The effect of heterocyclic π bridges on second order nonlinear optical properties of compounds formed between ferrocenyl and corannulenyl

Abstract

Compounds containing a donor (D) connected to an acceptor (A) via a heterocyclic π bridge are known as novel materials with excellent nonlinear optical (NLO) properties. Using the ferrocenyl donor and corannulenyl acceptor, we introduce three five-membered azole rings, including imidazole, oxazole, and thiazole compared with phenyl to systematically determine the contribution of the π bridge to the NLO response of the compounds. By calculating the electronic structures, UV-visible absorbance spectra and first hyperpolarizabilities using density functional theory, we found that the stabilities of the endo and exo conformers are almost the same. Two conformers are easy to convert. Furthermore, the maximum absorption wavelengths of heterocyclic compounds show an apparent red-shift compared with the compound with phenyl as π bridge. This trend indicates that the heterocyclic compounds possess the smaller crucial transition energy, which might lead to a larger NLO response. Interestingly, the first hyperpolarizability values of exo conformers are all larger than those of endo conformers because of the larger degrees of charge transfer transition of the exo conformers compared to those of the corresponding endo conformers. Our research provides important evidence for using five-membered heterocycles as π bridges in enhancing the NLO properties of D–π–A type compounds. Therefore, controlling the kinds of π bridges is an important method in the design of new appealing NLO compounds.

---

1. Instruction

Organic molecules with an electron-donating group (D) and an electron-withdrawing group (A) at the opposite ends of a π bridge system are often referred to as the D–π–A model of push–pull systems. These D–π–A type compounds are known as novel materials with excellent nonlinear optical (NLO) applications.^1–11 They serve as critical components for NLO devices¹¹ and can be found as active substances in organic electronics, optoelectronics and conductors.⁴ D–π–A type compounds taking the ferrocenyl (Fc) group as a donor have been the subjects of recent research, due to their attractive electronic and physical properties which make them suitable candidates for NLO-phores.^12–19 Moreover, Fc is bulkier than typical organic NLO donor groups, making the Fc moiety a donor with distinct properties, resulting in novel compound architectures.¹⁴ On the other hand, because of its noteworthy electro-acceptor ability, corannulene (C₂₀H₁₀), the simplest of the fragments of fullerene has been widely studied.^20–29

In 2012, Lentz et al. synthesized the compound with Fc as donor and corannulenyl as acceptor.³⁰ Subsequently, they introduced the phenyl between the Fc and corannulenyl forming D–π–A type compounds. Interestingly, they got two conformers of the D–π–A type compounds,³¹ namely endo and exo (Scheme 1) which show negligible energy difference. It demonstrates that the two conformers can exist stably. This drew our attention, and we will investigate here the NLO properties of this kind of D–π–A type compound and find a method to improve it.

Scheme 1 The endo and exo conformers of the D–π–A type compounds.

It is well-known that five-membered heterocycles such as furans, thiophenes, pyrazoles, and azole derivatives possess considerably less aromatic stabilization than benzene.³² Thus, while the cost in energy to break the aromatic delocalization is less than in benzene, substantial thermodynamic stability remains. This observation is important for organic electronic and optical materials.³³ Many people have studied the compounds containing heterocycles in place of phenyl and obtained larger second-order nonlinearities.^32–35

To improve the NLO response of this kind of D–π–A type compounds (using the ferrocenyl donor and corannulenyl acceptor), we compared three azoles (imidazole, oxazole, thiazole) with phenyl as π bridges, respectively. As a comparison, we named original D–A compounds as endo-1 and exo-1 (endo-1 represents that the attached Fc is in endo conformation with respect to the corannulenyl, whereas exo-1 represents the direct corannulene congener bears the Fc substituent in exo conformation) and named the compound with phenyl as π bridge as endo-2 and exo-2. The three compounds with imidazole, oxazole, thiazole as π bridge are named as endo-3, exo-3, endo-4, exo-4, endo-5, exo-5, respectively. Using Fc donor, corannulenyl acceptor and varied bridges makes it possible to systematically determine the contribution of the π bridge to the NLO response. Continuing our interests in the research of organic molecular materials with substantial NLO properties,^22,36–40 we performed density functional theory (DFT) to calculate the electronic structures, UV-visible absorbance spectra and first hyperpolarizabilities of the endo-n and exo-n (n = 1–5) compounds to mainly address the following issues: (1) if the stabilities of the endo and exo conformers were also nearly the same by the replacement of phenyl with five-membered heterocyclic azole rings? (2) Is there any significant difference between the endo and exo conforms for the NLO properties of the systems? (3) If the first hyperpolarizability values will be improved when the five-membered heterocyclic azole ring as π bridge? We hope to report in this article the design of a novel class of D–π–A compounds.

2. Computational details

Long-range corrected hybrid density functional (ωB97XD) with damped atom–atom dispersion correction is employed for the geometry optimization of all compounds.⁴¹ We have adopted the 6-31G* basis set for C and H and the SDD basis set for Fe in connection with effective core potential (ECP) which reduces the required basis set and includes the scalar relativistic effect (there are 10 electrons in the ECP vs. 16 valence electrons for Fe). Each of the geometries belongs to the minima on the potential energy surface. The high level method and the larger basis set calculations are necessary for the accuracy of the results, especially important for determining the energy orders of endo and exo isomers. Hence, we performed the energy calculation on all compounds through B2PLYP/6-311+G* (Def2-TZVPP for Fe) level.⁴²

It is well-known that the NLO properties depend on the selected computational methods. Therefore, when it comes to the calculation of the first hyperpolarizability, four functionals (CAM-B3LYP, BHandHLYP, ωB97XD and M06-2X) have been chosen. The Coulomb-attenuated hybrid exchange–correlation (CAM-B3LYP) functional is a hybrid functional with improved long-range properties which is Handy and coworkers' long range corrected version of B3LYP using the Coulomb-attenuating method.⁴³ The ωB97XD functional and the Becke's half-and-half LYP (BHandHLYP) functional which have been tested with notable success in the calculation of molecular (hyper)polarizability.^44,45 The fourth one is M06-2X, which presents a high percentage of HF exchange, is an excellent hybrid DFT functional for applications in the first hyperpolarizability of transition metal compounds.⁴⁶ Besides, we also calculated the polarizability (α₀) which is defined as follows:

The first hyperpolarizability (β_tot) is noted as:

The time-dependent density functional theory (TD-DFT) method has nowdays been extensively used to investigate the electronic transition property owing to its efficiency and accuracy.^47–49 To gain an insight into the description of the trend of the first hyperpolarizability, most important transitions and transition energies between the ground and excited states were calculated at the TD-BHandHLYP/6-311+G* (Def2-TZVPP for Fe) level. Corresponding, the UV-vis absorption spectra of compounds endo-n and exo-n (n = 1–5) were investigated by the same method.

In this work, all calculations were carried out using Gaussian 09W program package.⁵⁰

3. Results and discussion

3.1 Geometrical structure

The optimized geometric structures of endo-n and exo-n (n = 1–5) are shown in Fig. 1 (endo and exo conformers were shown for comparison). It is worthy of note that the difference of energy between the endo and exo conformers is small as shown in Table 1, showing that two conformers are easy to convert and both conformers are stable (the detail data are summarized in Table S1†). In order to further validate the results, MP2 method⁵¹ is employed and we obtain the same result which is shown in Table S2.† The bowl depths of endo-1, endo-2 and exo-2 are 0.867 Å, 0.826 Å and 0.904 Å, respectively. Which are virtually identical to those of their experimental values of 0.864 Å, 0.832 Å and 0.895 Å, respectively,^30,31 confirming that the ωB97XD functional is an efficient and reliable functional to optimize the compounds.

Fig. 1 Optimized structures of endo-n and exo-n (n = 1–5).

Table 1 The relative energies (E_rel, kcal mol⁻¹) between endo and exo conformers obtained by the B2PLYP functional

Compound	Erel	Compound	Erel
endo-1	0	exo-1	0.18
endo-2	0	exo-2	−0.06
endo-3	0	exo-3	−0.94
endo-4	0	exo-4	0.26
endo-5	0	exo-5	1.46

---

The bowl depth of pure corannulene is 0.875 Å in experiment,⁵² indicating that the introduction of phenyl between Fc and corannulenyl induces a little deformation of corannulene (0.826 Å and 0.904 Å). However, the bowl depth of corannulene remains almost unchanged by the introduction of imidazole, oxazole and thiazole (more information can be seen in Table S3†). The results demonstrated that the addition of heterocyclic π bridges nearly has no effect on the molecular structure of D–π–A type compound. However, the addition of heterocyclic π bridges whether has effect on the other properties of D–π–A type compound.

3.2 UV-vis absorbance spectrum

Since the π bridges are different, we initially investigate the effect of π bridges on the observed spectral properties. The spectra of endo and exo conformers are nearly the same (Fig. S1†) which indicates the similar electron transition of the isomers (like endo-1 and exo-1). Hence, we only take the data of the endo conformers for example. The UV-visible absorbance spectra of endo-n (n = 1–5) compounds are depicted in Fig. 2 (the corresponding spectra of exo-n (n = 1–5) compounds are depicted in Fig. S2†). The spectral shape of all compounds are similar. However, by introduction of the carbocyclic phenyl and heterocyclics imidazole, oxazole, thiazole spacer between the donor and acceptor, the high-energy absorption peak and low-energy absorption peak (the maximum absorption wavelength (λ_max)) of the compounds show different level of red-shifts. This behavior is clearly demonstrated by comparing endo-n (n = 2–5) with endo-1 in Fig. 2. The red-shifts of D–π–A compounds compared with the D–A compounds indicate that the introduction of phenyl and heterocycle will lead to the smaller transition energy of the D–π–A compounds, which might lead to the larger NLO response of D–π–A compounds. What's more, for D–π–A systems, the λ_max of heterocyclic compounds shows apparent red-shift comparing with the compounds with phenyl as π bridge. This trend indicates that the crucial transition energy of heterocyclic compounds become smaller, which might lead to a larger NLO response of heterocyclic compounds. Thus, it is enough to suggest that heterocyclics imidazole, oxazole, thiazole spacer are more suitable as π bridges compared to the phenyl for the enhancement of NLO properties.

Fig. 2 UV-vis spectra of endo-n (n = 1–5) compounds.

3.3 The polarizability and first hyperpolarizability

In order to investigate the effect of different π bridges on linear and nonlinear optical properties of the systems 1–5, the α₀ and β_tot values were calculated using the CAM-B3LYP, ωB97XD, BHandHLYP and M06-2X functional to confirm the accuracy of results in this work. Fig. 3 depicts the variation of α₀ and β_tot values as the function of the compounds using the four methods. All four functionals show the same trend for α₀ and β_tot values. However, two parameters show different variations. For example, the introduction of π bridges increase the α₀ values. However, the difference of α₀ values of the D–π–A type compounds is very small. Therefore, the effect of different kinds of π bridges on α₀ values is very slight. In contrast, the introduction of π bridges significantly increase the β_tot values. Different kinds of π bridges effectively change the β_tot values which means the π bridges effectively change the NLO response. The β_tot values of systems 3–5 are all larger than that of system 2, the replacement of phenyl between ferrocenyl and corannulenyl by the familiar five-membered rings, such as imidazole, oxazole, thiazole, improves the NLO response. This is consistent with the results obtained in the UV-visible absorption spectra section. Importantly, among the heterocyclic compounds, the β_tot values of system 3 is the biggest. It shifts by a factor of 2 relative to system 2 which we can clearly see from Fig. 3, meaning that system 3 has excellent NLO properties and the imidazole is the best bridge among the three five-membered heterocyclic rings. Besides, because of the smallest β_tot values of system 1 and we mainly studied the NLO properties of D–π–A type compounds rather than the D–A type compound, we won't discuss the system 1 in the latter discussion.

Fig. 3 Polarizability (α₀, a.u.) and first hyperpolarizability (β_tot, a.u.) of endo-n and exo-n (n = 1–5) at CAM-B3LYP, ωB97XD, BHandHLYP, and M06-2X functions.

Since hyperpolarizabilities are derivatives of the molecular energy with respect to the strength of the applied electric field, thus their theoretically calculated values may be sensitive to basis-set feature. Hence, we now turn our attention to the basis set extension effect. We calculated the β_tot values at the BHandHLYP level of theory using the Def2-TZVPP basis set for the Fe (as regards the Fe, one should always have in mind the difficulties in computing NLO properties of molecules containing transition metal atoms⁵³) and the 6-311+G* basis set and its extensions 6-311+G**, 6-311++G*, 6-311++G** basis sets for non-metal atoms, respectively (Table 2). Obviously, the difference of β_tot among different basis sets is very small which could be negligible. Thus, adding polarization and diffuse functions on the 6-311+G* basis set have a very slight influence on the first hyperpolarizabilities of the compounds. Luckily, the choice of the 6-311+G* basis set is adequate.

Table 2 The β_tot values (10⁻³⁰ esu) calculated at the BHandHLYP level using the 6-311+G* basis set and its extensions

Compound	6-311+G*	6-311+G**	6-311++G*	6-311++G**
endo-2	15.2	15.1	15.2	15.2
exo-2	15.6	15.6	15.6	15.6
endo-3	23.3	23.2	23.4	23.3
exo-3	26.5	26.4	26.6	26.5
endo-4	18.0	18.0	18.1	18.1
exo-4	18.4	18.4	18.4	18.4
endo-5	15.9	15.9	16.0	16.0
exo-5	18.8	18.8	18.8	18.8

---

Interestingly, for D–π–A type compounds, the β_tot values of endo and exo conformers show the visible difference that the β_tot values of exo conformers are all larger than the corresponding endo conformers as shown in Fig. 3. However, the differences between the endo and the exo conformers could be negligible for the energetic and spectral properties of the systems which have been discussed in the first two sections. Therefore, we comment in detail on the reasons for it in the latter paragraphs.

To further rationalize the origin of the first hyperpolarizability responses, the two-level model proposed by Oudar and co-workers that links the β_tot values and the low-lying charge transfer transition has been considered.^9,54 According to the two-level model, the β_tot value is proportional to the difference between the dipole moments of the ground state and the crucial excited state (Δμ_gm), as well as the oscillator strength (f_gm), but inversely proportional to the third power of the transition energy (ΔE_gm³). The relative data and major molecular orbital (MO) transition for endo-n and exo-n (n = 2–5) in singlet excited-state transitions are summarized in Table 3, respectively. As we can see from Fig. 4, Δμ_gm/ΔE_gm³ values are proportional to β_tot values in agreement with the two-level model.

Table 3 Excited state transition energies (ΔE_gm, eV), transition dipole moments (Δμ_gm, debye), and major molecular orbital contributions of endo-n and exo-n (n = 2–5) compounds are calculated at the BHandHLYP/6-311+G* (Def2-TZVPP for Fe) level (H = HOMO, L = LUMO, etc.)

Compound	ΔEgm	fgm	Δμgm	MO transition
endo-2	3.870	0.118	3.187	H → L (53%), H−2 → L (14%), H → L+1 (10%)
exo-2	3.868	0.127	3.190	H → L (53%), H−2 → L (13%), H → L+1 (10%)
endo-3	3.635	0.272	8.888	H → L (78%)
exo-3	3.619	0.311	8.863	H → L (79%)
endo-4	3.627	0.309	5.561	H → L (80%)
exo-4	3.601	0.321	5.423	H → L (81%)
endo-5	3.685	0.397	4.770	H → L (78%)
exo-5	3.667	0.434	4.726	H → L (80%)

---

Fig. 4 Relationship between the β_tot (black line) values and the corresponding Δμ_gm/ΔE_gm³ (blue line) values for endo-n and exo-n (n = 2–5) compounds.

The MOs of the crucial transition states were systematically analyzed. To clearly demonstrate the results, the crucial transition MOs of endo-n (n = 2–5) are depicted in Fig. 5. The electronic transition of endo-2 arise from HOMO → LUMO and HOMO → LUMO+1 excitation, which could be attributed to large charge transfer from the D and phenyl π bridge to the A fragment. Since the electronic distribution of HOMO is delocalized over the whole molecule, while the LUMO and LUMO+1 is mainly located at the A moiety. Except for that, there is a little charge transfer from A to phenyl π bridge (HOMO−2 → LUMO), this will offset part of the charge transfer which is from the D and phenyl π bridge to the A fragment. Similar to endo-2, the electronic transition of exo-2 (Fig. S3†) is ascribed as the D and phenyl π bridge to the A transition. Whereas the degree of charge transfer of exo-2 is larger than endo-2. Besides, the charge transfer in the opposite direction (the charge transfer from A to phenyl π bridge) of exo-2 is smaller than the endo-2. The large charge transfer transition from D and phenyl π bridge to the A fragment contribute to the large β_tot values of exo-2. This is the reason for that the β_tot value of exo-2 is larger than that of endo-2.

Fig. 5 Molecular orbitals corresponding to the dominant electron transitions of endo-n (n = 2–5) compounds (H = HOMO, L = LUMO, H−2 = HOMO−2, L+1 = LUMO+1).

For endo-3, the complexity of the charge transfer transition decreases. There is only the large charge transfer transition from the D and imidazole π bridge to the A fragment (HOMO → LUMO). There is no opposite direction charge transfer transition. The crucial transition (HOMO → LUMO) contributed to endo-3 exceed 60% of overall the first hyperpolarizability (Table 3), and the degree of charge transfer transition is even larger than that of endo-2. Similarly, the electronic transition of endo-4 and endo-5 are also viewed as the transition from D and π bridge to A. For exo-n (n = 3–5), the major form of charge transfer is the same with the corresponding endo-n (n = 3–5). Whereas the charge transfer extent of exo-n (n = 3–5) are all larger than that of endo-n (n = 3–5). The larger charge transfer degree result in larger β_tot values. So the β_tot values show the trend that exo-n > endo-n (n = 3–5). In addition, the larger charge transfer degree and different charge transfer form of systems 3–5 (with heterocyclics as π bridges) compared with system 2 (with phenyl as π bridge) resulted in their larger β_tot values.

These results show that the NLO properties of systems 3–5 are better than system 2. It suggests that heterocyclics imidazole, oxazole, thiazole spacer are more suitable as π bridges compared to the phenyl for the enhancement of NLO properties. Therefore, controlling the kinds of π bridges is an important way for the design of new appealing D–π–A type compounds with excellent NLO properties. We also expect that this insight into the relationship between the kinds of π bridges and NLO response will be applied to the design of new optical and photoelectric devices with good performances such as modulation and switching and optical data processing.

4. Conclusion

In this work, we found that the replacement of phenyl between ferrocenyl and corannulenyl by the familiar five-membered rings, such as imidazole, oxazole, thiazole, will improve the NLO response. One reason is that the λ_max of heterocyclic compounds show apparent red-shift. And the red shift of λ_max can lead to the smaller transition energy of the compounds, which might lead to larger NLO response. Hence, heterocycle are more suitable as π bridges compared to the phenyl for the enhancement of NLO properties. The other reason is that the larger charge transfer degree and different charge transfer form of heterocyclic compounds compared with the compound with phenyl as π bridge resulted in their larger β_tot values. Importantly, the β_tot values of system 3 shift by a factor of 2 relative to system 2 meaning that system 3 has excellent NLO properties and the imidazole is the best bridge among the three five-membered heterocyclic rings.

Interestingly, the differences between the endo and the exo conformers could be negligible for the energetic and spectral properties of the systems. However, the β_tot values of endo and exo conformers show the visible difference that the β_tot values of exo conformers are all larger than the corresponding endo conformers. Which is attributed to the larger degree of charge transfer transition of exo conformers compared to that of the corresponding endo conformers. Although the major form of charge transfer is nearly the same for the isomers. We hope that this work may be beneficial to experimentalists for designing large NLO D–π–A type compounds with five-membered heterocyclic ring as π bridge.

Acknowledgements

The authors gratefully acknowledge the financial support from the “12th Five-Year” Science and Technology Research Project of the Education Department of Jilin Province ([2016] 494) and the National Natural Science Foundation of China (No. 21173035).

References

1. Y. R. Shi, D. Frattarelli, N. Watanabe, A. Facchetti, E. Cariati, S. Righetto, E. Tordin, C. Zuccaccia, A. Macchioni, S. L. Wegener, C. L. Stern, M. A. Ratner and T. J. Marks, J. Am. Chem. Soc., 2015, 137, 12521–12538.
2. L. Beverina and G. A. Pagani, Acc. Chem. Res., 2014, 47, 319–329.
3. J. B. Wei, B. Y. Liang, X. Cheng, Z. L. Zhang, H. Y. Zhang and Y. Wang, RSC Adv., 2015, 5, 71903–71910.
4. F. Bures, RSC Adv., 2014, 4, 58826–58851.
5. M. C. Ruiz Delgado, J. Casado, V. Hernández, J. T. López Navarrete, J. Orduna, B. Villacampa, R. Alicante, J. M. Raimundo, P. Blanchard and J. Roncali, J. Phys. Chem. C, 2008, 112, 3109–3120.
6. R. L. Zhong, H. L. Xu, Z. M. Su, Z. R. Li, S. L. Sun and Y. Q. Qiu, ChemPhysChem, 2012, 13, 2349–2353.
7. Y. Bai, Z. J. Zhou, J. J. Wang, Y. Li, D. Wu, W. Chen, Z. R. Li and C. C. Sun, J. Phys. Chem. A, 2013, 117, 2835–2843.
8. T. Verbiest, S. Houbrechts, M. Kauranen, K. Clays and A. Persoons, J. Mater. Chem., 1997, 7, 2175–2189.
9. D. R. Kanis, M. A. Ratner and T. J. Marks, Chem. Rev., 1994, 94, 195–242.
10. S. Priyadarshy, M. J. Therien and D. N. Beratan, J. Am. Chem. Soc., 1996, 118, 1504–1510.
11. M. Zajac, P. Hrobárik, P. Magdolen, P. Foltínová and P. Zahradník, Tetrahedron, 2008, 64, 10605–10618.
12. J. H. Jia, Y. H. Cui, L. Han, W. J. Sheng, Y. J. Li and J. R. Gao, Dyes Pigm., 2014, 104, 137–145.
13. J. Kulhanek, F. Bures, W. Kuznik, I. V. Kityk, T. Mikysek and A. Ruzicka, Chem.–Asian J., 2013, 8, 465–475.
14. Y. Liao, B. E. Eichinger, K. A. Firestone, M. Haller, J. D. Luo, W. Kaminsky, J. B. Benedict, P. J. Reid, A. K. Y. Jen, L. R. Dalton and B. H. Robinson, J. Am. Chem. Soc., 2005, 127, 2758–2766.
15. G. Li, Y. L. Song, H. W. Hou, L. K. Li, Y. T. Fan, Y. Zhu, X. R. Meng and L. W. Mi, Inorg. Chem., 2003, 42, 913–920.
16. S. Barlow, H. E. Bunting, C. Ringham, J. C. Green, G. U. Bublitz, S. G. Boxer, J. W. Perry and S. R. Marder, J. Am. Chem. Soc., 1999, 121, 3715–3723.
17. L. Cuffe, R. D. A. Hudson, J. F. Gallagher, S. Jennings, C. J. McAdam, R. B. T. Connelly, A. R. Manning, B. H. Robinson and J. Simpson, Organometallics, 2005, 24, 2051–2060.
18. S. Barlow and S. R. Marder, Chem. Commun., 2000, 1555–1562,  10.1039/B004907G.
19. K. Kowalski, Ł. Szczupak, J. Skiba, O. S. Abdel-Rahman, R. F. Winter, R. Czerwieniec and B. Therrien, Organometallics, 2014, 33, 4697–4705.
20. P. Garcia-Novo, A. Campo-Cacharron, E. M. Cabaleiro-Lago and J. Rodriguez-Otero, Phys. Chem. Chem. Phys., 2012, 14, 104–112.
21. L. Zoppi, L. Martin-Samos and K. K. Baldridge, J. Am. Chem. Soc., 2011, 133, 14002–14009.
22. W. Y. Wang, N. N. Ma, S. L. Sun and Y. Q. Qiu, Organometallics, 2014, 33, 3341–3352.
23. W. B. Fellows, A. M. Rice, D. E. Williams, E. A. Dolgopolova, A. K. Vannucci, P. J. Pellechia, M. D. Smith, J. A. Krause and N. B. Shustova, Angew. Chem., Int. Ed., 2016, 55, 2195–2199.
24. C. Bruno, R. Benassi, A. Passalacqua, F. Paolucci, C. Fontanesi, M. Marcaccio, E. A. Jackson and L. T. Scott, J. Phys. Chem. B, 2009, 113, 1954–1962.
25. A. Ayalon, M. Rabinovitz, P. C. Cheng and L. T. Scott, Angew. Chem., Int. Ed. Engl., 1992, 31, 1636–1637.
26. V. Rajeshkumar, Y. T. Lee and M. C. Stuparu, Eur. J. Org. Chem., 2016, 2016, 36–40.
27. W. E. Barth and R. G. Lawton, J. Am. Chem. Soc., 1966, 88, 380–381.
28. R. G. Lawton and W. E. Barth, J. Am. Chem. Soc., 1971, 93, 1730–1745.
29. B. M. Schmidt and D. Lentz, Chem. Lett., 2014, 43, 171–177.
30. B. Topolinski, B. M. Schmidt, M. Kathan, S. I. Troyanov and D. Lentz, Chem. Commun., 2012, 48, 6298–6300.
31. B. Topolinski, B. M. Schmidt, S. Schwagerus, M. Kathan and D. Lentz, Eur. J. Inorg. Chem., 2014, 2014, 5391–5405.
32. R. D. Miller, V. Y. Lee and C. R. Moylan, Chem. Mater., 1994, 6, 1023–1032.
33. C. W. Dirk, H. E. Katz, M. L. Schilling and L. A. King, Chem. Mater., 1990, 2, 700–705.
34. A. K. Y. Jen, V. P. Rao, K. Y. Wong and K. J. Drost, J. Chem. Soc., Chem. Commun., 1993, 90–92,  10.1039/C39930000090.
35. P. R. Varanasi, A. K. Y. Jen, J. Chandrasekhar, I. N. N. Namboothiri and A. Rathna, J. Am. Chem. Soc., 1996, 118, 12443–12448.
36. L. Wang, W. Y. Wang, Y. Q. Qiu and H. Z. Lu, J. Phys. Chem. C, 2015, 119, 24965–24975.
37. L. Wang, W. Y. Wang, X. Y. Fang, C. L. Zhu and Y. Q. Qiu, RSC Adv., 2015, 5, 79783–79791.
38. W. Y. Wang, L. Wang, N. N. Ma, C. L. Zhu and Y. Q. Qiu, Dalton Trans., 2015, 44, 10078–10088.
39. M. Y. Zhang, C. H. Wang, W. Y. Wang, N. N. Ma, S. L. Sun and Y. Q. Qiu, J. Phys. Chem. A, 2013, 117, 12497–12510.
40. N. N. Ma, S. L. Sun, C. G. Liu, X. X. Sun and Y. Q. Qiu, J. Phys. Chem. A, 2011, 115, 13564–13572.
41. J. D. Chai and M. Head-Gordon, Phys. Chem. Chem. Phys., 2008, 10, 6615–6620.
42. S. Grimme, J. Chem. Phys., 2006, 124, 1–15.
43. T. Yanai, D. P. Tew and N. C. Handy, Chem. Phys. Lett., 2004, 393, 51–57.
44. A. D. Becke, J. Chem. Phys., 1993, 98, 1372–1377.
45. P. Karamanis, R. Marchal, P. Carbonniere and C. Pouchan, J. Chem. Phys., 2011, 135, 1–15.
46. D. Tzeli, I. D. Petsalakis and G. Theodorakopoulos, J. Phys. Chem. A, 2011, 115, 11749–11760.
47. Y. Wang, Y. K. Wang, J. M. Wang, Y. Liu and Y. T. Yang, J. Am. Chem. Soc., 2009, 131, 8839–8847.
48. I. Ciofini, P. P. Lainé, F. Bedioui and C. Adamo, J. Am. Chem. Soc., 2004, 126, 10763–10777.
49. G. Pescitelli, N. Sreerama, P. Salvadori, K. Nakanishi, N. Berova and R. W. Woody, J. Am. Chem. Soc., 2008, 130, 6170–6181.
50. M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman and G. Scalmani, et al., Gaussian 09, Revision D.01, Gaussian, Inc., Wallingford, CT, 2009.
51. M. Head-Gordon, J. A. Pople and M. J. Frisch, Chem. Phys. Lett., 1988, 153, 503–506.
52. C. Dubceac, A. S. Filatov, A. V. Zabula and M. A. Petrukhina, Cryst. Growth Des., 2015, 15, 778–785.
53. T. R. Cundari, H. A. Kurtz and T. Zhou, Chem. Phys., 1999, 240, 205–214.
54. J. L. Oudar and D. S. Chemla, J. Chem. Phys., 1977, 66, 2664–2668.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra22020g

---