Doubly Encapsulated Perylene Diimides: Effect of Molecular Encapsulation on Photophysical Properties

Intermolecular interactions play a fundamental role on the performance of conjugated materials in organic electronic devices, as they heavily influence their optoelectronic properties. Synthetic control over the solid state properties of organic optoelectronic materials is crucial to access real life applications. Perylene diimides (PDIs) are one of the most highly studied classes of organic fluorescent dyes. In the solid state, π-π stacking suppresses their emission, limiting their use in a variety of applications. Here we report the synthesis of a novel PDI dye that is encapsulated by four alkylene straps. X-ray crystallography indicates that intermolecular π-π stacking is completely suppressed in the crystalline state. This is further validated by the photophysical properties of the dye in both solution and solid state, and supported by theoretical calculations. However, we find that the introduction of the encapsulating ‘arms’ results in the creation of charge-transfer states which modify the excited state properties. This article demonstrates that molecular encapsulation can be used as a powerful tool to tune intermolecular interactions and thereby gain an extra level of control over the solid state properties of organic optoelectronic materials.


INTRODUCTION
Perylene diimides (PDIs) are one of the most highly studied classes of organic fluorescent dyes. 1 They were initially primarily used as high-grade colorants in the pigment and textile industry, but have recently attracted great interest as materials for application in optoelectronic devices, such as organic solar cells (OSCs), light-emitting diodes (OLEDs), transistors and lasers. [1][2][3] The widespread use of PDI derivatives can be mainly attributed to their desirable optoelectronic properties (e.g. high photoluminescence quantum yields (PLQYs), tunable absorption and emission and high electron mobilities), photothermal stability and synthetic versatility. 3 However, PDI chromophores have an inherently strong propensity to aggregate through π-π stacking. 4 While sometimes desirable for certain properties, such as charge transport, this also limits their potential applications because it can greatly diminish their photoluminesence at higher concentrations or in the solid state. [4][5][6] As a result, it is extremely important to enhance our understanding of the structure-property relationships in PDIs towards obtaining higher levels of control over their aggregation behavior and photophysical properties in both the solution and solid state.
Overall, there are three major strategies to suppress aggregation through π-π stacking in PDI dyes. The first approach relies on the incorporation of steric bulk at the bay positions (1,6,7,12) of the perylene core. Sterically demanding moieties can be incorporated via the facile nucleophilic aromatic substitution (SNAr) of halogenated PDI cores. 7,8 Although, incorporating steric hinderance at the bay positions reduces π-π stacking, it commonly imparts a twist in the planar aromatic PDI core. 9,10 This can undesirably influence the photophysical properties of the dye. 11 Alternatively, it is also possible to shield the PDI core from intermolecular communication without disrupting its planarity through functionalizing the imide positions with bulky substituents that shield both faces of the aromatic perylene. 12 This method was recently demonstrated by Wong and coworkers 4 . They reported the synthesis of a series of PDI derivatives with varying degrees of bulk on the imide positions, ranging from simple phenyl moieties to more congested 3,5-di-tert-butylphenyl and trityl groups. 4 While their X-ray crystal data confirm that increasing the bulk of the substituents leads to more spatially shielded dyes, the UV-vis absorption data still indicate significant aggregation for these molecules. 4 Therefore, incorporating bulky groups at the imide positions of PDIs does not completely preclude the formation of aggregates, and hence only offers limited control over their photophysical properties.
As a third option, we propose a strategy whereby the PDI core is surrounded by encapsulating alkylene straps, as has been demonstrated in other conjugated materials. [13][14][15][16][17][18][19][20][21][22] This method makes use of recently developed chemistry on the ortho-positions (2,5,8,11) of PDIs 23,24 to install 1,3-dimethoxybenzene groups. These are then further functionalized to fully sheath the PDI core with two encapsulating rings. This article discusses the synthesis of a novel, doubly-encapsulated PDI and investigates the structural influence of molecular encapsulation on the photophysical properties of the PDI chromophore.

NMR Spectroscopy
Both PDI-OMe and PDI-Encap were fully characterized via NMR, X-ray crystallography, HRMS and/or elemental analysis. The aromatic regions (6.6-8.4 ppm) in the 1 H NMR spectra of PDI-OMe and PDI-Encap are very similar (Figure 1, top), displaying three signals (labelled a, b and c) corresponding to the protons of the PDI core and peripheral phenyl groups, respectively. However, a significant difference is observed in the alkoxy region (3.6-4.1 ppm). PDI-OMe displays a singlet (labelled OMe) which integrates to 24 protons, as expected. For PDI-Encap the same signal is split into two multiplets (labelled e) which each integrate to 8 protons. Further information was gathered through 2D NMR experiments. 1 H-13 C HSQC data indicate that both proton environments (e) correlate to the same carbon atom ( Figure S12), suggesting that they undergo slow conformational/chemical exchange. This is confirmed by 1 H-1 H ROESY NMR ( Figure 1, bottom left) and is indicative of conformational restriction of the alkoxy chains as a result of encapsulation. Furthermore, the proton signals corresponding to the alkylene straps are highly shielded (< 0.9 ppm), indicating that these environments are proximal to the π system of the PDI chromophore, as expected upon encapsulation ( Figure S8).

X-Ray Crystallography
Single crystal X-ray diffraction analysis of PDI-Encap unambiguously proved that the alkylene moieties of the peripheral aryl groups engage in horizontal encapsulation, rather than in vertical looping or cross encapsulation (Figure 2 & 3). Surprisingly, neither of these possible side-products could be isolated ( Figure 2), although their formation cannot be precluded due to the low yield of PDI-Encap. PDI-Encap crystallizes in the triclinic P1̅ space group with one half of a PDI-Encap molecule and one CHCl3 molecule in the asymmetric unit. The PDI-Encap molecules stack along the crystallographic a axis in an offset manner, exhibiting 9.9 Å centroid-to-centroid separation distances between adjacent PDI cores ( Figure 4). The axes and planes of the PDI cores, belonging to neighboring PDI-Encap molecules, are aligned in parallel. Notably, the packing of the PDI-Encap molecules is not driven by π···π interactions, which is in contrast to conventional PDI dyes that regularly engage in π-stacking in the crystalline state. 26,27 Instead, it appears that the chromophore orientation is dictated by the crystallization of the encapsulating chains ( Figure 4). The absence of π···π interactions in the PDI-Encap crystal structure is clearly attributable to steric hindrances caused by the peripheral aryl groups and the horizontally aligned alkylene chains.
The structurally related tetra-arylated perylene (PDI-OMe) crystallizes in the monoclinic P21/c space group with one PDI-OMe and two CHCl3 molecules in the asymmetric unit. The PDI-OMe molecules stack along the (011) Miller plane ( Figure  4), whereby its PDI cores display 10.2 Å centroid-to-centroid separation distances. The planes of PDI-cores, belonging the neighboring molecules, close an angle of 58° ( Figure 4). The molecular stacks are stabilized by C−H···π and C−H···O interactions in the absence of π···π interactions, which is, similar to PDI-Encap, caused by steric hindrances of the peripheral aryl groups.

Photophysical Results
The solution and thin film absorption spectra of PDI, PDI-OMe and PDI-Encap are presented in Figure 5. The spectral profiles of the solution absorption spectra are near-identical for the three dyes. They consist of four well-resolved vibronic bands at ~430, ~460, ~485 and ~530 nm (λmax) corresponding to the 0-3', 0-2', 0-1' and 0-0' transitions, respectively. For the literature "PDI" dye, some vibronic fine structure is lost in the thin film spectrum and the spectrum is red-shifted compared to the solution data. The observed broadening for PDI is ascribed to aggregation in the solid-state; a phenomenon that is typical for most perylene diimides. 26,27 These observations are in stark contrast to what is observed for PDI-OMe and PDI-Encap. In spin coated thin films, PDI-OMe and PDI-Encap retain more or less the same absorption profile as in solution, indicating unimolecular behavior and showing that the formation of aggregates is suppressed. The sharper onset of absorption for the thin film of PDI-Encap compared to PDI-OMe suggests that it is more rigid and less prone to conformational disorder in the solid state. 13,28,29 The absorption and photoluminescence (PL) spectra of the three dyes in solution ( Figure 5 & 6) display the typical mirror image behavior that is established for rigid luminophores that emit from π-π* states. 30 However, we note that the PL spectra for PDI-OMe and PDI-Encap are broader than that of PDI with more poorly resolved vibronic fine structures and shallower onsets ( Figure 6). This is suggestive of a degree of intramolecular charge transfer (ICT) character in their excited states, specifically between their electron-rich peripheral aryl rings and electron-deficient PDI cores. This is supported by the density functional theory (DFT) calculations discussed below, as well as by time-resolved PL data ( Figure S17-22 & Table S1).
In solution, PDI displays a simple monoexponential decay with a lifetime ( of ~ 4 ns which can be ascribed to fluorescence from a π-π* state. In contrast, PDI-OMe and PDI-Encap display biexponential decays, consisting of a fast-decaying component (i.e. on a sub-ns timescale) followed by the decay of a longer-lived state ( = 3.4 and 2.3 ns for PDI-OMe and PDI-Encap, respectively). These data indicate that multiple states/processes contribute to the PL decay of PDI-OMe and PDI-Encap. The presence of a fast (non-radiative) decay component for PDI-OMe and PDI-Encap is responsible for their low solution PLQYs (< 1%) compared to that of PDI (67%). As these spectra were taken from solutions we propose that PL quenching (for the partially and doubly-encapsulated dyes) originates from fast formation of ICT states with reduced luminescence efficiency, as supported by DFT calculations (vide infra).
The profiles of the thin film PL spectra for the dyes differ strongly among one another ( Figure 6). In the case of PDI, the PL is much broader (~550-800 nm) and significantly less structured in the solid state than in solution, but retains a PLQY of ~20%. This can be ascribed to emission from intermolecular species, whose formation is confirmed by the UV-Vis data (Figure 5). However, the biexponential nature ( = 1.6 ns,  = 5 ns) of the PL decay (Table S1) suggests that the nature of such emissive aggregates is heterogeneous and might be partially excimeric. Indeed, formation of excimers is supported by the literature on similarly stacked perylene bisimides. 31 -34 In such systems, excimers can favor fast (sub-ns) charge separation/radical formation, 35 which is ultimately responsible for the unusually fast (~5 ns) excimeric emission observed from PDI films.
PDI-OMe and PDI-Encap display low thin film PLQYs (< 1%) in-line with their low intrinsic solution PLQYs. Therefore, their thin film PL is vestigial, which complicates unambiguous assignment of their spectral features. Nevertheless, we note that the PL spectrum of PDI-OMe spectrum displays vibrational fine-structure and suppressed aggregate/excimer emission. Such a suppression is likely due to the orthogonal crystal packing of PDI-OMe observed via X-ray crystallography (Figure 4), which prevents intermolecular interactions both in the ground state (in agreement with thin film UV-vis data) as well as in the excited state. In contrast, the PL spectrum of PDI-Encap displays some fine structure but also a significantly broadened band at longer wavelengths. As thin film UVvis data suggest that aggregation is largely suppressed for PDI-Encap, this is ascribed to an admixture of π-π* and ICT emission, which is also in agreement with the DFT data below.

Theoretical DFT Calculations
DFT/time-dependent DFT (TD-DFT) calculations at the level of B3LYP/6-31G* 36 were carried out to gain further insight into the surprising photophysical properties of PDI-OMe and PDI-Encap. For the gas phase optimized geometries of both dyes, the highest occupied molecular orbital (HOMO) and lowest occupied molecular orbital (LUMO) are each delocalized across the core of the PDI chromophore, while the first eight lower lying occupied molecular orbitals (HOMO-1 − HOMO-8) are all localized on the electron rich peripheral dialkoxybenzene groups (HOMO-LUMO DFT results, SI). TD-DFT shows that the transitions to the 5 lowest singlet states of PDI-Encap can all be predominantly assigned to charge transfer from the peripheral aryl rings to the PDI core (HOMO-1~5 → LUMO), with no significant local excitation (HOMO→LUMO) of the PDI chromophore (Table S6). This greatly contrasts with the data calculated for some highly emissive literature PDI derivatives. 23 For both a simple N-alkylated PDI and a PDI tetraarylated with electron poor p-benzonitriles at the 2,5,8,11positions, the S0→S1 transition is predominantly assigned to local excitation of the PDI chromophore (HOMO→LUMO) (Table S3 & S4). Therefore, we propose that fluorescence from the luminophoric PDI core of PDI-Encap and PDI-OMe is quenched by intramolecular charge transfer (ICT) between the electron-rich peripheral aryl groups and the PDI core, leading to the low observed PLQYs.
To gain a deeper understanding of the thin film PL results we also performed TD-DFT calculations on the crystal structure geometries (Table S5 & S7). We expect the gas phase structures to be representative of the average ground state geometry in solution, while the crystal structure geometries should be more representative of the thin film. For PDI-Encap in the gas phase optimized geometry, the lowest energy transition displaying a strong oscillator strength (ca. 0.42) is the local HOMO → LUMO π-π* transition (S0 → S5, Table S6). The same is predicted for PDI-OMe (S0 → S3, Table S4). These data correlate well with the solution PL spectra which display well-resolved vibronic fine structures ( Figure 6). There is greater contrast between the TD-DFT data obtained for the crystal structure geometries of PDI-Encap and PDI-OMe. The data for PDI-OMe are very similar for the gas phase and crystal structure geometries i.e. the lowest energy transition with a strong oscillator strength (ca. 0.34, S0 → S4, Table S5) is π-π* in nature. Therefore, the spectral features of the PL for PDI-OMe are very similar in solution and film ( Figure 6). Conversely, while the lowest energy π-π* transition (S0 → S5, Table S7) retains the gas phase oscillator strength (ca. 0.48) and approximate energy in the crystal structure geometry of PDI-Encap, an ICT transition (S0 → S3, Table S7) at lower energy with a high oscillator strength (ca. 0.19) is also predicted. This transition is ascribed to the broad low energy PL band that is observed for the thin film of PDI-Encap.
To further differentiate between PDI-OMe and PDI-Encap, we also analyzed the low frequency vibration modes (< 20 cm -1 ) of both molecules (SI, Vibration Modes and Frequencies). It was found that PDI-OMe has 6 different low frequency vibration modes which are mainly attributed to the peripheral aryl rings, whereas PDI-Encap only has 3 modes which arise from vibrations caused by the alkylene straps. These data suggest that there is a higher rate of non-radiative decay for PDI-OMe than for PDI-Encap, in agreement with what can be reasonably inferred from the fluorescence lifetime data. Hence, the encapsulating straps are preferable.

CONCLUSIONS
In conclusion, we have successfully prepared a novel, doubly encapsulated perylene diimide molecule, in which intermolecular aggregation is suppressed in both solution and the solidstate. This study therefore demonstrates that molecular encapsulation can be used as a three-dimensional tool to control intermolecular interactions. Furthermore, as opposed to shielding with bulky substituents, encapsulation can provide additional advantages because it can also rigidify molecular architectures as well as direct solid-state packing or 3D crystal growth. However, since the encapsulation strategy used electron-rich peripheral aryls, we have introduced charge transfer character into our molecule which thereby led to photoluminescence quenching. The work therefore highlights both the power of encapsulation as a synthetic tool for chromophore control but also their potential non-innocence with regards to underlying photophysical properties.