The Hunter Falls Prey: Photoinduced Oxidation of C60 in Inclusion Complex with Perfluorocycloparaphenylene.

Perfluorocycloparaphenylenes (PFCPPs) are cycloparaphenylenes (CPPs) in which all hydrogen atoms have been replaced by fluorine atoms. Like CPPs, PFCPPs are highly strained, hoop-shaped π-conjugated molecules. In this article, we report a computational modeling of photoinduced electron transfer processes in the inclusion complex of PF[10]CPP with C60 fullerene. Its unique feature is the favorable electron transfer from C60 to the host molecule. The photooxidation of C60 is predicted to occur on a sub-nanosecond timescale. The PF[10]CPP⊃C60 dyad is the first nanoring-fullerene complex in which C60 acts as an electron donor in the photoinduced charge separation.

Introduction

The importance of converting sunlight into electricity and chemical fuels cannot be overstated. A lot of attention and efforts have been paid to design and prepare model compounds that mimic natural photosynthetic systems. Generation of long‐lived charge‐separated (CS) states with a high quantum yield and a long distance between radical ions to prevent their recombination is of key essence for such donor‐acceptor systems. Since their discovery in 1985, fullerenes have attracted considerable attention as efficient electron acceptors in photovoltaics due to their properties such as a narrow HOMO–LUMO gap and good electron transportability. However, there are several known disadvantages of fullerenes as a component of photovoltaic devices, which can dramatically decrease the efficiency of photocurrent generation. In particular, some restrictions in functional tuning at the molecular level and a tendency towards self‐aggregation make it difficult to develop materials with desired properties. On the other hand, due to their chemical nature and spherical shape, fullerenes are an ideal guest molecule to form stable complexes with various macrocyclic host molecules such as cyclodextrines, calix[8]arenes or cycloparaphenylenes (CPP) because of the concave‐convex complementarity. The formation of host‐guest complexes effectively suppresses fullerene aggregation and thereby facilitates the conversion of light into electricity. The significant advances in CPP chemistry over the past decade have enabled the synthesis of CPPs with the diameter from 7 to 28 Å, that are suitable hosts for many fullerenes.[ , ] A number of supramolecular donor‐acceptor (DA) complexes of CPP with various fullerenes including endohedral species have been reported to be capable of photoinduced electron transfer (PET). In such complexes, CPPs act as an electron donor.

Recently, Itami et al. reported the synthesis and X‐ray characterization of perfluorocycloparaphenylenes (PFCPPs) of different sizes, which form a new class of ring‐shaped perfluoroarenes. PFCPPs are CPPs, in which all hydrogen atoms are replaced by fluorine atoms. This substitution strongly influences electronic properties of CPPs. Suitable size of ring cavity and increase in electron affinity through perfluorination of CPP encouraged us to investigate PF[10]CPP in combination with C. Electronic and photoinduced electron transfer properties of PF[10]CPP⊃C host‐guest complex were investigated using time‐dependent density‐functional theory (TD‐DFT) and compared with those of the well‐studied [10]CPP⊃C complex. Our results clearly demonstrate that PF[10]CPP⊃C has unique PET properties. In the complex, PF[10]CPP acts as a strong electron acceptor while C is an electron donor, in contrast to what was previously observed in nanoring‐fullerene dyads.

Computational Details

Geometry optimizations were performed employing BLYP exchange−correlation functional, Ahlrichs’ def2‐SVP basis set, and the resolution of identity approximation as implemented in the program ORCA 4.1.2. The formation energy of the complexes as well as their strain energies were computed with the BLYP/def2‐TZVP//BLYP/def2‐SVP scheme. The canonical energy decomposition analysis (EDA) was performed using the Amsterdam Density Functional (ADF) program. Electronic structure calculations and vertical excitation energies were calculated using the Tamm‐Dancoff approximation (TDA), the range‐separated functional CAM‐B3LYP, the def2‐SVP basis set, and the program Gaussian 16 (rev. A03). Recently, some of us have benchmarked different density functional approximations and showed that CAM‐B3LYP functional is the best choice for evaluation of charge‐transfer rates in fullerene‐based materials. The empirical dispersion D3 correction was computed using the Becke‐Johnson damping scheme. The population analysis and calculation of Hirshfeld and CM5 charges were carried out with the Gaussian 16 program. The topological analysis of the electron density distribution was conducted using the “Quantum Theory of Atoms in Molecules” (QTAIM). The AIMALL suite of programs was applied to evaluate the bond critical point properties and associated bond descriptors. To visualize molecular structures and frontier molecular orbitals, the program Chemcraft 1.8 was used. Details on the analysis of excited states, calculation solvent effects, electron transfer rates, reorganization and interaction energies can be found in the Supporting Information.

Results and Discussion

Structure and Ground State Properties

First, we compared the electronic properties of PF[10]CPP and [10]CPP, and their binary complexes with C (Figure 1). As seen in Figure 1, HOMO and LUMO energies of PF[10]CPP and [10]CPP nanorings differ significantly. Replacing hydrogen atoms with fluorine atoms lowers the HOMO energy by more than 2 eV, while the LUMO energy decreases by about 1 eV. In the complexes, a significant difference in the electronic properties is observed despite their structural similarity. The HOMO of [10]CPP⊃C is localized on the host unit, while the HOMO of PF[10]CPP⊃C is localized on C. In both complexes, LUMO is localized on C. The formation of complexes has a rather weak effect on the HOMO and LUMO energies of the individual fragments. In particular, HOMO energies remain almost unchanged (within 0.1 eV). A shift in the LUMOs does not exceed 0.3 eV (Figure 1). These small differences in orbital energies by passing from the individual fragments to the dyads indicate the lack of charge separation in the ground state (GS). Also, the population analysis does not reveal any significant charge transfer between the host and guest molecules in both systems (Table S1 in SI).

Figure 1

HOMO and LUMO energies of PF[10]CPP⊃C and [10]CPP⊃C and their subunits.

To estimate the stability of the complexes, the interaction energy (ΔEint) between the nanorings and fullerene was computed. For PF[10]CPP⊃C and [10]CPP⊃C systems, ΔEint was found to be −47.7 and −56.8 kcal/mol. It can be assumed that PF[10]CPP⊃C complex is formed in solution, since fullerene complexes with a noticeably lower interaction energy are known. For example, ΔEint for the complex of C with π‐extended corannulene bowl is only −31 kcal/mol, but the existence of such a complex in solution has been established experimentally. The dispersion interactions play a major role in the stability of the complex, as shown earlier for systems of a similar nature.[ , ] To get more insight, we performed the energy decomposition analysis (EDA) of the interaction energy using a Morokuma‐type method. Within this scheme, the interaction energy is divided into four components: Pauli repulsion (ΔEPauli), electrostatic (ΔEelstat), orbital interaction (ΔEoi), and dispersion (ΔEdisp) terms (see computational details in SI). The EDA results are merged in Table 1. The destabilizing term (Pauli repulsion) for PF[10]CPP⊃C is significantly larger than for [10]CPP⊃C (131.1 kcal/mol vs. 82.1 kcal/mol). Among the intermolecular attraction (electrostatic, orbital, and dispersion interactions), the dispersion term dominates and amounts to 54 and 62 %, while the electrostatic term is 32 and 25 % for PF[10]CPP⊃C and [10]CPP⊃C, respectively. The orbital interactions term is similar (14 and 12 %).

Table 1

EDA results for PF[10]CPP⊃C and [10]CPP⊃C complexes.[a]

Complex	Energy terms, [kcal/mol]
	ΔEPauli	ΔEelstat	ΔEoi	ΔEdisp	ΔEint
[10]CPP⊃C60	82.07	−35.20 (25.3 %)	−17.25 (12.4 %)	−86.46 (62.3 %)	−56.84
PF[10]CPP⊃C60	131.08	−56.28 (31.5 %)	−25.32 (14.1 %)	−97.20 (54.4 %)	−47.73

[a] The percentage contributions to the sum of attraction energies (ΔEelstat+ΔEoi+ΔEdisp) are given in parentheses.

[10]CPP⊃C

PF[10]CPP⊃C

To explain the noticeable difference in the Pauli repulsion, we compared the geometrical structures of the complexes. The guest molecule in both complexes is the same (C). Its effective radius, i. e. mean distance from center to each carbon atom, Reff=5.15 Å. Thus, we focused our attention on the hosts, PF[10]CPP and [10]CPP. The most stable structure of the nanorings corresponds to an alternating zigzag orientation of phenylene units. The radii for PF[10]CPP and [10]CPP are equal to 6.86 and 6.89 Å, respectively. The formation of the complexes insignificantly affects their size; the radius of PF[10]CPP increases from 6.86 to 6.90 Å, while the radius of [10]CPP decreases from 6.89 to 6.87 Å (Figures S1 and S2 in SI). Also, there are some changes in the twist angle between phenyl rings. For [10]CPP, they do not exceed 1.5° (the angles change from 30° to 28.5–29°). In the free PF[10]CPP, the twist angle is equal to 53°, it decreases to 47° due to the inclusion of C which leads to a deformed structure of the host. In turn, a noticeable difference in the distance between fullerene and hydrogen/fluorine atoms is observed by increasing the twist angles. For example, the difference for close and distant hydrogen/fluorine atoms in [10]CPP⊃C is about 1.2 Å, while for PF[10]CPP⊃C it becomes 2 Å (Figure S3 in SI). The strain energy of the considered nanorings was estimated using a homodesmotic reaction proposed by Itami and co‐workers. The strain energies in [10]CPP and [10]CPP⊃C are 52.8 and 53.4 kcal/mol, respectively. For PF[10]CPP, the strain energy changes from 54.9 to 57.5 kcal/mol (Table S1 in SI). Thus, the weaker intermolecular interactions in PF[10]CPP⊃C compared to those in [10]CPP⊃C can be explained by stronger host‐guest repulsion (see ΔEPauli in Table 1) caused by the change of H by F atoms, the latter with three more electron pairs in the PF[10]CPP fragment.

A topological analysis based on the QTAIM theory was used to obtain additional information about the host‐guest interaction in the complexes. The electron density, its Laplacian, and other topological parameters at bond critical points (BCPs) were calculated (see Table S2). The analysis confirmed that the interactions between fragments in the PF[10]CPP⊃C and [10]CPP⊃C complexes correspond mostly to π⋅⋅⋅π interactions. PF[10]CPP⊃C is characterized by a larger number of BCPs as compared to [10]CPP⊃C (20 vs. 17). In [10]CPP⊃C, BCPs were found only between carbon atoms of subunits. However, in PF[10]CPP⊃C, BCPs were additionally detected between fluorine atoms and carbon atoms of fullerene, which is associate with different twist angles between phenyl rings in the complexes. The descriptors of BCPs indicate that the π⋅⋅⋅π interaction between fragments in [10]CPP⊃C is stronger than in PF[10]CPP⊃C. The characteristics of BCPs for C⋅⋅⋅C and F⋅⋅⋅C interactions in PF[10]CPP⊃C are similar. The QTAIM molecular graphs for PF[10]CPP⊃C and [10]CPP⊃C are given in Figure S4, SI. The topology of the host‐guest interactions in the complexes was also described using non‐covalent interaction index (NCI). The shape of NCI isosurfaces clearly demonstrates the difference between the intermolecular interactions in PF[10]CPP⊃C and [10]CPP⊃C. In the case of [10]CPP⊃C, the isosurface is fairly uniform between C and [10]CPP, while for PF[10]CPP⊃C it has a wavy character with maxima in the region of closely spaced fluorine atoms. The reduced density gradient (RDG) plots and NCI isosurfaces are presented in Figures S5 and S6, SI.

Singlet Excited States

The significant differences in HOMOs and LUMOs described above for the PF[10]CPP⊃C and [10]CPP⊃C complexes suggest that PET properties of the complexes should also be different. The contributions of the guest and host molecules to the electronic density of the lowest 80 excited states were analyzed for each complex. Three types of excited states were identified: (1) locally excited (LE) states, where exciton is mostly localized either on the guest (LEGuest) or on the host molecule (LEHost) and charge transfer is smaller than 0.1 (CT<0.1  ); (2) charge transfer (CT) states showing significant charge separation (CT>0.8 e); and (3) mixed states, where both LE and CT states contribute substantially (0.1 

In the gas phase, the 80 lowest vertical singlet excitation energies of [10]CPP⊃C range from 2.51 to 4.54 eV. The lowest‐lying excited state at 2.51 eV is LEGuest state, which is associated with a HOMO‐3→LUMO+2 transition (Table 2). The lowest LEHost state localized on [10]CPP was found at 3.43 eV (1 eV higher in energy compared to LEGuest) and relates to a HOMO→LUMO+6 transition. Among the studied excited states only one type of CT was found: [10]CPP , generated by electron transfer from [10]CPP to C. This CT state associated with a HOMO→LUMO transition is characterized by CT=0.99 and lies about 0.2 eV higher than the first excited state.

Table 2

Excitation energies (Ex, eV), main singly excited configuration (HOMO(H)–LUMO(L)) and its weight (W), oscillator strength (f), extent of charge transfer (CT, e) or localization of exciton (Χ) computed for PF[10]CPP⊃C and [10]CPP⊃C complexes in the gas‐phase (VAC) and dichloromethane (DCM).

	Supramolecular host‐guest systems
	[10]CPP⊃C60		PF[10]CPP⊃C60
	VAC	DCM	VAC	DCM
	LEGuest (C60 )
Ex	2.508	2.501	2.486	2.476
Transition (W)	H‐3−L+2 (0.63)	H‐3−L+2 (0.27)	H‐4−L (0.84)	H−L (0.76)
f	<0.001	<0.001	<0.001	<0.001
Χ	0.952	0.948	0.937	0.936

[a] mixed states with significant contributions of LE and CT; [b] states of interest are not found within 80 lowest excited states.

[10]CPP⊃C

PF[10]CPP⊃C

LEGuest (C)

LEHost ([10]CPP/PF[10]CPP)

3.428

3.433

4.056

4.057

H−L+6 (0.30)

H‐4−L+5 (0.09)

H‐5−L+3 (0.49)

H‐5−L+3 (0.50)

0.939

0.938

0.905

0.906

Most absorptive (MA) transition[a]

3.817

3.823

4.462

4.430

H−L+7 (0.23)

H−L+7 (0.28)

H‐5−L+4 (0.15)

H‐5−L+4 (0.19)

1.303

2.624

0.647

1.444

Localization

[10]CPP

[10]CPP

PF[10]CPP

PF[10]CPP

CT

0.237

0.217

0.250

CT1 ([10]CPP/PF[10]CPP→C)

2.736

4.138

H−L (0.96)

H−L (0.96)

H‐5−L (0.70)

H‐5−L (0.92)

0.008

CT

0.982

0.983

0.855

0.946

CT2 (C→[10]CPP/PF[10]CPP)

n/f[b]

3.818

H‐4−L+3 (0.76)

H−L+3 (0.67)

CT

0.803

The 80 lowest‐lying vertical singlet excited states of PF[10]CPP⊃C are located between 2.49 eV and 4.53 eV. The energy of the LEGuest states in both complexes is almost identical. However, the energy of the LEHost in PF[10]CPP⊃C, 4.06 eV, is significantly higher than in [10]CPP⊃C. In this complex, two types of CT states were found. The first state, PF[10]CPP (CT1), is similar to that in [10]CPP⊃C, while the other CT state, PF[10]CPP (CT2), is generated by electron transfer from C to PF[10]CPP. Important to note that the energy of CT2 state is about 0.3 eV lower than the energy of CT1. Figures S7 and S8 in SI show natural transition orbitals that represent the lowest LE and CT states in the complexes.

Effects of Environment and Electron Transfer Rates

A well‐proven COSMO‐like model with dichloromethane (DCM) as the solvent was applied to estimate the effect of polar environment on electronic excitations. The GS solvation energy of PF[10]CPP⊃C and [10]CPP⊃C complexes is equal to −0.44 and −0.40 eV, respectively. The dipole moment is 0.04 and 0.01 D, respectively. It is well known that the solvation energy of LE states is quite similar to that of the ground state, but it is usually much stronger for CT states. The changes in the dipole moment, when going from GS to LE states, are quite small (∼0.1 D), and, as expected, the computed solvation energies of the GS and LE states are very similar. Considering the high symmetry of the studied complexes leading to efficient charge delocalization, it can be assumed that the solvation energy of the CT states is not much higher than that of the GS. Besides, the difference in the dipole moment of CT and GS states is small and does not exceed 0.04 and 0.20 D for PF[10]CPP⊃C and [10]CPP⊃C complexes, respectively. Consequently, the solvation energy of the CT1 state is only −0.70 eV for PF[10]CPP⊃C and −0.62 eV for 10]CPP⊃C. The solvation energy of the CT2 state of PF[10]CPP⊃C is −0.48 eV. The stronger solvent stabilization of the CT1 state compared to the CT2 is due to the different CT values. Indeed, the CT1 state is characterized by almost complete charge transfer (CT=0.95 ), while in the CT2 only 0.80 is transferred. Figure 2 displays the energies of the GS, LE, and CT states in the gas phase and DCM. The changes in the dipole moment and solvation energies for all considered excited states are given in Table S3.

Figure 2

(a) Energies of LE and CT states (in eV) computed for PF[10]CPP⊃C and [10]CPP⊃C complexes in vacuum (VAC) and dichloromethane (DCM); (b) Energy levels and ET rates for PF[10]CPP⊃C in DCM.

For both systems, the CT states are characterized by a very weak oscillator strength and, therefore, cannot be directly populated by light absorption. However, they can be generated by a decay of the lowest LE states. For the PF[10]CPP⊃C and [10]CPP⊃C complexes, most absorptive transitions are localized on the host unit. The MA states extremely fast decay by internal conversion to the lowest LEHost. Thus, charge separation process was considered between LEHost and CT state of interest. The rates of charge separation (k) and charge recombination (k) were calculated using the semi‐classical method by Ulstrup and Jortner. Within this approach, the intramolecular relaxation associated with ET is described by an effective vibrational mode, and the rate is controlled by four parameters: electronic coupling of the initial and final states Vij, solvation reorganization energy λs, reaction Gibbs energy ΔG0, and effective Huang‐Rhys factor Seff. The computed parameters and rates in DCM are listed in Table 3. The rates were estimated using the effective frequency of 1600 cm−1, which corresponds to the stretching of C=C bonds. Note that the calculated rates of charge separation do not change significantly by varying the effective frequency from 1400 to 1800 cm−1.[ , ] For PF[10]CPP⊃C, we additionally checked the influence of the frequency choice on the estimated ET rate (Table S4, SI) and found it to be insignificant.

Table 3

Gibbs energy ΔG0 (in eV), electronic coupling Vij (in eV), solvent (λs) and internal (λi) reorganization energy (in eV), Huang‐Rhys factor (S ) and rates k (in s−1) for CS and CR processes computed for PF[10]CPP⊃C and [10]CPP⊃C complexes in DCM.

Complex	Transition	ΔG0[a], eV	|Vij|, eV	Reorg. energy, eV		Seff [b]	k_X , s−1
				λi	λs
[10]CPP⊃C60	LEHost→CT1	−0.914	1.66 ⋅ 10−4	0.158	0.181	0.796	2.20 ⋅ 107
	CT1→GS	−2.519	4.12 ⋅ 10−3	0.134	0.181	0.675	3.90 ⋅ 101
PF[10]CPP⊃C60	LEHost→LEGuest	−0.052	2.05 ⋅ 10−6	0.184	0.009	0.932	1.01 ⋅ 105
	LEHost→CT1	−0.177	6.70 ⋅ 10−5	0.125	0.081	0.630	1.13 ⋅ 108
	LEHost→CT2	−0.276	8.84 ⋅ 10−4	0.185	0.096	0.933	1.67 ⋅ 1010
	CT1→GS	−3.880	2.01 ⋅ 10−4	0.114	0.081	0.575	[2.85 ⋅ 10−13]
	CT2→GS	−3.781	3.99 ⋅ 10−5	0.196	0.096	0.988	[1.18 ⋅ 10−9]

[a] Gibbs energy difference between the given states. [b] Effective value of the Huang‐Rhys factor Seff=λI /ħωeff, where ħωeff is set to 1600 cm−1.

, s−1

[10]CPP⊃C

PF[10]CPP⊃C

The charge separation in [10]CPP⊃C is characterized by a large negative Gibbs energy value. This reaction takes place in the inverted Marcus region (|ΔG0  |>λ) and the estimated CS rate is quite low (2.2 ⋅ 107 s−1). The electron transfer in PF[10]CPP⊃C complex occurs in the normal Marcus regime (|ΔG0|<λ) and is fast. Note that PET, in which C fullerene acts as electron donor generating PF[10]CPP (CT2), is by two orders of magnitude faster than the formation of PF[10]CPP (CT1, C is electron acceptor). The charge recombination reactions take place in a deep inverted Marcus region, and their rates are dramatically lower compared to the charge separation rates, although the rates determined in the deep inverted Marcus region are not reliable. To get more insight, we additionally considered exciton transfer between LEHost and LEGuest states in PF[10]CPP⊃C,. The reorganization energy for exciton transfer is relatively small (Table 3). Thus, the rate of this process determined by the Franck‐Condon factor will be maximum at relatively small negative values of the Gibbs energy. The fastest rate for the exciton transfer between the fragments was found to be equal to 1.0 ⋅ 105 s−1, when the difference in energy between LE states is −0.052 eV. As seen, the exciton transfer process is much slower than the related charge separation reaction and, thus, cannot compete with ET.

Since the electron transfer in the PF[10]CPP⊃C complex is characterized by a rather small negative ΔG0 (Table 3), the charge recombination CT1/CT2→LEHost state can be considered as an alternative deactivation channel of the CT state. As seen in Table S5, the back electron transfer in PF[10]CPP⊃C complexes is significantly slower than charge separation. Thus, we can be sure that back electron transfer will not compete with generation of CT state.

Conclusions

In this work, we have studied the ground and excited‐state properties of the complex formed by the fully fluorinated carbon nanoring and C fullerene. The low LUMO of PF[10]CPP and its ability to delocalize an excess electron make this nanoring an efficient electron acceptor. In the PF[10]CPP⊃C complex, we found an unusual photoinduced electron transfer from C to PF[10]CPP. This seems to be the first example of photooxidation of C in DA complexes with carbon nanohoops.

Author Contributions

O. A. S. Investigation, Formal analysis, Writing – original draft, Writing – review & editing; A. J. S. Investigation, Formal analysis, Writing – original draft, Writing – review & editing; M. S. Supervision, Writing – review & editing, Funding acquisition; A. A. V. Supervision, Writing – review & editing.

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.