Electroluminescent photoresists extending lithographic scaling to OLEDs

The monolithic integration of high-performance light sources with silicon-based complementary metal–oxide–semiconductor (CMOS) electronics is a cornerstone for next-generation optoelectronics¹, promising to revolutionize augmented-reality microdisplays², optical computing³ and high-speed on-chip interconnects^4,5. Although inorganic III–V semiconductor devices have dominated research in this field, their integration remains fundamentally constrained by lattice mismatch, the requirement for high-temperature epitaxy^6,7 and the yield-limiting complexity of heterogeneous transfer processes⁸. Furthermore, as pixel size scales into the sub-10-μm regime, inorganic emitters suffer from severe efficiency roll-off owing to non-radiative recombination at sidewall defects^17,18,19,20. In this regard, organic semiconductors emerge as a compelling alternative^21,22. For example, in organic light-emitting diodes (OLEDs), their amorphous emissive layers, coupled with low-temperature and lattice-free fabrication, enable direct back-end-of-line compatible processing on CMOS wafers while preserving high external quantum efficiencies (EQEs) independent of device dimensions²³.

Despite these material virtues, a viable technical roadmap towards the monolithic integration of multicolour OLEDs remains elusive²⁴. The primary bottleneck lies in the chemical fragility of organic semiconductors, which are readily degraded by the solvents and developers inherent to standard photolithography^21,25. Consequently, current manufacturing methods, such as fine-metal-mask evaporation and inkjet printing, are fundamentally limited by their restricted spatial resolution and overlay accuracy^9,10,26. More critically, these techniques are inherently decoupled from subsequent photolithographic processes, preventing OLED technology from aligning with the rigorous integration densities and scaling laws dictated by modern silicon circuitry.

The development of directly photolithographically patternable organic emissive layers (EMLs)—a material system we formalize here as electroluminescent photoresists (ELPRs)—was first envisioned in the early 2000s to enable the multicolour integration of solution-processed polymer OLEDs^11,12. Initial attempts utilized π-conjugated polymers with oxetane side groups synthesized via random polycondensation, forming insoluble networks through acid-catalysed crosslinking. Nevertheless, these early ELPRs were rather underperforming, with EQEs typically below 2% (refs. ^11,12). This performance gap resulted from: (1) the intrinsic limitations of early fluorescent polymers, which harvest only singlet excitons; and (2) the stochastic distribution of emissive units and residual photoacid additives, which created potent quenching sites. In the decades following these early reports, although synthetic efforts successfully incorporated the advanced OLED emitters^13,21, such as thermally activated delayed fluorescence (TADF) molecules^14,15, into polymer chains^{16,27,28,29,30}, the development of ELPRs has remained frozen. This persistent performance gap stems from the pervasive nature of crosslinking reactions; without precise structural control, these photochemical processes inevitably introduce non-radiative decay channels or disrupt charge-transport pathways, compromising the optoelectronic performance.

Notably, the early struggles of ELPRs coincided with the infancy of controlled radical polymerization^31,32, which now offers an advanced synthetic toolset for tailoring molecular architecture. We hypothesized that by encapsulating individual emitters within a three-dimensional host matrix, exciton quenching induced by peripheral crosslinking could be fundamentally suppressed. Here, leveraging atom transfer radical polymerization (ATRP)³³, we engineered a multi-arm star polymer architecture centred on a single TADF emitter. This core serves as the nucleus for star polymer arms consisting of two functionally distinct blocks: an internal host shell designed for exciton confinement and energy transfer, and a terminal crosslinkable periphery for structural definition (Fig. 1a). This site-specific, core–shell compartmentalization ensures that the radical-mediated crosslinking is restricted to the polymer boundary, effectively shielding the emissive cores from chemical and electronic perturbations (Fig. 1b).

Consequently, after ultraviolet patterning, our TADF-active ELPRs achieve a high EQE of 13.3%, representing a major leap for directly photolithographically patterned OLEDs. Furthermore, we demonstrate the robust processability and exceptional solvent orthogonality of these materials by fabricating monolithic, multicolour OLED micro-pixel arrays via three consecutive photolithographic cycles (Fig. 1c). By incorporating electron-beam (e-beam)-sensitive crosslinker moieties, we extend this capability to the nanometre regime, achieving bright emissive patterns with critical dimensions down to 110 nm—bridging the gap between microscale display technology and nanoscale integrated photonics.

We selected three representative TADF emitters—red (R), green (G) and blue (B)—as the central cores for our ELPRs. These scaffolds share a common donor moiety, 9,9-dimethyl-9,10-dihydroacridine (DMAC), coupled with colour-specific acceptor groups (naphthalimide³⁴, triazene (TRZ)³⁵ and diphenyl sulfone³⁶) to span the visible spectrum. The DMAC moiety was site-specifically functionalized at the 2,7-positions with 2-bromo-2-methylpropanoate initiating groups via a phenyl linker³⁷, ensuring that each polymer chain grows outwards from a single emitting core via ATRP (Fig. 2a).

Utilizing this central initiation site, we first synthesized self-hosted TADF polymers, the ‘P-series’ precursors, by polymerizing host monomers with varied aromatic structures as the first block. All the synthesized molecular monomers and polymers were thoroughly characterized (Supplementary Notes 1–10). Among the designed host monomers, an N-phenylcarbazole derivative bearing an acrylate group (ACz; Fig. 2a) was selected throughout this study owing to its superior photophysical and electroluminescene performance³⁸ (Supplementary Note 11 and Supplementary Table 1). To achieve optimal TADF activity, the emitter doping concentrations were precisely tuned between 5 wt% and 15 wt% by simply controlling the polymer chain length. A defining feature of this approach is the retention of dormant bromide chain-ends following this initial stage, enabling the subsequent re-initiation and growth of a secondary functional block. To ensure clarity, these intermediate polymers are designated with a prefix (R-, G- or B-) for the emission colour, followed by the substitution position (ortho-, meta- or para-), and the suffix ‘P’ to denote the pure self-hosted polymer before crosslinker attachment (Fig. 2a). All synthesized polymers exhibited well-controlled degrees of polymerization (DP) and narrow dispersities (Đ), owing to the efficient ATRP method (Fig. 2a, Supplementary Note 12 and Supplementary Table 2). For example, the G-ortho-P series showed a DP of 20–65 with a Đ of 1.16–1.25, enabling tunable doping concentration of 4.6–15.0 wt%.

Building on these precursors, a second-block synthesis was executed to incorporate a terminal crosslinkable periphery, transforming the P-series polymers into functional ELPRs. Methacrylates with cinnamoyl³⁹ (MAc) or allyl⁴⁰ (MAl) groups were used to confer ultraviolet and e-beam patternability, respectively (Fig. 2a). This modular approach allows for the independent tuning of emission characteristics and crosslinking density, and the multi-arm star architecture ensures the TADF core remains spatially isolated from reactive polymer boundaries.

Time-resolved photoluminescence (TRPL) spectroscopy was first used to evaluate the impact of molecular architecture on excitonic characteristics. The results of G-ortho-ELPR in toluene solution first confirm that the TADF activity is intrinsically preserved within the polymer framework, as evidenced by the prompt and prominent delayed fluorescence lifetimes observed under deoxygenated conditions (τ_PF = 27.5 ns and τ_DF = 1.14 µs) (Fig. 2c and Supplementary Table 3). The TADF characteristic of G-ortho-ELPR is confirmed by a high total photoluminescence quantum yield (Φ_tot) of 50%, which is dominated by the delayed fluorescence quantum yield (Φ_DF) of 39%. The favourable exciton dynamics are attributed to the fact that the non-radiative rate constant of the triplet state (k_nr^T = 6.18 × 10⁵ s⁻¹) is outpaced by the rapid reverse-intersystem-crossing rate (k_RISC = 3.64 × 10⁶ s⁻¹) and radiative rate constant (k_r^S = 3.93 × 10⁶ s⁻¹). The photophysical and kinetic characteristics of the self-hosted TADF polymers and ELPRs are summarized in Supplementary Table 8. In the solid film, G-ortho-ELPR exhibits a delayed fluorescence lifetime of τ_DF = 0.7 µs, comparable to that of the parent TADF molecule³⁸ and the polymer in solution. The delayed fluorescence also reveals strong temperature dependence, corroborating that the TADF mechanism is reserved in our ELPRs (Supplementary Note 13). The temperature-dependent TRPL responses of the thin-film sample of G-ortho-deBr reveal that τ_DF increases with decreasing temperature from 298 K to 220 K, exhibiting typical TADF characteristics⁴¹ (Supplementary Fig. 76 and Supplementary Table 12).

Although the emission profiles are primarily dictated by the central TADF core, we found that linker engineering at the initiation site also has an important role in determining emission spectra and quantum yields. For example, the substitution position of the DMAC donor considerably influences the photophysical performance: G-ortho-P (DP_ACz = 20) exhibited a photoluminescence quantum yield (Φ_PL) of 74% with an emission maximum at 505 nm, whereas G-para-P showed a reduced Φ_PL of 48% at 522 nm; both were measured in solid films under ambient conditions (Extended Data Fig. 1 and Supplementary Table 5). These findings reveal the enhanced steric hindrance imposed by the ortho-linkers^42,43, which effectively preserves the reverse-intersystem-crossing process within the TADF core and minimizes non-radiative decay, thereby maximizing the radiative decay of excited states in the self-hosted polymers. Building on these insights, linker engineering in combination with donor–acceptor design have enabled multicolour photoluminescence emission from the synthesized ELPRs, covering nearly the full visible spectrum (Fig. 2d).

We further investigated the evolution of optoelectronic properties upon incorporating photocrosslinkers by comparing photophysical properties of the self-hosted TADF polymers (P-series) with those of their functional ELPR counterparts. Upon the growth of the second block on the periphery, a reduction in Φ_PL was observed; for example, the Φ_PL value decreased from 81% for G-ortho-P (DP_ACz = 45) to 40% for G-ortho-ELPR (DP_ACz = 45, DP_MAc = 12); both were measured in solid films under inert conditions (Supplementary Note 14 and Supplementary Table 6). This initial drop is probably attributed to the increased separation between emissive units and the change in local dielectric environment introduced by the methacrylate periphery, and the potential charge-transfer processes induced by the electron-donating cinnamoyl groups³⁷. Notably, subsequent ultraviolet photocrosslinking did not further degrade the quantum efficiency, but instead slightly enhanced the Φ_PL to 44%. This enhancement suggests that the formation of a dense, crosslinked network rigidifies the polymer matrix, thereby suppressing non-radiative decay channels. Crucially, these findings confirm that the photochemical crosslinking process itself does not introduce additional quenching sites—a significant departure from traditional ELPR systems where radical-mediated reactions typically induce severe photoluminescence quenching⁴⁴.

To better understand the ability of the host monomer units to shell the emitter cores, classical molecular dynamics simulations were performed. Before film formation, single polymers solvated in toluene exhibited behaviour where the host monomer units would tend to aggregate, exposing the emitter cores to toluene, but largely shielding them from the crosslinker monomer units (Fig. 3a–c). As the toluene is evaporated to form films, the emitter cores largely remain shielded by the host monomer units, and the crosslinker monomer units of adjacent polymers are brought into contact where they can crosslink efficiently (Fig. 3d–f).

Centre-of-mass radial distribution functions were extracted to quantify the ability of the host shell monomers to shield the TADF emitters from the photocrosslinking moieties (Fig. 3g,h). For the green emitter, as revealed in Fig. 3g, each emitter was surrounded by 3.8 host monomers on average at a centre-of-mass distance with a maximum at 8.4 Å. Although the majority of photocrosslinking moieties were spatially distant from the emitter core, with a first peak maximum at 9.5 Å, a small number of emitter cores (12.5%) were in direct contact (centre-of-mass peak at 5.9 Å) with a single photocrosslinking moiety. For the photocrosslinking moieties themselves, each was in close contact (centre-of-mass peak maximum of 5.5 Å) with 0.8 other photocrosslinking moieties on average, rising to 2.7 on average within a radial distance of 10 Å. Similar results were found for the red and blue emitters. For example, for the red emitter, each emitter had 2 host monomers in close contact (centre-of-mass peak at 5.8 Å), and 4.2 within 10 Å. Approximately 25% had a photocrosslinking moiety in close contact (centre-of-mass peak at 7.2 Å). The photocrosslinking moieties were in close contact (centre-of-mass peak at 5.3 Å) with 0.6 other crosslinking moieties, rising to 1.8 at 10 Å.

In general, these results demonstrate that the crosslinker moieties are spatially separated from the emissive cores, and that the photocrosslinking moieties have a high probability of interacting with photocrosslinking moieties in adjacent polymer chains. Owing to the initial self-aggregation of the host monomers in toluene, there is a small probability that the emissive cores can remain solvent exposed and come into contact with photocrosslinking moieties; however, the results suggest that this occurs for only a minority of emissive cores.

We first examined the photopatternability of our synthesized ELPRs using ultraviolet photolithography. The ELPRs loaded with an optimized number of ultraviolet photocrosslinker units, cinnamoyl groups, were processed via negative-tone direct photolithography. The ELPR solutions in toluene were spin-coated onto silicon substrates to form solid films and thermally annealed to remove the solvent traces. These films were selectively exposed to 365-nm ultraviolet light through a photomask in vacuum contact mode. The ultraviolet-triggered [2 + 2] cycloaddition of neighbouring cinnamoyl groups (Fig. 4a) occurs exclusively within the exposed regions. Subsequent development with toluene effectively dissolved the unexposed areas, yielding high-fidelity negative-tone patterns with bright photoluminescence. We systematically examined the resulting patterns by fluorescence microscopy, scanning electron microscopy (SEM), and atomic force microscopy (AFM). For example, with G-ortho-ELPR (DP_ACz = 45, DP_MAc = 12), we successfully pushed the resolution limit to line/space (L/S) pitches of 4/2 μm with a precise film thickness of 36 nm (Fig. 4b–d). Beyond single-colour resolution, the multicolour patterns were nicely integrated with three sequential photolithographic processes, demonstrating the robust solvent orthogonality of our ELPR platform (Fig. 4e–h).

To further push the resolution beyond the diffraction limit of optical lithography, we synthesized e-beam-sensitive ELPRs incorporating allyl moieties (Fig. 4i). By leveraging the ultrashort de Broglie wavelength of accelerated electrons, we achieved unprecedented spatial control. Representative e-beam-patternable ELPRs, such as G-ortho-ELPR (DP_ACz = 35, DP_MAl = 30) and R-ortho-ELPR (DP_ACz = 23, DP_MAl = 20), were successfully patterned into nanodoughnuts with 300-nm features via direct electron-beam lithography (EBL; Fig. 4j–l). By further lowering the thickness of the spin-coated film of G-ortho-ELPR, we extended the resolution to L/S pitches of 110/160 nm (Fig. 4m,n).

The ELPR platform demonstrates exceptional robustness, enabling seamless integration with consecutive EBL processes and conventional lift-off processes of other functional materials. Crucially, these patterns remain intact even after prolonged exposure to aggressive chemical environments, for instance, warm acetone solution. This is exemplified in Extended Data Fig. 2, which showcases periodic nanodisk arrays composed of G-ortho-ELPR and aluminium, with precisely controlled diameters and pitches ranging from 600 nm down to 150 nm. Despite the high-energy electron exposure and subsequent solvent immersion, these nanopatterns retain intense photoluminescence with high contrast and fidelity. This validates that our single-emitter core–shell architecture effectively shields the TADF cores from both e-beam-induced degradation and chemical attack during post-processing. As a key demonstration of this nanoscale integration, we fabricated a multicoloured ‘macaw parrot’ micrograph via two consecutive EBL cycles (Fig. 4o–r). With individual pixel dimensions down to 200 nm, this represents, to our knowledge, the highest-resolution multicolour fluorescent image realized through direct lithographic patterning, enabled by the robust solvent orthogonality that facilitates high-density, multi-step integration without cross-contamination.

Before evaluating the electroluminescence of the ELPRs, we first investigated and optimized the device performance of their self-hosted TADF precursors (the P-series; Extended Data Fig. 3). These polymers initially possess dormant bromide end groups, a remnant of the ATRP process. We hypothesized that these reactive moieties may be activated under the high electric fields present during OLED operation, potentially acting as deleterious charge traps or quenching sites. To validate our hypothesis, we have deactivated the end groups by debrominating all the self-hosted TADF polymers, and matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectroscopy was used to confirm the successful reaction (Extended Data Fig. 4b and Supplementary Note 15). For instance, the green self-hosted polymer before (G-ortho-P-Br) and after (G-ortho-P-deBr) debromination showed distinct characteristic peaks; G-ortho-P-Br exhibited periodic mass peaks corresponding to dormant bromide end groups, whereas G-ortho-P-deBr did not. Both G-ortho-P-Br and G-ortho-P-deBr preserved the characteristic charge-transfer absorption peak in ultraviolet–visible spectra, indicating that the emitter core remained intact before and after debromination (Extended Data Fig. 4c). Furthermore, electrochemical analysis revealed that the highest occupied molecular orbital of G-ortho-P is consistent with the parent TADF emitter unit, DMAC-TRZ (Supplementary Note 16 and Table 7). G-ortho-P-deBr exhibits robust thermal stability, with a decomposition temperature (T_d, defined as a temperature account for 5% weight loss of material) of 350 °C. This high thermal robustness ensures reliable and uniform film formation during spin-coating and subsequent processing by preserving polymer integrity and mitigating thermally induced morphological instability⁴⁵ (Supplementary Note 17). Interestingly, after debromination, G-ortho-P-deBr exhibited a shorter τ_DF of 1.26 μs, compared with 1.60 μs for G-ortho-P-Br in thin-film TRPL measurements (Supplementary Note 13 and Table 3).

Finally, we compared the OLED performance of G-ortho-P-Br and G-ortho-P-deBr. Strikingly, deactivation led to an 18-fold and 17-fold enhancement in maximum EQEs and current efficiencies, surging from 1.6% to over 30.3% and from 4.25 cd A⁻¹ to 75.44 cd A⁻¹, respectively (Fig. 5a and Supplementary Fig. 18). Notably, the EQE of the solution-processed polymer OLED based on G-ortho-P-deBr is slightly higher than that for the control TADF OLEDs fabricated via vacuum evaporation of parent molecules (26.5%)³⁵, indicating the critical role of covalent molecular design of single-emitter TADF polymers. This strategy is also effective for self-hosted TADF precursors with red- and blue-emitting cores. The OLED performance of all self-hosted TADF precursors, including EQE and luminance, were markedly enhanced after the end-group deactivation (Extended Data Fig. 5). We believe that this approach could serve as a universal strategy for developing highly efficient electroluminescent polymers via controlled radical polymerization.

Leveraging this critical insight, we characterized the fully functionalized ELPRs in a standard OLED architecture, where the emissive layer was ultraviolet-crosslinked before the deposition of the electron-transport layer (Fig. 5b). Notably, G-ortho-ELPR (DP_ACz = 45; DP_MAc = 12) achieved a peak EQE of 13.3% (Fig. 5b,c). This represents a transformative advancement for directly photolithographically patternable organic emitters, which have been historically limited to EQEs below 2% (refs. ^21,22). Furthermore, through linker engineering at the TADF core, we demonstrated nearly multicolour tunability across the visible spectrum while maintaining robust electroluminescence performance (Fig. 5d).

To showcase the potential for high-density integration, we fabricated a miniaturized multicolour OLED array featuring 30-µm pixels. The device architecture utilized a common indium-tin-oxide anode with pixels defined by a photolithographically patterned silicon dioxide (SiO₂) passivation layer, which also served as an interlayer dielectric (Fig. 5e). Exploiting the precise photopatternability and solvent orthogonality of our ELPRs, we sequentially deposited blue, green and red pixels to form the letters ‘ETH’ through three consecutive lithographic cycles. Upon completion with common electron-transport and cathode layers, the multicolour array exhibited uniform and bright electroluminescence at 6 V (Fig. 5f,g). This demonstration of monolithic, micrometre-scale multicolour integration unlocks the scalability of our ELPR platform for next-generation microdisplays and on-chip photonic integration.

We further examined the operational lifetimes (LT₅₀; time required for the luminance to decay to 50% of its initial value) of our devices (Supplementary Note 19). Devices using poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS) showed pronounced degradation, probably owing to its acidic nature interacting with the EML during ultraviolet crosslinking^46,47. Conversely, using poly[bis(4-phenyl)(2,4,6-trimethylphenyl)amine] (PTAA) as the sole hole-transporting layer (HTL) significantly improved robustness. Although absolute lifetimes are currently limited by trace-metal impurities, we demonstrated that halving the residual copper content via intensive dialysis and optimizing the transport layer architectures extended the LT₅₀ of G-ortho-P-deBr devices to 172 min at a constant current density of 2.4 mA cm⁻² (Supplementary Fig. 63). A parallel stability enhancement (LT₅₀ of 146 min) was observed for the crosslinked G-ortho-ELPR system. We anticipate that device stability can be further improved by minimizing copper residues among the ELPRs through regenerated⁴⁸ or organocatalysed⁴⁹ ATRP synthesis or through multiple dialysis cycles.

Although these findings highlight that the extrinsic factors, including HTLs and metal impurities, significantly limit current device operational stability, the potential contributions of intrinsic degradation mechanisms, such as those related to the dual-functional polymer architecture and crosslinkable groups, remain to be independently evaluated. In addition, it should be noted that a distinction must be made between peak high-current operation and continuous stable operation. Although our measurements demonstrate that the devices can sustain transient current densities well over 100 mA cm⁻², continuous long-term operation is currently restricted to 10 mA cm⁻² or below to mitigate excessive Joule heating and prevent premature electrical breakdown.

End-group deactivation of TADF polymers

The bromo end groups of ATRP-synthesized TADF polymers were removed via hydrogenation. In a typical procedure, G-ortho-P (1.0 equiv.) and copper(I) bromide (CuBr; 2 equiv.) were charged into a Schlenk flask and evacuated for 20 min, followed by 3 vacuum–argon cycles. Next, 1,1,4,7,7-pentamethyldiethylenetriamine (PMDETA; 2.0 equiv.), degassed via argon sparging for 10 min, was injected into the flask, and the resulting mixture was stirred at room temperature for 30 min under an argon atmosphere. Tributyltin hydride (2.0 equiv.) was then injected into the Schlenk flask under argon. The flask was placed in an oil bath at 60 °C for 2 h with stirring. After a specific time, the Schlenk flask was cooled to room temperature to stop the reaction. Then the viscous mixture was dissolved in tetrahydrofuran, filtered through a column filled with activated basic Al₂O₃, precipitated once in hexane and twice in methanol, and then dried under vacuum overnight. Final solid powders were obtained as the product and ready for further use after characterization with NMR, size-exclusion chromatography and MALDI-TOF.

Block copolymerization affording ELPRs

ELPRs were synthesized via ATRP using the self-hosted precursor polymers as macroinitiators. The crosslinker loading was modulated by varying the stoichiometric feed ratio of the crosslinkable monomer (MAc or MAl) to the macroinitiator. In a representative synthesis of G-ortho-ELPR, G-ortho-P (1.0 equiv.) and CuBr (1.5 equiv.) were charged into a flame-dried Schlenk flask and subjected to 3 vacuum–argon cycles. Separately, a solution of MAc monomer and PMDETA (1.5 equiv.) in toluene was degassed via argon sparging for 20 min. The degassed monomer/ligand solution was then transferred to the Schlenk flask via syringe under argon. The polymerization was conducted at 60 °C and subsequently stopped by cooling the flask to room temperature. The resulting viscous mixture was diluted with tetrahydrofuran and filtered through a column filled with activated basic Al₂O₃ to remove the residue copper ions. The final block copolymer was isolated by precipitation once in hexane and twice in methanol, followed by drying under vacuum overnight to yield solid powders of G-ortho-ELPR.

Direct photolithography

In a typical direct photo-patterning process, a certain amount of ELPR polymers were dissolved in toluene and filtered through a 0.22-µm PTFE membrane, affording 10–25 mg ml⁻¹ precursor solution. Then this ELPR precursor can be spin-coated either on silicon wafer for photo-pattern demonstration or on top of the HTL layer for OLED device fabrication. The thickness of spin-coated ELPR films can be tuned by changing spin-coating speed and/or the precursor concentration of the ELPR. The mask contact aligner exposure was performed on an ABM ultraviolet mask aligner through a 4-inch quartz photomask. The 405–365 nm exposure mirror was installed on an ABM aligner that allows i-line exposure. The typical dose range used for ELPR patterning is 6,000–8,500 mJ cm⁻². The development was performed by rinsing the substrate with toluene to remove the un-crosslinked regions.

Direct e-beam lithography

The film deposition of ELPR polymers for direct EBL is the same process as that of direct ultraviolet photolithography. ELPR solution concentration can be varied between 10 mg ml⁻¹ to 25 mg ml⁻¹ if thickness tuning is needed. The e-beam exposure was performed on EBPG5200, Raith with a fixed 100 keV voltage from the Binnig and Rohrer Nanotechnology Center (BRNC) at the IBM research centre in Zurich. The optimal dose for each single-coloured ELPR is different, but typically ranges from 100 μC cm⁻² to 500 μC cm⁻². The development was performed by rinsing the substrate with toluene to remove the un-crosslinked regions. The aluminium nanodisks were patterned with a conventional poly(methyl methacrylate) and methacrylic acid (PMMA/MMA) bilayered e-beam lift-off process.

OLED fabrication

For pursuing the best OLED performance, all the synthesized polymers used as the EML underwent end-group deactivation before device integration. Indium-tin-oxide-coated glass substrates were cleaned and treated with oxygen plasma. For bulk devices, a PEDOT:PSS blend (Clevios AI 4083/CH 8000 mix) was spin-coated (4,000 rpm, 50 s) with adjustable relative ratio to a thickness range of 40–80 nm and annealed at 150 °C for 15 min on a laminar flow bench. Subsequently, the devices were transferred to a nitrogen-filled glovebox to deposit an optional HTL and polymer EML via spin-coating. For photo-patterned devices, a PTAA layer (5 mg ml⁻¹ in toluene) was spin-coated (4,000 rpm, 50 s) and annealed at 120 °C for 10 min. To understand the role of the hole-injection layer and the HTLs on the operational stability of the OLED devices, we used a hole-injection material, nickel oxide (NiO_x), and two other hole-transport materials, N4,N4′-di(naphthalen-1-yl)-N4,N4′-bis(4-vinylphenyl)biphenyl-4,4′-diamine (v-NPB) and poly(9-vinylcarbazole (PVK). Subsequently, a 20-nm TADF polymer EML was spin-coated (2,000 rpm, 50 s). In the case of ELPRs, the EML was crosslinked via 365-nm ultraviolet exposure. For the multicolour patterning device (ETH logo), each ELPR layer was ultraviolet-irradiated through a photomask selectively and developed in toluene; this process was repeated for additional colours to achieve pixelated EMLs. Finally, the substrates were transferred into a physical vapour deposition chamber, and a 50-nm 1,3,5-tri(m-pyridin-3-ylphenyl)benzene (TmPyPB) or 1,3,5-tris(1-phenyl-1H-benzimidazol-2-yl)benzene (TPBi) electron-transport layer, a 1.5-nm 8-hydroxyquinolinolatolithium interface layer, and a 100-nm aluminium cathode were sequentially deposited via thermal evaporation through a shadow mask at a vacuum level of 7.5 × 10⁻⁸ torr.

Device encapsulation

Following the thermal deposition of the aluminium electrode, the fabricated devices were immediately transferred without atmospheric exposure into an integrated, dry-nitrogen-filled glovebox (O₂ < 0.1 ppm, H₂O < 0.1 ppm) for hermetic encapsulation. The devices were sealed using a chemically etched aluminosilicate cavity cover glass integrated with a pre-conditioned desiccant label to scavenge residual moisture. Am ultraviolet-curable NOA 61 epoxy resin sealant was precisely dispensed along the perimeter of the cover glass. The encapsulation architecture was then pressed securely against the device substrate and cured under ultraviolet radiation (λ = 365 nm) for 30 s. A pixel-aligned shadow mask was utilized during exposure to completely shield the active polymer emissive area from ultraviolet-induced degradation.

Device operational lifetime optimization

We have further examined the operational lifetimes (LT₅₀; time required for the luminance to decay to 50% of its initial value) of our polymeric OLED devices incorporating spin-coated G-ortho-P-deBr (P-series) and G-ortho-ELPR (ultraviolet-crosslinked) EMLs deposited on different underlying HTLs, namely, PEDOT:PSS, PEDOT:PSS/PTAA, and PTAA (Supplementary Note 19). Devices with the EML deposited directly on PEDOT:PSS showed the lowest LT₅₀, where the LT₅₀ of G-ortho-ELPR was reduced by approximately 50% compared with the LT₅₀ of G-ortho-P-deBr. This pronounced degradation is attributed to the acidic PSS component of PEDOT:PSS layer^46,47, which is likely to chemically interact with the EML during ultraviolet crosslinking. Notably, devices using PTAA as the sole HTL showed decent robustness, exhibiting a nominal approximately 4% decrease in LT₅₀ for the crosslinked ELPR compared with the parent P-series counterpart.

Although the absolute lifetime remains limited at this stage, with the best-performing crosslinked G-ortho-ELPR devices reaching an LT₅₀ of approximately 70 min at a constant driving current density of 1 mA cm⁻² (Supplementary Fig. 53), this is probably associated with trace impurities in polymers and potential weak chemical bonds. We anticipate that device stability can be further improved by (1) using alternative controlled polymerization methods and (2) systematically optimizing the hole- and electron-transporting layer materials and architectures. By halving the residual copper content via intensive dialysis (Supplementary Fig. 62 and Supplementary Table 9) and replacing the baseline PTAA/TmPyPB transport layers with crosslinkable v-NPB and TPBi, respectively, the optimized G-ortho-P-deBr device showed an extended LT₅₀ of 172 min at a constant current density of 2.4 mA cm⁻² (Supplementary Fig. 63). This represents a 135% lifetime improvement over the control device, notably achieved under a 2.4-fold higher operating current. Furthermore, under elevated driving currents of 5.0 mA cm⁻² and 10 mA cm⁻², the devices recorded solid LT₅₀ values of 108 min and 30 min, respectively (Supplementary Figs. 63 and 65). A parallel stability enhancement was observed for the sister system, where the dialysed G-ortho-ELPR device achieved an LT₅₀ of 146 min (Supplementary Fig. 63d).

We observed a reduction of Φ_PL upon the integration of crosslinkable moieties. This reduction, coupled with limitations in EML conductivity and peak luminance, highlights the inherent complexity of balancing photochemical functionality with high-performance optoelectronics. The observed quenching suggests that the local dielectric environment or potential charge transfer induced by the crosslinker periphery still poses challenges to the radiative efficiency of the core. In addition, more detailed study is required to elucidate the effects of ATRP residues to the device operational stability, including the optimization of device architecture for long-term stress.

Molecular dynamics simulations

The polymer systems were constructed for molecular dynamics simulations using PolyConstruct⁵⁰, with CHARMM CGenFF force-field parameters generated using the online tool (www.cgenff.com)⁵¹. Each polymer system consisted of crosslinker monomer units (purple) with Br-atom termini, host monomer units (cyan) and initiator emitter cores (green, red and blue). The green polymer comprised the initiator with 2 polymer branches of 20 cyan monomers and 5 purple monomers each. The red polymer comprised the initiator with 3 polymer branches; 2 of 13 cyan and 3 purple monomers each, and one of 14 cyan and 4 purple monomers. The blue polymer comprised the initiator with 4 polymer branches 15 cyan and 2 purple monomers each. Initially, individual polymers were solvated in 100 × 100 × 100 ų boxes of toluene constructed using PACKMOL⁵², and after at least 20 ns of equilibration, multiple polymer systems of the same type were packed into a single box of the same dimensions, and the toluene was sequentially evaporated in a manner similar to our previous work⁵³. All molecular dynamics simulations were performed using the GROMACS 2024.1 in the isothermal–isobaric (NPT) ensemble with the v-rescale thermostat and c-rescale barostat to maintain temperature at 298.15 K and pressure at 1 bar⁵⁴. Pairwise electrostatic interactions were cut off at 12 Å, with a particle mesh Ewald treatment of long-range interactions, and van der Waals interactions were reduced to 0 between 10 Å and 12 Å using a switching function. Covalent bonds involving hydrogen were constrained using the LINCS algorithm. A time step of 2 fs was used, with simulation snapshots saved every 10 ps. Radial distribution functions and corresponding integrals were calculated using VMD 1.9.3⁵⁵.

1. Errando-Herranz, C. et al. MEMS for photonic integrated circuits. IEEE J. Sel. Top. Quantum Electron. 26, 1–16 (2020).

   Article  Google Scholar 

2. Xiong, J., Hsiang, E.-L., He, Z., Zhan, T. & Wu, S.-T. Augmented reality and virtual reality displays: emerging technologies and future perspectives. Light Sci. Appl. 10, 216 (2021).

   Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

3. Shastri, B. J. et al. Photonics for artificial intelligence and neuromorphic computing. Nat. Photon. 15, 102–114 (2021).

   Article  ADS  CAS  Google Scholar 

4. Zhou, Z., Yin, B. & Michel, J. On-chip light sources for silicon photonics. Light Sci. Appl. 4, e358–e358 (2015).

   Article  ADS  CAS  Google Scholar 

5. Tran, M. A. et al. Extending the spectrum of fully integrated photonics to submicrometre wavelengths. Nature 610, 54–60 (2022).

   Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

6. Liang, D. & Bowers, J. E. Recent progress in heterogeneous III-V-on-silicon photonic integration. Light Adv. Manuf. 2, 59–83 (2021).

   Article  Google Scholar 

7. Xiang, C. et al. High-performance silicon photonics using heterogeneous integration. IEEE J. Sel. Top. Quantum Electron. 28, 1–15 (2022).

   Article  ADS  Google Scholar 

8. Nagarajan, R. et al. 2.5D heterogeneous integration for silicon photonics engines in optical transceivers. IEEE J. Sel. Top. Quantum Electron. 29, 1–10 (2023).

   Google Scholar 

9. Kim, C. et al. Fine metal mask material and manufacturing process for high-resolution active-matrix organic light-emitting diode displays. J. Soc. Inf. Disp. 28, 668–679 (2020).

   Article  CAS  Google Scholar 

10. Zeng, S., Tian, T., Oh, J., Lin, Z.-H. & Shih, C.-J. Direct nanopatterning of complex 3D surfaces and self-aligned superlattices via molecular-beam holographic lithography. Nat. Commun. 16, 3436 (2025).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

11. Müller, C. D. et al. Multi-colour organic light-emitting displays by solution processing. Nature 421, 829–833 (2003).

    Article  ADS  PubMed  Google Scholar 

12. Gather, M. C., Köhnen, A., Falcou, A., Becker, H. & Meerholz, K. Solution-processed full-color polymer organic light-emitting diode displays fabricated by direct photolithography. Adv. Funct. Mater. 17, 191–200 (2007).

    Article  CAS  Google Scholar 

13. Hong, G. et al. A brief history of OLEDs—emitter development and industry milestones. Adv. Mater. 33, 2005630 (2021).

    Article  CAS  Google Scholar 

14. Uoyama, H., Goushi, K., Shizu, K., Nomura, H. & Adachi, C. Highly efficient organic light-emitting diodes from delayed fluorescence. Nature 492, 234–238 (2012).

    Article  ADS  CAS  PubMed  Google Scholar 

15. Hirata, S. et al. Highly efficient blue electroluminescence based on thermally activated delayed fluorescence. Nat. Mater. 14, 330–336 (2015).

    Article  CAS  PubMed  Google Scholar 

16. Wei, Q., Ge, Z. & Voit, B. Thermally activated delayed fluorescent polymers: structures, properties, and applications in OLED devices. Macromol. Rapid Commun. 40, 1800570 (2019).

    Article  Google Scholar 

17. Olivier, F., Daami, A., Licitra, C. & Templier, F. Shockley–Read–Hall and Auger non-radiative recombination in GaN based LEDs: a size effect study. Appl. Phys. Lett. 111, 022104 (2017).

    Article  ADS  Google Scholar 

18. Park, J. et al. Electrically driven mid-submicrometre pixelation of InGaN micro-light-emitting diode displays for augmented-reality glasses. Nat. Photon. 15, 449–455 (2021).

    Article  ADS  CAS  Google Scholar 

19. Wang, X., Zhao, X., Takahashi, T., Ohori, D. & Samukawa, S. 3.5 × 3.5 μm² GaN blue micro-light-emitting diodes with negligible sidewall surface nonradiative recombination. Nat. Commun. 14, 7569 (2023).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

20. Shin, J. et al. Vertical full-colour micro-LEDs via 2D materials-based layer transfer. Nature 614, 81–87 (2023).

    Article  ADS  CAS  PubMed  Google Scholar 

21. Forrest, S. R. Organic Electronics: Foundations to Applications (Oxford Univ. Press, 2020).

22. Huang, Y., Hsiang, E.-L., Deng, M.-Y. & Wu, S.-T. Mini-LED, micro-LED and OLED displays: present status and future perspectives. Light Sci. Appl. 9, 105 (2020).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

23. Marcato, T. et al. Scalable nanopatterning of organic light-emitting diodes beyond the diffraction limit. Nat. Photon. 20, 31–39 (2026).

    Article  ADS  CAS  Google Scholar 

24. Miao, W.-C. et al. Microdisplays: mini-LED, micro-OLED, and micro-LED. Adv. Opt. Mater. 12, 2300112 (2024).

    Article  CAS  Google Scholar 

25. Malinowski, P. E. et al. High resolution photolithography for direct view active matrix organic light-emitting diode augmented reality displays. J. Soc. Inf. Disp. 26, 128–136 (2018).

    Article  CAS  Google Scholar 

26. Joo, W.-J. et al. Metasurface-driven OLED displays beyond 10,000 pixels per inch. Science 370, 459–463 (2020).

    Article  CAS  PubMed  Google Scholar 

27. Nobuyasu, R. S. et al. Rational design of TADF polymers using a donor–acceptor monomer with enhanced TADF efficiency induced by the energy alignment of charge transfer and local triplet excited states. Adv. Opt. Mater. 4, 597–607 (2016).

    Article  CAS  Google Scholar 

28. Wei, Q. et al. Conjugation-induced thermally activated delayed fluorescence (TADF): from conventional non-TADF units to TADF-active polymers. Adv. Funct. Mater. 27, 1605051 (2017).

    Article  Google Scholar 

29. Shao, S. et al. Blue thermally activated delayed fluorescence polymers with nonconjugated backbone and through-space charge transfer effect. J. Am. Chem. Soc. 139, 17739–17742 (2017).

    Article  ADS  CAS  PubMed  Google Scholar 

30. Wang, T., Cheng, Y. & Yang, C. Thermally activated delayed fluorescence polymers and their application in organic light-emitting diodes. Prog. Polym. Sci. 158, 101892 (2024).

    Article  CAS  Google Scholar 

31. Matyjaszewski, K. & Xia, J. Atom transfer radical polymerization. Chem. Rev. 101, 2921–2990 (2001).

    Article  CAS  PubMed  Google Scholar 

32. Kamigaito, M., Ando, T. & Sawamoto, M. Metal-catalyzed living radical polymerization. Chem. Rev. 101, 3689–3746 (2001).

    Article  CAS  PubMed  Google Scholar 

33. Harrisson, S., Whitfield, R., Anastasaki, A. & Matyjaszewski, K. Atom transfer radical polymerization. Nat. Rev. Methods Primer 5, 2 (2025).

    Article  CAS  Google Scholar 

34. Zeng, W. et al. Achieving nearly 30% external quantum efficiency for orange–red organic light emitting diodes by employing thermally activated delayed fluorescence emitters composed of 1,8-naphthalimide-acridine hybrids. Adv. Mater. 30, 1704961 (2018).

    Article  Google Scholar 

35. Tsai, W.-L. et al. A versatile thermally activated delayed fluorescence emitter for both highly efficient doped and non-doped organic light emitting devices. Chem. Commun. 51, 13662–13665 (2015).

    Article  CAS  Google Scholar 

36. Zhang, Q. et al. Efficient blue organic light-emitting diodes employing thermally activated delayed fluorescence. Nat. Photon. 8, 326–332 (2014).

    Article  ADS  CAS  Google Scholar 

37. Ye, S. et al. Continuous color tuning of single-fluorophore emission via polymerization-mediated through-space charge transfer. Sci. Adv. 7, eabd1794 (2021).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

38. Polgar, A. M. et al. Thermally assisted fluorescent polymers: polycyclic aromatic materials for high color purity and white-light emission. ACS Appl. Mater. Interfaces 12, 38602–38613 (2020).

    Article  CAS  PubMed  Google Scholar 

39. Remmers, M., Neher, D. & Wegner, G. Photo-cross-linkable poly(p-phenylene)s. Synthesis, Langmuir–Blodgett multilayer film properties and pattern formation. Macromol. Chem. Phys. 198, 2551–2561 (1997).

    Article  CAS  Google Scholar 

40. Tan, Z. C. H. & Georgia, S. S. Evaluation of poly(allyl methacrylate)-co-(hydroxyethyl methacrylate) as negative electron-beam resist. Polym. Eng. Sci. 23, 963–967 (1983).

    Article  CAS  Google Scholar 

41. Tsuchiya, Y. et al. Temperature dependency of energy shift of excitonic states in a donor–acceptor type TADF molecule. Nat. Commun. 16, 4815 (2025).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

42. Wang, R. et al. Effects of ortho-linkages on the molecular stability of organic light-emitting diode materials. Chem. Mater. 30, 8771–8781 (2018).

    Article  ADS  CAS  Google Scholar 

43. Rao, J. et al. Sterically-locked donor–acceptor conjugated polymers showing efficient thermally activated delayed fluorescence. Angew. Chem. 133, 9721–9727 (2021).

    Article  ADS  Google Scholar 

44. Green, J. A. II, Singer, L. A. & Parks, J. H. Fluorescence quenching by the stable free radical di-t-butylnitroxide. J. Chem. Phys. 58, 2690–2695 (1973).

    Article  ADS  Google Scholar 

45. Burroughes, J. H. et al. Light-emitting diodes based on conjugated polymers. Nature 347, 539–541 (1990).

    Article  ADS  CAS  Google Scholar 

46. Crispin, X. et al. The origin of the high conductivity of poly(3,4-ethylenedioxythiophene)−poly(styrenesulfonate) (PEDOT−PSS) plastic electrodes. Chem. Mater. 18, 4354–4360 (2006).

    Article  CAS  Google Scholar 

47. Cameron, J. & Skabara, P. J. The damaging effects of the acidity in PEDOT:PSS on semiconductor device performance and solutions based on non-acidic alternatives. Mater. Horiz. 7, 1759–1772 (2020).

    Article  CAS  Google Scholar 

48. Matyjaszewski, K. et al. Diminishing catalyst concentration in atom transfer radical polymerization with reducing agents. Proc. Natl Acad. Sci. USA 103, 15309–15314 (2006).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

49. Theriot, J. C. et al. Organocatalyzed atom transfer radical polymerization driven by visible light. Science 352, 1082–1086 (2016).

    Article  ADS  CAS  PubMed  Google Scholar 

50. Munaweera, R., Quinn, A., Morrow, L., Morris, R. A. & O’Mara, M. L. PolyConstruct: adapting biomolecular simulation pipelines for polymers with PolyBuild, PolyConf, and PolyTop. J. Chem. Inf. Model. 65, 4918–4931 (2025).

    Article  CAS  PubMed  Google Scholar 

51. Vanommeslaeghe, K. et al. CHARMM general force field: a force field for drug-like molecules compatible with the CHARMM all-atom additive biological force fields. J. Comput. Chem. 31, 671–690 (2010).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

52. Martínez, L., Andrade, R., Birgin, E. G. & Martínez, J. M. PACKMOL: a package for building initial configurations for molecular dynamics simulations. J. Comput. Chem. 30, 2157–2164 (2009).

    Article  ADS  PubMed  Google Scholar 

53. Meftahi, N., Manian, A., Christofferson, A. J., Lyskov, I. & Russo, S. P. A computational exploration of aggregation-induced excitonic quenching mechanisms for perylene diimide chromophores. J. Chem. Phys. 153, 064108 (2020).

    Article  ADS  CAS  PubMed  Google Scholar 

54. Abraham, M. J. et al. GROMACS: high performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX 1–2, 19–25 (2015).

    Article  ADS  Google Scholar 

55. Humphrey, W., Dalke, A. & Schulten, K. VMD: visual molecular dynamics. J. Mol. Graph. 14, 33–38 (1996).

    Article  CAS  PubMed  Google Scholar 

Download references

Authors and Affiliations

1. Institute for Chemical and Bioengineering, Department of Chemistry and Applied Biosciences, ETH Zürich, Zurich, Switzerland

   Shao-Wei Lo, Sunil B. Shivarudraiah, Zhan-Hong Lin, Patrick Helbling, Lifei Song, Jiachen Wang, Miguel Nouman, Gerrit Stemmler, Donato Maria Carretta, Mei-Nung Chen, Jiayi Zhu, Sudhir Kumar, Yinyin Bao & Chih-Jen Shih

2. School of Information Technology and Electrical Engineering, ETH Zürich, Zurich, Switzerland

   Fuze Jiang & Hua Wang

3. Department of Chemical Engineering, National Taiwan University of Science and Technology, Taipei, Taiwan

   Mei-Nung Chen

4. School of Science, STEM College, RMIT University, Melbourne, Victoria, Australia

   Andrew J. Christofferson

5. Department of Chemistry, Faculty of Science, University of Helsinki, Helsinki, Finland

   Yinyin Bao

Contributions

C.-J.S. and Y.B. conceived of the project and supervised the study. S.-W.L., S.K., Y.B. and C.-J.S. designed the experiments. S.-W.L., P.H. and L.S. performed the polymer synthesis. S.-W.L., S.B.S., F.J. and H.W. designed and fabricated the miniaturized multicolour OLED devices. S.B.S., J.W. and S.-W.L. carried out bulk OLED fabrication and electroluminescence characterization under advice from S.K. S.-W.L. and Z.-H.L. designed and fabricated the experiments of direct e-beam lithography. S.K. carried out the electrochemical and thermal stability characterization. Z.-H.L. and S.-W.L. carried out SEM characterization. S.-W.L. performed TRPL measurement under advice from S.K. S.-W.L. and M.N. designed the layout design of ‘macaw parrot’ pattern. A.J.C. carried out the molecular dynamics simulations. S.-W.L. and J.Z. engineered the host monomer synthesis. S.-W.L. and M.-N.C. prepared Fig. 1. G.S. and S.-W.L. carried out the impedance spectroscopy measurement for OLED devices. D.M.C. and S.-W.L. carried out the AFM characterization. All authors contributed to the preparation of the paper, commented on the paper, and agreed to the contents of the paper and supplementary materials.

Corresponding authors

Correspondence to Yinyin Bao or Chih-Jen Shih.

Competing interests

A patent application about the concept of ELPRs and their synthesis methods has been submitted to the European Patent Office (application number EP25223816.7) by S.-W.L., Y.B. and C.-J.S. The other authors declare no competing interests.

Peer review information

Nature thanks Chihaya Adachi, Jeehwan Kim and Jung-El Ryu for their contribution to the peer review of this work. Peer reviewer reports are available.

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/.

Reprints and permissions