Planar OLED Microcavities - Color Shift and Extraction Efficiency

This example calculates the extraction efficiency and angular-dependent color shift from organic light-emitting diodes (OLEDs) using STACK.  In display applications, it is important to reduce the angular dependence of the output light spectrum, as this allows the widest range of viewing angles with minimal color distortion. At the same time, it is important to increase light extraction efficiency. As we will see trade-offs must be made between these two goals. We will recreate the results from Tan et al [1], exploring the emission patterns of unpatterned micro-cavity pixel designs in STACK. Since STACK solves planar devices, it is extremely fast and can be used to rapidly iterate over device layer thicknesses and materials to optimize for spectral uniformity and light extraction efficiency. Finally, the results are output for use in Ansys Speos where designers can directly experience how nanoscale design choices affect human perception.

download example

Overview

Understand the simulation workflow and key results

Step 1: Recreate test cavity results with STACK

In this step, we simulate the devices from Tan[1] with weak and strong microcavity effects. The radiation density into air and chromaticity is provided and compared with published results. A discussion of the trade-offs between color purity, optical efficiency and angle-dependent color shift is provided.

Step 2: Optimized RGB devices results are calculated

The emission properties of the final optimized RGB pixels are plotted.  We show how the angular dependence of color coordinates of each pixel is minimized, and discuss and demonstrate how to export these STACK results to Ansys Speos to better understand how these will affect human perception.

Run and results

Instructions for running the model and discussion of key results

Step 1: Recreate test cavity results with STACK

1. Open [[oled_simulation.fsp]]
2. Run [[OLED_simulation_Tan_comparison.lsf]]

Running this script will set-up the devices used in [1] by calling [[Tan2017_results.lsf]]  which defines several helper functions that are not described here. The published results are then compared with STACK.  For an introduction to this method, see STACK GUI - OLED Device Introduction.

Below is a basic OLED schematic, see important model settings for more information and the material and purpose of each layer.

The following gif depicts the indices of the test device variations up to the glass.

Devices 1-3, are described as weak cavities, due to having transparent ITO as the anode layer. The addition of an Aluminum layer below the ITO anode in devices 4-6 enhances the microcavity resonances and these are referred to as strong cavity devices. In both sets of devices, the ETL layer thickness is varied [40,60,80] nm to demonstrate the emission sensitivity.

After running the script Figure 1, and Figure 2 are displayed and give the radiant power density into the air from stackpurcell.

We can see in Fig 2 that the strong microcavities peak emission wavelength changes significantly. The peak bends towards shorter wavelengths with increasing angle. This is known as the as the blue shift and is the main source of angular dependent color shift in strong microcavities. In Figure 1 the peak emission angle is centered at 520nm; however, we also see that the emission spectra are quite broad, and the emission decreases significantly at higher viewing angle. This diffuse emission spectra from the weak microcavities is undesirable as it is indicative of poor color purity. Furthermore the rapid decay in  emission intensity along Y for devices 1-4 limits the viewing angles. On the other hand the strong microcavities 5-6 address these issues but the color shift in the strong cavities presents additional challenges. This demonstrates the trade-offs between color purity and color distortion.

In Figure 3 the normalized radiance from stackdipole is plotted in polar coordinates. The blue curves are the results from STACK which are compared with the green curves from Tan[1] . Excellent agreement is obtained, and we observe how microcavity effects in devices 4-6 influence the range of viewing angles.

Next the X, Y, Z tristimulus values are returned from STACK and are transformed into u’, and v’ for direct comparison with the paper. The are coordinates that precisely describe the emission color. For more information on these colorspaces see STACK GUI - OLED Device Introduction

The color shift as a function of viewing angle is returned as a trajectory in the chromaticity diagram.  In the weak microcavities we see very little shift in the perceived color as a viewing angle is increased. The strong cavities exhibit large shifts in the color coordinates.  The microcavity resonance tends to shift toward shorter wavelengths at higher viewing angles, this is known as the blue shift of a pixel, an undesirable behaviour in devices 4-6. We do notice that in the chromaticity diagrams that although the points shift they tend to stay close to the boundary which is a measure of the color purity.

The Color coordinates are very sensitive to the luminance spectrum of the emitting layer, as well as the material properties. Neither of these parameters was available from Tan[1], and slight discrepancies in the exact values used can explain most of the variation between the results. For accurate calculations precise knowledge of the material properties and EL spectra are necessary. In the next section, we will look at optimized designs for red, blue, and green pixels.

Step 2: Optimized RGB devices results are calculated

1. Run [[OLED_Tan_RGB_optimized.lsf]]

In this section, we look at the strong microcavity devices that have been optimized to enhance color purity while reducing color shift. We will not elaborate on the optimization method, but simply present the optimized results. The devices red, blue and green pixels are depicted in the following gif.

Below we plot the emission power density in the air as a function of both wavelength and theta using stackpurcell for the RGB pixels. Additionally, the El luminescence spectra are plotted.

We can clearly see how each device create a microcavity that favors the respective emission range. A flat dipole emission spectrum over the bandwidth is used for radiated power density images; however, the actual emission spectra will depend on the overlap of the EL luminescence spectra according to the formula at stackdipole . Note that even though the red cavity does have a secondary emission in the blue, the EL emission spectrum does not overlap; thus, no blue light would be emitted from this pixel.

Once more the X, Y, Z color coordinates are returned from stackdipole, and converted into u’, and v’ coordinates. These are plotted in the chromaticity diagram below.

All other colors will be produced by mixing RGB, and the area defined by these points (known as the gamut) will give the space of possible colors that can be represented by this display. We can see that the pure RGB pixels are adjacent to the boundary. These pixels provide excellent color purity and consequently the gamut of this device, will provides extensive coverage of human perceptible colors. Colors outside the area subtended by these points cannot be displayed but are real colors. Colors outside the boundary are not perceptible.

Let us now take a closer look at the angular dependent color shift. As we have discussed for RGB pixels a shift of the peak emission towards shorter wavelengths explains the angular dependent color shift. This blue shift is seen as the trajectories in the plot. They move in a counter clockwise direction (towards blue) as the angle increases. For mixed colors if the relative blue shift for the RGB pixels is different then this will contribute an additional color distortion. To display white, a balanced contribution from of each pure color is needed. If for example the blue, and green pixel had a greater color shift than red, then at high viewing angles whites would appear reddish.

In the plots of u’, and v’, we can see that there is some unavoidable color shift. An ideal pixel would have constant color coordinates as the viewing angle is increased, but the optimized devices perform well, compared to the test devices we analyzed last step. The color shift (du,dv) is printed to the script prompt.

Next we will export the results to Ansys Speos. This is done through functions defined in the write_rayset_to_file_functions.lsf. To produce a spectral density map we need to consider the overlap of power spectral density of the cavity with EL spectra. Below the exported raysets are plotted. Notice that the peak emission for the green pixel is not at normal incidence. It is at an angle due to the imperfect alignment at theta=0 of cavity resonance with the emission spectra peak.

Additionally to convert to dimensional quantities we should use the same parameters as those in stackpurcell. Finally the total current of each sub-pixel should be defined and here we use 100x100um pixels with a current density of 1 A/m ² , equivalent to 10 nA of current per sub-pixel. The emitted power of each sub-pixel can be adjusted later in Ansys Speos.

Important model settings

Description of important objects and settings used in this model

Materials

The following material stacks are used in step 1 and 2.

Test Devices

RGB Devices

Electrical Contacts

Indium tin oxide is ITO a common anode material due to its transparency at optical frequencies. It has reasonable conductivity and a low work function allowing it to inject holes efficiently. In the test devices from steps 1 and 2 a reflective Aluminum layer is used as the backside cathode to direct light out of the cavity. For the optimized RBG devices a three-layer ITO and Ag reflective anode was used on the backside, and 10:1 Mg:Ag semi-transparent cathode on top.

Hole\Electron Injection Layer HIL\EIL

The injection layers typically have work functions that are intermediate between the electrodes, and charge transport layers. In steps 1-2 a hole injection layer MoO3 is used in the strong microcavity devices, to enhance the injection from Al and a very thin LiF layer is used as the electron injection layer in some devices. The RGB devices 2T-NATA as hole injection layer, but do not use an electron injection layer.

Hole\Electron Transport Layer HTL\ETL

Transport layers are designed to bring the electrons and holes into the electro-luminescent layer. To maximize the devices efficiency the charge should be balanced, as this will minimize the recombination probability through non-radiative channels like heat. In OLED materials the hole mobility is typically larger then for electrons; thus, the HTL will usually be thicker. For electron transport (BPhen) is used as the hole transporting layer.

Capping Layer

OLED materials will react with moisture when exposed to the environment, so these devices are usually encapsulated. The capping layer can be used to enhance extraction efficiency. In the test devices a thick glass substrate is used, while for the RGB devices a thin film encapsulation of polymer and sapphire Al03 is used in addition to thick glass layer.

Emitter Layer, and Spectra:

The electro-luminescent EL layer is the region of electro-optic conversion where the charge carriers combine to form an exciton which may subsequentially decay to produce a photon. Reducing the likelihood of other possible decay mechanism will increase the efficiency. OLEDS often employ AlQ3 as a green emitter, or as a medium that is infused with dye molecule of the desired emission spectra

STACK solver does not attempt to model charge transport itself but uses some simple parameters and an analytic formulation to convert charge density (in Amps per m ² ) j into the emitted power density instead see stackdipole . Additionally it is important to specify information about the conversion process. These include the exciton fraction the number of carriers that result in an exciton ef. The singlet excitation fraction st is the ratio of excitons that are singlets.  Singlets and triplets are spin states, but one just needs to know that singlet states are useful for producing light and triplet states typically are not. By default this will be 0.25, since for every singlet state there are 3 triplet states. Finally, rd the relative decay rate is the percentage of excitons that decay radiatively vs all other singlet decay channels.

To high accuracy we can model the Quantum Emitters in the photoluminescent layer as tiny dipole antennas in this layer, which is what STACK does. Although a dipole emits radiation continuously and a quantum emitter QE can radiate only discrete photons it can be shown that the probability of the QE radiating and the power output of the dipole are both enhanced in a cavity by the same amount; the Purcell factor. For example a higher Purcell factor will increase the likelihood of radiative decay compared to non-radiative decay when there are non-radiative decay channels.

In the test devices, steps 1-2, we used a AlQ3 green emission spectra. The emission spectra used for the RGB devices was plotted in step, for more information see [1].

Cavity Effects:

Cavity effects refer to coherent interaction of light in the cavity. When light undergoes multiple reflections in a layer it can interact with itself, causing the fields to buildup. As we have seen cavity effects may degrade or enhance the optical performance depending on what exactly you are referring to. The microcavity can increase the Purcell factor which will result in more photons being emitted. In this example we have seen that the color purity is increased by using stronger microcavities, at the expense of the angular dependent color shift. In the strong cavities the devices are much more sensitive to the layer thicknesses and materials between the electrodes due to the cavity enhancement in this region. This sensitivity presents a compelling reason to optimize microcavities, which can be easily explored using STACK. Additionally, by exporting these results we can experience how design choices at the nanoscale such as layer thickness are realized in real devices using Ansys Speos. More information on this deep topic can be found at Greens function and local density of states of a dipole source .

Incoherence in STACK:

Coherent effects are very important in the cavity, but as the layer thicknesses become larger, they become less important. Once the layer thicknesses are on the same order as the coherence length of the light it is no longer physical to treat the light coherently. STACK, like FDTD, is inherently coherent, and special care must be taken when looking at incoherent sources or propagation distances beyond the coherence length. In the thick glass substrate\superstrate this will be the case, and by assuming coherent multipath interactions in these thicker layers the results display oscillatory artifacts.

In 2020R1 we introduced a line break feature that will allow you to specify where to break the coherence. For example we may want to consider coherent effects in an AR coating on the top of the substrate, and microcavity emission from below. Imagine a thick glass layer between these features, where the coherence will break down. To model this system just pass an optional “incoherent_propogation” argument to specify the glass as an incoherent layer.

Additional resources

Additional documentation, examples, and training material

Related publications

Guanjun Tan, Jiun-Haw Lee, Sheng-Chieh Lin, Ruidong Zhu, Sang-Hun Choi, and Shin-Tson Wu, "Analysis and optimization on the angular color shift of RGB OLED displays," Opt. Express 25 , 33629-33642 (2017)

See also

• STACK GUI - OLED Device Introduction
• Planar OLED - ZEMAX Interoperabillity
• stackdipole
• stackpurcell
• OLED Methodology
• FDTD vs STACK
• Coherence in FDTD
• Green’s Function and LDOS
• Fluorescence Enhancement

Related Ansys Innovation  Courses