Suppression of Dexter transfer by covalent encapsulation for efficient matrix-free narrowband deep blue hyperfluorescent OLEDs

Hyperfluorescence shows great promise for the next generation of commercially feasible blue organic light-emitting diodes, for which eliminating the Dexter transfer to terminal emitter triplet states is key to efficiency and stability. Current devices rely on high-gap matrices to prevent Dexter transfer, which unfortunately leads to overly complex devices from a fabrication standpoint. Here we introduce a molecular design where ultranarrowband blue emitters are covalently encapsulated by insulating alkylene straps. Organic light-emitting diodes with simple emissive layers consisting of pristine thermally activated delayed fluorescence hosts doped with encapsulated terminal emitters exhibit negligible external quantum efficiency drops compared with non-doped devices, enabling a maximum external quantum efficiency of 21.5%. To explain the high efficiency in the absence of high-gap matrices, we turn to transient absorption spectroscopy. It is directly observed that Dexter transfer from a pristine thermally activated delayed fluorescence sensitizer host can be substantially reduced by an encapsulated terminal emitter, opening the door to highly efficient ‘matrix-free’ blue hyperfluorescence.

Hyperfluorescence shows great promise for the next generation of commercially feasible blue organic light-emitting diodes, for which eliminating the Dexter transfer to terminal emitter triplet states is key to efficiency and stability.Current devices rely on high-gap matrices to prevent Dexter transfer, which unfortunately leads to overly complex devices from a fabrication standpoint.Here we introduce a molecular design where ultranarrowband blue emitters are covalently encapsulated by insulating alkylene straps.Organic light-emitting diodes with simple emissive layers consisting of pristine thermally activated delayed fluorescence hosts doped with encapsulated terminal emitters exhibit negligible external quantum efficiency drops compared with non-doped devices, enabling a maximum external quantum efficiency of 21.5%.To explain the high efficiency in the absence of high-gap matrices, we turn to transient absorption spectroscopy.It is directly observed that Dexter transfer from a pristine thermally activated delayed fluorescence sensitizer host can be substantially reduced by an encapsulated terminal emitter, opening the door to highly efficient 'matrix-free' b lu e h yp er fl uo re scence.
The 'blue OLED problem'-the quest for a stable and efficient blue organic light-emitting diode (OLED), has plagued academic and industrial researchers alike for over 20 years 1,2 .Thermally activated delayed fluorescence (TADF)-sensitized fluorescent OLEDs (coined as Hyperfluorescent in other work 3,4 ) are considered one of the most promising contemporary solutions, combining the high efficiency of a TADF donor with the stability and colour purity of a conventional fluorescent acceptor 2,5,6 .

Article
https://doi.org/10.1038/s41563-024-01812-4energy transfer is accessible via FRET.FRET follows an R −6 distance dependence, making it efficient over rather long distances-on the molecular scale-of up to around 10 nm (ref.18).Conversely, as the S 0 → T n transitions of fluorescent emitters are quantum mechanically forbidden, energy transfer to their triplet states is restricted to a Dexter transfer mechanism.Dexter transfer proceeds via direct host-emitter orbital overlap and is therefore restricted to the Ångstrom length scale, with its efficiency dropping off exponentially with distance 24,25 .Increasing the host-emitter distance in a system with good spectral overlap is, therefore, a plausible strategy to suppress Dexter transfer while preserving efficient FRET, which would prevent triplets accumulating on the terminal emitter to simultaneously improve the device efficiency and stability [18][19][20] .
The solution absorption and PL spectra of NB-1 are shown in Fig. 1b alongside the neat film PL of DMAC-DPS-an established blue TADF material with high reported non-doped efficiency (Φ PL = ~100%, EQE = ~20%), which functions as one of our host materials 29 .The photophysical properties of NB-1 are clearly beneficial for maximizing the spectral overlap with the TADF host to promote efficient FRET while preserving pure-blue emission.Our design also allows a subtle tuning of the emitter bandgap for matching with different host materials.Modifying the position of the peripheral tert-butyl groups of NB-1 affords the regioisomer NB-2 (Fig. 1c), for which the solution absorption and emission spectra are redshifted by 6-7 nm (Supplementary Figs.59 and 60).Supplementary Note 5 presents a full photophysical characterization of the emitters.
Second, we focused on suppressing the Dexter transfer.Previous evidence indicates that alkylene encapsulation can suppress the intermolecular interactions leading to concentration quenching and aggregation [30][31][32][33][34][35] .We designed NB-1 and NB-2 hypothesizing that the encapsulating straps would similarly shield the luminophore core from the short-range coupling required for Dexter transfer (Fig. 1a) [36][37][38] .The alkylene straps were installed by either Grubbs metathesis (NB-1) or nucleophilic substitution chemistry (NB-2) (Fig. 1c).Both routes crucially proceed via intermediates that can be obtained in high purity using conventional laboratory techniques.This was achieved by employing bulky alkene (NB-1) or benzyl (NB-2) groups to circumvent the poor solubility of the indolocarbazole core.The luminophore cores of NB-1 and NB-2 are completely enveloped by a continuous alkylene macrocycle (Supplementary Note 2), in contrast to designs that decorate emitters with peripheral steric bulk 39,40 .Despite high molecular weights (MW = 1,122), NB-1 and NB-2 remain sufficiently volatile for processing via thermal evaporation (240-260 °C at 10 −7 mbar; Supplementary Fig. 58).This suggests a suppression of intermolecular interactions, in line with calculations indicating that the non-conjugated straps of NB-1 contribute to over half its molecular volume (Supplementary Table 2).Additionally, although the shielding of emitters with side groups could lead to redshifted and broadened emission 1 , we note that our design has a negligible effect on the mDICz core photophysics (Supplementary Table 4).

Organic light-emitting devices
OLEDs were fabricated to explore the effect of incorporating encapsulated emitters into MFHF devices (Fig. 2).The EL spectrum of an OLED In current hyperfluorescent OLEDs, a high triplet energy matrix is implemented alongside the TADF sensitizer and emitter molecules, affording a three-component emissive layer (EML) (or even four in the case of exciplex/mixed matrices 7 )-a device structure that we term matrix-containing hyperfluorescence (MCHF) 2 .Progress in blue MCHF has been swift, with efficient devices now approaching the Commission Internationale de l'Éclairage (CIE xy ) colour coordinates for display requirements, owing to the concurrent development of efficient TADF sensitizers and narrowband terminal emitters 2,[7][8][9][10][11][12][13][14][15] .However, compared with the typical two-component host-guest EML already adopted by industry, the additional matrices required in MCHF devices unavoidably increases the complexity of device fabrication.Power efficiency is also reduced due to higher turn-on and driving voltages.A simple and efficient matrix-free hyperfluorescent (MFHF) 16 device (first introduced in another work 17 ) with a two-component EML solely consisting of a TADF sensitizer (hereafter referred to as the host) alongside a narrowband terminal emitter, would afford a new paradigm in blue OLED technology from a device fabrication perspective.
The underlying obstacle to MFHF is maintaining high device performance and simplifying the device structure.Dilution into a high triplet energy matrix has so far been inescapable to suppress Dexter transfer to long-lived terminal emitter triplets.Although Dexter transfer to terminal emitter triplets has been challenging to directly observe, it has been hypothesized to be highly disadvantageous in hyperfluorescent devices from both efficiency and stability perspectives [18][19][20] .First, due to their non-radiative nature, the population of terminal emitter triplet states is an efficiency loss pathway.Second, it has been suggested that the operational stability of hyperfluorescent OLEDs is adversely affected by the population of terminal emitter triplets due to a substantial increase in the exciton residence time 21 .Multiresonant (MR) TADF materials are a popular class of terminal emitter that could, in principle, alleviate the efficiency loss associated with Dexter transfer through their ability to triplet harvest.However, to the best of our knowledge, such a process has not been unequivocally observed 1,22 .The problem of an elongated exciton residence time through the Dexter population of terminal emitter triplets also persists, as MR TADF reverse intersystem crossing rates are typically orders of magnitude slower than for conventional donor-acceptor TADF materials 11,23 .
In this work, molecular design provides a new approach to suppress loss through the population of terminal emitter triplet statesthe encapsulated ultranarrowband blue emitter.When doped into a pristine TADF host with no additional matrix, transient absorption (TA) studies directly indicate that Dexter transfer can be substantially curtailed by decorating a terminal emitter with insulating alkylene straps.Owing to Dexter transfer suppression, negligible efficiency loss is observed in MFHF OLEDs compared with non-doped TADF devices, with external quantum efficiencies (EQEs) as high as 21.5% and narrow true-blue electroluminescence (EL) (14-15 nm full-width at half-maximum (FWHM)).Hence, simply structured MFHF devices with narrowband blue emission and high efficiency have been realized for the first time.

Molecular design
The design and syntheses of new terminal emitters NB-1 and NB-2 are illustrated in Fig. 1.In hyperfluorescence, the working mechanism is an energy cascade from a triplet-harvesting TADF donor (our host) to a fluorescent terminal emitter.To achieve efficient blue MFHF OLEDs, we sought to design emitters with structural and photophysical properties that promote singlet transfer via Förster resonant energy transfer (FRET) 18 while suppressing Dexter transfer (Fig. 1a) 24,25 .
FRET proceeds via dipole-dipole coupling, with its efficiency linked to the spectral overlap between the host photoluminescence (PL) and the emitter absorption.As the S 0 ↔ S 1 transitions of the TADF host and fluorescent emitter are spin allowed, the desirable singlet where the EML consists of a DMAC-DPS host doped with 1 wt% NB-1 is shown in Fig. 2b, alongside that for a non-doped reference device (System A; Fig. 2a).Pertinent device metrics are listed in Table 1.The non-doped DMAC-DPS OLED demonstrates peak EL at 476 nm with a broad FWHM of 88 nm.Owing to singlet energy transfer via FRET, the 1 wt% NB-1 MFHF device displays a far superior EL spectrum, with a desirable 449 nm peak wavelength and an extremely narrow FWHM of 14 nm.Harmful high-energy emission of <450 nm is also reduced.We note that the incorporation of 1 wt% NB-1 is accompanied by only a slight decrease in EQE compared with the non-doped device (16.4% versus 17.0%).
Perylene, the archetypal blue terminal emitter in hyperfluorescent systems, was also encapsulated to afford En-Per (System B; Fig. 2a).In an identical fashion to the results for System A, the 1 wt% En-Per in the DMAC-DPS device affords a maximum EQE of 16.4%, suggesting that encapsulation may be generally applicable to obtain high-MFHF EQEs from different luminophore cores.This efficiency is comparable with the best previously reported values for organic hyperfluorescent systems incorporating perylene terminal emitters, even when matrices are included 3,[41][42][43] .
In System C (Fig. 2a), the slightly redshifted emitter NB-2 exhibits good spectral overlap with the TADF material DPAc-DCzBN (Supplementary Fig. 76), a high-performing emitter in non-doped devices 44 .In analogy to System A, the incorporation of 1 wt% NB-2 into DPAc-DCzBN in System C affords a large decrease in FWHM compared with the host-only device (86 → 15 nm) (Fig. 2d).The 1 wt% NB-2 device also crosses the 20.0%EQE threshold, affording a high maximum EQE of 21.5% comparable with the non-doped device (maximum EQE = 20.3%).This is a 1.5 times improvement over the best industry-published triplet-triplet annihilation (TTA) blue OLEDs 45 .Systems A and C are the first examples of blue narrowband MFHF OLEDs, which-to the best of our knowledge-display the highest EQEs and lowest CIE xy coordinates yet reported for matrix-free systems 36,42 .For reference, an MFHF device based on the DMAC-DPS host and an archetypal bis(diphenylamino) pyrene fluorescent emitter affords a maximum EQE of merely 7% in sky blue (Supplementary Fig. 112) 46,47 .Additionally, even compared with contemporary MCHF and TTA devices, our 14-15 nm FWHM values are record equalling 12,27 .Supplementary Note 7 presents further detailed device characterization results.
An important observation is that for all the three MFHF systems, there is a negligible difference in the OLED EQEs between doped and non-doped devices.The same trend is additionally seen for a further encapsulated structure based on 9,10-diphenylanthracene (Supplementary Table 21).The lack of any substantial decrease in performance implies that the emitters introduce minimal loss pathways to MFHF OLEDs; for example, quenching via Dexter transfer to the terminal emitter may not be substantial, which would validate the encapsulated design.This prompted us to study the energy transfer dynamics of MFHF blends in further detail.

PL study
To determine the effect of an encapsulated acceptor structure on the energy transfer processes in a TADF host, non-encapsulated analogue NB-3 (Fig. 3a) was synthesized for comparison with NB-1   3, obtained on excitation of the host at 330 nm.Aggregation can plague narrowband emitters, leading to broadened emission and quenched luminescence even in dilute evaporated films 14,27,40,[48][49][50][51] .For NB-3, broad unstructured PL is recorded on increasing the doping concentration (Fig. 3c).In contrast, for NB-1, PL in DMAC-DPS progresses towards its narrow-structured solution spectrum (Fig. 3b), indicating that alkylene encapsulation effectively suppresses the aggregation of mDICz.Next, Φ PL and time-resolved PL experiments were conducted.Data for the films of NB-1 and NB-3 doped into DMAC-DPS are presented in Table 2.In solution, the emitters display only prompt fluorescence with a lifetime (τ) of around 5.0 ns (Fig. 3d and Supplementary Fig. 63), in agreement with previous reports for mDICz (ref.27), and expectedly shorter than the prompt fluorescence lifetime of pristine DMAC-DPS (13.8 ns).In contrast, when doped into DMAC-DPS, clear biphasic decay is observed, as singlet excitons from the TADF host are transferred to the terminal emitters on both prompt and delayed timescales.
Although the prompt component of Φ PL (Φ Prompt ) for both terminal emitters is similar, the overall Φ PL is substantially smaller for NB-3 compared with NB-1 due to a large reduction in the delayed PL (Φ Delayed ), as also evidenced by a shorter delayed PL lifetime (τ Delayed ).As triplets are only present in a PL experiment for TADF hosts on the delayed timescale after intersystem crossing, the selective decrease in delayed PL for NB-3 compared with NB-1 supports the hypothesis of a triplet-mediated pathway such as Dexter transfer, which is suppressed by the encapsulated structure of NB-1.Metrics describing FRET and Dexter transfer were next calculated from the time-resolved PL data (Supplementary Note 5).Crucially, in DMAC-DPS, the rate of Dexter transfer (k DET ) for NB-1 is reduced by over a factor of three (3.2) compared with NB-3 at the same wt% doping, supporting the fact that encapsulation can suppress Dexter transfer.Parameters such as FRET rate (k FRET ), Förster radius (R 0 ) and average donor-acceptor distance (R Average ) also suggest that FRET is more efficient for NB-3 than NB-1, in line with the encapsulated structure of NB-1 increasing the average intermolecular distances.

TA study
Although supportive of the hypothesis that the encapsulated structure of NB-1 can suppress Dexter transfer to a terminal emitter from a TADF host, calculations based on time-resolved PL data invariably require  assumptions stemming from the dark nature of organic triplet states.TA spectroscopy is not restricted to bright states, and therefore, we turned to this technique in the hope of directly observing Dexter transfer to further understand the energy transfer dynamics in DMAC-DPS-hosted films.Experiments were pumped at 355 or 400 nm.The TA spectra of the diluted emitters and a neat film of DMAC-DPS were first assigned (Fig. 4a and Supplementary Figs.92-97).In a toluene solution, NB-1 and NB-3 display very similar spectra: photoinduced absorption (PIA) centred at around 480 nm, which decays with a lifetime of around 10 ns, is assigned to S 1 → S n , whereas a slower-rising PIA centred between 540 and 560 nm with a lifetime of tens of microseconds and clear oxygen quenching is assigned to T 1 → T n (Supplementary Figs.92-97).A neat film of DMAC-DPS displays a broad PIA between 600 and 700 nm when probed on timescales between 0.2 ps and 100 μs (Fig. 4a and Supplementary Fig. 101).The kinetics roughly fit biphasic prompt-delayed behaviour, as previously reported for TADF materials 52 (Fig. 4b and Supplementary Fig. 104).Fortunately, the PIA of DMAC-DPS has poor overlap with the T 1 PIA of the emitters, displaying essentially no long-lived signal of ≤550 nm, which we envisaged should make it possible to directly observe Dexter transfer in doped films.
Next, 1 wt% NB-1 and NB-3 films in DMAC-DPS were investigated.Owing to the low concentration of the dopants, essentially an exclusive excitation of the DMAC-DPS host is expected.Short-time experiments (<2 ns) indicate a faster initial decay of the DMAC-DPS PIA for the NB-3-doped film compared with NB-1, in agreement with the larger k FRET calculated for NB-3 above (Supplementary Fig. 102).
Microsecond-timescale experiments were next performed to probe delayed processes (Fig. 4c,d).Normalized decays of the 540-560 nm region for the doped films are shown in Fig. 4b, with neat DMAC-DPS for comparison.For the NB-3-doped film, the prompt-delayed biphasic kinetics of DMAC-DPS are substantially quenched when probed at either 540-560 nm or 600-700 nm (Fig. 4b and Supplementary Fig. 108), suggesting that delayed fluorescence, and perhaps T → S reverse intersystem crossing, are suppressed (in agreement with the relatively low Φ Delayed value of the NB-3-doped film).Meanwhile, the 540-560 nm PIA levels out within ~200 ns and persists onto the microsecond timescale at an intensity appreciably larger than that for neat DMAC-DPS, consistent with terminal emitter triplets.Although it is plausible to simply ascribe delayed fluorescence quenching to FRET 20 , this mechanism cannot explain the concomitant population of the terminal emitter triplet state, implying a different quenching pathway.Such a result is consistent with what would be expected for Dexter transfer, which-to the best of our knowledge-is being directly observed for the first time in a hyperfluorescent system.
After validating the potential of TA for directly observing Dexter transfer with NB-3-doped films, DMAC-DPS films doped with the encapsulated NB-1 emitter were investigated.For NB-1-doped films, the prompt-delayed biphasic kinetics of DMAC-DPS are more strongly retained than for NB-3 (Fig. 4b and Supplementary Fig. 106), in agreement with the higher Φ Delayed mentioned above.The more appreciable delayed fluorescence of NB-1-doped films is accompanied by a weaker long-lived 540-560 nm PIA compared with NB-3, suggesting

Outlook
Current blue hyperfluorescence technology requires OLEDs with complex EMLs of at least three components-a TADF sensitizer, narrowband emitter guest and wide-gap matrix (matrices).High triplet energy matrices have been considered essential to suppress Dexter transfer to the terminal emitter towards high device efficiency and stability.
Here we developed ultranarrowband blue emitters encapsulated by bulky alkylene straps.The ideal spectral properties of the emitters facilitates efficient FRET of singlets from pristine blue TADF hosts.Through comparison with a non-encapsulated analogue using TA spectroscopy, the population of terminal emitter triplets via a Dexter mechanism was directly observed in a hyperfluorescent system for the first time, enabling us to reveal that encapsulating straps can efficiently suppress the Dexter loss pathway to terminal emitter triplets.
Consequently, the new molecular design circumvents the need for a matrix, facilitating efficient blue MFHF OLEDs, with EMLs consisting of TADF sensitizer hosts and encapsulated terminal emitters only.Owing to the efficient suppression of Dexter transfer, MFHF OLEDs demonstrated negligible drops in EQE compared with non-doped reference devices, with a maximum EQE of 21.5%.Simultaneously, the EL FWHM of devices was decreased by a factor of six compared with the non-doped devices from >80 nm to ultranarrow values of 14-15 nm, with desirable deep blue peak wavelengths (449 and 458 nm).Non-doped deep blue TADF research is steadily progressing [53][54][55] , and with improved TADF hosts, we envisage that this encapsulated ultranarrowband emitter design should enable efficient MFHF OLEDs satisfying BT.2020, similar to the CIE y ≤ 0.05 that can be obtained in TTA-sensitized devices with the NB-1 emitter (Supplementary Table 16).More broadly, it has been suggested that the unsatisfactory stability of current blue MCHF OLEDs is directly linked to the population of terminal emitter triplets 21 , making the new strategy outlined here for suppressing Dexter transfer highly relevant.
S3.A solution of S2 (5.81 g, 14.2 mmol, 1.00 eq.) in dry tetrahydrofuran (THF) (80 ml) was cooled to −78 °C in a dry ice-acetone bath under argon.n-BuLi (2.5 M in hexane, 7.37 ml, 18.4 mmol, 1.30 eq.) was added dropwise and the resulting mixture was stirred for 1 h.B(OMe) 3 (5.54ml, 49.7 mmol, 3.50 eq.) was added dropwise, and the resulting mixture was warmed overnight to room temperature.The reaction mixture was quenched with 1 M HCl (30 ml) and stirred for 1 h.The layers were separated and the aqueous layer was extracted with dichloromethane (DCM) (3 × 50 ml).The extracts were combined, dried over MgSO 4 , filtered and the solvent was removed under reduced pressure.The residue was purified by flash chromatography on silica gel (eluent:gradient, 0:1-1:0 DCM/n-hexane v/v) to afford S3 as a clear oil that solidified on standing (4.52 g, 12.1 mmol, 85%). 1 H NMR data were in accordance with those reported in the literature 57 . 1  S4.S1 (1.00 g, 1.48 mmol, 1.00 eq.), S3 (3.33 g, 8.90 mmol, 6.00 eq.), Pd(OAc) 2 (20 mg, 88.9 μmol, 6 mol%) and SPhos (73 mg, 178 μmol, 12 mol%) were combined in a 500 ml round-bottomed flask that was evacuated and backfilled with argon five times.Degassed toluene (250 ml, with five drops of Aliquat 336) and a degassed solution of K 3 PO 4 (1.89 g, 8.90 mmol, 6.00 eq.) in water (32 ml) were sequentially added, and the resulting mixture was refluxed overnight in a preheated 130 °C oil bath.The reaction mixture was cooled to room temperature and acidified with 1 M HCl (100 ml).The layers were separated, and the aqueous layer was extracted with DCM (3 × 50 ml).The extracts were combined, dried over MgSO 4 , filtered and the solvent was removed under reduced pressure.The residue was purified by flash chromatography on silica gel (eluent:gradient, 0:1-35:65 DCM/n-hexane v/v) and then recrystallized (100 ml ethanol and minimal CHCl 3 , reflux → −20 °C) to afford S4 as yellow needles (950 mg, 0.81 mmol, 55%). 1  S5.S4 (800 mg, 0.68 mmol) and Grubbs second-generation catalyst (36 mg, 41 μmol, 6 mol%) were stirred overnight in degassed dry DCM (200 ml) under argon at reflux.The solvent was evaporated and the residue was purified by flash chromatography on silica gel (eluent:gradient, 0:1-35:65 DCM/n-hexane v/v, a single yellow band was collected) to afford S5 as a yellow powder sufficiently pure for the next step after trituration with methanol and drying under suction (753 mg, 0.66 mmol, 96%), yielding a mixture of E and Z isomers. 1 1 H NMR spectrum are assigned to a mixture of E and Z isomers, considering that no shoulder peaks are present after alkene reduction in the next step.Shoulders are also present in the 13 C NMR spectrum.Only peaks for the major compound are listed.

X-ray crystallography
X-ray data were collected on a Bruker D8 QUEST diffractometer, equipped with an Incoatec IμS Cu microsource (λ = 1.5418Å) and a PHOTON III detector operating in the shutterless mode.Crystals were mounted on a MiTeGen crystal mount using inert polyfluoroether oil and the analysis was carried out under an Oxford Cryosystems open-flow N 2 Cryostream operating at 180(2) K.The control and processing software was Bruker APEX4 (v.2022.1-1) 60.The diffraction images were integrated using SAINT in APEX4, and a multiscan correction was applied using SADABS.The final unit-cell parameters were refined against all reflections.Structures were solved using SHELXT 61 and refined using SHELX 62 .All of the crystal structures include CHCl 3 solvent molecules, and the crystals were generally prone to solvent loss and degradation on removal from the mother liquour.In some cases, the SQUEEZE algorithm within PLATON was applied to complete the refinement 63 .Supplementary Table 1 provides summary details of the data collection and structure/refinement parameters.

Computational details
Computations were performed using (time-dependent) density functional theory as implemented in ORCA 5.0 (ref.64).Ground-and excited-state structure optimizations, Hessians and vibronic coupling https://doi.org/10.1038/s41563-024-01812-4parameters were computed at the ωB97X-D3/def2-SVP level of theory 65,66 .Vertical excitations are reported at the LC-BLYP/def2-TZVP level using a range separation parameter of μ = 0.1 a.u.chosen to experimentally reproduce the observed singlet and triplet excitation energies 67 .Absorption and emission spectra were computed using a path integral approach 68 as implemented in the ORCA ESD module, employing ωB97X-D3/def2-SVP Hessians along with LC-BLYP/def2-TZVP excitation energies.A linear vibronic coupling model [69][70][71] was used for interpreting the spectral lineshapes.Here the Huang-Rhys factor for normal mode i was computed as where κ i is the vibronic coupling parameter and ω i is the wavenumber for mode i, both inserted in atomic units and computed with respect to the ground-state modes.These computations were carried out using the linear vibronic coupling functionality of SHARC 70 .Natural transition orbitals 72 were computed using the TheoDORE program 73 .Molecular volumes were calculated via a marching tetrahedron model 74 .

Electrochemical measurements
Cyclic voltammetry was carried out using a PalmSens EmStat4S at a scan rate of 100 mV s −1 .Solutions were prepared in dry, degassed THF with 0.1 M n-Bu 4 NPF 6 as the supporting electrolyte.All the experiments were run under argon with a glassy carbon working electrode, Ag/AgCl as the wire quasi-reference and a Pt wire as the counter electrode.
The potentials were internally referenced to the half-potential of an fcH/fcH + redox couple.

Photophysical measurements
Solution absorption spectra were measured using a Shimadzu UV-1800 instrument.Molar extinction coefficients were determined from triplicate runs using a method intended to minimize weighing and dilution error.For each run, >3 mg of the compound was accurately weighed into a 25 ml volumetric flask to make a stock solution.Then, 100 μl of the stock solution was titrated five times into a 1-cm-path-length cuvette (2 ml starting blank solvent volume), measuring the absorption spectra after each addition.Fluorescence and time-correlated single-photon-counting experiments were carried out on Edinburgh Instruments FS5.Solution PL quantum yield measurements were carried out in an integrating sphere on Edinburgh Instruments FS5.For organic films for photophysics, glass substrates (steady-state PL and absorption, PL quantum efficiency and transient PL) and quartz substrates (TA) were prepared, and they were cleaned with acetone and isopropyl alcohol with sonification for 10 min before loading them into an evaporator (Angstrom Engineering).Then, 50-nm-and 100-nm-thick films were formed on glass and quartz substrates, respectively.The thermal evaporating process was conducted in a vacuum chamber under <5 × 10 −7 mbar.Steady-state PL spectra were measured by an Edinburgh Instruments fluorescence spectrometer (FLS980) with a monochromated xenon arc lamp at λ Ex = 330 nm under nitrogen flow.FLS980 with an integrating sphere under nitrogen flow was used to measure the PL quantum efficiency, and the films were excited by a 330 nm laser.Transient PL was recorded by using an Andor electrically gated intensified charge-coupled device (iCCD) with 330 nm laser excitation; the decay kinetics were obtained from the integration of the total spectrum at each time.
To determine the emitter triplet energies, the samples were prepared via drop casting onto a sapphire substrate from a mixture of the compound at 1 wt% to ZEONEX in toluene.The films were then dried in a vacuum oven at room temperature for 1 h to remove any trace solvent.The time-resolved PL spectra were recorded using nanosecond gated luminescence and lifetime measurement setup (from 400 ps to 1 s).The sample was loaded into a Janis Research VNF-100 cryostat, which was placed under a vacuum and kept at room temperature.The excitation pulses were provided by an Ekspla Nd:YAG laser at the third harmonic of 355 nm and the emission was collected after passing through a spectrograph on a Stanford Computer Optics iCCD camera to produce the time-resolved emission spectra.
For photostability measurements, the samples were prepared in degassed toluene and irradiated with a 400 nm upconverted pump laser.The PL intensity was monitored as a function of time with an iCCD camera.
TA spectroscopy was performed on a setup powered using a commercially available Ti:sapphire amplifier (Spectra Physics Solstice Ace).The amplifier operates at 1 kHz and generates 100 fs pulses centred at 800 nm with an output of 7 W.For the ultrafast TA measurements, a portion of the laser fundamental was frequency doubled using a 1-mm-thick β-barium borate crystal for sample excitation at 400 nm, whereas the third harmonic (355 nm) of an electronically triggered, Q-switched Nd:YVO 4 laser (Innolas Picolo 25) provided the ~1 ns pump pulses for the nanosecond-microsecond (1 ns to 100 μs) TA measurements.The probe was provided by a broadband visible (525-775 nm) non-collinear optical parametric amplifier.The probe pulses are collected with a Si dual-line array detector (Hamamatsu S8381-1024Q), driven and read by a custom-built board from Entwicklungsbüro Stresing.For determining the triplet-excited-state absorption of the terminal emitters, the probe was instead generated by a LEUKOS Disco 1 ultraviolet low-timing-jitter supercontinuum laser (STM-1-UV), which was then electronically delayed relative to the femtosecond 400 nm excitation by an electronic delay generator (Stanford Research Systems DG645).

Organic light-emitting devices
For the fabrication of OLED devices, indium-tin-oxide-coated substrates (~15 Ω cm -2 ) were cleaned with acetone and isopropyl alcohol, and then O 2 plasma treatment was applied to align the energy level with a hole-transporting layer.All the layers, including the organic layers and a LiF/aluminium cathode, were thermally deposited in a high vacuum (~10 −7 torr).The performance of the OLED devices was measured by a Keithley 2635 source meter and a calibrated Si photodiode.The EL spectra were recorded by an Ocean Optics Flame spectrometer.

Fig. 1 |
Fig. 1 | Molecular design.a, Schematic and Jablonski diagram for the suppression of Dexter triplet transfer from a TADF host via emitter encapsulation.b, Structures of the mDICz and encapsulated NB-1 luminophores; relevant absorption and PL spectra for NB-1 and DMAC-DPS.c, Synthesis schemes and structures of NB-1 and NB-2 with X-ray single-crystal structures (H atoms are omitted for clarity).
Article https://doi.org/10.1038/s41563-024-01812-4 in photophysical studies on System A. The PL spectra for the concentration series of NB-1 and NB-3 evaporated into DMAC-DPS are shown in Fig.

Fig. 2 |
Fig. 2 | Organic light-emitting devices.a, Material combinations for matrix-free Systems A, B and C. b, EL spectra of System A, with photographs of OLED emission shown in the insets.c, EL spectra of System B, with En-Per single-crystal X-ray structure shown in the inset.XRD, X-ray diffraction.d, EL spectra of System C, with photographs of OLED emission shown in the insets.

4 Extended Data Fig. 3 |
https://doi.org/10.1038/s41563-024-01812-Synthesis of NB-3.Scheme for the synthesis of NB-2 via acid-catalysed condensation and Cu-catalysed intramolecular Ullmann chemistry.Extended Data Fig.4| Synthesis of En-Per.Scheme for the synthesis of En-Per via chemistry analogous to that employed in the synthesis of NB-2.complete.

Table 2 | Φ PL and time-resolved PL data with relevant energy transfer parameters Host Dopant (1 wt%) Φ PL (±0.05) Φ Prompt Φ Delayed τ Prompt (ns) τ Delayed (μs) k FRET (10 7 s −1 ) R 0 (nm) R Average (nm) k DET (10 6 s −1 )
a lower population of terminal emitter triplets.Hence, TA provides direct evidence that the encapsulated molecular structure of NB-1 can appreciably suppress loss via Dexter transfer to terminal emitter triplets in a pristine TADF host, supporting the minimal OLED EQE drop encountered for the MFHF OLED.
Article https://doi.org/10.1038/s41563-024-01812-446.Lim, H., Woo, S. J., Ha, Y. H., Kim, Y. H. & Kim, J. J. Breaking the efficiency limit of deep-blue fluorescent OLEDs based on anthracene derivatives.Adv.Mater.34, 2100161 (2022).47.Ahn, Y. et al.The positional effect of arylamines on pyrene core in a blue fluorescent dopant significantly affecting the performance of organic light emitting diodes.Dyes Pigm.205, 110505 (2022).48.Qu, Y.-K.et al.Steric modulation of spiro structure for highly efficient multiple resonance emitters.Angew.Chem.Int.Ed. 61, e202201886 (2022).49.Cheon, H. J., Shin, Y., Park, N., Lee, J. & Kim, Y. Boron-based multi-resonance TADF emitter with suppressed intermolecular interaction and isomer formation for efficient pure blue OLEDs.Small 18, 2107574 (2022).50.Jiang, P. et al.Quenching-resistant multiresonance TADF emitter realizes 40% external quantum efficiency in narrowband electroluminescence at high doping level.Adv.Mater.34, 2106954 (2022).51.Meng, G. et al.Highly efficient and stable deep-blue OLEDs based on narrowband emitters featuring orthogonal spiro-configured indolo[3,2,1-de]acridine structure.Chem.Sci. 13, 5622-5630 (2022).52.Cui, L. S. et al.Fast spin-flip enables efficient and stable organic electroluminescence from charge-transfer states.Nat.Photon.14, 636-642 (2020).53.Han, J. et al.Narrowband blue emission with insensitivity to the doping concentration from an oxygen-bridged triarylboron-based TADF emitter: nondoped OLEDs with a high external quantum efficiency up to 21.4%.Chem.Sci. 13, 3402-3408 (2022).54.Xia, G. et al.A TADF emitter featuring linearly arranged spiro-donor and spiro-acceptor groups: efficient nondoped and doped deep-blue OLEDs with CIE y <0.1.Angew.Chem.Int.Ed. 60, 9598-9603 (2021).55.Shi, Y. Z. et al.Recent progress in thermally activated delayed fluorescence emitters for nondoped organic light-emitting diodes.Chem.Sci. 13, 3625-3651 (2022).Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material.If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.To view a copy of this licence, visit http://creativecommons. org/licenses/by/4.0/.