Wasayansi Avula Zithando Zokubonisa Ezithambekayo Zisebenzisa Graphene Ngetitaneyamu Dioksaidi

Authors: Jaeho Lee Tae-Hee Han Min-Ho Park Dae Yool Jung Jeongmin Seo Hong-Kyu Seo Hyunsu Cho Eunhye Kim Jin Chung Sung-Yool Choi Taek-Soo Kim Tae-Woo Lee Seunghyup Yoo Abstract Graphene-based organic light-emitting diodes (OLEDs) have recently emerged as a key element essential in next-generation displays and lighting, mainly due to their promise for highly flexible light sources. However, their efficiency has been, at best, similar to that of conventional, indium tin oxide-based counterparts. We here propose an ideal electrode structure based on a synergetic interplay of high-index TiO2 layers and low-index hole-injection layers sandwiching graphene electrodes, which results in an ideal situation where enhancement by cavity resonance is maximized yet loss to surface plasmon polariton is mitigated. The proposed approach leads to OLEDs exhibiting ultrahigh external quantum efficiency of 40.8 and 62.1% (64.7 and 103% with a half-ball lens) for single- and multi-junction devices, respectively. The OLEDs made on plastics with those electrodes are repeatedly bendable at a radius of 2.3 mm, partly due to the TiO2 layers withstanding flexural strain up to 4% via crack-deflection toughening. Introduction Form-factor advantages such as flexibility and stretchability have brought organic light-emitting diodes (OLEDs) great attention for use in emerging devices like wearable, roll-type or foldable displays that call for mobility, deformability and/or expandability , . Full realization of such potential, however, is non-trivial and requires significant effort in various aspects such as development of low-temperature thin-film transistor technologies and/or flexible encapsulation . Another important aspect to consider for highly flexible displays is to find transparent electrodes (TEs) that can replace indium tin oxide (ITO) electrodes, which typically suffer from limited flexibility and potential cost issues associated with high demand and/or unstable supply . In this regard, several kinds of flexible transparent electrodes such as metal nanowires, carbon-based materials (for example, carbon nanotubes or graphene) and metal grids coupled with other types of TEs were proposed for OLEDs , , , , , . Among them, graphene—an atomically thin sheet of sp2-hybridized carbon atoms—has been regarded as promising because the unique combination of its inherent thinness and superior electrical characteristics is expected to allow for ideal transparent electrodes that have low or no optical haze, smooth surface topology preventing roughness-induced electrical short and degradation, high transmittance, low sheet resistance and a high degree of flexibility all at the same time , . Facile preparation of high-quality graphene on large-scale plastic substrates has recently been demonstrated for transparent electrodes with mass-production-compatible methods by industrial sectors, demonstrating their practical viability , . Both single-layer graphene (SLG) and multi-layer graphene (MLG) were shown to have performance adequate for TEs in OLEDs , . In particular, efficient graphene-based OLEDs were achieved with the help of hole injection layers (HILs) used to overcome their relatively low work function, which is known to be around 4.5 eV. Such injection layers include conducting polymers with gradient electronic properties called a self-organized gradient HIL (GraHIL) , and thin MoO3 buffer layers coupled with conducting polymers of poly(3,4-ethylenedioxythiophene): polystyrene sulfonate (PEDOT:PSS) . Nevertheless, the power efficiency and external quantum efficiency (EQE) of these state-of-the-art graphene-based OLEDs were still on a par with those of ITO-based OLEDs, unless a bulky hemispherical lens was used for outcoupling enhancement . Securing the highest possible efficiency is particularly important for flexible displays applied to highly portable or wearable applications, as they often have to rely on batteries with a fairly limited energy capacity due to constraints in size, weight or form factors. This calls for the development of a device architecture that can maximize the efficiency of graphene-based OLEDs. A key challenge is to develop a structure or methodology to unlock their full optical potential yet retains graphene’s merits in form factors as much as possible. 1 2 3 4 4 5 6 7 8 9 10 11 11 12 13 14 13 14 14 To this end, we here explore a simple electrode architecture based on high-index TiO2 layers and low-index HILs sandwiching graphene electrodes. With the optical design that takes a full advantage of the synergetic collaboration between the high- and low-index layers controlling both cavity resonance enhancement and loss to surface plasmon polariton (SPP), the proposed graphene-based OLEDs exhibit ultrahigh EQE that is unprecedented in those using graphene as a transparent electrode. Furthermore, unusually high resistance of TiO2 to flexural strain is revealed that enables plastic OLEDs that are not only efficient but also highly flexible. Results Optical design of high-efficiency graphene-based OLEDs The efficiency of OLEDs is limited ultimately by a finite outcoupling efficiency that is typically around 20% at best for devices with isotropic emitters. Among various outcoupling enhancing schemes proposed to date , a method based on microcavity resonance has advantages in that it maintains a planar geometry without using any micro/nano structuring or lenticular structure so that the OLEDs made thereof are not subject to electrical shorts and optical artefacts unfavourable for display application (for example, blurring, diffraction and so on) yet can exhibit highly pure, saturated red (R), green (G), blue (B) primary colours with the enhanced efficiency , . With a weak but still significant microcavity effect, ITO-based OLEDs can also be optimized for maximum efficiency, without incurring any optical haze or blurring, simply by adjusting the thickness of ITO layers; in this scheme, first-order, 3 /4 cavity design, where refers to wavelength, is achieved with the open end of the cavity mode placed at the ITO/substrate interface , . In the case of ITO-free, graphene-based OLEDs, however, graphene electrodes cannot be made thick enough to control the cavity length, which makes it challenging to enhance efficiency by the resonance effect . 15 16 17 λ λ 18 19 20 As an alternative measure, one may place a thin layer with high refractive index ( H) underneath graphene layers as shown in in a way similar to dielectric capped thin-metal electrodes , , , . Because light can pass through the graphene with little phase difference due to the inherent thinness of a graphene electrode, 3 /4 cavity design can be achieved conveniently, for example, with the optical thickness of the organic plus injection layers and the high refractive index layer set at approximately 2 /4 and /4, respectively ( ). In particular, the ‘ /4-thick’ high-index layer functions also as a metal-free, dielectric mirror that enables significant reflectance from bottom electrode assembly ( bot) for light incident from organic layers, as can be confirmed in the graphs shown at the bottom of . This provides graphene-based OLEDs with opportunities to enhance their efficiency via the resonance effect, which would not be possible otherwise. It is shown that bot and concomitant resonance enhancement increase with H ( ). The high refractive index is also beneficial in terms of flexibility because the target optical thickness ( H H) can be achieved with low physical thickness H, and because bending-induced crack of a given film generally forms at a lower onset strain with a larger physical thickness . A successful high-index layer in this scheme should thus have as large H as possible. Furthermore, it should be transparent, and, most of all, should not be damaged by a given graphene transfer process. Among various candidates, a sputtered TiO2 layer is transparent in a visible spectral range and has a relatively high refractive index ( =2.5). Unlike other high-index layers we tried, the TiO2 layer turns out to have chemical endurance robust against graphene transfer processes that use solvent, such as acetone or isopropyl alcohol (IPA), allowing for formation of a high-quality graphene electrode on its top ( ). n Fig. 1a 17 20 21 22 λ λ λ Fig. 1b λ R Fig. 1c,d R n Supplementary Fig. 1 n d d 23 n n Supplementary Fig. 2 ( ) Schematic device structure of the proposed OLEDs. ( ) Electromagnetic field intensity distribution (dashed lines) of the OLEDs under study for their respective first-order cavity design. The field-intensity distribution of the graphene-based OLED without TiO2 is also shown for comparison for the case where the thickness values of organic layers are same as the OLEDs with TiO2 but without low-index HILs. ( , ) Calculated maximum external quantum efficiency ( EQE) of graphene-based OLEDs with TiO2 under-layer as a function of TiO2 and organic layer thickness ( TiO2 and org, respectively). Dashed lines represent contour lines for EQE of 10, 20, 30 or 40% as indicated in the graph. ( ) Without a low-index HIL; ( ) with GraHIL as a low-index HIL. On the bottom of each case, the reflectance ( bot) from the bottom electrode assembly (TiO2/graphene/(HIL)) as a whole is presented as a function of TiO2 for light incident from the organic layers. a b c d η d d η c d R d It is noteworthy that a higher order micro-cavity structure (for example, second order, 5 /4 cavity) based on a thick organic stack may also be possible. In this work, however, a first-order cavity design was first chosen because it results in a stronger Purcell effect and fewer waveguide modes than the higher-order design, ultimately leading to larger outcoupling efficiency and EQE , . Challenges still remain for the first-order cavity design because the ideal location of an emission zone is limited at the first antinode from the organic/metal interface, making it difficult to reduce SPP modes through longer emitter-to-metal distance , . An alternative method typically used for SPP reduction is to incorporate internal wavy structures or corrugations , but tolerance to electrical shorts can often be compromised by such approaches. The method we adopt here is to include a layer with a low refractive index, which has been shown to effectively reduce SPP modes , . Fortunately, GraHIL or a self-organized gradient hole-injection layer composed of PEDOT:PSS and tetra-fluoroethylene-perfluoro-3, 6-dioxa-4-methyl-7-octenesulphonic acid copolymer (PFI), which was previously proposed by authors and shown to be very effective in improving hole injection in graphene-based OLEDs , has a refractive index around 1.42 at of 550 nm ( ), which is much lower than those of typical organic layers ( =1.8). With the synergetic interplay of high- and low-index layers sandwiching graphene layers, enhancement of the cavity-resonant effect and reduction of SPP modes may thus be done simultaneously so that the EQE can be significantly boosted even without outcoupling structures. λ 24 25 25 26 27 28 29 13 λ Supplementary Fig. 3 n present the optical simulation results based on an advanced classical electromagnetic theory summarized by Furno . over the whole visible spectral range, and at of 550 nm. The formalism takes into account Purcell factor, dipole orientation effect, and excitations to SPP and waveguide modes. A comprehensive inclusion of all these factors turns out critical for a precise analysis and quantitative design as under- or overestimation can occur when one uses a simplified approach that do not account for one or more of those effects (see ; and for examples). The results indicate that the use of GraHIL redistributes the relative power contents among various modes towards smaller in-plane wave vectors such that the power coupled to SPP modes is reduced and the outcoupled portions are further enhanced ( ). It can also be seen in that use of the TiO2 layer underneath graphene effectively suppresses the amount of the power coupled to waveguided and substrate modes via resonance enhancement. This leads to a significant increase in outcoupling efficiency, although loss to evanescent modes is also shown to increase due to the shift of transverse-electric waveguided modes towards higher normalized in-plane wave vectors ( ). The combined use of high-index (H) TiO2 and low-index (L) GraHIL layers, results in an ideal situation where resonance enhancement is large, yet loss to SPP/evanescent modes is mitigated. In addition, the combined use creates a situation similar to HL-stacks, typically used for multilayer thin-film coating , and even further enhances bot ( ), allowing for additional enhancement in the Fabry–Perot resonance effect. What makes the proposed technology unique and advanced with respect to conventional technologies is indeed the synergetic collaboration of these high- and low-index layers that enable optical management of both resonance effect and SPP loss to the advantage of maximal outcoupling; and additional resonance enhancement via HL stacking. Together with their native compatibility with desired electrical properties (for example, efficient hole injection), all these optical benefits allow one to fully unlock even the hidden potential of what planar OLEDs can truly offer. Contour plots obtained for optically available maximum EQE ( EQE (max)) as a function of organic ( org) and TiO2 thickness ( TiO2) presented in show that EQE (max) can be as high as 44% in devices with both GraHIL and TiO2 layers while it would be limited to 38% without GraHIL or to 31% without TiO2. Optimal conditions do correspond to a situation where H (with H being TiO2) gets close to /(4 H) (55 nm for =550 nm). It can also be noted that EQE (max) of 44% is higher than those expected for optimized, conventional thin-metal based cavity OLEDs with the same emitter. This can be attributed to the fact that the proposed architecture is less subject to the loss due to photon absorption and/or SPP mode excitation within semi-transparent electrodes (see and for comparison with those of conventional thin-metal based cavity OLEDs). Figure 2a,b et al 19 λ Supplementary Fig. 4 Supplementary Table 1 Supplementary Note 1 Supplementary Fig. 5a Supplementary Table 2 Supplementary Fig. 5b 30 R Fig. 1d η d d Fig. 1c,d η d λ n λ η Supplementary Fig. 6 Supplementary Table 3 ( , ) Calculated power dissipation spectra weighted with the emitter spectrum in arbitrary units (mapped as a colour defined in the colour bars.) versus in-plane wave vector: ( ) with TiO2 but without GraHIL; ( ) with both TiO2 and GraHIL. The black dashed lines indicate border lines dividing representative optical modes including outcoupled, substrate-confined (subs), waveguided (wg) and evanescent modes. ( ) Calculated power dissipation versus normalized in-plane wave vector at =550 nm for various electrode structures under study. For the devices with TiO2, TiO2 was fixed at 55 nm and org was chosen for optimal conditions in each case. For the devices without TiO2, org was set at the same value as that of each counterpart in the devices with TiO2. a b a b c λ d d d Device performance of fabricated graphene-based OLEDs Inspired by the simulation results shown above, we fabricated green OLEDs based on phosphorescent emitters of bis(2-(2-pyridinyl- )phenyl-C)(acetylacetonato) iridium (III) (Ir(ppy)2acac) in configuration of glass/anode/(HIL)/OS1/LiF/Al with different anode/(HIL) structures of TiO2/graphene/GraHIL; graphene/GraHIL; and ITO (185 nm)/GraHIL, where OS1 refers to an organic multilayer stack defined in Methods. Experimental results presented in show that the maximum EQE, power efficiency and current efficiency obtained for the TiO2/graphene/GraHIL-based device are as high as 40.8%, 160.3 lm W−1 and 168.4 cd A−1, respectively, while those values are limited to 31.7%, 112.6 lm W−1 and 119.0 cd A−1 for the graphene/GraHIL-based device, and 27.4%, 104.3 lm W−1 and 106.2 cd A−1 for ITO/GraHIL-based device. The observed high EQE in the TiO2/graphene/GraHIL-based device confirms the synergetic role of both TiO2 and GraHIL layers, and is consistent with the simulation results. With a half-ball lens optically coupled to the back of a substrate, the TiO2/graphene/GraHIL-based device exhibits EQE, power efficiency and current efficiency as high as 64.7%, 250.4 lm W−1 and 257.0 cd A−1, respectively, further illustrating the promising potential of the proposed graphene-based OLEDs. We then further extended the proposed scheme to a tandem, multi-junction OLED, in which the net EQE is improved (at the expense of voltage) because it is eventually given as the sum of those of its individual OLEDs ; this approach is popular as the improved EQE generally leads to longer operation lifespan . TiO2/graphene/GraHIL-based multi-junction OLEDs containing a charge-generation layer (CGL) of 2, 9-dimethyl-4, 7-diphenyl-1, 10-phenanthroline (BCP):Li/MoO3 exhibit EQE and power efficiency as high as 62.1% and 120.8 lm W−1 (103.2% and 183.5 lm W−1 with the half-ball lens), demonstrating the state-of-the-art performance and the versatile applicability of the proposed electrode structure. The simulation indicates that the observed value corresponds to the sum of EQE values of equivalent second-order cavity structures (one with the emission zone near second antinode from the cathode; and the other with the emission zone near first antinode from the cathode) with the electrical balance factor of 0.89 (see , and for the detailed structure for multi-junction OLEDs as well as the detailed discussion on the simulation results). N Fig. 3 31 32 Supplementary Fig. 7 Supplementary Table 4 Supplementary Note 2 ( – ) EQE ( ) power efficiency ( ) and current efficiency ( ) versus luminance characteristics of graphene-based OLEDs with and without TiO2 under-layers. The data for ITO-based control OLEDs are also shown for comparison. All the devices in this figure contain GraHIL as low-index hole-injection layers and are based on glass substrates. In ( – ) legends with (L), (M) and (L,M) correspond to the data obtained with a half-ball lens attached, with a multi-junction architecture, and with both a multi-junction architecture and a half-ball lens, respectively. ( ) Current density ( )–voltage ( )–luminance ( ) characteristics. The detailed structure and characteristics of a multi-junction device are represented in . ( ) The picture of an OLED with the proposed TiO2/graphene/GraHIL electrode in operation. For the devices with TiO2, TiO2 was fixed at 55 nm and org was chosen as the optimal condition for maximum EQE. For the devices without TiO2 and the ITO-based devices, org was set at the same value as that of the corresponding devices with TiO2. a c a b c a c d J V L J–V–L Supplementary Fig. 7 e d d d Mechanical properties of the proposed electrodes and OLEDs The success of the graphene-based OLEDs would not be complete unless the advantage of graphene electrodes in flexibility can be utilized to a significant degree. To test the suitability of the proposed structure for flexible OLEDs, OLED devices were made in configurations of (TiO2)/graphene/PEDOT:PSS/OS2/LiF/Al on 50-μm-thick polyethylene terephthalate (PET) substrates (see Methods for the multilayer stack used for flexible OLEDs (OS2)). It is noteworthy that the commonly available PEDOT:PSS (AI4083 by Clevios; =1.56 at λ of 550 nm) can play essentially the same optical role as GraHIL, only with a little compromise in its optical effect due to its refractive index being slightly higher than that of GraHIL (see and ). As can be seen in , the proposed flexible OLEDs exhibit high EQE and power efficiency comparable to their glass-based counterparts (shown as black triangles in ), demonstrating that the proposed structure and its processes can be extended to plastic substrates without problems (see for full summary of the performance of all the devices studied in this work). Furthermore, these devices are shown to remain intact and operate well even after 1,000 bending cycles at a radius of curvature ( C) as small as 2.3 mm. This can be considered remarkable because the proposed OLEDs contains oxide layers, which are often regarded as brittle and thus prone to bending-induced fracture even at relatively low strain. n Supplementary Fig. 8 Supplementary Table 5 Fig. 4 Fig. 4 Table 1 r ( ) EQE versus luminance characteristics of graphene-based flexible OLEDs with and without TiO2 under-layers. Those of an OLED with TiO2 on a glass substrate are also shown for comparison. For devices presented in this figure, 50-nm-thick PEDOT:PSS (AI4083) was used as a low-index HIL in all cases. ( ) Power efficiency versus luminance characteristics. ( ) Luminance versus current density characteristics measured after repeated bending at radius of curvature ( C) of 2.3 mm. Inset: the photograph of an OLED bent at C of 2.3 mm placed on a custom-made cyclic bending tester. The OLED is located on the top side of the substrate in the picture and thus is under tensile strain. ( ) Photograph of the proposed flexible TiO2/graphene OLEDs in operation. a b c r r d