Top-emitting OLEDs are the workhorses of high-resolution active-matrix displays because their light exits away from the substrate, decoupling the emitting aperture from the transistor backplane beneath. They also lend themselves naturally to a Fabry–Perot microcavity, formed between a reflective bottom electrode and a semitransparent top cathode. The cavity narrows the electroluminescence spectrum and concentrates emission toward the forward direction, boosting frontal efficiency and color purity. But the same physics that sharpens the spectrum also makes extraction design unforgiving: much of the generated light ends up confined in waveguide modes traveling laterally through the organic stack and in surface plasmon polaritons hugging the metal electrodes. Fixing the air-side interface alone cannot recover those internal losses.

Conventional internal outcoupling structures, such as wrinkles, corrugations, and strongly curved microlens textures, scatter trapped modes back toward useful directions, but they carry a hidden cost. When the underlying texture is steep, the organic layers deposited on top fail to follow it conformally, so the local optical thickness of the cavity varies across the device and the resonance condition drifts from point to point. Stronger scattering then does not translate into a better device, because the cavity-defined forward emission benefit is simultaneously eroded. The key insight of the new work is that the geometry of an internal extraction layer must be judged not just by how much light it scatters but by how gently it does so.

The researchers call their solution a gentle internal corrugation layer, or GICL. Atomic force microscopy over a 50 by 50 micrometer area revealed a peak-to-valley height of about 1.12 micrometers with root-mean-square roughness of 144.5 nanometers, but the defining feature is the slope: representative surface angles of only 2.1 to 6.9 degrees along one axis and 1.8 to 3.5 degrees along the other, spread over lateral spans of nearly 9 micrometers. Cross-sectional electron microscopy showed the entire multilayer stack following this soft undulation without discontinuities. The contrast with reference structures is instructive. A steep imprinted random microlens array reached surface angles of 25.3 degrees with a peak-to-valley height of 7.87 micrometers, while one wrinkle reference achieved low roughness yet still exhibited angles above 53 degrees, demonstrating that roughness amplitude alone says little about geometric severity.

Finite-difference time-domain simulations confirmed that gentler is genuinely better. Among the compared structures, GICL yielded the highest relative total simulated power, outperforming both wrinkle references and the steep microlens array, even though the latter produced far stronger local field perturbation. Ray tracing sharpened the picture: GICL directed 77.3 percent of input power into the forward collection region, versus 43.5 percent for the microlens array, which instead diverted light into wide-angle, internally reflected, and backward trajectories. Under identical conditions, the calculated light extraction efficiency rose from 37.6 percent for the planar microcavity device to 74.1 percent with GICL, an increase of nearly a factor of two. An intermediate-index clear resin with a refractive index of 1.59 was applied at the outer surface to soften the air interface and let the redistributed light actually escape.

The emissive system matters as much as the geometry. In a microcavity, only the spectral portion of the emitter that overlaps the resonance band contributes efficiently to the enhanced external output, so a broad emitter wastes much of its spectrum on off-resonant wavelengths. The team used a phosphor-assisted thermally activated delayed fluorescence sensitization scheme, in which a TADF host and a phosphorescent assistant funnel excitation energy to a terminal multi-resonance TADF emitter called tCzphB-Fl, producing green emission with a full width at half maximum of only about 19 to 22 nanometers. When the microcavity was formed, the narrowband system’s peak shifted just 2 nanometers and its FWHM narrowed from 22 to 19 nanometers, whereas a conventional phosphorescent comparator shifted 13 nanometers and narrowed from 65 to 25 nanometers, evidence that most of the narrowband emission already sat inside the resonance band.

Device results bore out the combined strategy. All GICL devices shared the same 2.25-volt turn-on voltage and essentially unchanged current density, ruling out electrical effects, and the forward spectra showed only minor changes, ruling out spectral reshaping. Under matched resin conditions, adding GICL raised the external quantum efficiency from 34.8 to 69.2 percent, current efficiency from 302.7 to 378.4 candela per ampere, and power efficiency from 199.3 to 396.7 lumens per watt, with maximum luminance reaching nearly 120,000 candela per square meter. Across five samples the efficiencies averaged 68.2 plus or minus 0.7 percent, indicating good reproducibility. The angular color shift at 60 degrees off-axis also fell from 0.0197 to 0.0086 in CIE 1976 coordinates, and the luminance distribution became more Lambertian, both valuable for display uniformity.

Polarization-resolved measurements identified where the recovered light had been hiding. Because waveguide modes are predominantly transverse-electric like and surface plasmon polaritons are transverse-magnetic like, separating the s- and p-polarized emission provides a mode-sensitive diagnostic. Both polarized components, and the non-polarized total, increased across the viewing-angle range in the corrugated device, confirming genuine extraction rather than mere redistribution between channels. Quantitative modal analysis showed the outcoupled fraction rising from 29.7 to 38.6 percent, with the waveguide fraction dropping from 40.0 to 32.0 percent and the plasmon-related fraction changing only modestly. The roughly 30 percent simulated gain in outcoupled power matched the approximately 25 percent uncorrected device-level increase closely, lending quantitative coherence to the extraction interpretation.

Notably, the same corrugation helped the broad phosphorescent system far less, lifting its current efficiency by only about 15.7 percent compared with 25.0 percent for the narrowband system, a difference the authors attribute to spectral overlap with the cavity resonance rather than to bandwidth alone, since the two emissive chemistries differ in other respects as well. The gains did come with a small price: efficiency roll-off at high brightness was moderately larger in the corrugated devices, possibly reflecting local current crowding on the textured surface, and operational lifetime at 1,000 candela per square meter was slightly shorter, though still comparable at about 101 hours to 95 percent of initial luminance. The broader lesson stands: in microcavity OLEDs, internal texture geometry, outer-interface index matching, and emitter-resonance alignment must be engineered together, and gentleness, it turns out, extracts more light than force.

Subject of Research: Cavity-compatible internal light extraction in top-emitting microcavity OLEDs using a gentle corrugation layer and resonance-matched narrowband emission

Article Title: Cavity‐Compatible Light Extraction in Top‐Emitting Microcavity OLEDs via a Gentle Internal Corrugation Layer and Resonance‐Matched Narrowband Emission

Article References: Kim, Y. R., Nimbalkar, A., & Suh, M. C. (2026). Cavity‐Compatible Light Extraction in Top‐Emitting Microcavity OLEDs via a Gentle Internal Corrugation Layer and Resonance‐Matched Narrowband Emission. Advanced Science, Article e78013. https://doi.org/10.1002/advs.78013

Image Credits: AI Generated

DOI: 10.1002/advs.78013

Keywords: OLED, microcavity, light extraction, external quantum efficiency, corrugation, TADF, narrowband emission, waveguide modes, surface plasmon polariton, display technology, FDTD simulation, phosphorescence