Introduction

Organic light emitting diodes (OLEDs) have attracted significant attention in the modern display industry owing to their low power consumption, fast response time, and high contrast ratio. As self emissive devices, OLEDs do not require a backlight unit, which is essential in liquid crystal displays, and their thin and simple architecture enables their broad applicability to flexible electronics. These features allow OLEDs to provide an enhanced visual experience and versatile functionality in portable electronic devices such as smartphones, smart watches, tablet PC, automotive displays, and wearable devices, spanning both flat panel and flexible formats1,2,3,4,5. However, low external light extraction efficiency (LEE) remains a major limiting factor in achieving high performance OLEDs. Although their internal quantum efficiency can reach 100% through optimized device architectures, only about 20% of the generated light is extracted externally. This limitation results from optical confinement in waveguide modes, surface plasmon polaritons, and substrate modes, caused by refractive index mismatches within the multilayer structure6,7,8,9. Such light loss directly increases power consumption, thereby reducing the battery life of portable electronic devices and posing a particularly critical challenge for flexible OLEDs10,11.

To enhance the external LEE of OLEDs, various strategies have been extensively investigated to reduce refractive index mismatches and control the light path between the OLED device and air. These include polymer dispersed liquid crystal (PDLC) scattering films12,13,14, microlens arrays (MLA)15,16, random surface structures17, porous cellulose18, and MLA patterned parylene substrates19 or visible light films20. These external light extraction technologies have demonstrated substantial efficiency improvements, achieving external quantum efficiency (EQE) enhancements of 20–40% compared with reference OLEDs, and have proven particularly effective in the flat panel display sector. However, these approaches are largely optimized for rigid glass substrates, and their fabrication typically involves complex, multistep processes such as photolithography, plasma etching, and high temperature or high vacuum deposition. Such methods increase manufacturing costs and processing times, thereby limiting their applicability to next generation flexible OLEDs.

To overcome these challenges, technologies compatible with flexible substrates, such as salmon DNA based scattering layers and nanocellulose/polymer composites, have been proposed. Despite their advantages, their EQE enhancements remain limited to 29% and 22%, respectively21,22. Although these methods offer greater simplicity compared with conventional processes, they require specific processing conditions such as precise temperature control at 0 °C or slow tensile speeds which constrain their scalability for mass production. These limitations highlight the need for a novel external light extraction technology that is simple, cost effective, and highly efficient to accelerate the commercialization of flexible OLEDs. Furthermore, most prior studies have relied primarily on experimental and empirical approaches, while theoretical models capable of quantitatively correlating surface morphology, size, and electro-optical properties with light extraction efficiency have not yet been fully developed23,24,25.

This study proposes an MLA imprinted PDLC (MIP) substrate that synergistically combines strong volumetric scattering in a PDLC with interfacial refraction from an MLA. In the PDLC, micrometer scale liquid crystal droplets with refractive index contrast to the polymer act as Mie scatterers, randomizing the propagation direction of incident light, while the polymer matrix provides mechanical robustness and flexibility26,27,28,29. In the MIP structure, this volumetric scattering first converts guided and substrate modes into broadly diffusive light, and the MLA at the substrate-air interface then refracts a portion of this diffusive light out of the substrate, reducing total internal reflection and enhancing LEE30,31,32. As a result, the MIP achieves a 21% enhancement in EQE. The MIP can be fabricated through a simple room temperature solution process consisting of spin coating followed by UV curing, eliminating the complex, multistep procedures required for conventional external light extraction technologies and without requiring any special temperature conditions or mechanical deformation, actuation steps. In addition to experimental characterization, the feasibility and design space of the MIP were validated using Monte Carlo ray-tracing simulations. Overall, the MIP provides a mechanically flexible, process simple, and simulation guided light extraction strategy, offering a promising pathway for next generation flexible OLEDs.

Results and discussion

Figure 1a presents a schematic of the MIP, which functions as a light extraction layer positioned beneath the OLED. This structure combines microlens induced surface refraction with droplet induced internal scattering to enable multidirectional light extraction and enhance photon outcoupling. Figure 1b presents a scanning electron microscope (SEM) image demonstrating the successful replication of a uniform, periodic MLA layer on the PDLC surface, highlighting the high reproducibility and fidelity of the imprinting process. The periodic MLA structure acts as a wavelength scale optical element that effectively redirects light trapped in waveguided modes, thereby improving extraction efficiency. Figure 1c presents a distinct hexagonal diffraction pattern obtained under 632 nm laser irradiation. This pattern arises from the periodic hexagonal lattice of the imprinted MLA and provides optical evidence that the MLA structure is not only physically replicated but also optically functional. The observed diffraction confirms that the MLA actively modulates light propagation rather than serving as a passive surface feature. Figure 1d illustrates the fabrication process of the MIP substrate, which involves imprinting the MLA mold onto a polymer surface, followed by spin coating of the PDLC and a planarization layer, and subsequent UV curing. This solution based process is conducted under ambient conditions without the need for high temperature or vacuum, yielding a flexible substrate with sufficient surface flatness for OLED integration. Furthermore, the process is compatible with large area, roll-to-roll manufacturing systems.

Fig. 1
Fig. 1
Full size image

Schematic of a flexible OLEDs structure on an MIP. (a) Schematic illustration of high efficiency flexible OLEDs on an MIP. The green arrow indicates the direction of light emission. (b) SEM image of the MIP bottom side. (c) Image of a red laser (λ = 632 nm) beam hexagonal pattern through the MIP. (d) Schematic fabrication process of the MIP and OLEDs deposition.

Figure 2a presents the simulated structure consisting only of the MLA on glass, where the volumetric scattering of the PDLC droplets is not modeled. This simplified structure is used to evaluate the effectiveness of the periodic surface geometry. The simulation employed lens dimensions and periodicity identical to those of the fabricated MLA. As shown in the ray-tracing results in Fig. 2b, the flat reference substrate exhibited substantial optical losses due to total internal reflection and waveguide modes. In contrast, the MLA substrate emitted approximately 62.5% more light externally than the reference substrate. This enhancement is attributed to the curved surface of the MLA, which redirects the propagation of internally trapped light to facilitate its extraction. These results underscore the critical role of the MLA structure in mitigating optical confinement compared with a planar surface. Figure 2c presents the angular emission characteristics of the device based on a surface light source. At normal emission (0°), the total power of the reference and MLA devices was 21.3 W and 25.5 W, respectively, representing a 19.7% increase. Furthermore, stable and uniform emission characteristics were maintained at this enhancement level across a wide viewing angle of up to ± 30°. Collectively, these results confirm that the curved MLA structure effectively improves forward light extraction by efficiently redirecting confined photons, making it a promising approach for high brightness flexible OLED applications.

Fig. 2
Fig. 2
Full size image

Simulation of the microstructure surface. (a) Microstructure simulation model design and (b) ray-tracing simulation of light extraction through different substrates, with light propagation through a flat substrate (top) and an MLA substrate (bottom). (c) Comparison of light extraction performance with respect to viewing angle between the reference and MLA substrate (w = 20 μm, h = 10 μm, p = 20 μm).

Figure 3 presents the optical properties of bare glass, PDLC, MLA and MIP substrate. Figure 3a presents the total transmittance (Tt). At 550 nm, bare glass, MIP, MLA and PDLC are 92.7%, 58.9%, 20.0%, and 84.7%, respectively. Figure 3b presents the parallel transmittance (Tp), which is 91.9% for bare glass, 0.8% for the MIP, 15.1% for the MLA and 5.9% for the PDLC. Figure 3c presents compare with haze, which are 0.84%, 98.3%, 24.4%, and 92.9% for bare glass, MIP, MLA, and PDLC, respectively. Haze was calculated as \(Haze~=~\left( {Tt~ - ~Tp} \right)~/~Tt~ \times ~100\), respectively. In PDLC with randomly oriented nematic droplets, the scattered light field is known to be nearly isotropic, and multiple scattering tends to drive the transmitted intensity distribution toward a quasi-Lambertian profile31. These results indicate that the PDLC already acts as a strong volumetric scattering layer, it maintains a relatively high total transmittance while exhibiting a strongly reduced parallel transmittance and a haze exceeding 90%. In contrast, the MLA presents moderate haze and a partial reduction in parallel transmittance, suggesting that the surface microstructure redistributes the propagation directions but does not generate scattering as strong as that of the PDLC layer. The MIP substrate exhibits the highest haze and the lowest parallel transmittance among all samples, while still retaining total transmittance values on the order of 60%, indicating a synergistic structure in which the volumetric scattering of the PDLC and the surface structuring of the MLA combine to provide both strong diffusion and sufficient transmittance. This strong volumetric scattering plays a crucial role in randomizing the propagation angles of photons trapped in waveguide modes, thereby increasing the probability that they will exit the surface-air interface33,34,35.

Fig. 3
Fig. 3
Full size image

Optical Properties of bare glass, MIP, MLA, PDLC. (a) Total transmittance. (b) Parallel transmittance. (c) Haze.

Figure 4a presents a polarizing optical microscope (POM) image obtained using a first-order retardation plate (λ = 530 nm) inserted at a 45° angle between crossed polarizers. The observed magenta regions indicate liquid crystal molecules aligned parallel to the polarizer axes, isotropic alignment, or wall defects at the droplet interface. Various features were observed depending on the depth of focus. Near the surface, droplets surrounding the MLA structure were clearly visible, while in the bulk region, overlapping droplets created a more uniform background. Polarization optical microscope observations confirmed that the liquid crystal droplets were contained throughout the film, maintaining the MLA shape. Figure 4b and c present the results of SEM and 3D surface analyses, respectively, and demonstrate that the MLA mold pattern was replicated on the MIP substrate with the same periodicity and morphology. This result indicates that the solution based imprinting process successfully integrated the surface MLA structures with internal PDLC droplets to form a composite light extraction architecture. Overall, the MIP provides a dual optical structure in which internal droplet induced scattering and surface microlens induced refraction scattering act synergistically.

Fig. 4
Fig. 4
Full size image

Characteristic analysis of (a) POM images of the MIP by focal length. The black arrows indicate the polarizer and analyzer axes, and the pink arrow represents the direction of the retardation plate. (b) SEM images of the MLA substrate (top) and MIP (bottom). (c) Optical profiler 3D surface images of the MLA substrate (top) and MIP (bottom).

Figure 5a presents a photograph of a bent fabricated MIP sample. The lack of visible structural damage demonstrates that the film maintained its mechanical integrity and thus could be integrated into flexible devices. The surface appears opaque, which can be attributed to the combined scattering effects of the PDLC droplets inside the film and MLA surface features. Figure 5b presents the atomic force microscopy (AFM) images used to evaluate the surface roughness in relation to OLEDs deposition. The root-mean-square roughness (Rq) of the MIP was 0.808 nm, which is comparable to that of the polyethylene naphthalate (PEN) substrate (0.821 nm). This result suggests that the surface of the MIP is as flat as that of a standard PEN substrate, indicating that it is compatible with thin film deposition processes. Figure 5c presents the contact angle measurements obtained using deionized water. A low surface roughness is critical for preventing electrical shorts or current leakage in subsequently deposited ultrathin organic layers; therefore, the smooth topography and favorable wetting characteristics of the MIP highlight its compatibility with high performance OLEDs fabrication. The MIP surface exhibits a lower water contact angle than either glass or PEN, which generally indicates a higher surface free energy for polymer and liquid crystal based coatings36. A higher surface free energy promotes the wetting and adhesion of subsequently deposited layers, which is favorable for further coating steps37. Collectively, these observations suggest that the MIP combines flexibility and optical functionality with good process compatibility, making it a practical substrate for flexible OLEDs fabrication. Furthermore, we acknowledge that a more quantitative decomposition of the surface free energy into polar and dispersive components, for example using the Owens–Wendt method, would require contact angle measurements with multiple probe liquids38. This comprehensive analysis will be an important subject for future work.

Fig. 5
Fig. 5
Full size image

MIP surface characteristics. (a) Photograph of a fabricated flat and flexible MIP. (b) AFM images of an MIP (left) and PEN (right) substrate surface. (c) Deionized water contact angle measurement of bare glass, MIP, and PEN.

Figure 6a presents the device structures of OLEDs fabricated on a bare glass substrate and an MIP substrate to compare their electro-optical characteristics. Figure 6b presents a comparison of the current density–voltage–luminance (J–V–L) characteristics of the devices. After being turned on, the MIP OLEDs exhibited a slightly lower current density but delivered a higher luminance than the reference device. At 3 V, the current density of the MIP OLEDs was 0.19 mA/cm2, which is slightly below the value of 0.23 mA/cm2 for the reference, whereas at 5 V, the luminance reached 251.8 cd/m², which is approximately 14% higher than the reference value of 219.6 cd/m2. Figure 6c presents compare with the integrated EL spectral intensities, which correspond to the total area under the curve, the reference device exhibited a value of 58.25, whereas the MIP device exhibited 69.29. This result represents a 19% increase in total emission for the proposed device compared to the reference and thus demonstrates a greater overall LEE for the MIP. Figure 6d presents the current efficiencies. The MIP OLEDs reached a value of 4.34 cd/A compared with 3.77 cd/A for the reference, representing a 15% improvement. EQE result in The EQE results in Fig. 6e follow the same trend, increasing from 1.022% for the reference OLEDs to 1.238% for the MIP OLEDs, which is equivalent to a 21% increase. This experimentally measured 21% EQE enhancement is in strong agreement with the 19.7% luminance enhancement at a normal viewing angle that was predicted by our ray-tracing simulations. This agreement validates our design approach and confirms the significant contribution of the MIP structure. Taken together, these results demonstrate that the MLA modifies the photon paths and the PDLC droplets promote multiple scattering such that the light trapped inside the device is effectively extracted. Through this mechanism, the MIP design improves luminance, spectral output, current efficiency, and EQE, which overall demonstrates its promise as a light extraction strategy for flexible OLEDs.

Table 1 summarizes the substrate type, light extraction structure, relative EQE enhancement, and key processing characteristics for each method. As shown in Table 1, the relative EQE enhancement of the flexible OLEDs on the MIP substrate (~ 21%) is only 1–8% lower than that reported for recent flexible scattering layers based on salmon DNA and nanocellulose/polymer composites. In contrast, our approach relies solely on a simple room temperature spin coating and UV curing process, without any special temperature control or mechanical deformation steps. This comparison demonstrates that the proposed hybrid MIP offers a competitive balance among efficiency improvement, mechanical flexibility and process scalability.

Fig. 6
Fig. 6
Full size image

Electro–optical characteristics of OLEDs with and without an MIP. (a) Schematic illustration of the emission path of OLEDs without (left) and with (right) an MIP. (b) Current density–voltage and luminance characteristics. (c) Normalized EL spectra. (d) Current efficiency–current density characteristics. (e) EQE–current density characteristics.

Table 1 Summary of key performance metrics and process characteristics.
Full size table

Figure 7 presents photographs of a flexible OLEDs device fabricated on an MIP during operation. The device produced uniform green emission under bias, which indicates that the MIP is compatible with both OLEDs fabrication and driving. When the substrate was bent, the emission remained stable, which suggests that the polymer matrix maintained its mechanical flexibility while the liquid crystal domains continued to function optically. The observation of homogeneous emission under bending supports the view that the MIP architecture provides effective external light extraction and reliability in flexible displays. These experiments confirm that the MIP can serve as a practical substrate platform for flexible OLEDs. As shown in Fig. 7, both the device observed from the front and the bent device observed from the side exhibited the same color based on visual inspection. Therefore, the angular color shift is considered negligible.

Fig. 7
Fig. 7
Full size image

Reliability of a flexible OLEDs on an MIP.

Conclusion

In summary, we designed, fabricated, and validated a novel hybrid optical substrate referred to as the MIP that synergistically integrates an MLA with a PDLC layer to overcome key limitations in OLEDs light extraction. By utilizing a straightforward solution process involving room temperature imprinting and subsequent UV curing, our approach offers a practical and scalable route to the low cost, large area fabrication of advanced flexible optoelectronics, avoiding the use of complex lithographic steps. The effectiveness of this hybrid design was validated through both ray-tracing simulations and experimental measurements, which confirmed a 21% enhancement in EQE, significant gains in current efficiency and luminance, and improved spectral quality based on the relaxation of the microcavity effect. This work not only yielded a highly effective substrate for current flexible OLEDs but also introduced a broader design strategy involving the integration of deterministic surface structures with random bulk media as a powerful method for multifunctional light management in next generation optical systems.

Methods

Simulation modeling

Monte Carlo ray-tracing simulations were performed using LightTools (Synopsys, USA). The effect of the MLA on the external light extraction efficiency of the OLEDs was analyzed by tracing the photon trajectories emitted through glass substrates with and without MLA structures. The surrounding medium was set to air (n = 1.0). The substrate was modeled using NBK7 glass (n = 1.52) with dimensions of 4 × 4 × 1 mm and a three-dimensional textured surface to replicate the MLA geometry. The MLA structure was defined using the 3D Texture function built into the LightTools software. The simulated microlens array was designed as a concave lens with a hexagonal arrangement, and its geometric structure was set to match the dimensions of the fabricated device a diameter of 20 μm, a height of 10 μm, and a pitch of 20 μm. The light source was a 4 × 4 mm rectangular area emitter with a Lambertian angular distribution and normalized output power of 100 W.

Fabrication of MIP

The PDLC solution was prepared by combining nematic liquid crystal E7 (Qingdao QY Liquid Crystal Co., Ltd., China, ne = 1.74, no = 1.52) with the UV curable epoxy NOA 63 (Norland Products, USA, np = 1.56) as a polymer matrix. The two materials were mixed at a 40:60 weight ratio and stirred at 80 °C for 1 h. The MIP was fabricated on a commercial MLA mold with a diameter of 20 μm, height of 10 μm and pitch of 20 μm. The MLA mold was ultrasonically cleaned in acetone for 15 min and dried in a vacuum oven at 100 °C for 15 min. The PDLC film was deposited on the MLA mold via spin coating at 500 rpm for 5 s, then at 2000 rpm for 20 s. The film was then cured under UV irradiation (UV ozone cleaner AC-6, Ahtech LTS, Korea) for 15 min at 32 mW/cm² to form the MIP. For OLEDs deposition, a planarization layer of the same epoxy was applied on top of the PDLC via spin coating at 500 rpm for 5 s, then at 2000 rpm for 20 s, followed by UV curing for 5 min. This planarization layer is crucial for providing an optically smooth and flat surface, which is essential for the uniform deposition and reliable performance of the subsequent multilayer OLEDs.

Fabrication of OLEDs devices

Bottom emission flexible OLEDs were fabricated via thermal evaporation. The device structure was as follows: MoO3 (20 nm) / Ag (30 nm, anode) / MoO3 (10 nm, hole injection layer) / TAPC (50 nm, hole transport layer) / Alq3 (50 nm, emissive layer) / LiF (1.2 nm, electron injection layer) / Al (100 nm, cathode). Both bare glass and an MIP were tested for comparison.

MIP characterization

The optical properties of the MIP were characterized using an UV–VIS NIR spectrometer (Lambda 1050, PerkinElmer, USA) to measure the total transmittance, parallel transmittance, and haze. POM (BX53, Olympus, Japan) was used to observe the droplet distribution and texture of the PDLC layer. SEM (EM-30 N, COXEM, Korea) was used to confirm the MLA imprint fidelity on the film surface. Confocal microscopy (VK-X3000, Keyence, Japan) was used to measure film thickness, droplet diameter, and MLA height. AFM (SPM-9600, Shimadzu, Japan) was used to evaluate surface roughness. Contact angle measurements were conducted with deionized water using a surface contact angle analyzer (Phoenix-300 Touch, S.E.O., Korea) to assess surface wettability.

Characterization of OLEDs devices

The electrical and optical characteristics of OLEDs devices with and without an MIP were measured using a source meter (Keithley 2400, Keithley, USA) and spectroradiometer (LMS PR-650, Photo Research, USA). Specifically, the J–V–L characteristics, EQE, and normalized EL intensity were evaluated.