Organic room-temperature phosphorescence from halogen-bonded organic frameworks: hidden electronic effects in rigidified chromophores

Development of purely organic materials displaying room-temperature phosphorescence (RTP) will expand the toolbox of inorganic phosphors for imaging, sensing or display applications. While molecular solids were found to suppress non-radiative energy dissipation and make the RTP process kinetically favourable, such an effect should be enhanced by the presence of multivalent directional non-covalent interactions. Here we report phosphorescence of a series of fast triplet-forming tetraethyl naphthalene-1,4,5,8-tetracarboxylates. Various numbers of bromo substituents were introduced to modulate intermolecular halogen-bonding interactions. Bright RTP with quantum yields up to 20% was observed when the molecule is surrounded by a Br⋯O halogen-bonded network. Spectroscopic and computational analyses revealed that judicious heavy-atom positioning suppresses non-radiative relaxation and enhances intersystem crossing at the same time. The latter effect was found to be facilitated by the orbital angular momentum change, in addition to the conventional heavy-atom effect. Our results suggest the potential of multivalent non-covalent interactions for excited-state conformation and electronic control.


Introduction
Room temperature phosphorescence (RTP) has received increasing interest due to the potential it presents for photonic devices, bio-imaging, anti-counterfeiting, and night-vision applications. [1][2][3] Until recent years, the main sources of RTP luminophores have been inorganic or organometallic complexes, due to the presence of metal atoms being able to promote singlet-to-triplet intersystem crossing (ISC) in the excited states. However, heavy metal complexes or inorganic materials can oen be toxic and expensive; through the study of purely organic phosphors, the applications of phosphorescence materials can expand by becoming more biocompatible, cheaper to acquire, and environmentally safer. 4,5 While there are many benets of organic phosphors compared to those containing heavy metals, achieving RTP from purely organic molecules has proven a challenge on account of slow ISC rates and competitive non-radiative processes, in particular.
In recent decades, organic phosphorescence has become a more widely explored topic due to the discovery of long-lasting RTP by utilising crystallisation, 6-8 aggregation, 9,10 halogen bonding, [11][12][13][14] heavy atoms, 15 and carbonyl substituents [16][17][18] to circumvent the aforementioned issues. [19][20][21][22][23][24][25][26][27][28] Although spin-orbit coupling (SOC) in organic molecules is usually small (on the order of 1 cm À1 , cf. 10 2 to 10 3 cm À1 for organometallic complexes), the introduction of a carbonyl functionality to aromatic rings oen opens up a 1 (n-p*) / 3 (p-p*) (or 1 (p-p*) / 3 (n-p*)) channel with SOC $100 cm À1 . [29][30][31][32][33] Such a small increase is sufficient to allow efficient ISC and populate the triplet of, for instance, benzophenone or benzaldehyde with a near-unitary quantum efficiency. 34,35 The structure of the asgenerated triplet states can be rigidied in the solid state with the aid of non-covalent interactions (e.g. hydrogen and halogen bonds) [11][12][13]20 to suppress non-radiative vibrational relaxation, resulting in nearly quantitative RTP quantum yields in the solid state. 19,36 Combining these design principles, the Kim group reported seminal work on efficient RTP luminophores based on 2,5bis(hexyloxy)-4-bromobenzaldehyde. 37 The linear C]O/Br halogen-bonding interactions 38,39 present in the solid state were suggested to be the major reason to avoid energy dissipation through vibrational motions. The proximity of a fourth-row Br element to the C]O group, where the non-bonding electrons originate in the n-p* transition, is believed to enhance SOC as well. 40,41 In fact, in a later study by Kim and Dunietz, it was found that moving the Br substituent from the para to the ortho position, closer to the triplet-producing carbonyl functionality, in benzaldehyde increases SOC on the single-molecule level, which enhances both the rates of ISC k ISC and phosphorescence k Phos signicantly by 5-15 fold. 40 Inspired by these ndings as well as other successful demonstration of halogen-bond-induced phosphorescence in the solid state, we exploited the naphthalene scaffold, a prototypical building block in organic optoelectronics, to study the effect of the halogen substitution and the role of halogen bonding in mediating triplet formation. Compared to the previously studied bromobenzaldehydes, this system is expected to have less carbonyl-originated n-p* character in the low singlet excited states to drive ISC, thus offering a platform to highlight the halogen effects. Well-developed synthetic methodologies 42-45 were used to introduce multiple halogen-bond donors (e.g. Br) and acceptors (e.g. O) in naphthalene to permit multiple non-covalent interactions to occur synergistically, enabling phosphorescence from halogen-bonded frameworks. 46

Results and discussion
Naphthalene derivatives with halogen-bond accepting carbonyl functionalities and a various number of halogen-bond donating Br atoms can be prepared readily from 1,4,5,8-naphthalenetetracarboxylic dianhydride (NDA) (Scheme 1). [47][48][49] Bromination of  NDA with 1,3-dibromo-5,5-dimethylhydantoin is slow and can produce a mixture of NDA with various numbers of Br substituents. 47 However, the application of excess reagents at elevated temperatures for a prolonged reaction time gives tetrabrominated NDA (Br 4 NDA) as the sole product. Esterication of Br n -NDA with ethyl iodide in alkaline ethanol gave a mixture of naphthalene tetracarboxylic ethyl esters, Br n NTE (n ¼ 0, 1, 2, 4; n ¼ 3 can be isolated but it is not discussed here for simplicity). 50,51 The individual compounds were isolated by column chromatography on SiO 2 and their identity was conrmed by NMR, MS, and single-crystal X-ray crystallography.
In the solid state, an extended Br/O network can be observed for Br 2 NTE 50 (Fig. 1). For each molecule, each pair of the peri ester groups interacts with a Br atom of a neighbouring molecule to establish bifurcated, slightly asymmetric halogen bonds with 152.40(9) and 149.21 (8)). 38 Reciprocally, each Br atom is interacting with two peri ester groups of a nearby molecule of Br 2 NTE. Being symmetrically substituted with four Br and four ester functionalities, Br 4 NTE is also embedded in a framework of halogen bonds in the solid state. However, likely due to the steric requirement of large Br atoms, the same arrangement in  1) ).
With only one Br atom per molecule, the ester groups in Br 1 NTE do not engage in extended halogen-bonded networks. In fact, the shortest d(C-Br/O-C 2 H 5 ) distance is measured to be 3.32(2)Å (q(C-Br/O) ¼ 151.7(9) ), barely shorter than the van der Waals contact distance. At last, no p-stack or short contact between C-H and naphthalene was found in the crystals of Br 0 NTE.
While naphthalene and its 2-brominated derivatives display electronic absorption <300 nm, ester substitution induces a bathochromic shi of the naphthalene-centred transitions by ca. 50 nm, extending the absorption bands to 350 nm with maxima at $300 nm (Fig. 2). [54][55][56] Compared to pristine naphthalene, which has an appreciable uorescence quantum yield of 23% (40% triplet formation yield), 55 no emission was detected from all Br n NTE in deaerated CH 2 Cl 2 up to 0.02 M (near saturation) excited at 330 nm.
Despite the non-radiative energy dissipation in solution, crystalline solids of the brominated molecules display visible phosphorescence in the 500-700 nm region with millisecond lifetimes, whereas non-brominated Br 0 NTE remains nonemissive ( Fig. 2 and Table 1). Powdery crystalline solid samples of Br n NTE, whose powder X-ray diffraction proles match the pattern based on their single-crystal data, were used in all phosphorescence measurements. Phosphorescence of crystalline Br 2 NTE and Br 4 NTE feature clear vibrational progression with a quantum yield of F Phos ¼ 19.6% and 9.3%, respectively. Much weaker and structureless emission was observed for Br 1 NTE (F Phos ¼ 1.4%).
The varying luminescent behaviours suggest that the excitedstate dynamics were modulated in a subtle way by Br-specic properties, which is however not directly related to the number of Br atoms in the molecule. It is conceivable that multi-point halogen bonding provides a geometric framework to strengthen the rigidity of Br n NTE in the crystalline state. This effect is especially substantial for Br 2 NTE where all the peripheral substituents engage in the directional Br/O interactions, providing the additional factor to the solid-state effect 6,8 of RTP to impede competitive non-radiative relaxation through intramolecular motions. The highest phosphorescence quantum yield was thus observed for the crystalline sample of Br 2 NTE.
The weaker and non-structured phosphorescence observed for Br 1 NTE (and Br 0 NTE) seems to be originated from its looser solid-state packing. If we dene the volumetric index V i as the ratio between the Voronoi volume (V Vor ) 57,58 and the van der Waals volume (V wdW ) of a molecule in the crystal, smaller V i ¼ V Vor /V wdW would suggest denser packing. V i of 1.27-1.30 were found for Br 2 NTE and Br 4 NTE embedded in halogen-bonded frameworks, but the values are signicantly larger for Br 0 NTE and Br 1 NTE (1.36-1.38). The larger free space available to each molecule in the Br 0 NTE and Br 1 NTE crystals allows the excited molecules to decay radiatively and non-radiatively on various points of the triplet potential energy surface.
The signicance of the inter-Br n NTE Br/O interactions is further supported by comparing the phosphorescence of crystalline Br n NTE with that of the dispersed molecules in poly(methyl methacrylate) (PMMA, M w $996 kDa; 2 wt% doping). The rigid polymer matrix is expected to constrain the molecular motion at room temperature but disrupt inter-Br n NTE Br/O halogen bonds. The phosphorescence spectra of Br 1 NTE remained identical in either environment (Fig. 2), indicating that the triplet decay in Br 1 NTE is largely intrinsic to the monomeric molecule. However, the vibrational progression of Br 4 NTE, a signature of chromophore rigidity, became less pronounced, and that of Br 2 NTE completely disappeared and the overall emission prole resembles very well to that of Br 1 NTE.
Additional support for the efficient population of the triplet excited state was provided by transient absorption measurements. Spectroscopically, all Br n NTE exhibit similar excitedstate dynamics: following the initial formation of the singlet excited state, which displays excited-state absorption (ESA) peaking at ca. 490 nm and a broad feature in the near infrared region of 800-1000 nm ( Fig. 3 and ESI Section 5 †), a new excitedstate species with ESA at ca. 480 nm appears with microsecond lifetimes. This long-lived species was assigned to the triplet of each chromophore based on the lifetime and spectral similarity to the triplet-triplet absorption of methyl 1-naphthalate 59 and 2bromonaphthalene. 60 Therefore, the decay of the initial state can be ascribed to singlet-to-triplet ISC; time constants on the order of tens of picosecond were observed for this process (Table 1). Compared to the typical uorescence lifetime (1 ns or longer) of naphthalene derivatives, 54,61 the fast ISC process suggests a high triplet forming efficiency. Such efficient ISC on  the molecular level is likely due to the combined results of bromo (cf. >90% triplet yield for 2-bromonaphthalene) 60 and carbonyl substitution. 62 Broadly speaking, the more bromo atoms in a molecule, the faster the S 1 / T n and T 1 / S 0 processes, in line with the stronger heavy-atom enhanced SOC. 63 Unexpectedly, however, the S 1 / T n ISC for Br 4 NTE is noticeably slower than its less brominated analogues.
Since the rate of S 1 / T n ISC is largely determined by the energy gap between the singlet and triplet states (DE ST ) and the magnitude of spin-orbit coupling (SOC), 64 we evaluated the matrix elements of hS 1 |Ĥ SO |T n i using the two-layer ONIOM (QM:MM) scheme to simulate the photophysical processes in crystals. The molecular geometry was computed at the level of uB97X-D/6-31G(d):OPLS-AA, and hS 1 |Ĥ SO |T n i calculated at the TDA-uB97X-D/6-311+G(d,p) level of theory based on the ONIOM geometries (Table 2 and ESI Section 7 †). The (TD-)DFT calculations were performed using Gaussian 16, 65 which was then interfaced with PySOC 30 to evaluate the SOC matrix elements. The Tamm-Dancoff approximation (TDA) was exploited to minimise triplet instability. 40,66 In all cases, the state energies are not signicantly affected by aggregation; thus results from the calculations with one molecule in the QM region are discussed here.
The ISC process for Br n NTE likely takes place between S 1 and the high-lying triplet states. Considering DE ST alone (<0.5 eV), ISC to T 2,3 for Br 1 NTE, T 2-4 for Br 2 NTE, and T 2 for Br 4 NTE should dominate in the respective molecules, whereas the large energy gap DE ST > 1.5 eV prevents direct ISC into T 1 (see ESI Section 7 † for the relative energies). Compared to Br 0 NTE, incorporating fourth-row Br elements into the naphthalene scaffold increases SOC by 1-2 orders of magnitude. Despite the larger number of Br atoms in the structure, smaller SOC was found for Br 4 NTE than Br 2 NTE, in line with the slower triplet formation found experimentally for the former molecule. The hS 0 |Ĥ SO |T 1 i calculated at the T 1 geometry, the key factor determining the rate of phosphorescence, was similarly found to be smaller for Br 4 NTE than Br 2 NTE.
A close examination of the electron density of the key states provided hints to the origin of the unexpected drop in SOC for Br 4 NTE. Fig. 4 shows the electron density difference between the selected excited states and the ground state for Br 2 NTE (S 1 and T 3 ) and for Br 4 NTE (S 1 and T 2 ). These transitions displayed a signicant naphthalene-centred p-p* character; the involvement of the Br atoms can be clearly seen and hence the higher SOC in brominated Br n NTE. Comparatively, the carbonyl n-p* contribution, the typical driver for the ISC process in aromatic ketones/aldehydes, appears to be much less substantial. In the case of Br 2 NTE, the Br-centred transition densities are roughly perpendicular to the naphthalene plane in the S 1 state but rotate distinctively in the T 3 state, facilitating the orbital angular momentum change for ISC (similar rotation found in T 4 ). In the case of Br 4 NTE, however, the Br-centred transition densities in S 1 and T 2 are both perpendicular to the naphthalene plane. The absence of the analogous rotated transition density for Br 4 NTE is understandable as unfavourable electron repulsion in the region between neighbouring Br atoms would be caused by such a change in density orientation. Taken together, the judicious heavy-atom positioning in Br 2 NTE results in the favourable structural and electronic contributions to its efficient RTP. The 2,6-dibromo substitution offers a lock-in mechanism through halogen bonding to inhibit non-radiative relaxation. Furthermore, high SOC and hence efficient ISC are made possible by adding the orbital angular momentum change to the heavy-atom effect in both the tripletgeneration (S 1 / T n ) and phosphorescence (T 1 / S 0 ) processes.

Conclusions
In summary, we have shown that simultaneously incorporating multiple heavy halogens and halogen-bond donor/acceptor pairs in aromatic molecules can enable bright phosphorescence from purely organic materials. The formation of halogenbonded frameworks in the solid states rigidies phosphorophores, favouring the radiative decay. However, our results indicate that a ne balance has to be struck in terms of the number and positioning of halogens. Too many large halogen atoms in proximity may prohibit structurally the access of halogen-bond acceptors and electronically the contribution of the non-bonding electrons of halogens for enhancing SOC. The latter effect is especially important to consider in the case of carbonyl-bearing polycyclic aromatic hydrocarbons, such as rylenes and its derivatives in the present study where the S 1 state is primarily p-p* in nature. It should be noted that the formation of halogen bonds cannot necessarily be correlated to the increase in ISC and phosphorescence rates; an excited-state analysis will be needed to elucidate the magnitude and origin of SOC when designing organic RTP materials.

Conflicts of interest
There are no conicts to declare.