Deuterated Radical Emitters Achieve 10.8% EQE at 800 nm with Remarkably Enhanced Operational Stability

Abstract

Near-infrared (NIR) radical emitters are promising for organic light-emitting diodes (OLEDs) but are limited by low emission efficiency and operational stability. Here, we demonstrate that C–H perdeuteration markedly enhances the performance of both the benchmark TTM-TPA radical and a red-shifted analogue, PyBTM-TPA. Photoluminescence quantum yields increase from 24.0% to 30.7% for TTM-TPA and from 7.3% to 10.4% for PyBTM-TPA. Deuterated devices also exhibit improved performance, achieving record maximum external quantum efficiencies (EQE_max) of 10.81% at peak wavelength 800 nm and 3.56% at 865 nm for perD-TTM-TPA and perD-PyBTM-TPA, respectively, together with extended operational lifetimes. We further reveal that the isotope effect is governed by molecular reorganization energy, determined by both the effective Huang–Rhys factor and the distribution of vibronic coupling among individual vibrational modes. These findings establish deuterium engineering as a general strategy for improving radical-based NIR-OLEDs.

Near-infrared organic radical emitters are promising, but their efficiency and stability remain limited. Wang et al. used deuterium substitution to enhance emission, device performance, and lifetime, and identified the factors governing isotope-driven improvements.

Introduction

Deuterium (D) possesses approximately twice the atomic mass of protium (H). In principle, because the zero-point energy of a molecule is the sum of the energies of all vibrational normal modes at the ground vibrational level (v = 0), substitution of C–H with C–D lowers the vibrational frequencies and thus reduces the zero-point energy. Consequently, C–D bonds exhibit higher effective bond dissociation energies than their C–H counterparts. As a result, the incorporation of C–D bonds in place of C–H bonds has been in ‘hot spots’ in drugs via modulation of the metabolism process^1–4, the development of catalysts^5,6, and materials in optoelectronics with less propensity to photodegradation^7–10.

From another perspective, the theory of radiationless deactivation via internal conversion (IC)—commonly referred to as the emission energy-gap law¹¹—states that, for a fixed reorganization energy, a decreasing emission energy gap strengthens vibronic (exciton–vibration) coupling, thereby accelerating the rate of IC (k_ic). Assuming k_ic to be the dominant nonradiative pathway, the PLQY, defined as PLQY = k_r/(k_r+k_ic) where k_r denotes the radiative decay rate constant, is then decreased as emission gap is reduced. This framework, originally formulated by Englman and Jortner¹¹ (eq. (1) in the Supplementary Information (SI)), highlights that suppressing k_ic is particularly critical for achieving efficient near-infrared (NIR) emission. Conventional understanding suggests that high-frequency C–H stretching modes require only a small number of vibrational quanta to bridge the reduced energy gap and therefore play a dominant role in quenching NIR emission, especially given the ubiquity of C–H bonds in organic molecules. Accordingly, substituting C–H with C–D has attracted considerable interest, as the lower vibrational frequency of C–D modes is expected to require more quanta to reach the gap, thereby suppressing nonradiative decay pathways. However, reported outcomes remain mixed, and the resulting pros^12–15 and cons¹⁶ have continued to fuel the debate. The point of contention stems from the fact that, in most organic and organometallic NIR emitters, the lowest excited states are typically π → π* and/or charge-transfer (including metal-to-ligand charge transfer) in nature, where C–H stretching modes couple weakly to the electronic transition. Therefore, the contribution of C–H stretching to the reorganization energy (λ_reg) between S₁ (or T₁) and S₀ is expected to be very small, implying minimal impact on vibronic coupling and negligible differences in PLQY between C–H and C–D analogues¹⁷_.

Recently, by integrating the Englman–Jortner formalism¹¹ (eq. (1)) with Lin’s treatment¹⁸, we clarified that the C–D isotope effect on k_ic is closely related to the overall exciton–vibration coupling strength (see eq. (2) in SI)¹⁹. In the weak coupling regime, where λ_reg is <$h\overline{\upsilon }$ ($\overline{\upsilon }$: the collective vibration frequency, with 2π$\overline{\upsilon }$ = $\overline{\omega }$ in eq. (2)), and the effective Huang–Rhys factor S_eff, defined as S_eff = λ_reg/h$\overline{\upsilon }$^20,21 is less than unity, k_ic can be expressed by eq. (2a) (see SI) and the C–D substitution is expected to induce only limited changes. Conversely, in the strong coupling regime, where λ_reg$\gg$ $h\overline{\upsilon }$ and S_eff $\gg$1, k_ic follows eq. (2b) (see SI), a pronounced C–D enhancement is anticipated. Here, S_eff is employed as an effective descriptor of the collective exciton–vibration coupling strength within a multimode vibronic framework rather than as a uniquely defined single-mode quantity.

On this basis, we demonstrated pronounced C–D benefit reported for self-assembled Pt(II) complexes exhibiting NIR phosphorescence^19,22. Unfortunately, for the self-assembled Pt(II) complexes, exciton delocalization and the intrinsically weak, formally spin-forbidden triplet absorption preclude reliable determination of λ_reg and, consequently, S_eff. This limitation hampers direct examination of the proposed coupling-strength-dependent picture.

More importantly, eqs. (2a) and (2b) are derived for two limiting cases, namely S_eff $<$ 1 and S_eff $\gg$ 1, respectively. In practice, however, many organic emitters fall within an intermediate regime (e.g., S_eff ≈ 0.3–1.5). For example, cyanine systems generally reside at the lower end of this range, whereas polyene-like scaffolds tend to lie toward the higher end^23,24. In this intermediate regime, neither limiting approximation is expected to apply rigorously. The isotope response may depend not only on the overall coupling strength but also on the distribution of vibronic coupling among individual vibrational modes.

To provide direct and compelling evidence for the above concept, a decisive experimental validation would be to examine the correlation between S_eff and the deuterium isotope effect on luminescence intensity under comparable energy gaps. Based on this objective, we extend isotope engineering, for the first time, to state-of-the-art organic radical NIR emitters^25–28. Notably, as summarized in Supplementary Table 3, which compiles literature reports of deuterated organic and organometallic emitters used in OLEDs, none of the reported systems involve deuterated radical emitters.

Specifically, the benchmark radical TTM-TPA (peak wavelength = 800 nm, PLQY = 24.0% in toluene)²³ meets requirements. Also, we developed further red-shifted analogue, PyBTM-TPA (peak wavelength = 905 nm, PLQY = 7.3%) (Fig. 1), both of which are characterized by λ_reg and S_eff (vide infra). Systematic deuteration results in an almost linear enhancement of the PLQY, reaching 30.7% and 10.4% for the fully deuterated compounds perD-TTM-TPA and perD-PyBTM-TPA, respectively. This enhancement translates directly into improved device performance, enabling perD-TTM-TPA–based NIR OLEDs to achieve a record external quantum efficiency (EQE) of 10.81% at 800 nm among radical emitters. These results are further corroborated by comparison with another known radical emitter TTM-3PCz, and the NIR organic emitter Y6, both of which were also systematically deuterated (Supplementary Table 1). These comparisons provide unambiguous evidence that C–D substitution effectively suppresses internal conversion as the exciton–vibration coupling strength increases. Together with the markedly prolonged operational device lifetime, this strategy pushes the performance frontier of organic radical NIR OLEDs. The detailed results and discussion are presented below.

Fig. 1. Structures and brief synthetic methods for the studied radicals.

The chemical structures of (a) TTM-TPA series radicals, and (b) PyBTM-TPA series radicals (c) General synthetic strategy for radicals.

Results and discussions

Molecular design and photophysical properties

In this work, the choice of TTM-TPA is motivated by its large Stokes shift between the lowest-lying absorption band (peak wavelength ~ 660 nm in toluene and in solid films) and the emission maximum (peak wavelength ~ 800 nm) due to the charge-transfer effect. Such a Stokes shift implies a substantial excited-state λ_reg and S_eff (vide infra), rendering TTM-TPA a sensitive platform for probing vibronically mediated nonradiative decay processes. At the same time, the appreciable emission efficiency provides a meaningful proof-of-concept for evaluating the deuterium isotope effect, with direct relevance to practical optoelectronic applications such as OLEDs. We then extended the synthesis to another radical emitter PyBTM-TPA. Owing to the electron-withdrawing nature of the dihalopyridine moiety²⁹, PyBTM-TPA is expected to exhibit further charge transfer, increasing λ_reg and further red-shifted emission relative to TTM-TPA.

The TTM-TPA radical emitter was synthesized based on previously reported procedures, with subsequent optimization of the synthetic conditions^23,30. However, its fully deuterated analogue, perD-TTM-TPA, failed to obtain under common one-step harsh hydrothermal approach due to severe steric hindrance arising from the propeller-like HTTM core and the TPA unit, which results in inefficient deuterium incorporation^31,32. In addition, conventional deuteration procedures involving Pd-catalyzed reactions conducted under a hydrogen atmosphere may lead to detrimental dechlorination of the TTM core³³. Therefore, alternatively, a stepwise synthetic strategy was adopted, in which fully deuterated 4-(diphenylamino)phenylboronic acid was first prepared and then coupled with the HTTM-d₆ precursor via a Suzuki cross-coupling reaction, following radical generation to afford perD-TTM-TPA. This route yielded an overall deuterium incorporation of up to 98%. Additionally, deuterium substitution was selectively introduced at either the TTM or the TPA moiety, yielding TTM-TPA-d₆ and TTM-TPA-d₁₄, respectively. This systematic variation in the degree of deuteration enables investigation of the relationship between deuterium content as well as whether deuteration on the donor or acceptor unit leads to distinct contributions to the deuterium effect. To demonstrate that deuteration substitution is also effectively enhance the emission on the further red-shifted regions, another radical emitter, PyBTM-TPA, and its per-deuterated analogue, perD-PyBTM-TPA, were synthesized.

Figure 2a and b show the solution-state absorption and emission spectra of TTM-TPA and its C–D isotopologues, as well as those of PyBTM-TPA and its systematically deuterated analogues, respectively. As shown in Fig. 2a, the absorption and emission spectral features of TTM-TPA including peak positions, 660 nm and 800 nm for absorption and emission maxima, respectively, and full width at half-maximum (FWHM) remain essentially unchanged upon deuteration. Similar behavior is observed for the PyBTM-TPA series (Fig. 2b), for which the lowest-lying absorption and emission maxima are located at approximately 720 nm and 905 nm, respectively, in toluene. These results indicate that deuteration does not perturb the electronic transition energies or the Franck–Condon factors governing the ground-to-excited-state transitions.

Fig. 2. Absorption, emission spectra and time-resolved photoluminescence decay of the studied radicals.

a, b Normalized absorption and photoluminescence spectra of (a) TTM-TPA series or (b) PyBTM-TPA series in toluene. c Photoluminescence intensity of the TTM-TPA series as a function of deuteration degree under the same absorbance of 0.15 at wavelength of 620 nm in toluene. d Photoluminescence intensity of the PyBTM-TPA series as a function of deuteration degree under the same absorbance of 0.15 at wavelength of 660 nm in toluene. e, f Time-resolved photoluminescence decay profiles of (e) TTM-TPA series or (f) PyBTM-TPA series, showing an increase in lifetime with increasing deuteration degree. Inset of (c) and (d): The plot of ratio of PLQY (versus non-deuterated reference) as a function of % of deuteration.

Despite their nearly identical spectral profiles, pronounced differences are observed in both emission intensity and the associated excited-state relaxation dynamics. For both TTM-TPA and PyBTM-TPA, progressive deuteration leads to a systematic enhancement of PLQY. Plotting PLQY as a function of the degree of deuterium substitution (see inset of Fig. 2c and d, and Supplementary Fig. 18 for details) reveals a strong linear correlation, indicating that the PLQY enhancement approximately scales with the number of incorporated deuterium atoms in the current molecular series. Moreover, the steeper slope of the linear correlation observed for the more bathochromic PyBTM-TPA system, compared with TTM-TPA, indicates a larger PLQY enhancement per C–D substitution at smaller emission energy gaps. Consequently, the PLQY increases from 24.0% for undeuterated TTM-TPA to 30.7% for fully deuterated perD-TTM-TPA, and corresponds to an overall enhancement of approximately 28%, and from 7.3% for PyBTM-TPA to 10.4% for perD-PyBTM-TPA (Fig. 2d), corresponding to an enhancement of up to 43% upon full deuteration. Notably, a concomitant change in the appearance of the sample, from deep red for TTM-TPA to deep green for perD-TTM-TPA, is observed. (Supplementary Fig. 19). This visual change arises from differences in the emission intensity distribution within the visible region around ~700 nm, which strongly modulates the spectral response of the human eye.

Table 1 summarizes the key photophysical parameters, including the absorption and emission maxima, PLQYs, observed lifetimes (τ_obs), and the corresponding radiative (k_r) and nonradiative (k_nr) decay rate constants derived therefrom. For both the TTM-TPA and PyBTM-TPA series, the s k_r remain essentially invariant within experimental uncertainty, regardless of the extent of C–D substitution. In contrast, the major differences among the deuterated analogues arise from changes in the k_nr. Given the small energy gap in the NIR region, surface crossing between the doublet ground state (D₀) and the emissive excited state (D₁) is expected to be negligible. Moreover, the weak temperature dependence of the lifetimes (Supplementary Figs. 13 and 14) further supports the conclusion that non-radiative decay is predominantly governed by internal conversion, which is mainly determined by Franck–Condon overlap and vibronic coupling. The observed reduction in k_nr upon deuteration, therefore, mainly indicates a suppression of k_ic, which directly accounts for the enhanced PLQY.

Table 1.

The photophysical properties of TTM-TPA, PyBTM-TPA, and Y6 with different degrees of deuteration

Compounds	λAbs (nm)	λEm (nm)	Φ (%) a	ΦD/Φ b	τobs (ns)	kr (107 s−1) c	knr (107 s−1) c
TTM-TPA	660	800	24.0	1.00	8.35	2.9	9.1
TTM-TPA-d6	660	800	25.4	1.06	8.80	2.9	8.5
TTM-TPA-d14	660	800	27.3	1.14	9.86	2.8	7.4
perD-TTM-TPA	660	800	30.7	1.28	10.34	2.9	6.7
PyBTM-TPA	720	905	7.3	1.00	2.72	2.7	34.1
PyBTM-TPA-d6	720	905	8.2	1.13	2.77	2.9	33.1
PyBTM-TPA-d14	720	905	8.8	1.21	3.21	2.7	28.4
perD-PyBTM-TPA	720	905	10.4	1.43	3.39	3.0	26.4

^aPLQYs were measured in toluene using known compound TTM-TPA as a standard.

^bIncrease in PLQY ratio (compare to parent radical).

^cEmission lifetimes were measured using TCSPC system with excitation at 375 nm, k_r = PLQY/τ_obs and k_nr = (1-PLQY)/τ_obs.

To further examine whether deuteration perturbs the underlying electronic structure, we analyzed the frontier radical orbitals of the protonated and deuterated emitters (Supplementary Fig. 21). The calculated SOMO and SUMO distributions remain essentially unchanged upon deuteration, accompanied by only negligible variations in orbital energies and SOMO–SUMO energy gaps. These results indicate that isotope substitution exerts little influence on the electronic-state character and excited-state assignment. Therefore, the observed enhancement is more reasonably attributed to modifications of vibronic coupling and associated nonradiative relaxation processes rather than electronic-state reorganization. We then performed computational analyses aimed at evaluating the reorganization energy for both TTM-TPA and PyBTM-TPA, which is evaluated through mode-resolved reorganization energy calculations based on the harmonic approximation. The vibronic coupling calculations were carried out using MOMAP based on Gaussian 16 optimized geometries and frequency analyses^34,35. In the present treatment, the Duschinsky rotation effect arising from normal-mode mixing between different electronic states was included in the calculation of vibronic quantities. The analysis focuses on the D₁ state, which is the experimentally observed emissive state. Although higher-lying excited states were examined, no experimental evidence, such as additional emission bands or anomalous spectral features, suggests their significant involvement in the observed photophysics. Furthermore, the nearly identical spectral profiles observed upon deuteration support a common D₁ emissive origin across all isotopologues. Therefore, higher excited states were not explicitly included in the vibronic analysis. Meanwhile, higher-order vibronic mixing beyond the harmonic approximation and spin-dependent effects, including spin–orbit coupling induced by multiple chloride substituents as well as spin-flip processes intrinsic to radical excited states, were assumed to be small and neglected. Upon summing the mode-resolved Huang–Rhys factors obtained from vibronic analysis (Supplementary Fig. 20, see computational detail in SI), the reorganization energies of 3.9 and 5.1 kcal mol⁻¹ are obtained for TTM-TPA and PyBTM-TPA, respectively (Supplementary Table 2). To estimate the effective Huang–Rhys factor, an effective collective vibrational frequency of 1500 cm⁻¹ was adopted following the reduced-mode treatment employed in the Englman–Jortner analysis and representative C = C skeletal vibrations commonly used to describe organic π-conjugated systems^22,23. S_eff were calculated to be 0.91 and 1.20, respectively.

For comparison, we examined another radical TTM-3PCz³⁰ (Supplementary Table 1 and Supplementary Fig. 15), which serves as an additional radical reference system for examining the applicability of the coupling-strength-dependent isotope picture, together with a typical NIR organic dye, Y6³⁶ (Supplementary Fig. 16). TTM-3PCz provides a radical counterpart with comparable electronic characteristics but emission in the visible-to-NIR boundary region (λ_abs ~ 600 nm and λ_em ~ 700 nm) whereas Y6 is expected to exhibit weak vibronic coupling, as reflected by its very small Stokes shift between the absorption peak (λ_abs ~ 720 nm) and emission peak (λ_em ~ 750 nm). In Supplementary Fig. 15, deuterated TTM-3PCz derivatives exhibit nearly identical absorption and emission spectra, consistent with the observations for deuterated TTM-TPA and PyBTM-TPA. However, the PLQY enhancement induced by deuteration in TTM-3PCz is much less pronounced than that observed for TTM-TPA and PyBTM-TPA. The PLQY ratio between perD-TTM-3PCz and the parent TTM-3PCz is only 1.11 (Supplementary Table 1), indicating that the contribution of each deuterium substitution to PLQY improvement is relatively minor. Similarly, Y6, Y6-d₂, and Y6-d₆ exhibit nearly identical emission characteristics, including peak positions (750 nm) and FWHM, as shown in Supplementary Fig. 16 and summarized in Supplementary Table 1. Compared with Y6, Y6-d₂ and Y6-d₆ show slight PLQY improvements of 2% and 4%, respectively. Based on the same method, the sum of the mode-resolved Huang–Rhys factors obtained from vibronic analysis of Y6 (Supplementary Fig. 20) deduces a reorganization energy of 1.24 kcal mol⁻¹. Assuming the same collective vibrational frequency of 1500 cm⁻¹, an S_eff value of 0.29 is calculated for Y6 in this study, which is close to the previously reported value of 0.33²⁴. Taken together, the radical emitters reside near or beyond the S_eff crossover regime, whereas Y6 remains in the weak-coupling region (S_eff = 0.29). Consistent with this classification, all radical systems exhibit larger isotope effects than Y6, the percentage increase in PLQY for the per-deuterated compounds follows the order of 4% (Y6-d₆) < 11% (perD-TTM-3PCz) < 28% (perD-TTM-TPA) < 43% (perD-PyBTM-TPA), showing a proportional enhancement across the series. However, the magnitude of enhancement does not scale strictly with S_eff alone. For example, despite comparable λ_reg and estimated S_eff values, perD-TTM-3PCz exhibits only minor enhancement (Φ_D/Φ=1.11) relative to perD-PyBTM-TPA. Inspection of the mode-resolved Huang–Rhys distributions (Supplementary Fig. 20) reveals distinct vibronic characteristics among the radical emitters. In particular, TTM-3PCz exhibits vibronic coupling concentrated into a few low-frequency modes, whereas PyBTM-TPA displays broader mode participation. This difference may contribute to the relatively small isotope enhancement observed for TTM-3PCz despite its comparable λ_reg and estimated S_eff. Therefore, while S_eff serves as an effective descriptor for coupling-regime classification, the detailed isotope response additionally depends on the distribution of vibronic coupling among individual vibrational modes.

Deuterium isotope promoting OLEDs

To evaluate the impact of the C–D isotope effect on the electroluminescence (EL) performance of organic radical emitters, we fabricated NIR OLEDs utilizing a standardized device architecture: ITO/HAT-CN (5 nm)/TAPC (60 nm)/EML (20 nm)/B3PYMPM (10 nm)/CN-T2T (60 nm)/Liq (0.5 nm)/Al (100 nm). The emitting layer (EML) comprised 3 wt% of the radical dopant (TTM-TPA, perD-TTM-TPA, PyBTM-TPA, or perD-PyBTM-TPA) host in MADN. This specific doping concentration and device architecture were deliberately adopted to enable a direct, 1-to-1 comparison with the established benchmark report for TTM-TPA³⁷. Maintaining strictly identical architecture and concentration guarantees that any divergence in EQE and lifetime stems exclusively from the intramolecular C–H versus C–D substitution. Furthermore, the direct translation of solution-state PLQY enhancements (Table 1) into solid-state device EQE confirms this boost originates fundamentally from intramolecular exciton–vibration decoupling rather than host-guest factors. HAT-CN (dipyrazino[2,3-f:2’,3’-h]quinoxaline-2,3,6,7,10,11-hexacarbonitrile) was used as the hole injection layer. TAPC (1,1-bis((di-4-tolylamino)phenyl)-cyclohexane) worked as the hole transport layer. MADN (2-methyl-9,10-bis-(naphthalene-2-yl) anthracene) served as the energy transfer host. B3PYMPM (bis-4,6-(3,5-di-3-pyridyl phenyl)−2-methyl pyrimidine) and CN-T2T (3’,3”’,3””’-(1,3,5-triazine-2,4,6-triyl)tris(([1,10-biphenyl]−3-carbonitrile))) acted as electron transport layers. Liq (lithium 8-hydroxyquinolinato) and Al were used for electron injection and as the cathode, respectively. Figure 3 shows the current density–voltage–radiance (J–V–R) curves, external quantum efficiencies–current density (EQE–J) plots, normalized EL spectra, and device lifetime for the NIR OLEDs. Table 2 summarizes the detailed electroluminescence (EL) data for devices.

Fig. 3. Device characteristics of NIR OLEDs based on the studied radicals.

a Current density–voltage–radiance (J–V–R) characteristics, (b) EQE as a function of current density, (c) EL spectra, and (d) Lifetime for the devices measured at 0.1 mA cm^-2 with different dopant.

Table 2.

EL performance parameters of NIR OLEDs based on the studied radicals

Dopant (3 wt%)	Von a [V]	R/J at 12 V b [mW sr-1m−2/mA cm-2]	EQEmax c [%]	EQE J=1/10/100 c [%]	λEL d [nm]	R0 e [mW sr−1m−2]	LT85f [hour]
TTM-TPA	2.7	60659/213	9.01 g	8.89/8.22/6.54	798	40	4.0
perD-TTM-TPA	2.7	84429/271	10.81	10.65/9.50/7.75	800	48	14.0
PyBTM-TPA	2.8	34662/469	2.66	2.64/2.49/2.04	866	10	8.5
perD-PyBTM-TPA	3.0	57913/561	3.56	3.42/3.50/3.00	865	11	76.8

^a The voltage at which the radiance reaches 1 mW sr^–1 m^–2.

^bBoth the radiance and current density are measured at 12 V.

^cThis value corresponds to the EQE_max and EQE determined at 1, 10, and 100 mA cm^–2.

^dEL spectra measured at 5 V.

^eThe initial radiance values at the applied constant current density of 0.1 mA cm⁻².

^fThe duration it takes for the EL intensity to decrease to 85% of its initial value.

^gAn EQE_max of 9.6% was reported in ref. ³⁷.

TTM-TPA, perD-TTM-TPA, PyBTM-TPA and perD-PyBTM-TPA.

As shown in Fig. 3c, all devices exhibit characteristic radical doublet emission in the NIR region. TTM-TPA series displays EL maxima at 798–800 nm, while PyBTM-TPA series shows a significant bathochromic shift to 865–866 nm. All devices turned on at low voltages of 2.7–3.0 V (Table 2), indicating efficient charge injection and exciton formation within the radical system. The influence of deuteration on device efficiency is strikingly evident in the J–V–R and EQE–J characteristics (Fig. 3a, b). For the benchmark 800 nm emitter, the TTM-TPA-based device achieved a maximum external quantum efficiency (EQE_max) of 9.01% and a radiance of 60,659 mW sr^-1 m⁻² at 12 V. This performance is in line with previous results using the MADN: TTM-TPA system³⁷. Upon full deuteration, the perD-TTM-TPA device reached a record EQE_max of 10.81%, representing a 20.0% improvement.

This isotope-induced benefit becomes even more pronounced as the emission energy gap narrows. For the further red-shifted PyBTM-TPA device (λ_EL ~ 865 nm), the EQE_max increased from 2.66% to 3.56% in the perD-PyBTM-TPA analogue—a substantial relative enhancement of 33.8%. The radiance at 12 V similarly jumped from 34,662 to 57,913 mW sr⁻¹ m⁻² (Table 2). This trend confirms that C–D substitution is a particularly powerful strategy for deeper NIR radical emitters, where suppressing nonradiative internal conversion is most critical due to the strong exciton–vibration coupling regime. Furthermore, both deuterated devices demonstrated suppressed efficiency roll-off. At a high current density of 100 mA cm⁻², the perD-TTM-TPA device maintained an EQE of 7.75% (compared to 6.54% for TTM-TPA), while the perD-PyBTM-TPA device held at 3.00% (compared to 2.04% for PyBTM-TPA). Beyond efficiency, importantly, deuteration significantly improved the operational stability of these radical NIR OLEDs. As shown in Fig. 3d, the device lifetimes were measured under a constant current density of 0.1 mA cm⁻². At this driving current, the initial radiance (R₀) values were recorded as 40 and 48 mW sr⁻¹ m⁻² for TTM-TPA and perD-TTM-TPA, respectively, whereas those for PyBTM-TPA and perD-PyBTM-TPA were 10 and 11 mW sr⁻¹ m⁻², respectively. (Table 2). This moderate current density was specifically selected to isolate intrinsic molecular degradation from macroscopic thermal breakdown (e.g., Joule heating), which typically plagues radical emitters at higher currents and currently precludes reliable long-term extrapolation modeling. The LT₈₅ (the time taken for EL intensity to decrease to 85% of its initial level) for the perD-TTM-TPA device was approximately 14.0 h, representing a 3.5-fold enhancement over the 4.0 h observed for the TTM-TPA device. Remarkably, the stability enhancement was even more pronounced for the deep NIR PyBTM-TPA series. The LT₈₅ for PyBTM-TPA device was 8.5 h. Upon full deuteration, the perD- PyBTM-TPA device exhibited an exceptionally long lifetime of 76.8 h, marking a nearly 9-fold increase in operational stability. Initially, we hypothesized that the suppressed efficiency roll-off and enhanced operational lifetime might have originated from differences in molecular packing or stacking density between the hydrogenated and deuterated films. To probe this possibility, we performed GIWAXS analysis on the film samples of the four compounds (TTM-TPA, PyBTM-TPA, perD-TTM-TPA, and perD-PyBTM-TPA). However, as shown in Supplementary Fig. 23, the GIWAXS patterns of all four films exhibited highly disordered and amorphous features, demonstrating that macroscopic film morphology and packing density are not the primary drivers behind the performance enhancement. Alternatively, we attribute the suppressed roll-off and enhanced stability to the higher bond dissociation energy of C–D bonds, which suppresses initial bond cleavage and subsequent defect formation. The concurrent improvements in efficiency and operational stability, particularly in the longer-wavelength NIR region, demonstrate the potential of deuterium-isotope engineering as an effective approach to reducing degradation in radical-based NIR optoelectronics.

In summary, this work delivers both fundamental insights and technological advances concerning the C–D isotope effect in NIR emitters and their corresponding OLED devices.

This study introduces deuterium-isotope engineering into radical optoelectronics by developing deuterated NIR radical emitters and establishes a general synthetic protocol for their preparation. The experimental work suggests a relationship between C–D isotope effects and the exciton–vibration coupling strength that governs non-radiative decay in radical emitters. The charge-transfer character of the series of TTM-TPA and PyBTM-TPA radical emitters leads to larger λ_reg and S_eff, resulting in more pronounced C–D–induced enhancement of emission efficiency. These results support the coupling-strength-dependent isotope picture proposed previously²², in which stronger exciton–vibration coupling generally favors larger isotope enhancement. Meanwhile, comparison with the additional radical reference TTM-3PCz indicates that the detailed isotope response does not scale strictly with S_eff alone and may additionally depend on the distribution of vibronic coupling among individual vibrational modes.

Moreover, record-breaking device efficiency is achieved together with markedly improved operational stability in the resulting OLEDs. The C–D substitution–induced enhancement in PLQY directly translates into improved device performance. The-EQE_max increases from 9.01% for TTM-TPA to record-high 10.81% for perD-TTM-TPA and from 2.66% for PyBTM-TPA to 3.56% for perD-PyBTM-TPA, with emission peaks at 800 nm and 865 nm, respectively. These improvements are accompanied by substantially extended operational lifetimes for the perdeuterated emitters. More broadly, this work establishes isotope engineering as a powerful strategy to simultaneously enhance efficiency and stability, providing a compelling proof-of-concept that it effectively modulates exciton-vibration coupling and mitigates operational degradation in organic radical NIR OLEDs.

Methods

General procedure for syntheses

3,5-dichloroaniline 1 was selected as the starting material. Hydrogen–deuterium exchange was carried out by stirring the substrate in deuterium chloride solution at high temperature in D₂O, affording the perdeuterated aniline 2 with excellent deuterium incorporation ($>$98% D). Subsequent Sandmeyer reaction of the perdeuterated aniline provided the key starting material 3 for the synthesis of HTTM-d₆ (Supplementary Fig. 1)^38,39. Another radical core, HPyBTM, was synthesized according to previous literature⁴⁰. The precursor 8 for HPyBTM-d₆ was obtained by deuteration of pyridine N-oxide 5 in D₂O and K₂CO₃⁴¹. Followed by deoxygenation with phosphorus trichloride and formylation using lithium diisopropylamide, affording compound 8 with a high level of deuterium incorporation ($>$ 98% D). Then, compounds 3 and 8 were reacted in the deuterated trifluoromethanesulfonic acid (TfOD) at 180  ^oC to afford HPyBTM-d₆ (Supplementary Fig. 2). It is worth mentioning that the purity of TfOD is essential; otherwise, low product yields (trace) and partially back-exchange of deuterium with protium was observed⁴². Finally, 4-(diphenylamino)phenylboronic acid underwent Suzuki coupling with HTTM or HPyBTM to afford HTTM-TPA and HPyBTM-TPA, respectively. Replacing the boronic acid with (9-phenyl-9H-carbazol-3-yl)boronic acid in the Suzuki coupling reaction afforded HTTM-3PCz. Perdeuterated (9-phenyl-9H-carbazol-3-yl)boronic acid 11 used for the synthesis of perD-HTTM-3PCz was prepared from commercially available 9H-carbazole−1,2,3,4,5,6,7,8-d₈. (Supplementary Fig. 4). Radicals with different degrees of deuteration were also generated through the combination of the corresponding protiated or deuterated moieties. Subsequent oxidation using KO^tBu and chloranil furnished the target radical compounds. Y6 and its deuterated derivatives, Y6-d₂ and Y6-d₆, were synthesized via a two-step procedure. Precursor pre-Y6 was first reacted with DMF-d₇ to afford pre-Y6-CDO, which subsequently proceeded through Knoevenagel condensation with synthesized 2FIC-d₂ (Supplementary Fig. 5 and 7) to yield the target compounds. Deuterium incorporation was determined by NMR spectroscopy. For compounds containing proton signals unaffected by deuteration (e.g., NH₂), the deuterium incorporation was calculated from the corresponding ¹H NMR integrals. Otherwise, 1,2-dibromoethane was used as an internal standard, and the deuterium incorporation was determined by comparing the ¹H NMR integrals of the parent and deuterated compounds (Supplementary Figs. 3 and 6). The deuterium incorporation of the NMR-silent radicals was determined by high-resolution mass spectrometry (HRMS).

Device Fabrication of NIR OLED

Prior to device fabrication, all chemicals underwent purification via vacuum sublimation to ensure high material quality. The substrates consisted of ITO-coated glass with a sheet resistance of 15 Ω sq⁻¹. These were prepared by ultrasonic cleaning in a sequential manner: first with acetone, followed by methanol and deionized water. The process concluded with treatment under N₂ plasma to achieve a pristine surface. OLED devices were then assembled using vacuum deposition of the materials onto the cleaned ITO substrates, maintained at a pressure of 10⁻⁶ torr. During this step, the deposition rate for each organic layer was controlled between approximately 1 and 2 Å·s⁻¹. Device characterization involved simultaneous measurement of current density–voltage–luminance (J–V–L) properties within a glove-box environment. A programmable source measurement unit (Keithley 2614B) supplied the driving voltage, while luminance was quantified using a calibrated silicon detector. Additionally, EL spectra were acquired through an optical fiber linked to a spectrometer (Instrument Systems CAS 125).

Author contributions

S.-F.W. designed and carried out all the synthesis of NIR radicals and Y6 series molecules reported in this work. Y.-C.K. and W.-Y.H. performed the OLED fabrication and analyzed the device data. C.-C.W. conducted the theoretical computation. P.-Y.H. carried out the optical measurements. W.-T.C. and Z.-H.S. performed the GIWAXS experiments and conducted the data analysis. P.-T.C. developed the theoretical approach and drafted the manuscript. All authors discussed the results and contributed to the preparation of the manuscript.

Peer review

Peer review information

Nature Communications thanks Dongdong Zhang, Lian Duan and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

P.-T.C. acknowledges the support from the National Science and Technology Council (NSTC 114−2639-M-002-008-ASP) and the National Taiwan University for their generous support. W.-Y. H. thanks the support from the National Science and Technology Council (114-2112-M-019-001). We appreciate the financial support of the National Science and Technology Council (NSTC114-2124-M-002-003).

Data availability

All data supporting the findings of this study are available within the manuscript and Supplementary Information. Additional data related to this study are available from the corresponding author upon request. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-76493-5.