoichiometries. The association constants for the three possible supramolecular complexes were determined by 1 H NMR titration. Furthermore, the stability of the three complexes was calculated by theoretical methods that also predict the photoinduced electron transfer from the curved nanographene to the electron acceptor C60 . Time-resolved transient absorption measurements on the ns-time scale showed that the addition of C60 to NG-1 solutions and photo-exciting them at 460 nm leads to the solvent-dependent formation of new species, in particular the formation of the one-electron reduced form of C60 in benzonitrile was observed.

Introduction

Graphene quantum dots are electron confined flakes of graphene with dimensions usually in the 3–20 nm range. They are considered as a singular family of zero dimensional (0D) carbon‐based materials, since, in contrast to pristine graphene, these nanomaterials are luminescent with a non‐zero band gap stemming from quantum confinement and edge effects. By virtue of such unique electronic features, they render outstanding materials for optoelectronic devices, energy storage systems, perovskite solar cells, and biomedical applications.

In recent years, bottom‐up approaches have enabled solution‐phase syntheses of these materials in a step‐by‐step fashion. This has resulted in an accurate control over size, morphology and, therefore, on‐demand fine‐tuning of the properties. Molecular nanographenes with a wide range of shapes, namely bilayers, belts, ribbons, propellers, helical, planar and curved, have been prepared by using controlled bottom‐up approaches (Figure 1).

Figure 1

Recent examples of nanographenes with different topologies synthesized by bottom‐up approach.

Synthetic implementation of defects such as non‐hexagonal rings into the 2D honey‐comb lattice endows Gaussian curvature to molecular nanographenes. Introducing five‐ or four‐membered rings leads to positive Gaussian curvature affording bowl shape morphologies. In contrast, inserting rings larger than hexagons involves the formation of negative Gaussian curvature and implies saddle shape structures. No doubt, the co‐existence of both curvature motifs in molecular nanographenes is extraordinary. Thus, the syntheses and properties of several nanographenes with a combined bowl and saddle shape have been reported in the literature.

Recently, our research group has described the synthesis of corannulene‐based nanographenes featuring both positive and helical or negative curvatures. This procedure is based on the π‐extension of positively curved bromocorannulene by sequential Sonogashira, Diels–Alder, and Scholl reactions. Oxidant‐dependent and selective Scholl closure of a seven‐membered ring leads to a positively and simultaneously negatively curved molecular nanographene NG‐1 or to a positive‐curved helical molecular nanographene NG‐2 when the ring closure does not take place (Figure 2).

Figure 2

π‐extended corannulene‐based curved nanographenes NG‐1 and NG‐2.

Curvature of nanographene NG‐1 renders it a unique host for the complexation of fullerenes, driven by concave‐convex interactions. As a matter of fact, complexation by any negatively curved nanographenes remains poorly explored. In addition, the electron‐accepting nature of C60 gives rise to discernible electronic communication with, for example, suitable electron donors, especially under light irradiation.

In the current contribution, we report on results regarding static and dynamic interactions between NG‐1 and C60. The combination of these two compounds creates a system which is solely based on carbon and hydrogen. Furthermore, theoretical calculations underpin the stability of the different complexes and predict that photoinduced electron transfer occurs in the picosecond timescale. Nanosecond time‐resolved assays pointed out the tunability of the dynamic processes. In polar benzonitrile, electron transfer from the curved nanographene to C60 leading to the formation of the respective radical‐ion pair is observed. In less polar chlorobenzene, the radical ion pair cannot be stabilized and triplet excited state energy transfer from NG‐1 to C60 remains.

Results and Discussion

As aforementioned, curved NG‐1 and NG‐2 have been synthesized by Sonogashira–Diels–Alder–Scholl sequential reactions. The aim of this work is focused on the supramolecular complexation of NG‐1 with C60 by taking advantage of their respective concave and convex geometries.

Experimental corroboration for the complexation of, for example, curved NG‐1 and C60 was performed by 1H‐NMR titration. In an NMR sample tube, a 5×10−4 M [D8]toluene solution of NG‐1 was prepared. Sequential additions of a solution of C60 (2×10−3 M) and NG‐1 (5×10−4 M) resulted in 1H‐NMR spectra, which are shown in Figure 3 a. Several signals shift down‐field and prompt to the successful formation of NG‐1⋅C60. In this way, signals at δ=9.36 (Ha) and δ=7.91 (Hd) underwent shifts to lower fields by Δδ Ha=−0.02 ppm and Δδ Hd=−0.05 ppm, respectively. Signals at δ=9.24 (Hb) and δ=8.76 (Hc) experienced shifts to higher fields Δδ Hb=0.04 ppm and Δδ Hc=0.05 ppm. 2D‐NMR experiments were used to assign these signals (Figure 3 b). Ha, Hc, and Hd match those protons, which are located at the interface between the corannulene and the rest of the π‐system. In other words, in close proximity to the seven‐membered ring, where the curvature is most significant, and where concave‐convex interactions between NG‐1 and C60 are strongest.

Figure 3

a) 1H NMR spectra of NG‐1 with sequential additions of C60. b) Signal assignment of NG‐1 by 2D NMR experiments.

The association constants have been calculated taking three different complex stoichiometries, namely 1:1 (NG‐1⋅C60), 1:2 (NG‐1⋅(C60)2), and 2:1 ((NG‐1)2⋅C60), into account. Considering a 1:1 stoichiometry in NG‐1⋅C60, complexes are formed between the π‐system containing the concave surface of corannulene and C60 by means of concave‐convex interactions. Here, the proton‐shifts for Ha, Hb, Hc, and Hd fit with the model with a low value of Sum of Squared Residuals SSR1:1=1.0024×10−4. From this, we conclude a rather tight fit of the model with the data. In this model, the calculated association constant is K a=1.17×103 M−1 and the mole fraction of NG‐1⋅C60 is 0.5 when 2.2 equivalents of C60 were added (Figure S2).

For NG‐1⋅(C60)2, additionally to the aforementioned interactions in NG‐1⋅C60, a second set of concave‐convex interactions occur at, however, the opposite face of NG‐1. It is based on the negative curvature that confers the saddle‐shape of NG‐1. CH‐π interactions between Bu groups of NG‐1 and C60 are likely to play an important role. The data fit brings SSR1:2=5.42×10−5, which is somewhat better than the 1:1 stoichiometry fittings. The calculated association constants are K a1=1.69×103 M−1 and K a2=1.16×103 M−1. Lastly, a 2:1 stoichiometry model was studied. (NG‐1)2⋅C60 implies a complexation between the π‐extended concave face of corannulene of two molecules of NG‐1 and one C60. Overall, the fullerene remains inside the nanographene cage (SSR2:1 is 5.14×10−5). From the calculated association constants of K a1=1.71×103 M−1 and K a2=3.17×103 M−1 we derive the coexistence of three plausible species NG‐1⋅C60, NG‐1⋅(C60)2, and (NG‐1)2⋅C60 in almost the same concentration when about 1.5 equivalents of C60 were added (Figure S4).

For NG‐1⋅C60, two conformers are possible. The first isomer is formed by non‐covalent interactions between C60 and the concave surface of corannulene, whereas in the second isomer C60 interacts with the saddle‐shaped surface on the opposite side of NG‐1. The latter is 7.9 kcal mol−1 less stable than the former according to our calculations (Figure S5, SI). In NG‐1⋅(C60)2 complex, each side of NG‐1 interacts with one C60, thus only one isomer is possible, while for (NG‐1)2⋅C60 multiple isomers with different mutual orientations of the fragments we found. Figure S5, SI, depicts several isomers found for these complexes. The relative stability of the conformers of NG‐1⋅C60, NG‐1⋅(C60)2, and (NG‐1)2⋅C60 complexes in the ground state (GS) was assessed by DFT calculations at the BLYP‐D3(BJ)/def2‐SVP level of theory (see SI for details). The structures of the most stable conformers are shown in Figure 4.

Figure 4

Optimized structures and interaction energies of the most stable conformers obtained at the BLYP‐D3(BJ)/def2‐TZVP//BLYP‐D3(BJ)/def2‐SVP level of theory.

To estimate the stability of each complex, the interaction energies (ΔE int) between NG‐1 and C60 were computed. For the most stable conformers of NG‐1⋅C60, NG‐1⋅(C60)2, and (NG‐1)2⋅C60, ΔE int is found to be −39.83, −66.98, and −84.94 kcal mol−1, respectively. We infer that these interaction energies are somewhat overestimated. For NG‐1⋅C60, accounting for the basis set superposition error (BSSE) reduces the interaction energy by ca. 2 kcal mol−1, whereas the use of hybrid and range corrected functionals results in lower stabilization energies, which vary in the range from 5 to 9 kcal mol−1, (Table S2, SI). Similar to previously reported carbon‐rich complexes, the interaction energy in (NG‐1)2⋅C60 is superadditive. In particular, the total stabilization energy is 5 kcal mol−1 larger than the sum of interaction energies in two NG‐1⋅C60 (Figure S6, SI).

Taking into account the electron acceptor properties of C60, we looked at the charge separation between NG‐1 and C60 in the ground state. The population analysis provides no evidence for significant charge transfer between the fragments (Table S3, SI). In addition, we performed an energy decomposition analysis (EDA), which enables to divide the interaction energy into four components: electrostatic (ΔE elstat), Pauli repulsion (ΔE Pauli), orbital (ΔE oi), and dispersion correction (ΔE disp) (Table S4, SI). Binding forces in these complexes were found to be similar. The largest contribution to the attraction between the two constituents is the dispersion term (around 60 %), which is followed by the electrostatic term (26–29 %) and the orbital interaction term (13–15 %). The topology of the non‐covalent interactions was described using the NCI calculations (Figures S7 and S8, SI).

The simulation of the lowest 80 excited states was carried out by the TDA‐DFT method with the CAM‐B3LYP‐D3(BJ)/def2‐SVP//BLYP‐D3(BJ)/def2‐SVP Scheme.[ , ] To characterize the excited states, the complexes were divided into NG‐1 and C60 fragments (two in the case of NG‐1⋅C60 or three in the cases of NG‐1⋅(C60)2 and (NG‐1)2⋅C60. Within this Scheme, three types of excited states can be identified: (1) locally excited (LE) states, in which the excitation is mainly localized only on a single fragment and the degree of charge separation is less than 0.1 (CS value <0.1 ); (2) charge separated (CS) states with a significant amount of electron density transferred between the fragments (CS value >0.8 ) and (3) mixed states with a significant contribution of both LE and CS (0.1 < CS value <0.8 ).

In the gas phase, the excitation energies range from 2.46 to 4.25 eV (Table S5, SI). In all complexes, the LE state on C60 (LE1) has the lowest energy. The LE states with an exciton on NG‐1 (LE2) are 0.58–0.65 eV higher in energy. Among the 80 lowest excited states, we found only CS states that correspond to the electron transfer from NG‐1 to C60. The energy of the lowest CS state varies slightly from 2.65 to 2.72 eV depending on the stoichiometry of the complex. The direct population of the CS states due to light absorption is unlikely due to their low oscillator strength (f) of <0.001. However, the CS states can be generated by the decay of LE states. The highly absorptive states, with an oscillator strength ranging from 0.32 to 0.74 are found in the complexes at 3.83–3.89 eV (Table S5, SI). The Kohn–Sham molecular orbitals representing the LE and CS states are shown in Figures S9–S11, SI.

A COSMO‐like model was applied to estimate the influence of a polar environment on the electronic excitations. Benzonitrile (BZN) was taken as the solvent. The ground state (GS) solvation energies of NG‐1⋅C60, NG‐1⋅(C60)2, and (NG‐1)2⋅C60 are found to be −0.35, −0.37, and −0.67 eV, respectively.

As expected, the dipole moment and solvation energy of the LE1 and LE2 states are similar to those in the ground state (Table S6, SI). In contrast, the molecular dipole moment increases considerably as a result of the CS excitations. For NG‐1⋅C60, NG‐1⋅(C60)2, and (NG‐1)2⋅C60, the ΔμCS is 22.5, 22.4, and 16.1 D, respectively. This in turn leads to a better stabilization of CS states by the polar environment as compared to the LE states (Figure 5). The calculations revealed that the CS state energy in the NG‐1⋅C60, NG‐1⋅(C60)2, and (NG‐1)2⋅C60 complexes decreases by 0.65, 0.63, and 0.89 eV, correspondingly. As seen in Figure 5, the solvent stabilization of the excited states makes the charge separation from LE1 energetically favorable. The LE state energies differ by less than 0.1 eV when compared to those of isolated C60 and NG‐1 units (Table S7, SI).

Figure 5

Relative energies of LE and CS states of NG‐1⋅C, NG‐1⋅(C, and (NG‐1) complexes computed in vacuum (VAC) and benzonitrile (BZN). The CS states correspond to electron transfer from NG‐1 to C60 fragment.

The rate of electron transfer (k ET) was calculated using the Marcus equation. Within this approach, the rate is controlled by three parameters: electronic coupling V ij of initial and final states, reorganization energy λ, and reaction Gibbs energy ΔG 0. Using the computed data listed in Table 1, we estimated the rates for the charge separation.

Table 1

Gibbs energy ΔG 0, electronic coupling |V ij|, reorganization energy λ, ET rate k ET, and characteristic time τ for charge separation in the NG‐1⋅C, NG‐1⋅(C, and (NG‐1) complexes in benzonitrile.

Complex	Transition	ΔG 0[a] [eV]	|V ij| [eV]	Reorg. Energy [eV]	k ET [s−1]	τ [ns]
NG‐1⋅C60	LE→CS	−0.063	6.01×10−3	0.545	1.30×1010	0.08
NG‐1⋅(C60)2	LE→CS	−0.091	4.35×10−3	0.440	1.98×1010	0.05
(NG‐1)2⋅C60	LE→CS	0.035	3.40×10−3	0.570	2.10×109	0.48

[a] Gibbs energy difference between LE and CS states in benzonitrile.

NG‐1⋅C

NG‐1⋅(C

(NG‐1)

The charge separation is characterized by a negative ΔG 0, except for (NG‐1)2⋅C60, where ΔG 0 is almost zero, and it occurs in the normal Marcus regime (|ΔG 0| < λ) on the picosecond time scale. Note that the charge transfer reaction in NG‐1⋅C60 and NG‐1⋅(C60)2 is significantly faster than in (NG‐1)2⋅C60.

To shed light onto the photo‐physical consequences that stem from the interaction between NG‐1 and C60 we turned to steady‐state and time‐resolved investigations. The starting point were titrations, in which variable amounts of C60 up to 4.0×10−5 M were added to a constant concentration of NG‐1 (2.0×10−6 M) and the absorption differences were subsequently recorded. No particular changes were, however, noted. As a matter of fact, the absorption spectra of NG‐1 in the presence of C60 are best described as the simple superimposition of the individual constituents. An independent confirmation came from subtracting any contributions from C60 along every step of the titrations. Here, it is simply the unaltered NG‐1 absorptions that are discernible. Quite similar is the outcome of the titrations when following the fluorescence rather than the absorption. In fact, the NG‐1 fluorescence is subject to a minimal decrease, without showing any new fluorescent characteristics. Contributions from inner filter effects rendered it impossible to quantify the static interactions between NG‐1 and C60. Considering the binding constants as they were determined in the NMR experiments (see above) complex formation under the conditions used in our spectroscopic experiments is disfavored.

Information regarding the dynamic interactions, that is, energy versus electron transfer between NG‐1 and C60, came from time‐resolved transient absorption measurements on the ns‐time scale (ns‐TAS). Initially, NG‐1 was probed in 460 nm photo‐excitation experiments in the absence of any C60 (Figure S12, SI). The presence of two dominant species was derived from the ns‐TAS 3D heat maps. It is, on the one hand, the singlet excited state of NG‐1 with a maximum at 665 nm. On the other hand, maxima at 550 and 690 nm as well as a minimum at 615 nm are ascribed to the triplet excited state of NG‐1.

All characteristics concluded from differential absorption spectra are in excellent agreement with the species associated spectra (SAS) obtained from Global Target Analysis. Our two‐species model is based on the lowest singlet and triplet excited states. Importantly, any branching coefficients were derived from FQY, while assuming that contributions from internal conversion are minor. All lifetimes are summarized in Table 2.

Table 2

Summary of the transient lifetimes determined by global target analysis for different molar ratios of NG‐1 and C60.[a]

NG‐1⋅C60	1:0	1:5	1:10	1:15	1:25
Chlorobenzene
S1 (NG‐1)	4.44 ns	4.44 ns	4.44 ns	–	4.44 ns
T1 (NG‐1)	22.3 μs	3.90 μs	2.07 μs	–	0.87 μs
T1 (C60)	–	26.8 μs	16.0 μs	–	31.4 μs
Benzonitrile
S1 (NG‐1)	5.59 ns	5.59 ns	5.59 ns	5.59 ns	5.59 ns
T1 (NG‐1)	>100 μs	6.07 μs	4.26 μs	3.49 μs	2.69 μs
NG‐1.+ ‐C60 .−	–	>100 μs	>100 μs	43.5 μs	22.9 μs

[a] In chlorobenzene and benzonitrile after laser excitation at 460 nm. The lifetime of the NG‐1 singlet excited state was fixed during the global target analysis, since it only showed small deviations.

–

–

–

–

–

Adding C60 to NG‐1 solutions to realize different NG‐1‐to‐C60 ratios of 1:5, 1:10 as well as 1:25 and photo‐exciting them at 460 nm led to some changes in the ns‐TAS 3D heat maps (Figure 6). Both the singlet and triplet excited states of NG‐1 dominate the 3D heat maps, albeit with somewhat shorter lifetimes for the triplet excited state. As a major change, another species evolved on the long timescales, that is, >1 μs. It is linked to the faster decay of the NG‐1 triplet excited state. Intriguingly, an analysis of the 3D heat maps reveals that the nature of this newly formed species is solvent dependent. In chlorobenzene, for example, the spectroscopic fingerprints of the C60 triplet excited state are seen to evolve at 750 nm. In benzonitrile, it is the one‐electron reduced form of C60 that is identified by its 1080 nm fingerprint (Figure 7). Global target analyses allowed deconvoluting the spectra for each component. Important is the fact that we took the ratio of the extinction coefficients of NG‐1 and C60 at the excitation wavelength to account for the population of the C60 triplet excited state via ISC into our model. Such a mechanistic refinement helped in terms of the SAS quality of the NG‐1 singlet and triplet excited states as the first and second species, respectively.

Figure 6

a) Differential transient absorption spectra of NG‐1 (1.0×10−5 M) and C60 (1.0×10−4 M) in a 1:10 molar ratio at time delays between 0 and 70 μs after laser excitation at 460 nm in argon purged chlorobenzene. b) SAS obtained from global target analysis. c) Determination of the bimolecular rate constant k 2 of the triplet excited state lifetime of NG‐1 with increasing amount of C60.

Figure 7

a) Differential transient absorption spectra of NG‐1 (1.0×10−5 M) and C60 (1.0×10−4 M) in a 1:10 molar ratio at time delays between 0 and 70 μs after laser excitation at 460 nm in argon purged benzonitrile. b) SAS obtained from global target analysis. c) Determination of the bimolecular rate constant k 2 of the triplet excited state lifetime of NG‐1 with increasing amount of C60.

The SAS of the third species bears in chlorobenzene the features of the C60 triplet excited state at 750 nm, while those of the C60 radical anion at 1080 nm were noted in benzonitrile. Using the pseudo first‐order rate constants and treating them as a function of C60 concentrations afforded second‐order rate constants for the energy transfer in chlorobenzene and the electron transfer in benzonitrile of 4.45×109 and 1.36×109 M−1 s−1, respectively. Both of them are nearly diffusion‐controlled.

Conclusion

The complexation between a corannulene‐based molecular nanographene NG‐1 with both, negative and positive curvature and C60 was achieved. The 1H NMR titration performed by the addition of a solution of C60 to a solution of NG‐1 showed the major shift of the protons close to the union between corannulene and the π‐extended system, where the curvature is the strongest. Concave‐convex interactions between the concave surface of the corannulene moiety of the molecular nanographene and C60 can result in the formation of the complex NG‐1⋅C60 with 1:1 stoichiometry and an association constant K a=1.17×103 M−1. Furthermore, the saddle shape gathered by the negative curvature allows the formation of concave‐convex interactions with another molecule of C60 conducing to the formation of the complex NG‐1⋅(C60)2 with 1:2 stoichiometry and association constants K a1=1.69×103 M−1 and K a2=1.16×103 M−1. However, the best fit for the data extracted from the 1H NMR titration is for the complex (NG‐1)2⋅C60 with 2:1 stoichiometry where a C60 molecule remains encapsulated between the concave surface of the corannulene of two molecules of NG‐1. The measured association constants are K a1=1.71×103 M−1 and K a2=3.17×103 M−1.

Theoretical calculations of the stability of these complexes were performed computing the interaction energies (ΔE int) between NG‐1 and C60. For the most stable conformers of NG‐1⋅C60, NG‐1⋅(C60)2, and (NG‐1)2⋅C60, ΔE int is found to be −39.83, −66.98, and −84.94 kcal mol−1, respectively. Furthermore, population analysis lacked any evidence for a significant charge transfer between the fragments in the ground state. However, in the calculations of the excited states for the complexes with the three stoichiometries, we found: 1) locally excited (LE) states, in which the excitation is mainly localized on one fragment 2) Charge separated (CS) states with a significant amount of electron density transferred between the fragments and, 3) mixed states with a significant contribution of both LE and CS. The charge transfer reaction for NG‐1⋅C60 and NG‐1⋅(C60)2 is significantly faster than that for (NG‐1)2⋅C60.

Titration absorption and fluorescence studies do not show the formation of any complex, because the small association energies for the complexation reactions determined by NMR do not allow the complexation under the high dilution conditions of these spectroscopic methods. Time‐resolved transient absorption measurements on the ns‐time scale of solutions of NG‐1 in the presence of C60 indicate solvent dependent, bimolecular processes. In chlorobenzene, it is triplet excited state energy transfer to yield the C60 triplet excited state. In polar benzonitrile, the detection of one‐electron reduced C60 confirms electron transfer.

This study validates the use of curved nanographenes as interesting and singular hosts for further concave‐convex supramolecular complexations and paves the way to the use of other nanographenes where shapes can play an essential role. Furthermore, molecular nanographenes exhibit appealing chemical, chiroptical and photophysical properties, which can be skillfully used in the formation of unprecedented non‐covalent complexes in the search for amazing controlled properties.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Click here for additional data file.