A Deep Blue B,N-Doped Heptacene Emitter That Shows Both Ther- mally Activated Delayed Fluorescence and Delayed Fluorescence by Triplet Triplet Annihilation

An easy-to-access, near UV-emitting linearly extended B,N-doped heptacene with high thermal stability is designed and synthesized in good yields. This compound emits thermally activated delayed fluorescence (TADF) at ambient temperature from a multi-resonant (MR) state and represents a rare example of a nontriangulene-based MR-TADF emitter. At lower temperatures triplet-triplet annihilation dominates. The compound simultaneously possesses narrow, deep-blue emission with CIE coordinates of (0.17, 0.01). While delayed fluorescence results mainly from triplet-triplet annihilation at lower temperatures in THF solution, where aggregates form upon cooling, the TADF mechanism takes over around room temperatures in solution when the aggregates dissolve, or when the compound is well dispersed in a solid matrix. The potential of our molecular design to trigger TADF in larger acenes is demonstrated through the accurate prediction of ∆EST using correlated wavefunction-based calculations. Based on these calculations, we predicted dramatically different optoelectronic behavior in terms of both ∆EST and the optical energy gap of two constitutional isomers where only the boron and nitrogen positions change. A comprehensive structural, optoelectronic and theoretical investigation is presented. Further, the ability of the achiral molecule to assemble on a Au(111) surface to a highly ordered layer composed of enantiomorphic domains of racemic entities is demonstrated by scanning tunneling microscopy.

In 2015, Hatakeyama et al. 45 introduced a one-step borylation method to generate polycyclic aromatic hydrocarbons (PAHs) with a 1,4-oxoborine substructure through ortho-lithiation of 1,4-diaryloxybenzenes and subsequent transmetalation to boron, followed by electrophilic borylation of the arene. Later, the same group demonstrated the versatility and the impact of the one-step borylation through the synthesis of boron-and nitrogen-containing PAHs, which they showed to be high-performance blue-emitting thermally activated delayed fluorescence (TADF) materials for organic light-emitting diodes (OLEDs). The synthetic protocols were further simplified and a series of B-N-doped nanographenes could be obtained in a two-step synthesis consisting of a key multi-borylation of a triarylamine intermediate. By adjusting the boron source, Brønsted base and reaction temperature, selectively doublely and triply borylated products could be obtained. [45][46][47][48][49][50] Among the striking advantages of these molecules is their narrow emission spectra (FWHM = 14- para with respect to each other, a regiochemistry that is necessary but not sufficient to turn on TADF in multiresonance emitter. As the conjugation length of the B,N-doped acenes is extended in the series where the relative positions of the boron and nitrogen atoms alternates, there is a pronounced observed red-shift of the photoluminescence (PL) spectrum from 9 to 3 to 8. Surprisingly, when there is a regioregular extension of the B,N-doping, as in 4, extended conjugation across the acene is suppressed, which leads to a very small red-shift of the PL spectrum upon extension of the acene (compare 3 versus 4); these compounds were identified as fluorescent materials, and a detailed photophysical study of these B.N-doped acenes was not provided.

RESULTS AND DISCUSSION
In the context of our previous theoretical study that predicted that higher-order B,N-doped acenes should show promisingly small DEST and high oscillator strength, and mindful of the limited prior art summarized above, we targeted a linearly extended B-N-doped heptacene. Compound α-3BNOH was obtained from a three-fold electrophilic borylation 49 of triamine 2 in good yield (Scheme 1), which itself was synthesized efficiently in two steps by a sequence of Ullman 54 and Buchwald-Hartwig 55 cross-coupling reactions. The OH functionalities are introduced during the isolation of α-3BNOH due to hydrolysis of the B-Br bonds. Extracting the crude reaction mixture in EtOAc followed by aqueous wash afforded crude α-3BNOH, which was then redissolved in MeCN and placed in an ultrasonic bath for 15 min. The formed off-white precipitate was then filtered and washed with acetonitrile, DCM and hexane and dried to afford pure α-3BNOH. The identity and purity of α-3BNOH was established from a combination of 1 H, 13 C NMR spectroscopy, high-resolution mass spectrometry, elemental analysis and single crystal X-ray diffraction analysis. High themal stability was noted for α-3BNOH on thermogravimetric analysis (TGA). Decomposition temperature (Td), defined as the 5% weight loss of the material was calculated to be 554 °C ( Figure   S10).  We then applied spin-component scaling second order approximate coupled-cluster (SCS-CC2) to model α-3BNOH as well as its alternative B,N-doped heptacene analog, β-3BNOH (Figure 3). The difference density plots of both the S1 and T1 states are reminiscent of other R-TADF emitters where there is an alternating pattern of increasing (in yellow) and decreasing (in green) density on adjacent atoms. A smaller ∆EST of 0.29 eV was calculated for α-3BNOH compared to that of β-3BNOH (DEST = 0.37 eV). Though somewhat higher than ideal for TADF emitters, we nevertheless expected that reverse intersystem crossing (RISC) would be thermally activated in a-3BNOH with potentially the assistance of higher-lying excited triplet states to the upconversion process (see Figure S12 for an energy diagral including higher-lying singlet and triplet excited states). A slightly larger oscillator strength, f, is predicted for β-3BNOH (0.15) compared to α-3BNOH (0.09). Most striking is the large change in the predicted energies of the excited states, with vertical excitation from S0 to S1 at 3.69 and 2.79 eV for α-3BNOH and β-3BNOH, respectively, corresponding to a difference of 0.9 eV. This prediction is consistent with the observed red-shifting of the emission in the previously reported B/N-doped heptacenes (vide supra). The HOMO and LUMO electronic distribution is provided in the ESI ( Figure S11). . Different density plots for lowest singlet and triplet excited states for 3BNOH (linear, a-, and alternating, b-isomers) calculated in the gas phase using SCS-CC2. Blue colour represents an area of decreased electron density and yellow represents an increased electron density between the ground and excited states.
Having completed an in silico study of a-3BNOH with the alternating hypothetical analogue b-3BNOH, we now turn our attention to the optoelectronic properties of the title compound. The electrochemical behavior of α-3BNOH was measured by cyclic voltammetry (CV) and differential pulse voltammetry (DPV) in deaerated THF with 0.1 M tetra-n-butylammonium hexafluorophosphate as the supporting electrolyte (Figure 4a). The oxidation potential, E ox , determined from the peak value of first oxidation wave of DPV curve is 1.17 V versus SCE corresponding to a HOMO energy level of -5.97 eV (EHOMO = −(E ox onset + 4.8) eV. 56,57 The oxidation is irreversible, and no reduction wave was found within the electrochemical window of the solvent. Therefore, the LUMO energy level of -2.81 eV was inferred by subtracting the optical energy gap, Eg, estimated from the onset of the absorption spectrum (Eg= 3.16 eV).  UV-vis absorption spectrum shows a strong absorption band at 3.84 eV (323 nm), assigned to a π-π* transition. The lowest-energy absorption band is found to be at 3.27 eV (379 nm) and is seen to arise from the combined contributions of the quasi-degenerate S1 and S2 excited states. Overall, this band appears to be significantly weaker compared to the low energy absorption bands reported by Takayuki et al. 53 for compounds 3, 4, 8 and 9 (Figure 1) where the most intense band was the lowest energy band. Quantitative agreement in the respective intensities and relative energy differences of the absorption bands is achieved between experiment and theory (see Figures S12). According to our SCS-CC2 calculations, S1 to S3 excited states are expected to contribute to the lowest-energy absorption bands due to their very close energies and oscillator strengths (S1, S2 and S3 oscillator strengths are 0.09, 0.02 and 0.11, respectively) while the highest-energy absorption band is assigned to S4 (oscillator strength of 0.65). The larger oscillator strength calculated for S4 than for S1 is explained by the larger overlap between the HOMO-2 and LUMO+1 orbitals (see Figures S12 for the orbitals plots), which are involved in the dominant one-electron transition (54%) associated with S4 resulting in a larger transition dipole moment in S4 than in S1. Similar to the emission profile reported for compound 8 (Figure 1), α-3BNOH shows structured emission. Both compounds are ladder-type heptacene-based molecules containing para-disposed boron and nitrogen atoms. Aside from the pattern of the boron and nitrogen atoms, structural differences include the substituents found at both nitrogen and boron; for compound 8, mesityl groups are attached to boron and methyl groups are attached to nitrogen, while on α-3BNOH hydroxyl are substituted on boron and tolyl rings are connected to nitrogen.
The photoluminescence spectrum at RT shows a structured profile with a sharp, near UV emission band peaking at 3.18 eV (390 nm) and a shoulder at 3.03 eV (409 nm). Noteworthy is the very small full width at half maximum (FWHM) of 0.24 eV (31 nm), which we attribute to the rigid nature of the compound that does not allow for torsional motions. [58][59][60][61][62] Due to the small Stokes' shift of this emission and its short lifetime (see Figures 4b), we can assign this blue emission peaking at 3.18 eV (390 nm) at RT to fluorescence. By contrast, heptacene and its substituted derivatives absorb and emit in the red. [63][64][65][66] The second emission band, peaking at 2.81 eV (442 nm ) with a shoulder at (465 nm) for 77 K, has a long lifetime of 310 ms (Figure 4b), as evident from its monoexponential decay curve (Supporting Information). Due to its wellresolved vibrational structure, the large shift from the fluorescence and the long lifetime, it can be assigned to a triplet state. When measuring under steady-state conditions at 77 K, we observe both the fluorescence and the phosphorescence components. The ∆EST, determined from the 0-0 peaks of the two bands at 77 K, is measured to be 0.31 eV, consistent with our SCS-CC2 calculations. The absolute photoluminescence quantum yield, FPL, in 7.6×10 -5 M THF at RT is 50%, which reduces to 38% upon exposure to air ( Figure   S13).
In order to obtain a deeper insight, we performed further photoluminescence measurements as a function of temperature, concentration, delay time after pulsed excitation and gate width. Figure 5 shows the relevant spectra. For a 7.6×10 -5 M solution at room temperature (Figure 5a), we observe essentially the fluorescence spectrum for all delay times for short gate times of 10 ns, albeit there is some small hypsochromic shifting of the emission with increasing delay times. We point out that the resolution of the detector used for the time-resolved measurements of Figure 5 is lower than that of the spectrometer used for Figure   4, in particular in the blue spectral range. After a longer delay time of 1 µs and gate width of 1 µs, we still observe the same deep blue emission, which clearly indicates an origin that involves intermediate longlived triplet states.
In order to observe phosphorescence, it is necessary to cool the solution, e.g. by forming a glass at 5 K (Figure 5b). Using a long delay time of 30 ms and gate width of 15 ms, the structure phosphorescence spectrum is clearly visible. The fluorescence band shows the same small hypsochromic shift with increasing delay times as already observed at room temperature. In addition, a broad, unstructured band emerges in the green-to-red spectral range at short times that has fully disappeared by the µs timescale. The broadness and lack of feature, along with the red-shift compared to the fluorescence band near 400 nm and the intermediate lifetime of less than 1 µs is suggestive of an origin from a physical dimer or aggregate that may have formed when cooling; the driving force for aggregate formation at high dilution may be due in part to intermolecular hydrogen bonding as was observed by STM on gold surfaces, vide infra. To address this, we measured the PL in a freshly prepared, more dilute solution of 7.6×10 -6 M, using a delay of 100 ns and a 400 ns gate width (Figure 5c). Around 500 nm, there is only a weak signal that can hardly be distinguished from the fluorescence tail (red curve). When repeating the measurement after one day (blue curve) and after 2 days (green curve), a band centred at 500 nm emerges. After sonication of the solution for a few minutes, this signal dissappeared (black curve). Based on this observation, we associate the broad emission centred at 500 nm to interactions between adjacent chromophores, such as those prevailing in excimers, physical dimers or aggregates. 62 As molecular motion or reorientation is severely restricted at 5K, the most likely origin of the emission are rather weakly bound H-type aggregates. Figure 5d displays the PL spectra taken during several time intervals that evidence that the broad emission at 500 nm decays within 1 µs. It is remarkable that, for these concentrations, no such broad emission at 500 nm can be observed beyond 1 µs (also in the glass of Figure 5b), even though phosphorescence prevails on a ms timescale. The lack of this broad emission and the rise of phosphorescence on a ms timescale implies that aggregated molecules do not act as trap sites for triplet states. We could observe long-lived emission from aggregates, i.e. aggregate phosphorescence, only when using more concentrated solutions or in thin films. This is detailed in the SI (Figure S14a, b). In passing, we mention that the singlet-triplet splitting in the aggregate is very small, on the order of about 70 meV. Figure 5. Emission spectra of α-3BNOH in THF at 7.6 × 10 -5 M as indicated a) at 300 K and b) at 5 K for different delay times d and gate width g (smoothed data). c) At different times after preparation and after sonication (son), taken a with a delay of 100 ns and a gate of 400 ns. d) At 300 K for different delay times, d, and gate width, g, taken in a solution 2 days after preparation. The solid lines are smoothed curves through the lighter shaded data. For all data, λexc = 355 nm. Figure 6 addresses the luminescence transients. In the blue spectral window from 380 -420 nm, the PL taken from a dilute 7.6×10 -5 M solution shows a decay from prompt fluorescence followed by a delayed fluorescence component from about 50 ns onwards (see Figure 6a). This DF decay has a very similar intensity at 5 K and 100 K, yet it reduces relative to the prompt fluorescence upon heating to 150 K and 200 K (see Figure 6b). We note that the glass transition temperature Tg of THF is about 120 K. The transient in the green-red spectral windows, 500 -600 nm, show a similar evolution insofar that the 5 K and 100 K decays coincide, yet the relative intensity reduces upon heating to 200 K and 300 K. The signal between 500 nm and 600 nm is due to aggregates, and we attribute the reduction of this signal upon heating to the dissociation of the aggregates with temperature. Recalling that for THF, Tg is about 120 K, it is clear that this process only sets in from about 150 K onwards. From the observation that the signal between 380 and 420 nm reduces in parallel to the signal between 500 to 600 nm upon heating we conclude that the existence of aggregates is a prerequisite for the delayed blue emission up to and including 300 K.
The most likely origin is that the blue delayed fluorescence results from triplet-triplet annihilation (TTA) that can occur in the aggregated material, which forms despite the high dilution. We recall that at these concentrations the emissive aggregates do not provide a significant number of traps for the (monomer) triplet states. We have measured the dependence of the steady-state emission, i.e. the blue signal, on the excitation power and found this to be linear (Figure S15). While a linear power dependence is often considered to indicate a TADF process, we point out that this is evidently not the case here, as (i) the delayed fluorescence reduces upon heating up to 250 K instead of being thermally activated, and (ii) its intensity evolution correlates with the aggregate concentration, while TADF should be independent of any aggregation effects. In fact, a linear power dependence is characteristic for TTA if the TTA-process dominates over other radiative or non-radiative triplet decay channels. 62 For pulsed excitation with triplet lifetimes of 30 ms or longer, this condition is easily fullfilled. This is confirmed by the time dependence of the DF signal.
The 200 K signals shows the plateau followed by the t -2 decay that is characteristic for the DF decay, which evolves as ( is the initial triplet concentration and is the annihilation constant. At lower temperatures, transport becomes increasingly dispersive thus reducing the slope towards t -1 . Having established that TTA prevails from 5 K to close to 300 K, we now consider the temperature regime from 300 K onwards (Figure 6c). It is evident how, upon heating, a thermally activated component emerges. Thus, it seems that there is a TADF component that becomes active above 300 K. The cross-over between the regimes where TTA and where TADF prevail is around room temperature. For the analysis of the TADF kinetics it is important to establish whether any contribution dominates at 300 K. We have therefore measured the dependence of the decay transient on excitation power. A higher TTA rate due to a higher triplet concentration would alter the time where the plateau merges into the t -2 decay. We observe no change in the decay transients upon increasing the fluence from 0.3 µJ to 30 µJ for a spot size of 4 mm diameter, even when using a more concentrated solution where aggregation should be stronger (See ESI, Figure S16).
This suggests that at 300 K and above, the decay kinetics in the "shorter" time range up to a few µs is controlled mostly by a monomolecular process such as TADF. At longer times above a few µs, where the decay merges into a t -2 dependence, TTA evidently still prevails. Figure 6d shows a fit to the PL decay considering a mono-exponential decay with a fixed lifetime of 450 ns to account for TADF and a decay according to [T0/(1+T1*t)] 2 to account for TTA. A fit to the prompt fluorescence yields a lifetime of about 9 ns. This, as well as fits with other TADF lifetimes are available in the ESI, Figure S16. It is not possible to derive a rate for the reverse intersystem crossing, #$%& , from the dynamics described above. Simply inserting the lifetime into the frequently used equations from the Monkman group, as is sometimes done, would give a nominal value of about 2×10 6 s -1 , which would be a high value for a ∆EST of 0.31 eV. 67,68 However, as has also been stressed by the Monkman group, in deriving these equations a number of approximations are made that only hold when the quantum yield for DF, Φ '( , is about four times larger than that of the PF, Φ )* . This is not the case for α-3BNOH. Rather, in Figure 6d, Φ '( ≈ 0.01Φ +( , consistent with the ∆EST value and previous MR-TADF emitters. 48,69 We note that the TADF component seems to have an activation energy of only about 70 meV while the singlet-triplet splitting is in the range of 300 meV. This discrepancy has previously been accounted for by mixing with higher-lying excited states. 68 SCS-CC2 calculations predict the presence of two higher-lying triplet states of energies intermediate between T1 and S1, each of which has MR-TADF character ( Figure S12). Here, the low value may also relate to the fact that the TADF component superimposes on the TTA component. The reduction of the TTA contribution with increasing temperature due to the dissolution of any aggregates also reduces the observed increase in TADF, and the two processes are difficult to disentangle.
For electroluminescent device applications, it is essential to know how these properties translate into films. Figure 7 shows the spectra and transients obtained for α-3BNOH dispersed (doped) at 1 wt% into poly(methylmetacrylate) (PMMA). The PL transients of the doped film show three distinct regimes. First, we observe a prompt decay component with a lifetime of 8.5 ns, identical to that in solution. Second, from about 100 ns to several µs we observe a decay of unclear origin. Finally, towards longer times there is an exponential decay that is temperature activated. At room temperature, the lifetime of this mono-exponential decay is about 260 µs (see Figure S18a). Evaluating the thermal activation of the long-lived decay gives an activation energy of 220 meV, which matches with the observed DEST. (see Figure S18c) It does not show any dependence on incident light intensity (see Figure S18b), identifying it as a monomolecular process. This, in combination with the thermal activation with an energy that matches the singlet-triplet splitting strongly suggests the 260 µs lifetime to be a TADF signal. If fitted with a single exponential decay, then the signal in the intermediate range from about 100 ns to several µs has a lifetime on the order of 0.5 µs, though an interpretation as a dispersive decay following a ,law would also be possible. It shows a small dependence on intensity (see Figure S18b). Evidently, more detailed follow-up studies beyond the scope of this investigation are required to unambiously assign this intermediate feature. The photophysical data in solution and film is summarized in Table 1. Solvatochromic measurements have also been carried out in solution ( Figure S19) yet they show no systematic shift with solvent polarity in agreement with the short-range charge transfer observed in the S1 excited state (see Figure 3). We next carried out scanning tunneling microscopy (STM) measurements to study the self-assembly process of α-3BNOH on a Au(111) surface. The focus was on the first layer which, through templating and modification of chemical interactions, is critical for the growth of thin films. [70][71][72] Samples prepared at two different temperatures (RT and 343 K) and a range of concentrations were investigated. The STM images at submolecular resolution of layers prepared from solutions of α-3BNOH (Figure 8) reveal a pronounced influence of concentration on the order of the layers as illustrated in Figure 8 whereas no noticeable influence of the preparation temperature was observed. Higher concentrations yield disordered layers as evidenced by Figure 8a. In contrast, a highly crystalline herringbone structure is formed at the lower concentration (Figure 8b)  Looking at the structural model of Figure 8c, the major intermolecular interactions in the monolayers are those resulting in dimer formation. Interactions between different dimers seem to be comparably weak, even though not negligible as the alignment of the tolyl groups of adjacent molecules suggest as illustrated in the top left corner of Figure 8b. Also sterical hindrance is likely to play a role (sites marked by arrows), which would require some conformational adjustment. Therefore, the interaction between the acene backbone and the metal surface is considered the major driving force to minimize the energy observed for a mixture of enantiomeric monomers. 76 As far as the assembly process is concerned it is not clear at present whether the dimers preform and attach to or individual molecules lock into a growing domain. Detailed concentration dependent studies will provide further insight into the mechanism of film formation.

CONCLUSIONS
In summary, a linearly extended B,N-doped heptacene (α-3BNOH) has been synthesized via electrophilic arene borylation. Starting from p-toluidine, we obtained α-3BNOH in three steps in good yields through easy-to-access chemical reactions. High thermalstabilty (Td = 554 °C) was exhibited by this compound. The multi resonant TADF property of α-3BNOH have ben investigated by combining results from state-of-the art quantum-chemical calculations and temperature dependent emission decay measurements.
Our heptacene derivative presented a near UV emission profile in solutions with a narrow emission band.