High-Efficiency, Blue, Green, and Near-Infrared Light-Emitting Diodes Based on Triple Cation Perovskite

Transcription

1 High-Efficiency, Blue, Green, and Near-Infrared Light-Emitting Diodes Based on Triple Cation Perovskite Hyeong Pil Kim, Jeongmo Kim, Byung Soon Kim, Hyo-Min Kim, Jeonggi Kim, Abd. Rashid bin Mohd Yusoff,* Jin Jang,* and Mohammad Khaja Nazeeruddin* Perovskite light-emitting diodes (PLEDs) have attracted intense research interest over the past few years; [1 3] however, the brightness, electroluminescence, lifetime, and reproducibility of PLEDs are associated with limitations that should be addressed before commercialization. Here, we explored the effects of the energy level, mobility, and morphology of the charge transport layers and the triple-cation mixed perovskite (formamidinium, methylammonium, and cesium) used in blue, green, and infra-red LEDs. By varying the composition of the halides, spectrally narrow electroluminescence over the entire wavelength is achieved. We fabricated red PLEDs that operate at infrared wavelengths with an unprecedentedly high external quantum efficiency of 9.23% and maximum radiant emittance of W sr 1 cm 2. By optimizing the thickness and morphology of the constituent layers of the devices, the green and blue PLEDs achieved improved electroluminescence (EL) efficiencies (23.7 and cd A 1, respectively), low turnon voltages (2.4 and 3.2 V, respectively), and maximum luminances (19 42 and 3567 cd m 2, respectively). Interestingly, the triple-cation perovskite-based devices exhibited long operational environmental stability, demonstrating that compositionally engineered perovskite is an attractive material for PLEDs; this finding should foster the development of next-generation displays and solid-state lighting technologies. Metal halide perovskites featuring unique and attractive properties, such as wide absorption and narrow emission spectral bandwidth, for use in optoelectronic devices have recently received numerous attention. [1 6] In addition, the emission wavelength can be tuned by adjusting the bandgap via the halide composition. Therefore, perovskite is considered one of the most practical candidate materials for use in next-generation light-emitting devices and solid-state lighting. [1 9] Since the first PLEDs were demonstrated, [1] the device performance H. P. Kim, J. M. Kim, B. S. Kim, H.-M. Kim, J. Kim, Prof. A. R. B. M. Yusoff, Prof. J. Jang Advanced Display Research Center Department of Information Display Kyung Hee University Dongdaemoon-gu, Seoul, South Korea abdr@khu.ac.kr; jjang@khu.ac.kr Prof. A. R. B. M. Yusoff, Prof. M. K. Nazeeruddin Group for Molecular Engineering of Functional Materials Institute of Chemical Sciences and Engineering École Polytechnique Fédérale de Lausanne CH-1951 Sion, Switzerland mdkhaja.nazeeruddin@epfl.ch DOI: 1.12/adom has improved tremendously, achieving high brightness ( 2 cd m 2 ) and external EL quantum efficiency (η EQE > 8%) as a result of multilateral efforts based on the fundamental understanding of the device physics as well as the development of advanced materials and interfacial engineering and device architectures. [2 4] Despite these recent advances, there are still substantial limitations (including low device efficiency, low η EQE and poor stability) in the use of PLEDs for applications of displays or solid-state lighting. Herein, we introduce robustly processable, highly efficient near-infrared (NIR), green and blue PLEDs with an inverted device structure that uses a low-temperature sputtering zinc oxide nanoparticle (ZnO NP) film as an electron transporting layer (ETL) and an electron injection layer (EIL). The advantages of the ZnO NP film are as follows: (i) the inherent electron injection and transport property enables the material to act as a common EIL/ETL, (ii) the film provides a robust platform for consecutive layer deposition, and (iii) the film enables the systematic engineering of HTLs featuring conventional organic materials with proven performance. Therefore, independent optimization of the charge injection, transport, and light emission for efficient PLEDs becomes feasible. For the emitter layer, we propose the use of a triple-cation perovskite that incorporates cesium (Cs) cation, into a mixed cation perovskite : MA.17 FA.83 Pb(Br x I 1 x ) 3. [1,11] By employing a common device structure and process and by varying the halide composition, we fabricated a color-saturated NIR (EL λ max. = 75 nm, FWHM = 15 nm) PLED with a maximum light output of W sr 1 cm 2 and η EQE of 9.23%. In addition, we fabricated a green PLED with an EL λ max. = 569 nm, FWHM = 32 nm and a blue PLED with an EL λ max. = 475 nm, FWHM = 28 nm. The green and blue PLEDs demonstrate luminance values of and 3567 cd m 2, maximum η EQE values of 7.3 and 1.7% and low turn-on voltages (V T ) of 2.4 and 3.2 V, respectively, which directly correspond to the optical bandgap of perovskites. The significant improvement in the balanced injection and transport of charge carriers into perovskites enables exciton formation and efficient recombination within the perovskite emissive layers (EMLs). The balanced injection of charges leads to high efficiency and brightness compared to previous reports of devices with conventional or inverted device architecture. [1 9] Specially, the blue PLEDs achieve a remarkably high luminous efficacy of 6.6 lm W 1, and the green and NIR PLEDs are among the highest value yet reported in the literature. Moreover, we demonstrate uniform defect/pinhole-free EL emission from the perovskite over a surface area of 2 inch, suitable for potential use in perovskite-based large-area flat panel displays. The devices also exhibit a long operational lifetime under 217 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com (1 of 9) 1692

2 a c Normalized XRD Intensity θ (Degree) Normalized PL Intensity b d Urbach Energy (mev) Normalized Absorption Intensity Iodide content Iodide Content FWHM (x 1 3 degree, ev) e f Figure 1. Optical, crystallographic, and morphological triple cation perovskite. a) UV vis absorption spectra; b) normalized photoluminescence spectra for Cs 1 (MA.17 FA.83 ) (1 x) Pb(Br x I 1 x ) 3 perovskite thin films with different iodide bromide ratios as indicated. Excitation for PL was performed with a pulsed laser system at 3.1 ev photon energy and 1 fs pulse length, and PL spectra have been normalized to the peak emission. c) XRD patterns showing the evolution of the (1) reflection as a function of composition for Cs 1 (MA.17 FA.83 ) (1 x) Pb(Br x I 1 x ) 3 perovskite thin films showing diffraction pattern shift to lower scattering angle, 2θ with increasing iodide content. d) The Urbach energy, FWHM of PL emission peak and (1) XRD diffraction peak versus iodide loading amount. e,f) Surface scanning electron microscopic images of e) Cs 1 (MA.17 FA.83 ) (1 x) PbI 3 and f) Cs 1 (MA.17 FA.83 ) (1 x) PbBr 2.97 I.3 spin-coated onto ZnO NPs films. electrical excitation, where the light output of NIR devices almost vanished after 3 min with an initial radiance of W sr 1 cm 2 ; for the green and blue devices, the lifetime is greater than 475 at an initial luminance of and 264 min at an initial luminance of cd m 2, respectively. Figure 1a shows XRD spectra of the perovskite films (Cs 1 ) with various iodide (I) fractions; these spectra indicate that the cubic phase space group structure Pm3m exists for all values of loading content. As the I fraction increases, the (1) diffraction peak shifts to a lower scattering angle, as anticipated when bromide (Br) ions (ionic radius of 1.96 Å) are replaced by larger I ions (ionic radius of 2.2 Å), indicating an increase in lattice parameter. From the XRD patterns, we observed a single perovskite phase without any noticeable widening of the (1) reflection. For the lower fractions (I.33 and I.44 ), the relative intensity of the (11), (21), and (211) peaks are even further reduced, indicating that these films are both oriented and in better agreement with a cubic rather than orthorhombic perovskite structure. Notably, only the (1) and (11) perovskite peaks are present as the I fraction decreases to.22. A comprehensive description regarding the structural analysis, morphology, and kinetics of Cs into the mixed cation perovskite (MA.17 FA.83 ) (1 x) Pb(I.83 Br.17 ) 3 is shown in Figures S1 S5 and Table S1 (Supporting Information). To gain deeper insights into the incorporation of Br and I, we investigated the absorption and steady-state photoluminescence properties of the perovskite. By varying the Br to I ratio, Cs 1 (MA.17 FA.83 ) (1 x) Pb(Br x I 1 x ) 3 [ 1], we tuned the bandgap of perovskite (from 1.6 to 2.3 ev) and the electronic quality of the perovskite film in terms of energetic disorder 1692 (2 of 9) wileyonlinelibrary.com 217 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3 and charge-carrier lifetime (Figure 1b,c, respectively). The significant redshift of the absorbance spectra, shown in Figure 1b, corresponds to the bandgap shifting with increasing I content, from 546 nm for the I precursor solution to the sharp optical band-edge of 783 nm for the I.99 precursor solution. This observation indicates that the bandgap decreases with increasing I fraction in the perovskite systems. A similar redshift in the PL spectra was also observed (Figure 1c), where a linear trend was found for I.66 to I.99 ; this observation indicates faster kinetics than those of films with low loading of I. Figure 1d shows the Urbach energy versus the I fraction versus the FWHM of the PL emission. For the I film, the Urbach energy is 47 mev and decreases with increasing I loading; the Urbach energy reaches the lowest value of 28 mev for I.99. A higher Urbach energy for the I film demonstrates that the magnitude of the disorder is significantly higher than that of the film loaded with I.99. An identical observation was made in the XRD FWHM (1), with a minimum and maximum of.262 for the I film and.16 for the I.99 film, respectively. In addition, the PL FWHM exhibits a minimum for I (26.29) and a maximum for I.99 (65.56). The similarity between the FWHM of the PL emission and the XRD diffraction peak suggests that the photoexcited species experienced random energetic disorder. The magnified surface SEM images of Cs 1 (MA.17 FA.83 ) (1 x) PbBr 2.97 I.3 and Cs 1 (MA.17 FA.83 ) (1 x) PbI 3 films on highly uniform ZnO NPs-coated ITO (indium tin oxide) glass are illustrated in Figure 1e. The Cs 1 (MA.17 FA.83 ) (1 x) PbI 3 film has a layered structure with relatively smaller crystals alongside an unclear grain boundary, whereas the Cs 1 (MA.17 FA.83 ) (1 x) PbBr 2.97 I.3 film features larger crystals with pronounced grain boundaries and compact layer formation on top of ZnO NPs. The relatively large crystal size is also indicated by the XRD patterns in Figure 1a. Optimization of the ETL and HTL plays a crucial role in terms of device performance. [12,13] Because the ZnO NPs layer can efficiently inject and transport electrons into the perovskite EML and is required to reduce the driving voltage, the performance of PLEDs depends greatly on the positioning of the HTL s highest occupied molecular orbital (HOMO) energy level. Thus, to confirm the validity of our hypothesis, a series of experiments were conducted to evaluate the dependence of the PLED performance on the hole injection capability by introducing several HTLs with different HOMO levels: N,N - bis(1-naphthyl)-n,n -diphenylbenzidine (α-npd, 5.4 ev), [14] poly(4-butylphenyl-diphenyl-amine) (poly-tpd, 5.2 ev), [15] and 2,7-bis(carbazol-9-yl)-9,9-spirobifluorene (spiro-2cbp, 5. ev). [16] In Figure 2a, we present the maximum η EQE values of NIR, green, and blue PLEDs versus the HOMO level of the HTLs of the PLEDs using the following device configuration: glass/ ITO/ZnO NPs/perovskite/HTL/MoO 3 /Al. We observed a similar trend for all devices, in which the maximum η EQE can be achieved when a HTL with lower HOMO levels is used. In this case, electron transport and hole transport are more balanced, resulting in efficient radiative recombination. For example, devices employing α-npd outperformed other devices, whereas the devices with spiro-2cbp demonstrated low efficiency because this HTL has the highest HOMO level, resulting in poor injection of holes into the perovskite EML. In addition, high V T and low luminance/efficiency were observed when spiro-2cbp was used because inefficient emission from the spiro-2cbp was involved in the EL process of the PLEDs. We note the following reasons for high η EQE : (i) the lower HOMO and higher LUMO levels of α-npd than those of poly-tpd and spiro-2cbp, enhancing hole injection and reducing recombination at the interface because the higher LUMO levels provide effective electron-blocking capability and (ii) efficient electron injection from the ETL resulting from the activationless barrier at the ZnO NPs/perovskite interface. Also, the hole mobility of α NPD (μ h cm 2 V 1 s 1 ) [17a] is higher than that of poly-tpd (μ h cm 2 V 1 s 1 ) [17b] and comparable to spiro-2cbp (μ h cm 2 V 1 s 1 ). In addition, the well-matched energy level of molybdenum trioxide (MoO 3 ) is responsible for efficient hole injection from the Al electrode (Figure S6, Supporting Information). The corresponding device performances are shown in Figure 2b f, in which the differences in V T, luminous efficiency, luminous efficacy, and light output are caused by differences in the energy barrier between the perovskite EML and the HTL. The maximum brightness values of α-npd-based PLEDs are 716 cd m 2 at a V T of 2.2 V (green) and 1117 cd m 2 at a V T of 3.4 V (blue), and the maximum radiance emittance is W sr 1 cm 2 at a V T of 2.8 V (NIR). Having established a relationship between the HOMO level and the device performance, we note that the thickness of the perovskite EML plays a decisive role in the current and power efficiencies of the PLEDs. Atomic force microscopy images shown in Figure S7 (Supporting Information) illustrate that the perovskite films deposited by the two-step spin-coating method are highly dense and pinhole free with relatively uniform grain sizes over the entire substrate. This result is completely different than that of MAPbI 3 films deposited by the same technique but exhibiting less desirable surface morphology. [18 21] To reveal the relationship between the grain size and the performance of PLEDs, a series of separate experiments were performed in which perovskites of different grain sizes were sandwiched between ZnO NPs and a stack of α-npd/moo 3 following the previous device architecture. Figure 3a illustrates the current density versus the driving voltage versus the luminance characteristics of green-based PLEDs featuring different perovskite grain sizes. All the parameters increased with increasing grain size; the highest luminous efficiency of cd A 1 was obtained for PLEDs using the grain size of 3 nm: luminance of cd cm 2, luminous efficacy of lm W 1 and η EQE of 4%. The maximum luminances achieved were 8986, 11 25, 9881, and cd m 2 at 8 V for the devices using perovskites of grain sizes of 25, 3, 35, and 4 nm and the values of V T of 2.5, 2.9, 2.9, and 3.9 V, respectively. Uniform perovskite coverage is likely to lead to minimization of V T. As the grain size increases from 25 to 3 nm, the perovskite films become uniform, flat and dense with high reproducibility. When the grain size is further increased to 35 and 4 nm, microsized rods are formed on top of the perovskite films, as demonstrated in Figure S8 (Supporting Information); the formation of microsized rods is due to the presence of Cs. The formation of Cs on the top of the 217 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com (3 of 9) 1692

4 a c e Maximum EQE (%) NIR Green Blue polyspiro-2cbp TPD NPD HOMO Level (ev) Green Emission 1 1 spiro-2cbp poly-tpd 1-1 α-npd Blue Emission spiro-2cbp poly-tpd 1-1 α-npd b d f Current Efficiency (cd A -1 ) Current Efficiency (cd A -1 ) NIR emission spiro-2cbp poly-tpd α-npd Radiance (W sr -1 cm -2 ) Power Efficiency (lm W -1 ) Power Efficiency (lm W -1 ) Figure 2. Characteristics of NIR-, green-, and blue-colored PLEDs. a) Maximum EQE of NIR-, green-, and blue-emission PLEDs using various HTLs with various HOMO levels. b) Characteristics of NIR-emission PLEDs using various HTLs. b) Current voltage radiance. c,d) Characteristics of greenemission PLEDs using various HTLs. c) Current voltage luminance. d) Current efficiency luminance power efficiency. e,f) Characteristics of blueemission PLEDs using various HTLs. e) Current voltage luminance. f) Current efficiency luminance power efficiency. perovskite film presumably retards the contact between perovskite and HTL, thereby possibly deteriorating the device performance and reliability. Figure 3b shows the current and power efficiencies for green-based PLEDs as a function of various grain sizes, demonstrating that the 3-nm grain size is optimal. Unfortunately, such a thick perovskite layer or large grain size alone does not improve the degree of electron hole recombination significantly, ultimately limiting the EL performance. A smaller grain size (probably 25 nm), however, results in a higher leakage current through the perovskite EML without radiative recombination because of the presence of defects, the small grains and the incomplete surface morphology of the perovskite film. At any given grain size, there is no easily discernible contribution from the oxide ETL/polymer HTL in the EL spectra of the PLEDs without deterioration of the color purity of the PLEDs (shown in Figure 3b inset). Figure 3b also reveals that no leakage of holes into the ETL occurs in the green PLEDs. By combining the results in Figure 3a,b, we determined that 3 nm is the optimal thickness of the perovskite layer for green PLEDs and is sufficient to enable efficient recombination of electron hole pairs to form excitons directly inside the perovskite layer. Figure 3c,d exhibits the difference of the EL spectra obtained from the blue and NIR PLEDs with various perovskite layer thicknesses, where the optimum thickness for the blue and NIR PLEDs is 22 and 27 nm, respectively. From these optimization experiments, we can safely conclude that large grain size does not always lead to a better device performance. A schematic diagram featuring a multilayered architecture of the patterned indium tin-oxide (ITO)/ZnO NPs (5 nm)/perovskite (27 or 3 or 22 nm)/α-npd (4 nm)/moo 3 (4 nm)/ Al (12 nm) is shown in Figure 4a. The energy level diagram of the constituent layers is depicted in Figure 4b, which also shows the complete PLEDs, in which the perovskite EML and the metal oxides are represented by the corresponding valence band (VB) and conduction band (CB), and the organic HTL is denoted by its HOMO and LUMO. The sputtered ZnO NPs of sizes of 1 4 nm (see Figure S9, Supporting Information) have several advantages: (i) the capability to avoid solvent penetration during layer-by-layer deposition, (ii) solution processability, (iii) high transparency in the visible regime and modest 1692 (4 of 9) wileyonlinelibrary.com 217 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

5 a c nm 3 nm 1 35 nm 4 nm nm 26 nm 27 nm b d Current Efficiency (cd A -1 ) nm 21 nm 22 nm EL Intensity (a.u.) 2 nm 18 nm 16 nm Power Efficiency (lm W -1 ) Figure 3. Effect of perovskite layer thickness on electroluminescence performance of the PLEDs. Device characteristics of green-colored PLED with different triple cation thicknesses. a) Current voltage luminance. b) Current efficiency luminance power efficiency. Inset, EL spectra of green-colored PLEDs operated at 1 cd m 2 with different perovskite thicknesses. c,d) EL spectra of blue and NIR-colored PLEDs with various perovskite thicknesses operated at 1 cd m 2 and 5 W sr 1 cm 2, respectively. electron mobility (μ e cm 2 V 1 s 1, obtained from field-effect-transistor measurements; Figure S1, Supporting Information) compared to commonly used organic EIL/ETL, (iv) deep conduction band ( 4.1 ev, see Figure S11, Supporting Information) (which is the same as the CB of the perovskite layer spin-coated from anhydrous dimethylformamide:dimethy lsulfoxide (DMF:DMSO) solutions) that facilitates the electron injection and transport from the ITO electrode into the perovskite EML, (v) a low-lying valence band of 7.4 ev that can efficiently suppress the hole leakage current to the ETL from the HTL through perovskite layer, and, most importantly, (vi) complete surface coverage upon perovskite deposition. Figure S12 (Supporting Information) shows the mechanically smooth and homogenous (root-mean-square roughness below 1 nm) surface of the ZnO NPs layer. The HTL, α-npd, has a HOMO level of 5.4 ev (see Figure S12, Supporting Information), similar to the VB of the perovskite layer spin-coated from anhydrous solutions; thus, the HTL can form an Ohmic-like contact that allows efficient hole transport to perovskite EML because of the negligible energy barrier. Furthermore, α-npd, in conjunction with the high energy level, efficiently blocks electron back-diffusion to the Al and eliminates electroluminescence, thereby guaranteeing the color purity of the EL spectra. A relatively high hole mobility of MoO 3 (μ h cm 2 V 1 s 1, obtained from field-effect-transistor measurements, Figure S14, Supporting Information) is highly desirable to achieve low V T and high power efficiency. Conclusively, a suitable combination of metal oxide and polymer charge transport layers is highly favoured to have free energy-band offsets relative to the perovskite layer; as a result, electrically injected electrons and holes into the perovskite EML are balanced. Current density-luminance, current efficiency, power efficiency, and current density-radiant emittance are plotted in Figure 4c e as a function of voltage, where the maximum luminance, current efficiency, and power efficiency for green (blue) PLEDs are cd m 2 (3,567 cd m 2 ), 23.7 cd A 1 (11.31 cd A 1 ), and lm W 1 (6.6 lm W 1 ), respectively (Table 1); to the best of our knowledge, the maximum power efficiency represents a record value for blue PLEDs. We estimated the η EQE of all emissions by measuring the light intensity in the forward direction and converting it into total external emission by assuming the external emission profile to be Lambertian, [22] and all emitting devices exhibit almost ideal Lambertian distributions [23] (Figure S15, Supporting Information). The calculated η EQE values reach 7.3 and 1.7%, respectively, and their V T values are found to be 2.4 and 3.2 V, respectively. Although all device structures are optimized, the higher bandgap of perovskite EML produces higher operating voltage and lowered current density, particularly for blue devices. Moreover, the NIR device turns on at 2.8 V (1 mw sr 1 cm 2 ) and has a maximum radiance of W sr 1 cm 2 at 8 V (Figure 4e) and maximum η EQE of 9.23%, which is slightly higher than findings in the previous work 3 and also the highest so far for NIR devices. High radiant emittance and η EQE indicate the effectiveness of perov skite in achieving the balanced carrier transport. At high-brightness display (1 65 cd m 2 ), the green and blue PLEDs exhibit relatively stable and high luminous efficiencies of and 8.56 cd A 1, respectively, at 1 cd m 2, and these value increase to and cd A 1, respectively, at 65 cd m 2. The efficiency rollovers for green and blue PLEDs are most likely due to the accumulation of space charge within the perovskite EML regime. The high luminous efficiency, 217 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com (5 of 9) 1692

6 a b c d 1 2 e f g Current Efficiency (cd A -1 ) j V 6 V 4 V 3 V Power Efficiency (lm W -1 ) h V 6 V 4 V 3 V NIR k Radiance (W sr -1 cm -2 ) Green Blue i Iodide content V 6V 4V 3V l Figure 4. Perovskite light-emitting diode design and electroluminescence performance of ZnO NP-based PLEDs with green, blue, and NIR emissions. a) Schematic diagram of the light-emitting diode illustrating the ITO cathode, the ZnO NPs ETL, the perovskite EML, the α-npd HTL, the MoO 3 HIL, and the aluminium anode. b) The flat band energy band diagram of the light-emitting diode. HOMO and LUMO energy levels of α-npd and MoO 3. Conduction band and valence bands of ZnO NPs and perovskite. The HOMO and LUMO energy levels of the materials involved are extracted from UPS spectra (see the Supporting Information). c) Current voltage luminance. The insets illustrate the photograph of green- and blue-colored PLEDs. d) Current efficiency luminance power efficiency. e) Current voltage radiance of NIR PLEDs. The inset displays the photograph of NIR-PLEDs. f) Normalized EL spectra of Cs 1 (MA.17 FA.83 ) (1 x) Pb(Br x I 1 x ) 3 perovskite thin film based LEDs with different iodide bromide ratios as indicated and measured at 7 K. g i) Electroluminescence spectra of the g) green-, h) blue-, and i) NIR-colored PLEDs operating at various voltages. j) CIE coordinates (NTSC) of green-, blue-, and NIR-colored PLEDs. k,l) Picture of large scale devices. k) Green-colored (1 inch) and l) NIR-colored (1 cm 2 ) PLEDs operating at 5 and 8 V, respectively. luminous efficacy, and radiant emittance achieved in these systems can be explained by three main advantages of the triplecation perovskite. First, solution processing allowed us to form a continuous thin film comprising grains of uniform size over the entire substrate, free of pinholes and defects; deposition of such a film has been reported as a considerable challenge for the MAPbX 3 films. [19] Reference devices without the perovskite layer (ITO/ZnO NPs/α-NPD/MoO 3 /Al) were also developed to probe the consequences of pinholes on device performance (Figure S16, Supporting Information). Second, the reduced bandgap of 1.6 ev in the perovskite-bearing small barrier for charge injection allows a large amount of current to be easily 1692 (6 of 9) wileyonlinelibrary.com 217 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

7 Table 1. Performance characteristics of multicolored perovskite light-emitting diodes with α-npd as the HTL. Color λ max [nm] FWHM [nm] injected into the device at lower applied bias. We would expect a corresponding increase in V T to accompany a larger bandgap. Finally, the ZnO NP film has outstanding capability for electron transport due to high VB and blocking exciton quenching. These features are very important to achieve high efficiency in our devices. We also conducted a separate experiments on green PLEDs using various ETLs (TiO 2 (titanium dioxide), ZnO:Mg (zinc oxide doped magnesium) and ZTO (zinc doped tin oxide)) and found that ZnO NPs provide superior performance to that of the TiO 2, ZnO:Mg and ZTO-based ETL devices in terms of luminance and power efficiencies (Figures S17 and S18 and Table S2, Supporting Information). Figure 4f displays the capability to manipulate the emission wavelength and bandgap of the perovskite films; these characteristics were measured at low temperature (7 K). The EL emission wavelength monotonically increases from 555 nm for I to 777 nm for I.99. The driving voltage dependence of the EL spectra is visualized in Figure 4g I, illustrating that light emission from the NIR, green, and blue PLEDs was achieved at a driving voltage as low as 3. V. Figure 4f h also reveal that electrons and holes can be efficiently injected into the perovskite EML at low driving voltages. At lower driving voltages, the electrons and holes accumulate at the interfaces of ZnO NPs/ perovskite and perovskite/α-npd, respectively, due to the negligible energy offset at these interfaces. As the driving voltage increases, the recombination zone shifts toward the EML and confines injected carriers in the EML for radiative recombination of electrons and holes; additionally, exciton diffusion to the adjacent carrier transport layers is halted in the devices, leading to enhanced emission from the perovskite layer. With a higher driving voltage, electrons accumulating at the interface possess higher energy after absorbing the energy released from the interfacial recombination of an electron hole pair. The resulting high-energy electron can overcome the injection barrier and recombine with a hole inside the perovskite EML to emit a photon. Figure 4g i also show negligible shift from (.91,.781) at 3 V (2.7 ma cm 2, 29 cd m 2 ) to (.93,.794) at 7 V ( ma cm 2, cd m 2 ) for green PLEDs, from (.137,.264) at 3 V (.4 ma cm 2,.15 cd m 2 ) to (.138,.251) at 7 V (16.67 ma cm 2, cd m 2 ) for blue PLEDs, and (.697,.288) at 3 V (.17 ma cm 2,.7 W sr 1 cm 2 ) to (.695,.29) at 7 V (46.1 ma cm 2, W Sr 1 cm 2 ) for NIR PLEDs. We also note that there is no detectable contribution from oxide ETL/polymer HTL in the EL spectra of the PLEDs. The Commission Internationale de l Enclairage (CIE) color coordinates are (.92,.779), (.138,.268), and (.78,.291) for green, blue, and NIR devices, respectively (Figure 4j). Note that the ideal green, blue, and NIR PLEDs for display V T [V] CE max [cd A 1 ] L max [cd m 2 ] PE max [lm W 1 ] Green ,.779 Blue ,.268 Color λ max [nm] FWHM [nm] V T [V] R max [W sr 1 cm 2 ] J max [ma cm 2 ] NIR ,.287 EQE [%] EQE [%] CIE [x, y] CIE [x,y] application should have narrow bandwidth and a wavelength, such that their color coordinates on the CIE chromaticity diagram lie outside the current National Television System Committee (NTSC), as depicted in Figure 4i. This restriction implies that using these saturated green, blue, and NIR PLEDs for display application would demonstrate a considerably large color triangle on the CIE color coordinates. To demonstrate manufacturing scalability for large-area display technologies, we have fabricated 1 inch (diagonal) green (Figure 4k) and 1 cm 2 red PLEDs (Figure 4l) operated at 8 and 5 V, respectively; these devices are the largest-area PLEDs reported to date. Since their first demonstration, PLEDs have been anticipated to pave the way for cost-effective, flexible, and large-scale displays for future applications. Flexible PLEDs displays are appealing because they enable the development of thin, ultralight products in various shapes to suit desired applications. Therefore, to demonstrate the versatility of triple-cation perovskite, we developed a large-scale (2 inch diagonal) flexible green PLED screen based on colorless polyimide. Video S1 (Supporting Information) shows a detached 2-inch flexible PLED screen. Videos S2 and S3 (Supporting Information) demonstrate bending measurements of green PLEDs at bending radii of 2 and 4 mm, respectively. Video S4 (Supporting Information) illustrates the folding test of our flexible PLEDs on CPI substrate using CPI encapsulation at a folding angle of 9. After 1 bending (at different bending radii) and folding cycles, the current density and brightness decreased significantly from its initial values. The combination of high-performance PLEDs, optimized structure and tuneable perovskite allow the fabrication of the largest flexible green PLED devices ever reported, with a dimension of 2 inches, constructed on a CPI substrate. Before we can seriously take the next logical steps toward commercialization, other production aspects, including fabrication yield and cost, must be addressed. These results could serve as a platform for future PLED developments. Despite the huge progress in PLEDs, the long-term stability of these devices under operating conditions remains a challenge, and we evaluated the green and blue devices at room temperature and a constant current density ( 3 ma cm 2 ) for initial luminance of and cd m 2, respectively, as shown in Figure 5. In addition, we fixed the voltage for initial EL radiance of W sr 1 cm 2 to collect the light output over time. Because of the encapsulation method used with all devices designed in this study, oxygen/moisture penetration is negligible and can be ruled out as a mechanism for increasing current density, voltage or luminance. Thus, the stability study of these devices at the same current density and fixed voltage suggest that the absolute value of the device lifetime is greatly 217 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com (7 of 9) 1692

8 a Time (min) dependent on the nature of the materials used. Figure 5a indicates that the light output decreases significantly over the first 15 min and becomes nonluminescent shortly after 3 min have passed. Note that the decrease in light output is accompanied by an enormous increase in current density, indicating that the perovskite EML becomes more conducting over time, likely because of ion migration in the perovskite layer under an applied electrical bias. [24] However, a distinctive difference between green and blue devices is observed. Both devices illustrate comparable roll-off trends regarding overshoot in luminance and voltage, although the luminance of the green PLEDs is almost twice that of the blue PLEDs (Figure 5b,c). For the first 135 min (green), the luminance increases from to 1,335.7 cd m 2 and then slowly decreases to 8.7 cd m 2 after 475 min. In contrast, for blue devices, during the first 25 min, the luminance increases to cd m 2 and then rapidly decreases over time, reaching 88.5 cd m 2 after 15 min. Additionally, the values of V T are slightly improved from 2.6 to 2.91 V for green and from 3.3 to 6.21 V for blue PLEDs after 475 and 15 min, respectively. The stability of triple-cation perovskite in comparison with mixed-cation perovskite (MAPbBr 3 ) and quasi 2D perovskite (CsPbBr 3 ) employing a similar device structure, i.e., ITO/ZnO NPs/perovskite/α-NPD/MoO 3 /Au is shown in Figure S19 in the Supporting Information. The lifetime measurements reveal that the triple cation perovskite is more stable than MAPbBr 3 and CsPbBr 3 in terms of the operational stability of the devices. These results indicate that the introduction of Cs is favourable for the long-term stability of NIR, green, and blue devices. The short lifetime demonstrated by the MAPbBr 3 and CsPbBr 3 is probably related to the intrinsic instability. Further lifetime improvements are anticipated by finding suitable charge transport layer combinations with conduction properties similar to the materials used here; such efforts to improve the lifetime of PLEDs are currently underway. In conclusion, we have reported high performance in both device characteristics and lifetime for NIR, green, and blue PLEDs. Our results demonstrate that tailoring the halide stoichiometry of the perovskite EML and careful selection of the ETL and HTL are crucially important in achieving such high performance PLEDs. The high-quality, defect-free, and complete surface coverage of the perovskite films shown here present high photoluminescence efficiencies, narrow emission peaks, high charge injection/transport and high radiative Voltage (V) b Time (min) Figure 5. Lifetime of our encapsulated multicolored PLEDs. a) Lifetime characteristics of the green- and blue-colored PLEDs. b) Lifetime characteristics of the NIR-colored PLEDs at fixed voltage. Radiance (W sr -1 cm -2 ) recombination. We believe that the work revealed in this study is an important step toward the realization of next-generation solid-state lighting and devices for use in fullcolor large area display applications. Supporting Information Supporting Information is available from the Wiley Online Library or from the author. Acknowledgements A.R.B.M.Y., H.-M.K., J.K., B.S.K., H.P.K., J.K., and J.J would like to thank MOTIE (Ministry of Trade, Industry and Energy (Grant No ) and KDRC (Korea Display Research Corporation) support program for the development of future device technology for display industry. M.K.N acknowledges financial support from European Union Seventh Framework Programme [FP7/27-213] under Grant Agreement No of the MESO project. Received: November 4, 216 Revised: December 8, 216 Published online: February 2, 217 [1] M. Era, S. Morimoto, T. Tsutsui, S. Saito, Appl. Phys. Lett. 1994, 65, 676. [2] H. Cho, S-H. Jeong, M.-H. Park, Y.-H. Kim, C. Wolf, C-L. Lee, J. H. Heo, A. Sadhanala, N. S. Myoung, S. Yoo, S. H. Im, R. H. Friend, T. -W. Lee, Science 216, 35, [3] M. Yuan, L. N. Quan, R. Comin, G. Walters, R. Sabatini, O. Woznyy, S. Hoogland, Y. Zhao, E. M. Beauregard, P. Kanjanaboos, Z. Lu, D. H. Kim, E. H. Sargent, Nat. Nanotechnol. 216, 11, 872. [4] J. Wang, N. Wang, Y. Jin, J. Si, Z-K. Tan, H. Du, L. Cheng, X. Dai, S. Bai, H. He, Z. Ye, M. L. Lai, R. H. Friend, W. Huang, Adv. Mater. 215, 27, [5] Y.-H. Kim, H. Cho, J. H. Heo, T.-S. Kim, N. S. Myoung, C.-L. Lee, S. H. Im, T.-W. Lee, Adv. Mater. 215, 27, [6] A. Sadhanala, A. Kumar, S. Pathank, A. Rao, U. Steiner, N. C. Greenham, H. J. Snaith, R. H. Friend, Adv. Electron. Mater. 215, 1, 158. [7] Y.-K. Chih, J.-C. Wang, R.-T. Yang, C.-C. Liu, Y.-C. Chang, Y.-S. Fu, W.-C. Lai, P. Chen, T-C. Wen, Y.-C. Huang, C.-S. Tsao, T.-F. Guo, Adv. Mater. 216, 28, [8] Y. Ling, Y. Tian, X. Wang, J. C. Wang, J. M. Knox, F. P. -Orive, Y. Du, L. Tan, K. Hanson, B. W. Ma, H. Gao, Adv. Mater. 216, DOI: 1.12/adma [9] J. C. Yu, D. Wo. Kim, D. B. Kim, E. D. Jung, J. H. Park, A-Y. Lee, B. R. Lee, D. D. Nuzzo, R. H. Friend, M. H. Song, Adv. Mater. 216, DOI: 1.12/adma [1] M. Saliba, T. Matsui, J-Y. Seo, K. Domanski, J. P. C. -Baena, M. K. Nazeeruddin, S. M. Zakeeruddin, W. Tress, A. Abate, A. Hagfeldt, M. Graetzel, Energy Environ. Sci. 216, DOI: 1.139/ C5EE3874J. [11] A. R. B. M. Yusoff, H. P. Kim, X. Li, J. M. Kim, J. Jang, M. K. Nazeeruddin, Adv. Mater. 216, DOI: 1.12/adma [12] K. -Y. Cheng, R. Anthony, U. R. Kortshagen, R. J. Holmes, Nano Lett. 211, 11, [13] N. K. Patel, S. Cina, J. H. Burroughes, IEEE J. Select. Top. Quantum Electron. 22, 8, 346. [14] J. Kwak, W. K. Bae, D. Lee, I. Park, J. Lim, M. Park, H. Cho, H. Woo, D. Y. Yoon, K. Char, S. Lee, C. Lee, Nano Lett. 212, 12, (8 of 9) wileyonlinelibrary.com 217 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

9 [15] D. H. Kim, T. W. Kim, Opt. Express 215, 23, [16] H. Lee, J. Kwak, C. M. Kang, Y. Y. Lyu, K. Char, C. Lee, Opt. Express 215, 4, [17] a) T. Noguchi, M. Mishima, US B2, 211; b) Q. Huang, J. Pan, Y. Zhang, J. Chen, Z. tao, C. He, K. Zhou, Y. Tu, W. Lei, Opt. Express 216, 24, [18] P. Gao, M. Graetzel, M. K. Nazeeruddin, Energy Environ. Sci. 214, 7, [19] G. E. Eperon, V. M. Burlakov, P. Docampo, A. Goriely, H. J. Snaith, Adv. Funct. Mater. 214, 24, 151. [2] M. Xiao, F. Huang, W. Huang, Y. Dkhissi, Y. Zhu, J. Etheridge, A. G. -Weale, U. Bach, Y. B. Cheng, L. Spiccia, Angew. Chem. 214, 126, 156. [21] Z. Xiao, Q. Dong, C. Bi, Y. Shao, Y. Yuan, J. Huang, Adv. Mater. 214, 26, 653. [22] Z. Wang, T. Cheng, F. Wang, S. Dai, Z. Tan, Small 216, DOI: 1.12/smll [23] N. C. Greenham, R. H. Friend, D. D. C. Bradley, Adv. Mater. 1994, 6, 491. [24] Z. Xiao, Y. Yuan, Y. Shao, Q. Wang, Q. Dong, C. Bi, P. Sharma, A. Gruverman, J. Huang, Nat. Mater. 215, 14, WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com (9 of 9) 1692