Secondary Transfer of Graphene Electrodes for Highly Stable Flexible OLEDs

Background

Graphene, a single layer of sp²‑bonded carbon arranged in a hexagonal lattice, offers exceptional conductivity, high optical transparency, and mechanical flexibility, making it a promising replacement for indium tin oxide in flexible optoelectronics [1–3]. Recent work has shown that large‑area graphene can be grown by chemical vapor deposition (CVD) on copper foils, achieving sheet resistances as low as 30 Ω/sq and transmittances near 90% [4]. Moreover, boron‑doped graphene has delivered record external quantum efficiencies of ~24.6% in FOLEDs [5].

While CVD on copper is the most mature route for high‑quality, large‑scale graphene, the polycrystalline nature of the foil introduces grain‑boundary cracks and surface roughness. These defects are transferred to the graphene during the typical bubbling or wet‑transfer processes, leading to sharp wrinkles that increase surface roughness and degrade device performance [13–17]. Prior studies have focused on reducing defect density during growth and transfer, but the specific problem of wrinkles arising from grain‑boundary replication has received limited attention.

To address this, we employ a fast, bubble‑based transfer that eliminates residual contaminants and allows us to probe the morphology of the first‑transfer graphene. We observe wrinkle heights reaching several hundred nanometers, which can cause short circuits in FOLEDs. We therefore introduce a secondary‑transfer step that uses two adhesives with distinct adhesion properties—PET coated with a heat‑release adhesive (HRA/PET) as the initial support and the high‑adhesion polymer NOA63 as the final support—to effectively reshape the surface topology from peaks to valleys without damaging the film.

Overview of the graphene synthesis and secondary‑transfer process. a CVD growth of graphene on Cu foil using CH₄. b First bubbling transfer onto PET/HRA. c‑d Secondary transfer onto NOA63, achieving a flattened surface.

Experimental Methods

The copper foil (25 µm) was pre‑annealed at 1040 °C for 30 min, followed by a 30 min growth step with 15 sccm H₂ and 60 sccm CH₄ at 1040 °C. Rapid cooling to room temperature completed the CVD process. For the first transfer, a 2 mol/L NaOH electrolyte was used, and the graphene/Cu stack was electrochemically bubbled off onto PET/HRA. The secondary transfer involved spin‑coating 300 µL of NOA63 onto the graphene/HRA/PET stack (300 rpm for 15 s, then 600 rpm for 15 s), followed by UV curing (350–380 nm, 4 min). During curing, the HRA’s adhesion drops sharply, allowing NOA63 to adhere strongly to graphene and complete the transfer.

Results and Discussions

Optical microscopy and AFM confirmed that the copper grain‑boundary cracks (50–200 µm) were faithfully reproduced as sharp wrinkles on the first‑transfer graphene, with heights up to ~300 nm. Raman spectra of graphene transferred onto SiO₂/Si showed a G′/G intensity ratio of 1.75 ± 0.015 and an I_D/I_G ratio of ~0.065, indicating high‑quality single‑layer graphene [29,30].

Three‑dimensional AFM maps of graphene on Cu foil and HRA/PET, and Raman spectra of transferred graphene.

After the secondary transfer onto NOA63, the previously sharp wrinkles were effectively flattened, as shown by AFM (Fig. 3b). Sheet resistance measurements (Van der Pauw, four‑point probe) over 20 mm × 20 mm areas showed negligible change: both HRA/PET and NOA63 films exhibited average sheet resistances of ~360 Ω/sq, with a narrow distribution (Gaussian fit). Transmittance spectra revealed that NOA63 (90.8 % at 550 nm) outperformed PET/HRA (88.1 %) while maintaining the high transparency of the graphene layer (96.6 %).

AFM and optical maps of graphene on HRA/PET (a) and NOA63 (b); cross‑sectional height profiles (c).

Sheet‑resistance histograms (a) and visible‑range transmittance spectra (b) for HRA/PET and NOA63 supported graphene.

To evaluate the impact on device performance, we fabricated FOLEDs with the following stack: 10 nm Hat‑CN (hole injection), 40 nm TAPC (hole transport), 30 nm CBP doped with 10 % PO‑01 (emissive), 30 nm TPBI (electron transport), 1 nm Liq and 100 nm Al (cathode). A 50 nm PEDOT:PSS layer (3 wt.% DMSO) was inserted between graphene and Hat‑CN to smooth the surface and lower the work‑function barrier. Kelvin probe measurements gave a graphene work function of 4.8 eV, matching well with the HOMO of PEDOT:PSS and the LUMO of Hat‑CN.

Device architecture (a), energy‑level alignment (b), and J‑V‑L, CE‑V curves for D1 (SLG/HRA/PET), D2 (SLG/NOA63), and D3 (PEDOT:PSS/SLG/NOA63) (c‑d). Photograph of a 4 mm × 4.5 mm × 6 mm D3 device (e).

The D1 device, based on the first‑transfer graphene, displayed a pronounced drop in luminance and current density above 13 V due to localized short circuits from the sharp wrinkles. In contrast, D2 (secondary‑transfer) maintained a steady increase in brightness, achieving ~15 000 cd/m² at 14.5 V, confirming the efficacy of wrinkle removal. Introducing PEDOT:PSS (D3) further enhanced performance: luminance reached 35 000 cd/m² and current efficiency peaked at 16.19 cd/A, surpassing D2’s 10.74 cd/A. These results demonstrate that the secondary‑transfer process preserves electrical properties while dramatically improving device stability and efficiency.

Conclusion

We have shown that grain‑boundary cracks in copper foils generate sharp wrinkles in CVD‑grown graphene that compromise FOLED performance. A secondary‑transfer method using a heat‑release adhesive followed by a high‑adhesion polymer effectively flattens these wrinkles with negligible impact on sheet resistance or transparency. Devices employing the PEDOT:PSS/SLG/NOA63 configuration achieved 35 000 cd/m² luminance and 16.19 cd/A current efficiency, establishing the technique as a scalable route for roll‑to‑roll fabrication of high‑quality graphene electrodes for flexible optoelectronics.

Abbreviations

CVD

Chemical vapor deposition

FOLED

Flexible organic light‑emitting device

HOMO

Highest occupied molecular orbital

HRA

Heat release adhesive

ITO

Indium tin oxide

LUMO

Lowest unoccupied molecular orbital

SLG

Single‑layer graphene