OLED Waveguide-Mode Suppression with an Ultrathin Metal Anode

The paper's 140 nm organic stack is thicker than the ITO device's 69 nm first-order transverse-magnetic mode (TM₁) cutoff, while it is thinner than the 208 nm fundamental transverse-electric mode (TE₀) and 236 nm TM₁ cutoffs of the 5 nm Cu-seeded Ag electrode. The reported Waveguide fraction consequently falls from 16.0% to 0%. This case tests that mode-transfer mechanism in Dreapex TMM at the 530 nm emission peak, using open optical constants and a disclosed effective electrode proxy. The Figure 1 cutoff consequence is the primary target; Table 1 is a semi-quantitative point of comparison because the paper integrates that table over the emission spectrum.

The red 140 nm marker falls in the unsupported region for both Cu-seeded Ag cutoffs. The same thickness remains above the ITO TM₁ cutoff, and the paper reports that the ITO TE₀ mode is guided even as organic thickness approaches zero.

Background

A planar OLED guides light when its organic and electrode stack can sustain an in-plane optical mode. A cutoff thickness marks the boundary below which a named mode is no longer supported. Jeong et al. identify two relevant branches at 530 nm: a dielectric-like TE₀ mode and a metal-sensitive TM₁ mode.

The ultrathin electrode changes the two branches through different physical levers. Its negative real permittivity weakens TE₀ confinement, while replacing 150 nm ITO with only 5 nm of metal reduces the total guiding thickness seen by TM₁. The practical consequence is the same for the paper's 5/40/20/75 nm organic stack: both branches are below cutoff with the ultrathin electrode.

The Mode result performs a complete power budget after an emitting dipole couples into the planar stack. Power Dissipation retains the in-plane-wave-vector view and can show whether a guided peak survives. These are emission calculations; externally incident reflectance or transmittance alone cannot establish the OLED mode budget.

Mapping the Paper to the TMM Model

The paper's Figure 2 optical model replaces the simplified CBP slab of Figure 1 with an archetypal multilayer OLED. The organic total remains 140 nm, so the Figure 1 cutoff comparison directly predicts the presence or absence of its waveguide channel.

Structure from the output side to Al

In the table, $R$ denotes power reflectance and $T$ denotes power transmittance; both are dimensionless fractions reported as percentages.

The Cu-seeded Ag film is approximately 5 nm Ag grown on a Cu seed, not a bulk Cu–Ag alloy. The effective value above reproduces the publicly reported $R$ and $T$ of the complete ultrathin film on glass to the reported precision. It is not the authors' original ellipsometry and is held constant from 550 to 530 nm for this single-wavelength test.

Emission and detector settings

The normalized in-plane wave vector is $u=k_{\parallel}/k_\mathrm{EML}=k_{\parallel}/(k_0n_\mathrm{EML})$, where $k_{\parallel}$ is the magnitude of the wave-vector component parallel to the layers, $k_0=2\pi/\lambda$ is the vacuum wave number, $n_\mathrm{EML}$ is the emissive-layer refractive index at that wavelength, $k_\mathrm{EML}$ is the emissive-layer wave number, and $\lambda$ is the vacuum wavelength. The wave numbers have inverse-length units, so $u$ is dimensionless.

Mode-channel mapping

Surface plasmon polariton (SPP) denotes a bound electromagnetic mode at a metal–dielectric interface. The paper uses this label for the high-in-plane-wave-vector power beyond the EML wave number.

The paper places all power above the EML wave number into SPP, and Dreapex TMM uses the same interval for Evanescent. Absorption is residual power below the EML light line that does not enter an escape channel, so it remains separate and cannot be added to SPP.

Reproduction Target and Acceptance Criteria

The following criteria were defined before the results were compared:

• Both models must complete the Mode and Power Dissipation calculations.
• Each seven-channel Mode budget must close to 100% within 0.01 percentage point.
• The Cu-seeded Ag model must return at most 2.0% Waveguide power and at least 10.0 percentage points less Waveguide power than ITO.
• The Cu-seeded Ag Substrate fraction must exceed the ITO Substrate fraction.
• Top Outcoupling and Substrate must each be within 10.0 percentage points of their Table 1 anchors for both models.
• ITO Waveguide must be within 10.0 percentage points of 16.0%; Cu-seeded Ag Waveguide uses the stricter 2.0% ceiling.
• SPP is compared directly with Evanescent, while Absorption is reported independently. SPP is not the sole first-pass criterion because this case uses one 530 nm point whereas Table 1 is spectrum-integrated.

A missed criterion remains a reported failure. Material constants, thicknesses, dipole settings, and thresholds are not tuned after seeing the result; any alternative is labeled as a separate sensitivity run.

Modeling Path in Dreapex TMM

Structure

Import the two benchmark inputs:

• 150 nm ITO reference model
• 5 nm Cu-seeded Ag proxy model

The layer sequence runs from the incoherent Glass output side to the optically thick Al cathode. The two inputs differ only in the anode layer. The EML remains emissive in both.

Optical Settings

In the Emission detector lane, keep Mode and Power Dissipation enabled. Set both wavelength controls to Single at 530 nm; set the Power Dissipation coordinate $u$ from 0 to 2 with a 0.002 step. Run each imported model once.

Example Setup

The table below is the compact reconstruction checklist for both inputs. The anode row is the only variant.

The screenshots show the settings contained in the two downloadable benchmark models.

Simulation Results and Comparison with Figure 1 and Table 1

Figure 1 supplies the cutoff mechanism, while Table 1 supplies an integrated mode budget. The primary numerical verdict uses the 530 nm Waveguide, Substrate, and Top Outcoupling channels.

Mode Budget

The single-wavelength calculation reproduces the direction of the paper's transfer from Waveguide power into Substrate and high-wave-vector channels, but not the reported complete elimination. The Waveguide reduction is 8.78 pp, leaving 15.52% rather than the paper's 0%.

In-Plane-Wave-Vector Evidence

The ITO result contains several narrow peaks between approximately $u=0.85$ and $1.03$. The effective ultrathin-electrode model has a TM-dominated narrow peak near $u=0.97$; this point remains inside the paper's Waveguide interval and cannot be dismissed as a channel-naming difference. The integrated Mode Waveguide fraction remains the numerical acceptance criterion.

Acceptance Verdict

The overall verdict is Fail: eight of ten checks pass. Both power budgets close to 100%, all four Air/Substrate values stay within 10 pp of Table 1, ITO Waveguide is within 10 pp, and the Cu-seeded Ag Substrate share exceeds ITO by 4.42 pp. The two failed checks are the ≤2% Cu-seeded Ag Waveguide ceiling and the required ≥10 pp Waveguide reduction; the calculated values are 15.52% and 8.78 pp, respectively.

Deviation Analysis

• The authors measured material permittivities by spectroscopic ellipsometry but did not publish a complete numeric table in the main article. Open single-wavelength values replace the unavailable TAPC, CBP, TPBi, ITO, MoO₃, and Al curves.
• The Cu-seeded Ag constant is an effective inversion of the reported calculated bare-film R and T. It is neither bulk Ag nor the authors' original dispersion, and holding its 550 nm value at 530 nm is a disclosed extrapolation. The fitted real permittivity is −13.70, compared with the article's approximate visible value of about −10; the difference remains an input uncertainty rather than being tuned away.
• The Figure 2 optical-stack description omits the fabricated 1.5 nm LiQ layer and 3 nm AZO passivation. The primary model follows that optical description; both thin layers belong in separate sensitivity checks.
• The paper's Table 1 is spectrum-integrated, whereas this primary comparison is monochromatic at the 530 nm peak. A broad EL spectrum is unnecessary for the cutoff verdict but necessary for exact Table 1 weighting.
• The paper's SPP and Dreapex TMM's Evanescent both cover the region above the EML wave number; Absorption is reported separately.
• The model is planar and one-dimensional. Roughness, scattering, electrode nonuniformity, electrical balance, exciton quenching, and substrate extraction structures are outside the optical comparison.

Further Extensions

• Sweep TPBi thickness while keeping the other layers fixed to reproduce the mode-transfer curves in Figure 2C and 2D.
• Load the measured Ir(ppy)₂acac EL spectrum and full dispersive optical constants, then repeat Mode with spectrum weighting for a stricter Table 1 comparison.
• Add 3 nm AZO and 1.5 nm LiQ as separately named sensitivity variants; do not replace the primary benchmark inputs.
• Vary the effective electrode constant within the uncertainty allowed by the public R/T anchors and report the resulting Waveguide ceiling.
• Compare the ITO and ultrathin-electrode Power Dissipation peaks across 100–260 nm total organic thickness to connect the integrated budget to the reported cutoff map.

References

• C. Jeong, Y.-B. Park, and L. J. Guo, “Tackling light trapping in organic light-emitting diodes by complete elimination of waveguide modes,” Science Advances 7, eabg0355 (2021). DOI
• Aulika et al., open optical constants for OLED materials, CC BY 4.0. Zenodo dataset
• M. N. Polyanskiy, refractiveindex.info database. Open database repository

Back to Case Studies