a Representative examples of Cu(I)-OLEDs and their performance during 25 years’ development; Φ_em, k_r, λ_EL, EQE, L_max, LT_x refer to the emission quantum yield, radiative decay rate constant, electroluminescence peak maxima, external quantum efficiencies, maximum luminance, and operational lifetime at x% of initial luminance, respectively. b Schematic mechanism of TSF-OLEDs; S₀, S₁ and T₁ refer to the ground state, singlet excited state, and triplet excited state, respectively, of the Cu(I) sensitizer (denoted as [Cu]) or the fluorescence emitter (denoted as [emitter]). c Examples of TSF/TST-OLEDs using CMA(Cu) sensitizers and organic fluorescence or TADF emitters.

CMA(Cu) complexes belong to donor-acceptor type metal-TADF materials, which usually have a broad emission band, resulting in poor color purity. This is due to their LL’CT properties leading to structural shifts between the ground and singlet excited states³². This problem can be ameliorated by adopting a metal-TADF-sensitized fluorescent/TADF (TSF/TST) strategy, in which a metal-TADF emitter is used as a sensitizer to harvest singlet and triplet excitons and transfer these excitons (through the Förster energy transfer (FRET) process) to organic fluorescent or TADF compounds that have narrower bandwidth emission (Fig. 1b)^{22,23,33,34,35,36}. Compared with organic-TADF emitters, CMA(Cu) emitter is a good TSF sensitizer because of its significantly higher rISC rate constant (k_rISC; ~10⁷–10⁸ vs. ~10⁵–10⁶s⁻¹)³². The improved k_rISC helps to suppress the competing Dexter energy transfer (DET) process, thereby reducing the non-radiative decay energy loss and triplet exciton lifetime in TST-OLEDs (Fig. 1b). It was reported that TSF/TST-OLEDs containing (MAC*)Cu(Cz) sensitizer and TBRb and BN3 emitters showed yellow electroluminescence with full-width-half-maximum (FWHM) as narrow as 30 nm and 43–46 nm, respectively (Fig. 1c)³⁶. The TST-OLED using (PyIPr)Cu(Cz) sensitizer and ν-DABNA emitter showed blue electroluminescence with a narrower FWHM of 19 nm (Fig. 1c)²². However, these reported Cu-sensitized TSF/TST-OLEDs are only at the proof-of-concept stage due to the short device operational lifetime (LT₉₀ or LT₅₀ of ~10 h at 1000 cd m⁻²; Fig. 1c)^22,36. Currently, mainstream sensitizers are still based on luminescent precious metal complexes (i.e., Ir and Pt), and the development of Cu-TADF sensitizers for practical TSF/TST OLED applications is hampered by their stability in OLED device operation.

In the present work, we utilized modified carbazole donor ligands 1,3-dihydro-3,3-dimethylindeno[2,1-b]carbazole (DMIC) for 3-H, perdeuterated carbazole for 1-D, 2-D, 3-D, 6^tBu-D and 7-D, 1,8-dimethylcarbazole for 4, or a quinoxaline-fused NHC acceptor for 5 and 6 to prepare CMA(Cu) complexes. Complexes 1–7 show blue to deep-red TADF emission (470–677 nm) in thin films with medium-to-high Φ_em of 0.39–0.90, short τ_em of below 1.2 µs, and large k_r of 1.2–2.7 × 10⁶s⁻¹. OLEDs based on these complexes as emitters exhibit EL λ_max of 488–722 nm and EQE_max as high as 24.0%, with efficiency roll-off as small as 0.6% at 1000 cd m⁻² and brightness up to 265,000 cd m⁻². Notably, the operating lifetime (LT₉₅) of these Cu-OLEDs can reach 3582 hours at an initial luminance (L₀) of 1000 cd m⁻². This LT₉₅ value is one of the few best values reported in the public domain for all organic TADF and metal-TADF emitters. Furthermore, efficient and robust TSF-OLEDs using Cu(I) sensitizer and α-NAICZ as fluorescent emitter achieved narrowband EL λ_max 612–614 nm (FWHM of 33–38 nm), high EQE (reaching 21.9%), brightness up to 162,000 cd m⁻², and LT₉₅ up to 3689 h at 1000 cd m⁻².

a Chemical structures of CMA(Cu) emitters in this work. b Emission spectra of emitters 1–7 doped in 1,3-Bis(N-carbazolyl)benzene (mCP) thin film (in a concentration of 2 wt%) by drop-cast method. c Emission spectra of 1-D, 2-D, 3-D, 7-D (solid line) and 1-H, 2-H, 3-H, 7-H (dashed line) measured in degassed methylcyclohexane (MeCy), toluene (Tol), or 2-methyltetrahydrofuran (MeTHF). Emission spectra of 1-H, 2-H and 7-H have been reported previously²². d Calculated hole and electron distribution, the overlap between hole and electron wavefunctions (O_h,e) and the distance between centroids of hole and electron (Δr) of coplanar structures of 2-H and 3-H in the S₁ excited state; blue and green isosurfaces represent hole and electron distributions, respectively. e Reorganization energy distribution of S₁ → S₀ vibrational relaxation for 1-H and 1-D, red lines represent major vibrational modes localized on the carbazole ligand, blue lines represent other major vibrational modes, and black lines represent minor vibrational modes; f Major vibrational modes 1-5 localized on the carbazole ligand, the corresponding vibrational frequencies (ω_j) for 1-H and 1-D are labeled.

Photophysical properties

The photophysical properties of 1–7 were studied (Fig. 2b, c; and Supplementary Fig. 8–12; Supplementary Table 5). Preliminarily, 2 wt% Cu(I) emitters were doped in 1,3-bis(N-carbazolyl)benzene (mCP) thin films via drop-cast (DC) method. These thin films showed blue to deep red emission with peak maximum (λ_max) ranging from 470 to 677 nm, emission decay lifetime (τ_em) below 0.46 µs, and medium-to-high emission quantum yields (Φ_em) of 0.39–0.75 (Fig. 2b; Supplementary Table 5). These Cu(I)-doped mCP thin films exhibit large radiative decay rate constants (k_r) in the range of 1.2–1.9 × 10⁶s⁻¹, which are comparable to previously reported CMA(Cu) emitters with π-extended NHC ligands and are among the highest Cu(I) TADF-emitters reported to date^10,11,22.

The deuterium isotope effect on the photophysical properties of these Cu(I) emitters is then explored. In degassed toluene, the emission λ_max of deuterated emitters 1-D, 2-D, 3-D, and 7-D are consistent with those of corresponding non-deuterated emitters (Fig. 2c)²². Notably, in toluene, red-emitting 1-D, 2-D, and 3-D have higher Φ_em (increased by 29–100%) and smaller non-radiative decay rate constants (k_nr; 1.3–3.5 folds of decrease) compared to their non-deuterated analogs, whereas their k_r values are comparable (Supplementary Table 5)²². In contrast, 7-D and 7-H exhibited green-emission with similar Φ_em, k_r and k_nr values in toluene (Supplementary Table 5)²². These observations suggest that deuterium substitution in the carbazole ligand does not affect the singlet ligand-to-ligand charge transfer (¹LLCT) energy but leads to a decrease in the k_nr of the red-emitting compounds. Variable temperature lifetime measurements reveal that the singlet-triplet energy gap (ΔE_ST) of 1-D is 51 meV and the singlet state decay time (τ_s1) is 17 ns (Supplementary Fig. 10a). According to the equation k_S1 = Φ_em/τ_S1, the S₁→S₀ radiative rate constant (k_S1) for 1-D is 3.5 × 10⁷s⁻¹. These TADF parameters for 1-D are very close to those of 1-H (ΔE_ST of 50 meV, k_S1 of 3.5 × 10⁷s⁻¹)²². The fs-TA spectra of 1-D and 2-D recorded in toluene show the same spectral changes as the non-deuterated analogs with comparable inter-system crossing (ISC) rate constants (k_ISC; 4.5–5.0 × 10⁹s⁻¹ for 1, 2.2–2.3 × 10⁹s⁻¹ for 2; Supplementary Fig. 12). These results indicate that the ISC process of these CMA(Cu) emitters is not affected by the deuterium isotope effect, which is consistent with the observations of other reported deuterium-labeled TADF emitters^27,28,29,30.

Theoretical calculations

To gain deeper insights, density functional theory (DFT) and time-dependent density functional theory (TD-DFT) calculations were performed. The optimized 3-H and 4’ ground state (S₀) geometries are in good agreement with the experimentally obtained crystal structures (Supplementary Table 6). Geometry optimization of the 1-H, 2-H, and 3-H complexes in the S₁ and T₁ excited states resulted in two structures with coplanar or orthogonal orientations between the carbazole and carbene ligands (Supplementary Fig. 13). This shows that 1-H, 2-H and 3-H can undergo flexible dihedral angle rotations in the excited states. However, geometry optimization of the 4’ excited state resulted in only confined structures with torsional angles (θ_N1-C1-N2-C2) of 30–40° due to inter-ligand repulsion between bulky substituents on the ligands (θ_N1-C1-N2-C2 = 41° for S₁ state and 31° for T₁ state).

The frontier molecular orbital diagrams of 3-H and 4’ are shown in Supplementary Fig. 14. Their HOMO and LUMO are localized on the carbazole and carbene ligands, respectively. The excitation of electrons from HOMO to LUMO results in the S₀ → S₁ ¹LLCT transition. To further elucidate the excited state characteristics, we calculate the hole (h⁺) and electron (e⁻) distribution diagrams of 2-H and 3-H in the S₁ excited state (Fig. 2d and Supplementary Fig. 15)^37,38. Compared with 2-H with a tert-butyl substituent on the carbazole ligand, the π-extended fluorene substituent on the carbazole ligand leads to a longer distance and less overlap between the hole and electron of 3-H (Fig. 2d). The calculated ΔE_ST value of 3-H is about 0.03 eV smaller than that of 2-H (in coplanar geometry, ΔE_ST = 0.10 eV for 3-H and 0.13 eV for 2-H). The narrower ΔE_ST would promote the fast rISC process of 3-H, leading to a higher k_r value and more efficient TADF of 3-H compared with 2-H (in toluene, k_r = 2.3 × 10⁶ s⁻¹ for 3-H and 1.3 ×  10⁶ s⁻¹ for 2-H; Supplementary Table 5)²².

To investigate the effect of ligand deuteration, we performed vibrational mode analysis of 1-H and 1-D. The total reorganization energy is projected into the vibrational relaxation from the S₁ state to the S₀ state. For 1-H and 1-D, there are 15 major vibrational modes that contribute significantly to the total reorganization energy of the S₁ → S₀ vibrational relaxation (Fig. 2e, blue and red lines), five of which mainly contain carbazole ligand vibrations (Fig. 2e, red lines; Fig. 2f, modes 1–5). Deuterium substitution on the carbazole ligand (1-H → 1-D) leads to a significant decrease in the vibrational frequency (ω_j) of the C-H bond of the carbazole ligand due to the increase in atomic mass (Fig. 2f), while the other major vibrational modes remain almost unchanged (Supplementary Fig. 16, modes 6-7). This is supported by the Raman measurements of 1-H and 1-D. While most of the Raman shifts of both emitters are comparable, shifts at ~745, 1010, 1285 and 1446 cm⁻¹ for 1-H were not observed in the Raman spectrum of 1-D, but shifts in the lower energy regions at ~650, 853, 1201 and ~1329–1340 cm⁻¹ were observed (Supplementary Fig. 17). Since the total reorganization energies (λ) of 1-H and 1-D are similar (λ = 317 meV for both emitters), the smaller the vibrational frequency of 1-D, the higher the effective Huang-Rhys factor (S_eff) of 1-D relative to 1-H (S = λ/?ω, where ? is the reduced Planck’s constant; S_eff = 0.00783 for 1-D and 0.00758 for 1-H)^27,39. According to the energy gap law (Eq. 1)^40,41,42, lower vibrational frequencies lead to a reduction of non-radiative decay. This explains why the k_nr values of 1-D and 2-D are lower than those of 1-H and 2-H in degassed toluene (Supplementary Table 5).

where ΔE is the adiabatic energy difference between the S₀ and S₁ states. ω_M and λ_M are the vibrational frequency and reorganization energy of the vibrational modes that induce non-radiative decay. l is the number of vibrational modes. C is the effective coupling constant.

TADF-OLED characteristics and operational stability

The electroluminescent (EL) properties of CMA(Cu) emitters were studied by fabricating and characterizing OLEDs based on emitters 1–7. Detailed information on the device structure, auxiliary organic materials, fabrication, and characterization process can be found in the Supplementary Fig. 18, Methods and Supplementary Methods. The EL spectra and EQE-luminance characteristics of OLEDs based on 1–7 are shown in Fig. 3a–d and Supplementary Fig. 19–28. Key device performance data are summarized in Table 1 and Supplementary Table 7.

Fig. 3: OLED characteristics and operational lifetime measurement.

a Normalized EL spectra and b EQE-luminance characteristics of OLEDs based on CMA(Cu) emitter 1–6 in RH host and 7-D in SiCzCz:SiTrzCz₂ co-host. c Normalized electroluminescence (EL) spectra and d EQE-luminance characteristics of OLEDs based on emitter 6 in RH host; inset of c displays the electroluminescence image of OLED containing 2 wt% 6. e Operational lifetime measurement of OLEDs with emitters 1–6 in RH host and 7-D in SiCzCz:SiTrzCz₂ co-host at L₀ of: 11500 cd m⁻² for 1-D (4 wt%), 20000 cd m⁻² for 2-D (6 wt%), 8500 cd m⁻² for 3-H (8 wt%), 20000 cd m⁻² for 3-D (2 wt%), 10000 cd m⁻² for 4 (4 wt%), 7000 cd m⁻² for 5 (2 wt%), 900 cd m⁻² for 6 (2 wt%), and 10000 cd m⁻² for 7-D (8 wt%).

As shown in Fig. 3a and Table 1, the EL λ_max of selected devices containing emitters 1-D, 2-D, 3-H, 3-D, 4, 5, and 7-D are located at 563, 596, 591, 610, 586, 636, and 493 nm, respectively. The maximum EQEs (EQE_max) of devices containing emitters 1-D, 2-D, 3-H, 3-D, 4, and 5 doped in the RH host are 16.4%, 21.2%, 17.5%, 23.1%, 16.5% and 11.4%, respectively (Fig. 3b). The device structures were further optimized by replacing the RH host in the 2-D and 3-D devices with NBP-BC: PCPF-Trz co-hosts and replacing the RH host in devices 4 and 5 with TCTA:TPBi co-hosts, which increased the EQE_max of these four devices to 22.7%, 24.0%, 24.0% and 17.5%, respectively (Table 1). The EQE_max of the device containing the blue emitter 7-D doped in SiCzCz:SiTrzCz₂ co-host is 23.5% (Fig. 3b; and Table 1).

For those devices containing red emitters (e.g., 2-D, 3-H, 5 and 6), the EQE_max of the examined devices decreases with increasing emitter doping concentration. For example, when the concentration of emitter 6 increases from 2 wt% to 6 wt% and 10 wt%, EL λ_max shifts from 690 (deep red region) to 720 nm and 722 nm (NIR region), while EQE_max decreases from 9.59% to 7.17% and 6.62% (Fig. 3c, d and Table 1). The red-shift in emission energy and decrease in efficiency may be due to the intermolecular interactions between these Cu(I) red emitters in the OLED emissive layer (EML), which can quench excitons at high doping concentrations. Nevertheless, among NIR OLEDs based on earth-abundant Cu(I) complexes, the NIR device containing 6 wt% 6 achieved the highest EQE value and was competitive with Au-based NIR OLEDs^23,41,43,44.

Notably, OLEDs based on emitters 1–7 exhibited good operational stability under our laboratory conditions. At initial luminance (L₀) of 11500, 20000, 8500, 20000, 10000, 7000, 900, and 10000 cd m^-2, the operational lifetimes at 95% of L₀ (LT₉₅) are 17.1, 16.6, 32.1, 16.8, 26.2, 5.98, 60.9 and 0.42 hours for the devices with emitters 1-D, 2-D, 3-H, 3-D, 4, 5, 6, and 7-D, respectively (Fig. 3e; and Table 2). Using the exponential decay model with the acceleration factor (n), the LT₉₅ lifetimes at 1000 cd m^-2 were estimated to be 1354 hours (1-D), 2398 hours (2-D), 1682 hours (3-H), 3582 hours (3-D), 1407 hours (4), 157 hours (5), and 13.6 hours (7-D) (Table 2; and Supplementary Fig. 29; Supplementary Methods).

Notably, the OLEDs containing 2-D emitters doped in RH hosts exhibited higher EQE_max values and had an operational lifetime more than four times longer than that of 2-H (EQE_max of 14.4% and 21.2% for 2-H and 2-D, respectively; LT₉₅ values of 532 h and 2398 h for 2-H and 2-D, respectively; Tables 1 and 2)²². The same trend was observed for 3-H and 3-D (Tables 1 and 2). These results show that deuterium-substituted carbazole ligands have a significant impact on improving the efficiency and stability of CMA(Cu)-based OLEDs. Additional experiments were also performed to further evaluate the differences in the photophysical properties and stability of 2-H and 2-D. On the one hand, in order to simulate the EML of OLED and evaluate the photophysical properties of these emitters under OLED conditions, films doped with 4 wt% and 8 wt% 2-H and 2-D in the RH host were prepared by vacuum-deposition (VD) method. It was found that 2-D in RH exhibited higher Φ_em and lower k_nr than 2-H (Supplementary Table 5), which can be attributed to the use of deuterated carbazole ligands to reduce the vibrational frequency (Fig. 2e, f). This is why 2-D-based OLEDs have higher device efficiency. On the other hand, the T_d of the deuterated emitter (1-D: 337 ^oC; Supplementary Table 3) is slightly higher than that of the non-deuterated emitter (1-H: 325 ^oC)²². The calculated bond dissociation energy (BDE) of the C-D bond in 1-D is, on average, 8.5 kJ/mol higher than that of the corresponding C-H bond of 1-H (Supplementary Fig. 30), indicating that the C-D bond is stronger than the corresponding C-H bond. In addition, we further performed photo-/electro-chemical stability tests to evaluate the intrinsic stability of these deuterated and non-deuterated emitters (Supplementary Fig. 31–32)^27,45,46. As shown in Supplementary Fig. 31a, the absorbance of the 2-H LLCT band decreased by 31% after continuous irradiation with a yellow LED for 36 min. In comparison, 2-D degraded more slowly, with a 28% decrease in absorbance after 36 min (Supplementary Fig. 31b). As shown in Supplementary Fig. 32a–b, after 60 voltametric cycles of repetitive scans, the E_pa of 2-H and 2-D decreased by 35.5% and 28.9%, respectively. This indicates that the 1e-oxidized species of 2-D has higher electrochemical stability than that of 2-H. In contrast, the E_pc of 2-D and 2-H both decreased by 11% after 60 voltametric cycles (Supplementary Fig. 32c–d). These stability tests show that 2-D is more stable than 2-H, which may explain the significantly extended operational lifetime of 2-D-based OLEDs.

In addition, OLEDs based on emitters 3-H and 4 also exhibited good device stability, with operational lifetimes approximately three times longer than that of OLEDs based on 2-H (Table 2). Compared with 2-H, the 3-H doped in RH VD film mimicking the OLED EML exhibits a shorter τ_em and a larger k_r (Supplementary Table 5), which is beneficial to improve the stability of OLEDs. This can be attributed to the fact that complex 3-H adopts the DMIC ligand with π-extended fluorene substituent, which has a larger electron-hole distance (Fig. 2d). The 3-D emitter with deuterated π-extended carbazole ligands enables OLEDs with a LT₉₅ lifetime of 3582 h at 1000 cd m^-2, which is the longest lifetime among the Cu(I) TADF OLEDs studied in this work and comparable to high-performance OLEDs using noble metal emitters (Supplementary Table 8). In contrast, the high device stability of the 4-based OLEDs may be attributed to the steric effect provided by the bulky 1,8-methyl substituted carbazole ligands (Supplementary Discussion and Supplementary Fig. 33)²⁶.

Förster energy-transfer (FRET) and TSF-OLEDs

Although TADF OLEDs employing Cu(I) emitters 1–7 exhibit high efficiency and good operational stability, their EL spectra have large full-width-half-maximum (FWHM) values (> 100 nm for the yellow and red OLEDs; Table 1), showing unsatisfactory color purity. To address this issue, we constructed a ternary system, selecting 1-D, 2-D, or 3-H as the sensitizer, the previously reported α-NAICZ⁴⁷ as the fluorescent emitter, and RH as the host (Fig. 4a).

Fig. 4: TSF-OLED characteristics and operational lifetime measurement.

a Chemical structures of CMA(Cu) sensitizers and α-NAICZ emitter. b Absorption spectrum of α-NAICZ in DCM and normalized photoluminescence (PL) spectra of binary system of CMA(Cu) sensitizers in RH and α-NAICZ emitter in RH, and ternary system of CMA(Cu) sensitizers and α-NAICZ emitter in RH; inset displays the PL decay of 2-D/α-NAICZ/RH ternary system, 2-D/RH or α-NAICZ/RH binary systems. c Normalized electroluminescence (EL) spectra, d EQE-luminance characteristics, e luminance-voltage (left) and current density-voltage (right) characteristics of TSF-OLED with CMA(Cu) complex: α-NAICZ and OLEDs with CMA(Cu) complexes or α-NAICZ only; insets of c display the electroluminescence images of OLEDs containing 2 wt% 3-H only (upper) and TSF-C (down). f Operational lifetime measurement of TSF-OLEDs; L₀ refer to the initial luminance. Composition of TSF-OLEDs: TSF-A contains 1-D (6 wt%):α-NAICZ (0.2 wt%); TSF-B contains 2-D (8 wt%):α-NAICZ (0.4 wt%); TSF-C contains 3-H (8 wt%):α-NAICZ (0.3 wt%); TSF-D contains 3-D (6 wt%): α-NAICZ (0.3 wt%).

The Förster energy transfer (FRET) process of the Cu/α-NAICZ/RH ternary system was investigated by studying the emission properties of the corresponding VD thin films. The absorption spectrum of α-NAICZ (in DCM)⁴⁷ overlaps well with PL spectra of doped 1-D, 2-D, and 3-H in RH VD films, resulting in large spectral overlap integrals J(λ) of 6.8–11.6 × 10¹⁴M^-1 cm⁻¹ nm⁴. The Förster radius distance of this ternary system was estimated to be 3.92–4.41 nm, and the theoretical FRET rate constant (k_FRET) was 3.1–6.3 × 10⁸s⁻¹ (see Supplementary Discussion for details). The FRET efficiency (E_FRET) reached 0.99, indicating that the ternary system can achieve a highly efficient FRET process. It is noteworthy that all Cu/α-NAICZ/RH ternary films exhibit narrowband emission with λ_max at 612–616 nm, which is the same as the emission of α-NAICZ/RH binary film and thus has higher color purity compared with the broadband emission of Cu/RH binary films (Fig. 4b). The emission quantum yields of these sensitized thin films were 0.54–0.79 (Supplementary Table 5). In addition, the delayed fluorescence lifetimes of these Cu/α-NAICZ/RH ternary films are shorter than those of the corresponding Cu/RH binary films (Fig. 4b, inset; Supplementary Table 5). This helps reduce harmful side reactions such as triplet-triplet annihilation and improves the operational stability of the device.

Based on these energy transfer studies, TSF-OLEDs were fabricated by doping 0.2–0.4 wt% α-NAICZ into the EML of Cu-based devices. To optimize the device performance, we introduced different concentrations of Cu complexes into these TSF-OLEDs. Figure 4c–e and Supplementary Fig. 34 show the EL spectra and OLED characteristics of these TSF-OLEDs, as well as fluorescence OLEDs containing only α-NAICZ. The addition of α-NAICZ effectively narrowed the FWHM of the EL spectrum from 103–122 nm to 33–38 nm, showing that the energy transfer from the Cu(I) sensitizer to the α-NAICZ emitter was quite efficient (Table 1 and Supplementary Table 7).

Taking TSF-OLEDs using sensitizer 2-D as examples, the FWHM of the EL spectrum of TSF-E containing 4 wt% 2-D was 38 nm (Supplementary Fig. 34 and Supplementary Table 7). When the concentration of 2-D increased to 6-8 wt% (TSF-B and TSF-F), the FWHM of the EL decreased to 37 nm, showing a more efficient energy transfer from 2-D to α-NAICZ. Thus, TSF-B and TSF-F achieved higher EQE_max values (20.9% and 20.4%, respectively) compared to TSF-E with an EQE_max of 17.1% (Table 1 and Supplementary Table 7). It is noteworthy that the efficiency of TSF-OLEDs increases with the increase of 2-D concentration (EQE_max of 17.1% for TSF-E containing 4 wt% 2-D, and 20.9% for TSF-B containing 8 wt% 2-D). In contrast, the efficiency of the 2-D-based OLED decreases as the 2-D concentration increases from 4 to 8 wt% (Table 1 and Supplementary Table 7). This is because the efficient energy transfer from Cu(I) complex to α-NAICZ reduces the population of excited states of Cu(I) complex, thereby alleviating the exciton quenching of Cu(I) complex at high concentration.

At L₀ of 1000 cd m^-2, the estimated LT₉₅ values of TSF-A, TSF-B, TSF-C and TSF-D were 2190, 2805, 1508, and 3689 hours, respectively (Fig. 4f, Table 2, and Supplementary Fig. 35). These values are comparable to the corresponding LT₉₅ values of 1354, 2398, 1682, and 3582 hours for OLEDs using only 1-D, 2-D, 3-H, and 3-D. To the best of our knowledge, Cu(I) emitter-sensitized OLEDs have not previously achieved simultaneously high efficiency and such a long device lifetime. This work highlights the great promise of earth-abundant metal-based OLEDs as efficient and stable alternatives to precious Ir(III)-/Pt(II)-based OLEDs.