Towards highly efficient and highly transparent OLEDs: advanced considerations for emission zone coupled with capping layer design

Strategies to achieve efficient transparent organic light-emitting diodes (TrOLEDs) are presented. The emission zone position is carefully adjusted by monitoring the optical phase change upon reflection from the top electrode, which is significant when the thickness of the capping layer changes. With the proposed design strategy, external quantum efficiency and transmittance values as high as 15% and 80% are demonstrated simultaneously. The effect of surface plasmon polariton (SPP) loss from thin metal electrodes is also taken into account to correctly describe the full scaling behavior of the efficiency of TrOLEDs over key optical design parameters. ©2015 Optical Society of America OCIS codes: (230.3670) Light-emitting diodes; (230.4170) Multilayers; (310.4165) Multilayer design; (310.6845) Thin film devices and applications; (310.6860) Thin films, optical


Introduction
Organic light-emitting diodes (OLEDs) are considered a promising candidate for transparent displays that can lead to futuristic applications that are not readily available through conventional technologies -augmented reality, mutual interactive displays, and invisible displays seamlessly integrated with various objects, to name a few [1,2]. Due to the transparency of the organic thin films used in OLEDs, transparent OLEDs (TrOLEDs) can be easily achieved by replacing their top, thick metal electrodes with a transparent one. A seemingly straight-forward transparent electrode for such a purpose may be transparent conductive oxides (TCOs), such as indium tin oxide (ITO) or indium zinc oxide (IZO) [3][4][5][6][7][8][9]; however, the sputtering deposition process used for TCOs can easily damage the underlying organic layers, making it difficult, though not impossible, to adopt them as a top electrode in TrOLEDs [4][5][6][7][8][9]. For this reason, thin metallic films (e.g. Ag or Au) have particularly been popular because they can be deposited by thermal evaporation with little damage to the underlying organic layers [10][11][12]. To improve the transmittance of the thin metal films, which exhibit finite visible-light transmittance typically in the range of 30% -60%, many groups have suggested depositing dielectric capping layers (CL) on top of the thin metal layers [13][14][15][16][17][18][19][20]. For example, Lee et al. made a systematic study showing the influence of organic CL on the luminous current efficiency (CE), external quantum efficiency (EQE), and bottom-to-top intensity ratio. Huh et al. also discussed in detail the optical effects of CL on TrOLEDs, including the transmittance and emission spectra, and they proposed a useful guideline for the design of efficient TrOLEDs [15,16,19]. Most of these previous works, however, have focused only on the modulation of transmittance and concomitant resonance enhancement achieved by changes in CL thickness, leaving room for further improvement in the efficiency of TrOLEDs.
This work goes one step further and carefully monitors the changes in optical phase upon reflection from the top electrode when the CL thickness is varied and how this impacts the design of TrOLEDs for maximum efficiency and transmittance. Advanced optical simulations based on a classical dipole model are also used to describe the full scaling behavior with respect to a capping layer design, which also has an impact on loss to surface plasmon polariton (SPP) modes.

Experiments
TrOLEDs were fabricated on ITO-coated glass substrates (EVA SNP, Korea; 12 Ω/sq.). The substrates were cleaned sequentially with soapy water, deionized water, acetone, and isopropyl alcohol (IPA) and were subsequently treated by air plasma (PDC-32G, Harrick Plasma) for 5 min. Poly (3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS; Clevios AI4083, Heraeus, Germany) was then spin-coated on top of ITO layers at 2000 rpm for 30 s, followed by drying at 120 °C for 10 min on a hotplate. The samples were loaded into a thermal evaporator (HS-1100, Digital Optics & Vacuum), where all the other layers were deposited consecutively. Figure 1 The electrical and optical characteristics were measured in an N 2 -filled glove box using a customized measurement setup composed of a source-measure unit (Model 2400, Keithley), a calibrated Si photodiode (FDS-100-CAL, Thorlabs), a fiber-optic spectrometer (EPP 2000, StellarNet. Inc.), and a motorized rotation stage for goniometric measurement. The transmittance was measured using a UV-VIS spectrometer (SV2100, K-MAC).
The transmittance, reflectance, and phase change upon reflection were calculated with custom MATLAB codes based on transfer-matrix formalism [21]. Optical analysis for outcoupled modes, power dissipation into substrate modes, waveguided modes, and SPP modes was done based on an advanced classical dipole model that takes into account dipole orientation, Purcell effect, etc [22]. A thin sheet-like emission zone was assumed to be located at the interface between the emission layer (CBP: Ir(ppy) 3 ) and the electron transport layer (Bphen) [22]. The optical constants of the materials used in calculations were measured by spectroscopic ellipsometry.

Effect of capping layer thickness on optical phase change and its implication for optimal emission zone location
In consideration of a OLEDs as a micro-cavity, its overall light output is determined by two major factors: (i) Fabry-Perot resonance enhancement due to multiple-beam interference and (ii) the two-beam interference effect, which is related to interference between directly emitted light and reflected light [23]. In the case of TrOLEDs, however, device structures typically involve a transparent electrode on at least one side, and thus strong multiple-beam resonance is hardly expected due to low reflectance from the transparent electrode. It may still be important to consider the two-beam interference effect even in TrOLEDs, provided that one electrode is semitransparent and thus has a finite reflectance. For light with wavelength λ propagating along the symmetry axis, the two-beam interference factor (f TI ) is proportional to I out or the axial light output toward a direction away from a semi-reflective electrode and is given by [23]: in which R r and φ r are the reflectance and the phase change for light reflecting from the semireflective electrode (e.g. top electrode for bottom emission or bottom electrode for top emission); n org is the refractive index of organic layers, which is assumed, for simplicity, to be the same for all the organic layers involved; and d emission is the distance between the emission zone and the semi-reflective electrode. [See Fig. 1 ) maximizing f TI , derived by setting Δφ TI in Eq. (1) as the integer multiple of 2π, varies from 20 nm to 85 nm to compensate the 0.87π variation in φ top for an OLED device with 10-nm-thick Ag covered with (22 ± 3) nm of d cap . These tendencies suggest that the emission zone should be positioned carefully when one wishes to design highly transparent TrOLEDs by using a thin Ag layer and d cap leading to maximum transmittance. While the discussion above illustrates well the significant influence of d emission in conjunction with d cap on f TI , the scaling behavior of actual light output over d cap and d emission can be complicated due to subtle interplay among various loss mechanisms -waveguide and SPP modes as well as parasitic absorption. For more precise optical optimization, therefore, advanced optical simulation based on a classical dipole model has been used to obtain EQE as a function of d cap and d emission . In this model, the excitation of waveguided and SPP modes as well as the effect of the Purcell factor are fully taken into account [22]. Figures 3(a)-3(c) and Fig. 3(d) show the EQE for the top and bottom directions, total EQE, and the transmittance (T OLED ; at the wavelength of 520 nm), respectively, calculated for an OLEDs device with d Ag fixed at 15 nm. In this calculation, d cap and d emission are varied within a practically reasonable range [21], while the thicknesses of CBP as a hole transport layer (HTL) and CBP:Ir(ppy) 3 as an emitting layer (EML) are fixed at 35 nm and 15 nm, respectively.
It can be first seen that the EQE for the top direction ( = EQE top ) is smaller than that for the bottom direction ( = EQE bot ), as typically expected for TrOLEDs with an asymmetric electrode structure in which the bottom electrode has a lower reflectance than the top electrode [13,15]. For this reason, the dependence of the total EQE ( = EQE total ) on d cap and d emission is shown to be dominated by EQE bot ; therefore, we focus on the trend of EQE bot unless otherwise noted. It may also be noted that EQE total and EQE bot depend more sensitively on d emission than on d cap , while T OLED exhibits a slightly more sensitive dependence on d cap than on d emission , at least for a specific range of d cap . The latter would be the case even for a wide range of d cap , provided that the total thickness of organic layers was fixed. Such an opposite trend can be beneficial from a practical perspective in that one can adjust d cap , more or less as a single independent parameter, to meet a target transmittance. Then, d emission may be chosen for a given d cap to maximize efficiency. This simplifies the overall design strategy and further illustrates the importance of emission zone position obtained as a function of d cap in TrOLEDs. As seen in the contour plots shown in Fig. 3(a), d emission (opt) yielding local maxima again varies in relation to d cap for both 1st-order and 2nd-order cavity designs. However, it is noted that the range of d emission (opt) obtained in this way [ = Method full ; shown as shapes in Fig. 2(c)] is not as wide as the range predicted from f TI [ = Method TI ; shown as lines in Fig. 2(c)]. For the 1st-order design, for example, the difference between the largest and smallest d emission (opt) [shown as circular shapes in Fig. 3(a)] is merely 16 nm ( = 70 nm for d cap of 40 nm minus 54 nm for d cap of 20 nm) in Method full , while it is ca. 31 nm ( = 66 nm for d cap of 35 nm minus 35 nm for d cap of 14 nm) in Method TI . Such a discrepancy is thought to come from SPP modes that take a significant portion when d emission decreases to a value less than a few tens of nanometers. Indeed, Fig. 3(a)-3(c) clearly show that EQE values rapidly drop as d emission becomes smaller than ca. 30 nm. The fact that the difference between the largest and smallest d emission (opt) for 2nd-order cavity TrOLEDs [shown as rectangular shapes in Fig. 3(a)] remains relatively large and becomes comparable to the value obtained with Method TI is also consistent with such a notion because the role of SPP can decrease significantly in a 2ndorder cavity design.
The power dissipation spectra given as a function of in-plane wavevector (k x ) shown in Fig. 4(a) and 4(b) confirm that a significant amount of power dissipation exists at k x even well beyond the boundary of k x that divides propagating and evanescent modes (e.g. 22 μm −1 at λ of 520 nm or photon energy of 2.38 eV) in TrOLEDs with a small d emission . The spectral power density at λ of 520 nm shown in Fig. 4(c) and 4(d) shows the dominant contribution of TM waves to these evanescent modes, being consistent with the fact that SPP modes are due to TM waves coupled with electron motion within metal.

Device characteristics of transparent OLEDs
To demonstrate the role of d emission in conjunction with top electrode design in TrOLEDs, working devices with d cap values of 15 nm, 30 nm, and 40 nm were prepared with varying d emission values. As seen in Fig. 5, all the fabricated TrOLEDs corresponded to a configuration yielding high transmittance, in which the device with a 30-nm-thick capping layer exhibited the highest and most balanced T OLED (λ) throughout the visible range. Those with 15 nm-and 40 nm-thick capping layers showed T OLED values that were relatively high at red and blue parts of spectra, respectively. This range of d cap leading to high T OLED was chosen not only to achieve highly transparent OLEDs but also because the associated change in φ top was shown to be the most significant in the thickness range leading to high T top and thus high T OLED in the previous section [see Fig. 2(a) and 2(b)]. To minimize complications from electrical characteristics, all the devices were prepared with a fixed total thickness of organic layers (110 nm) by adjustment of the HTL thickness to compensate the variation in d emission . The current density (J) -voltage (V) characteristics of the TrOLEDs with a d cap of 40 nm presented in Fig. 6(a) were virtually identical, regardless of the d emission tried, supporting the idea that the difference in efficiency shown in Fig. 6(b) can be attributed mainly to the optical effect.    Fig. 7(a); normalized to have the same value for the devices with d emission of 55 nm] fails to match the experimental results, demonstrating the importance of using the advanced model (Method full ) [22] adopted in this work, which includes the effect of SPP modes as well as other loss channels. In either case, it is clearly confirmed that the relative effect of d emission behaves in a quite different and nontrivial manner even for a relatively small variation in d cap . Based on these results, one may conclude that it is essential to optimize d emission in conjunction with a given d cap to utilize the full potential of TrOLEDs. With this approach, TrOLEDs can be realized with both high transmittance and efficiency, which are typically known to have a trade-off relationship [13,15]. TrOLEDs with (d cap , d emission ) given at (15 nm, 55 nm) and (30 nm, 55-80 nm) are good examples; their T OLED 's are quite high and even close to 80%, yet their EQE total values are only slightly smaller than that of the TrOLEDs with (d cap , d emission ) given at (40 nm, 80 nm), which has a T OLED below 70%.  [22] or on two-beam interference method described in Section 3.1 (Method TI ).

Conclusion
We proposed a strategy for achieving highly efficient TrOLEDs. Unlike opaque OLEDs, in which the location of the optimal emission zone (d emission ) is fixed with respect to the organic/ metal electrode interface, it was shown that it has to be adjusted in accordance with the capping layer design in TrOLEDs mainly due to the associated phase change for reflection from the top cathode structure, which turns out to be more significant for the capping layer thickness (d cap ) near the value leading to high transmittance (T top ). With the optimal emission zone position found for a given d cap , EQE total and transmittance values as high as 15% and 80%, respectively, were realized simultaneously for TrOLEDs based on Ir(ppy) 3 emitters. It was also shown that the effect of SPP loss from thin metal electrodes should also be taken into account to correctly describe the full scaling behavior of TrOLEDs over d cap and d emission as key design parameters. We believe the present work provides a rational guideline for the balanced design of TrOLEDs, opening up a way to unlock their full potential.