JOURNAL OF POLYMER SCIENCE WWW.POLYMERPHYSICS.ORG FULL PAPER Electrical and Optical Simulations of a Polymer-Based Phosphorescent Organic Light-Emitting Diode with High Efficiency Mahmoud Al-Sa’di,1 Frank Jaiser,1 Sergey Bagnich,1 Thomas Unger,1 James Blakesley,1 Andreas Wilke,2 Dieter Neher1 1 Institute of Physics and Astronomy, Soft Matter Physics, University of Potsdam, Karl-Liebknecht-Str. 24-25, D-14476 Potsdam, Germany 2 €t zu Berlin, Brook-Taylor-Str. 6, D-12489 Berlin, Germany Institute of Physics, Supramolecular Systems, Humboldt-Universita Correspondence to: M. Al-Sa’di (E-mail: alsadi@uni-potsdam.de) or D. Neher (E-mail: neher@uni-potsdam.de) Received 10 August 2012; accepted 13 August 2012; published online 11 September 2012 DOI: 10.1002/polb.23158 ABSTRACT: A comprehensive numerical device simulation of the electrical and optical characteristics accompanied with experimental measurements of a new highly efficient system for polymer-based light-emitting diodes doped with phosphorescent dyes is presented. The system under investigation comprises an electron transporter attached to a polymer backbone blended with an electronically inert small molecule and an iridium-based green phosphorescent dye which serves as both emitter and hole transporter. The device simulation combines an electrical and an optical model. Based on the known highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) levels of all components as well as the measured electrical and optical characteristics of the devices, we model the emissive layer as an effective medium using the dye’s HOMO as hole transport level and the polymer LUMO as electron transport level. By fine-tuning the injection barriers at the electron and hole-injecting contact, respectively, in simulated devices, unipolar device characteristics were fitted to the experimental data. Simulations using the so-obtained set of parameters yielded very good agreement to the measured current–voltage, luminance–voltage characteristics, and the emission profile of entire bipolar light-emitting diodes, without additional fitting parameters. The simulation was used to gain insight into the physical processes and the mechanisms governing the efficiency of the organic light-emitting diode, including the position and extent of the recombination zone, carrier concentration profiles, and field distribution inside the device. The simulations show that the device is severely limited by hole injection, and that a reduction of the hole-injection barrier would improve the device C 2012 Wiley Periodicals, Inc. J Polym efficiency by almost 50%. V Sci Part B: Polym Phys 50: 1567–1576, 2012 INTRODUCTION Electronic devices based on organic semiconductors, such as light-emitting diodes, field effect transistors, and solar cells have attracted much interest as possible inexpensive and flexible alternatives to inorganic devices. Since the seminal work of Tang and VanSlyke in the late 1980s where they reported the first green organic lightemitting diode (OLED) based on small organic molecules by thermal vacuum deposition,1 OLEDs have been intensively investigated as potentially promising candidates for the fabrication of thin and flexible displays and other novel optoelectronic applications. The prospects for wide angle viewing, high efficiency, and the fast response compared to liquid crystal displays have made OLED technology very attractive for display applications. Also, the small active layer thickness and the possibility to fabricate OLEDs on flexible substrates enable lightning applications that are not yet realizable with conventional light sources. Here polymer-based OLEDs are particularly suited as they allow for low-cost deposition of the active material on large areas. Besides optimization of the device fabrication techniques, continuous improvement of the OLED performance relies on the comprehensive understanding of the electrical and optical properties of the different components forming the OLED and the physical processes underlying the device operation. Simulation and modeling of OLEDs under various driving conditions can provide the necessary information and can avoid the tedious and costly process of device optimization by systematic experimental approach. KEYWORDS: conjugated polymers; high performance polymers; organic electronics; organic light-emitting diode; simulations; TCAD In an efficient OLED, the active layers need to combine different functions such as electron and hole transport and radiative recombination. The simplest approach is to blend materials with respective properties in one active layer. Indeed, OLEDs where a single blend layer is sandwiched between an anode and a cathode exhibited high efficiencies when using phosphorescent emitters.2,7 However, such blend layers might suffer from aggregation and/or phase separation. One simple route to avoid these problems is to C 2012 Wiley Periodicals, Inc. V WWW.MATERIALSVIEWS.COM JOURNAL OF POLYMER SCIENCE PART B: POLYMER PHYSICS 2012, 50, 1567–1576 1567 FULL PAPER WWW.POLYMERPHYSICS.ORG FIGURE 1 Chemical structures of the materials used in the EML, (a) electron transporting polymer, (b) conjugated small molecule ‘‘filler,’’ and (c) green emitting Iridium-based phosphorescent dye. covalently attach the active conjugated components to a nonconjugated backbone.8,10 It was also shown that the performance of polymer-based OLEDs can be substantially improved when adding a thin polymer interlayer between the anode and the active layer.11,15 These interlayers are usually formed by depositing a suitable high band-gap polymer onto the conducting polymeric anode, followed by annealing and washing. As a result, a very thin (typically 2–10 nm thick) insoluble polymer interlayer forms on top of the anode. It was proposed that these interlayers improve device properties by either preventing excitons reaching the conducting polymer anode,11 by blocking electrons at the interface between the active film and the interlayer13 or by improving hole injection.14 In this work, we investigate the electrical and optical properties of a new efficient system for OLEDs through comprehensive numerical device simulations using the Technology Computer Aided Design (TCAD) software from Silvaco.16 The OLEDs under study consist of a hole-injecting/electron-blocking interlayer and an emissive layer (EML) sandwiched between two metallic contacts. The EML under investigation comprises an electron transporter attached to a polymer backbone blended with an electronically inert small molecule and an iridium-based green phosphorescent dye which serves as both emitter and hole transporter. The chemical structures of all three components are shown in Figure 1. A detailed investigation of the electrical and optical properties of devices made from these materials will be published elsewhere.17 This combination of polymer and small molecules showed high efficiencies above 40 cd/A and 35 lm/W (see Fig. 2). This compares quite well to other green phosphorescent devices based on soluble systems published recently.8,18,20 While we observe an only weak roll-off of luminance efficiency with increasing bias, the power efficiency decreases rapidly. This deterioration is mainly caused by an only moderate increase in current with bias. To gain a comprehensive understanding of the processes determining the electrical and optoelectronic device characteristics, we have conducted an extensive simulation of the characteristic properties of single and dual carrier devices. 1568 JOURNAL OF POLYMER SCIENCE PART B: POLYMER PHYSICS 2012, 50, 1567–1576 JOURNAL OF POLYMER SCIENCE In addition to that, the simulation is used to extract physical parameters such as the width and location of the recombination zone that are otherwise difficult to determine experimentally. The physical models used in our simulations comprise the key optical and electronic processes governing OLED performance such as charge injection and transport, exciton formation, diffusion and decay, and the outcoupling of the generated photons to the outside. We find that the device performance is mainly limited by the barrier for hole injection between the polymeric anode and the hole-injection layer (HIL). Reducing the hole-injection barrier is predicted to enhance the power efficiency of the device by up to 50%, while lowering the barrier for electron injection will have an only small effect on the device efficiency. EXPERIMENTAL The OLED devices presented here comprise the following multilayer structure: Glass substrates (1 mm) are covered with prepatterned indium-tin oxide (ITO,150 nm) and PEDOT:PSS (Poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate)) (Clevios P VP AI 4083, Heraeus Precious Metals, 60 nm) serving as transparent anode. The next layer is a 5 nm HIL formed via the ‘‘interlayer’’ method as described FIGURE 2 (a) Current–voltage (closed symbols) and luminance–voltage characteristics (open symbols) of a device with 100 nm EML thickness. (b) Corresponding dependencies of the luminous efficiency (gL, closed symbols) and of the luminous power efficiency (gP, open symbols) on the current. Straight lines indicate the operation conditions for a luminance of 1000 cd/m2. JOURNAL OF POLYMER SCIENCE WWW.POLYMERPHYSICS.ORG FIGURE 3 Schematic band diagram of the investigated system. (a) HOMO energies and work functions as measured by UPS and LUMO energies as measured via cyclovoltammetry. (b) Effective energy scheme used in simulation. In the EML, dashed lines represent the green dye and full lines represent the electron transporting polymer. The inert ‘‘filler’’ component of the EML is neglected as described in the text. above, using a conjugated polymer based on triarylamine and fluorene-type units (polymer provided by Merck KGaA). The EML comprises three components with different values of the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO). The first component is a polymer with a non-conjugated backbone and conjugated side groups that act as electron transporter in the system. The second component is a (conjugated) small molecule and acts as ‘‘filler,’’ which due to its HOMO and LUMO levels has no electronically active role in the device operation. The third component is a green-emitting Iridiumbased phosphorescent dye. Polymer and ‘‘filler’’ are blended in a 1:2 ratio by weight with the emitter added at 17 wt %. The chemical structures of the organic materials used in the EML are shown in Figure 1. The polymer was synthesized by the group of Dr. Krueger at Fraunhofer Institute of Applied Polymer Research, Potsdam-Golm. Details on the synthesis of the electron transporting polymer will be published elsewhere.17 The small molecules were supplied by Merck KGaA. The cathode (Ba covered by Al) was subsequently thermally evaporated in high vacuum at pressures of 106 mbar. The complete layer sequence of the finished device is ITO (150 nm)/PEDOT:PSS (60 nm)/HIL (5 nm)/EML (100 nm)/Ba (5 nm)/Al with an active area of 16 mm2. The luminance–voltage and current–voltage characteristics as summarized in Figure 2 were measured in nitrogen atmosphere using a Konica Minolta CS-100 ChromaMeter and a Keithley 2400 SourceMeter. The OLED turns on slightly above 2 V as indicated by the steep increase in both current and luminance. A brightness of 1000 cd/m2 is reached at about 7.3 V. The efficiency increases strongly after turn on, reaching a maximum of 40 cd/A at a luminance of about 185 cd/m2. At 1000 cd/m2, there is only a small roll off of luminance efficiency to 39 cd/A. However, the peak power efficiency of 35 lm/W is reached at a luminance of 5.2 cd/m2 and then drops pronouncedly to only 17 lm/W at 1000 cd/m2. Mobilities of electrons measured by sensitized A detailed description published elsewhere.21 and holes in the active layer were transient electroluminescence (TEL). of this method and the setup is To measure electron mobility, the WWW.MATERIALSVIEWS.COM FULL PAPER OLED device structure was altered by replacing the HIL with an insoluble sensing layer of poly[2,5-dimethoxy-1,4-phenylene-1,2-ethenylene-2-methoxy-5-(2-ethylhexyloxy)-(1,4-phenylene-1,2-ethenylene)] (M3EH-PPV, H.H. H€ orhold, Jena) of 5 nm thickness formed by the ‘‘interlayer method.’’11,21 CsF covered by Al was used as the cathode. For hole mobility measurements, a red-emitting small molecule provided by Merck KGaA was evaporated as a sensing layer (10 nm) on top of the unaltered active layer stack and covered by Ba and Al as cathode. All devices for TEL measurements had an active area of 2 mm2. For unipolar devices, the contacts were adapted to ensure injection of only one type of charge carriers into the films. Hole-only devices were fabricated by replacing the low work function cathode (Ba) with high work function metal oxide MoO3 (/MoO3 ¼ 6.6 eV) resulting in a layer stack PEDOT:PSS/HIL/EML/MoO3. Similarly, electron-only devices were fabricated by inserting a aluminum layer (/Al ¼ 4.2 eV) between PEDOT:PSS and EML so that the devices comprised PEDOT:PSS/Al/EML/Ba/Al. In both cases, the unaltered contact was used for carrier injection. Steady-state spectra of the electroluminescence (EL) were measured with an Ocean Optics HR2000 spectrometer. Absolute PL efficiencies were determined with a Hamamatsu C9920 setup, including an integrating sphere combined with a photonic multi-channel analyzer. Transient photoluminescence measurements were performed with excitation at 420 nm by an EKSPLA NT-242 Nd:YAG/optical parametric oscillator system with pulse widths below 6 ns. The phosphorescence was recorded by a monochromator and a Becker & Hickl multiscaler PMS 400. The HOMO energies of the electron transporting unit, the filler, and the dye as well as the PEDOT work function were measured using ultraviolet photoemission spectroscopy (UPS) in vacuum. Corresponding LUMO energies were obtained from cyclovoltammetry measurements at Merck KGaA. A schematic band diagram of the measured HOMO and LUMO values is shown in Figure 3(a). The HOMO and LUMO energy levels of the filler are 6.6 eV and 2.3 eV, respectively, and have been neglected from the figure for clarity reasons. The real and imaginary parts of the refractive index of each organic layer were derived from transmission and reflection measurements of the respective materials on glass slides with a Varian Cary 5000 spectrophotometer equipped with an integrating sphere. The optical constants of the film have been iteratively fitted point by point with the Newton–Raphson method until the measured and theoretical reflection and transmission data converged.22 The measured refractive index values of the EML and the HIL are shown in Figure 4. The radiation pattern and spatial distribution of luminescent power of the OLEDs were measured at Fraunhofer Institute for Applied Optics and Precision Engineering IOF, Jena. The polarized angular radiation patterns were recorded utilizing a rotational stage, where the OLED is mounted, and a JOURNAL OF POLYMER SCIENCE PART B: POLYMER PHYSICS 2012, 50, 1567–1576 1569 FULL PAPER JOURNAL OF POLYMER SCIENCE WWW.POLYMERPHYSICS.ORG The physical equations are solved numerically to obtain solutions for the electron and hole current densities, carrier densities, electric field, electrostatic potential, and recombination rate. The boundary conditions used in the calculations will be discussed below together with a brief description of the models used in the electrical and optical simulations. FIGURE 4 Real and imaginary parts of the refractive index of HIL (squares) and EML (circles) as determined from absorption and reflection measurements of the organic layers. detection system comprising a polarizer, a retarder, and a fiber coupled spectrometer. During the measurements, the number of charge carriers in the devices was controlled and stabilized by driving the OLEDs at constant current (0.1 mA) using a current source.23 This current corresponded to a luminance of about 250 cd/m2. DEVICE MODEL AND PARAMETERS The Metal-Insulator-Metal model is used to describe the investigated OLED devices. Accordingly, the device is considered to consist of semiconductor layer(s) sandwiched between two metal contacts. From device measurements obtained in our laboratory it is clear that the electron transport occurs via the electron transporting polymer, while the hole transport is exclusively via the dye.17 This is supported by the UPS measurements of the system which show that the HOMO of both the electron transporter and the filler are very deep and well below the dye’s HOMO. At the same time, the polymer LUMO is the lowest energy state for excess electrons. Hence we model the EML layer as an effective medium using the dye’s HOMO as hole transport level and the polymer LUMO as electron transport level. UPS measurements showed a small vacuum level shift (D ¼ 0.1 eV) at the EML/HIL interface due to interface dipoles. As the simulator cannot include such a shift, a flat vacuum level was used. To restore the correct energy level alignment at the HIL/EML interface, the EML HOMO and LUMO were shifted by 0.1 eV accordingly. The effective work functions of the contact materials (/PEDOT and /Ba) were optimized to give optimum fits to the experimental current density–voltage (J-V) characteristics of single-carrier devices. A schematic of the band diagram used to simulate our system illustrating the relevant LUMO and HOMO levels as well as the effective work functions of the contact materials is shown in Figure 3(b). The OLED devices are simulated using a calibrated finite-element TCAD simulation software based on the drift-diffusion transport model, coupled to Poisson’s equation and the current continuity equations for electrons and holes. In addition to that, a Poole–Frenkel type field-dependent mobility and Langevin bimolecular recombination model are employed. 1570 JOURNAL OF POLYMER SCIENCE PART B: POLYMER PHYSICS 2012, 50, 1567–1576 Electrical Model Charge Transport Equations Charge transport in organic devices is described by the general semiconductor drift-diffusion model, where the continuity equations for electrons and holes in the drift-diffusion form for current density are coupled to Poisson’s equation. Assuming a trap free system, the model consists of rðerwÞ ¼ qðn pÞ dn ¼ qðG RÞ dt * dp rJ p þ q ¼ qðG RÞ dt * rJ n q (1) (2) (3) where w is the electrostatic potential, e is the product of the vacuum permittivity e0 and the relative permittivity er of the organic material, n(p) is the electron (hole) density, Jn (Jp) is the electron (hole) current density, R (G) denotes the recombination (generation) rate, and q is the elementary charge. The time derivatives in eqs 2 and 3 will be omitted due to the steady-state nature of this work. In the drift-diffusion model, the currents of electrons and holes are described as a sum of two contributions; the drift component, proportional to the electric field, and the diffusion component, proportional to the gradient of the charge density. The current equations for electrons and holes are given by * * J n ¼ qn ln E þ qDn rn * * J p ¼ qp lp E qDp rp (4) (5) with Dn (Dp) denoting the diffusion coefficient of electrons (holes). Assuming that the Einstein relation holds for the system, the diffusivities read Dn ¼ kB T l q n (6) Dp ¼ kB T l q p (7) where kB is Boltzmann constant, T is the lattice temperature, and ln (lp) is the electron (hole) mobility. The generation rate is neglected since it is not relevant for materials with an energy gap larger than 2 eV. The bimolecular recombination rate is described in the Langevin form, where recombination rate is given by R ¼ cðnp n2i Þ (8) with the intrinsic carrier density ni and the reduced Langevin recombination rate JOURNAL OF POLYMER SCIENCE WWW.POLYMERPHYSICS.ORG c¼a q ln ðEÞ þ lp ðEÞ z (9) Assuming Maxwell–Boltzmann statistics, the electron and hole concentrations are expressed as EF; n ELUMO n ¼ NLUMO exp kB T (10) EHOMO EF;p p ¼ NHOMO exp kB T (11) and where ELUMO and EHOMO are the energy levels of LUMO and HOMO, EF,n (EF,p) is the quasi-Fermi level for electrons (holes) and NLUMO (NHOMO) is the density of states in the LUMO (HOMO).26 The densities of states are taken as 1.0 1020 cm3 for all organic layers in this work. The mobilities are taken to be field dependent with the Poole–Frenkel form (12) (13) where ln0 (lp0) is the zero field electron (hole) mobility, E0n and E0p are constants related to the disorder in the material, and E is the electric field.27,28 Boundary Conditions The OLED structures are numerically simulated using Schottky contact boundary conditions between the organic layer and the anode or cathode metal, respectively. The barrier height that governs carrier injection at the contacts is given by /B ¼ /m ve (14) where /m is the contact work function, /B is the Schottky barrier height over which the carriers have to be injected and ve is the electron affinity of the organic material. The built-in voltage is taken to be the difference between the work functions of the two different contact materials used as anode and cathode. The injection is treated using the Scott–Malliaras modification of Schottky barrier boundary conditions.29 The net current density at the contact is given by /B exp f 1=2 J ¼ 4w2 N0 qlE exp kB T (15) where N0 is the density of chargeable sites in the polymer, f is the reduced electric field expressed by f ¼ e3 E 4peðkB TÞ2 WWW.MATERIALSVIEWS.COM and wðf Þ ¼ f 1 þ f 1=2 f 1 ð1 þ 2f 1=2 Þ1=2 where a 1 is a model parameter.24,25 rffiffiffiffiffiffiffi E ln ðEÞ ¼ ln0 exp En0 sffiffiffiffiffiffiffi! E lp ðEÞ ¼ lp0 exp Ep0 FULL PAPER (16) (17) Finally, the interface between the organic materials (EML and HIL) has been taken into account by employing thermionic and tunneling models at the organic materials interface when solving the current density equations.16,27,30 Optical Model Optical characteristics of the bipolar OLEDs were simulated using the transfer-matrix approach for multilayer systems in combination with the dipole emission source term method and reverse ray-tracing technique. The dipole emission model is based on the equivalence between photon emission due to an electrical dipole transition and the radiation from a classical electrical dipole antenna. The OLED is considered as an emission source embedded in a microcavity. Because the lateral dimensions are much larger than the thickness, the device is assumed to be a one-dimensional structure, and the excitons within the OLED are modeled assuming randomly oriented point dipoles driven by the reflected electromagnetic waves inside a microcavity. The model includes the emission of the dipole antenna into plane and evanescent waves in the emitting layer with transverse electric (TE) and transverse magnetic (TM) polarizations. The model accounts for the wide-angle and multiple-beam interference caused by partial reflection, total internal reflection, and absorption. A detailed description of the dipole emission source term method and the transfer-matrix approach has been published elsewhere.31,35 Simulation Parameters The main parameters required as input for the device simulation are the charge carrier mobilities and the energy levels of each organic layer as well as the effective work functions of the contacts. The effective work functions of the electrode materials in contact with the respective organic material were obtained by fitting experimental J(V) characteristics of electron- and hole-only devices and found to be consistent with literature values.36,38 For PEDOT:PSS, there is a small difference of the work function which yielded the best simulation results (4.7 eV) and the value measured by UPS (4.8 eV). This deviation will be addressed in the discussion. The mobility parameters were obtained by fitting the mobilities measured by sensitized TEL (see below). A summary of the parameters used in the electrical simulations are shown in Tables 1 and 2. The input parameters for the optical simulation comprised the layers thickness, the real (n), and imaginary (k) parts of the refractive index of each layer as a function of wavelength, and the photoluminescence spectrum of the EML. The measured refractive index values of the EML and HIL shown in Figure 4 are implemented in our simulator. The values of n and k for ITO were taken to be 1.85 and 0.0065, respectively.39 For glass, n and k were set to 1.5 and 0, respectively.39 Table 3 summarizes the additional parameters required for the optical simulations.40,41 The triplet radiative JOURNAL OF POLYMER SCIENCE PART B: POLYMER PHYSICS 2012, 50, 1567–1576 1571 FULL PAPER WWW.POLYMERPHYSICS.ORG TABLE 1 Electrical Simulation Parameters Parameter er EML TABLE 3 Parameters Required for the Optical Simulation HIL 3.0 JOURNAL OF POLYMER SCIENCE Units 3.0 – Parameter Triplet-polaron quenching constant Value Units 13 cm3 s1 12 cm3 s1 3 10 EHOMO 5.1 5.2 eV Triplet-triplet annihilation constant 4 10 ELUMO 2.7 1.9 eV Intersystem crossing constant 1 1013 s1 6 Eg 2.4 3.3 eV Triplet radiative decay lifetime 1.25 10 s NLUMO 1 1020 1 1020 cm3 80 % NHOMO 1 1020 1 1020 cm3 Dye photoluminescence quantum efficiency ln0 9.534 108 7 cm2/Vs 0 4 lp0 1.12 10 2.5 10 cm2/Vs En0 42,711.96 – V/cm Ep0 176,233.5 176,233.5 V/cm The energy values have been determined by UPS and cyclovoltammetry. For the densities of states and the dielectric constant, typical values for organic semiconductors have been taken. Mobility parameters were determined by fitting measured mobilities to a Poole–Frenkel type model (see text). The electron mobility in HIL is discussed in the main text. decay time was calculated from the measured triplet lifetime of 1 106 s and the photoluminescence quantum efficiency of the dye in the EML (80%). RESULTS AND DISCUSSION Mobility parameters obtained experimentally by TEL on single-carrier devices are used as input to model the electrical characteristics. The measured mobilities of electrons and holes for the EML at room temperature are shown in Figure 5. The field dependence of ln and lp can be well explained by Poole–Frenkel type field-assisted hopping transport.42 The fitted mobility parameters as summarized in Table 1 were used for the simulation of electron- and hole-only devices as well as bipolar OLED. As in the experiments, the unipolar devices were obtained by changing the appropriate contacts. Regrettably, the measured data for both types of unipolar devices suffers from a relatively large uncertainty due to deviations between different pixels and devices. This limits the accuracy of the work functions extracted from the simulation to 60.1 eV. For the hole-only device with PEDOT:PSS anode, the best agreement with the mean experimental data was obtained for /PEDOT:PSS ¼ 4.7 eV. For the system under investigation, UPS measurements have shown a PEDOT:PSS work function of 4.8 eV and a barrier of 0.4 eV for hole injection into HIL [Figure 3(a)]. Considering the problems mentioned above, the PEDOT:PSS work function used here to simulate the measured characteristics agrees well with these measurements. Also, it is known that the work function of PEDOT:PSS decreases when exposed to water vapor.43 Hence, the different environments—vacuum for UPS measurements and device preparation in nitrogen atmosphere—are an additional explanation for this slight deviation. For the electron-only device with Ba cathode, the best fit to the experimental data was obtained with /Ba ¼ 2.8 eV. This value is slightly higher than the standard Ba work function of 2.7 eV, which we attribute to partial oxidation of the metal during device fabrication and the above mentioned uncertainty of the measured data. The effective work function of the contact materials and the mobility parameters obtained from unipolar devices are used without further adjustment as input to model the electrical and optical characteristics of the bipolar OLEDs. Figure 6 shows simulated and experimental J(V) curves at room temperature for hole-only, electron-only, and bipolar devices. The figure shows very good agreement between simulation and experimental results, especially for the unipolar devices. In addition to that, the simulation model correctly predicts that the bipolar OLED current density is almost three orders of magnitude larger than either of the unipolar current densities in the same applied voltage range. The deviation at low currents (currents smaller than 103 mA/cm2) is explained by the experimental leakage current which dominates at low bias. In addition to the J(V) characteristics, the simulation yields the spatial distribution of carrier density, electric field, and recombination rate inside the devices. For the bipolar device TABLE 2 Effective Work Function of the Contact Materials as Obtained by Fitting Experimental J(V) Characteristics of Electron- and Hole-Only Devices 1572 Parameter Ba PEDOT:PSS Units Effective work function 2.8 4.7 eV JOURNAL OF POLYMER SCIENCE PART B: POLYMER PHYSICS 2012, 50, 1567–1576 FIGURE 5 The measured (symbols) field-dependent mobilities of electrons and holes for the EML and the corresponding fits to a Poole–Frenkel type field dependence (lines). Experimental mobilities were obtained by sensitized TEL (see text). JOURNAL OF POLYMER SCIENCE WWW.POLYMERPHYSICS.ORG FIGURE 6 J(V) characteristics for a hole-only device (circles), electron-only device (squares), and bipolar device (triangles). EML thicknesses are 200 nm for the unipolar devices and 100 nm for the bipolar device. The symbols denote experimental data and the solid lines represent simulation results. under forward bias, the density of electrons injected into the EML is much higher than the density of the injected holes due to the larger barrier offset at the anode side (0.5 eV) compared to the cathode side (0.2 eV). There is a strong FULL PAPER FIGURE 8 Comparison of simulated (lines) and measured (symbols) normalized radiation pattern for the wavelength range (450–850) nm for an OLED with EML thickness of 100 nm. TE and TM denote transverse electric and transverse magnetic modes, respectively. electron accumulation at the EML/HIL interface, as the electrons cannot penetrate the HIL due to the large LUMO offset. Simulations with an arbitrary, non-zero electron mobility in the HIL did not change the overall behavior. Due to the lower mobility of holes in the EML, the hole density in the EML is much higher than in HIL. In the bulk of the EML, both carrier densities are relatively low and almost constant, with a factor of 10 lower hole density. The electric field is found to be high across the HIL and at the EML/HIL interface, and uniformly distributed across the EML as shown in Figure 7(a). In agreement with the charge distribution, we found that the recombination zone is located close to the EML/HIL interface and extends only slightly into the EML as shown in Figure 7(b). The width of the recombination zone obtained from the electrical simulations, which is determined to be approximately 5 nm, was used in the optical model to simulate the optical characteristics of the bipolar OLED device. FIGURE 7 (a) Simulated electric field and carrier density distributions and (b) recombination rate profile for the OLED with EML thickness of 100 nm at an applied voltage of 15 V. Electrons are injected from the left-hand side (x ¼ 0) and holes are injected from the right-hand side (x ¼ 105 nm). WWW.MATERIALSVIEWS.COM The optical performance of the bipolar OLED has been simulated using the optical models described in the text. The simulated and measured results of the normalized radiation pattern and the luminescent power as a function of the viewing angle for the wavelength range 450–850 nm are shown in Figures 8 and 9, respectively. The simulation closely resembles the measured data, further supporting the validity of the implemented electrical and optical models. As is to be expected for a planar substrate, the results show that light emitted at higher angles to the normal direction is out-coupled less efficiently than light emitted at smaller angles, resulting in a reduced spectral power density. However, the close agreement between measured and simulated emission properties supports the simulated recombination profile discussed above. Figure 10 compares simulated and JOURNAL OF POLYMER SCIENCE PART B: POLYMER PHYSICS 2012, 50, 1567–1576 1573 FULL PAPER WWW.POLYMERPHYSICS.ORG FIGURE 9 Simulated (line) and measured (points) normalized luminescent power as function of viewing angle (to the normal) for an OLED with 100 nm EML thickness. measured luminance–voltage characteristics. Again, our simulation results are in good agreement with the experimental data, verifying the validity and accuracy of the model and the parameters used. Note that the simulation overestimates the luminance at low bias. This trend is also seen when comparing the measured and simulated bipolar current in Figure 6. It is possible to reduce this discrepancy by allowing the parameters describing injection and transport in the bipolar device to derivate from the parameters deduced from fits to the unipolar currents, but this procedure would introduce an additional uncertainty. Also, the simulation reproduces well the experimental data for relevant brightness levels (above 100 cd/m2). The impact of the effective barrier heights for carrier injection at the anode and the cathode sides were investigated by calculating the J(V) characteristics while varying the work function of either the anode or the cathode with respect to the reference device, with the work function of the other contact kept at the reference values. The effective barrier for hole injection was varied between 0.7 and 0 eV by setting the work function of the anode to values ranging from 4.5 eV to 5.2 eV. Similarly, the influence of the effective barrier for electron injection was investigated by varying the effec- FIGURE 10 Simulated (line) and measured (points) luminance– voltage characteristics of an OLED with 100 nm EML thickness. 1574 JOURNAL OF POLYMER SCIENCE PART B: POLYMER PHYSICS 2012, 50, 1567–1576 JOURNAL OF POLYMER SCIENCE FIGURE 11 Influence of different hole or electron injection barriers on the current of a bipolar device with EML thickness of 100 nm. The barrier for the other contact was kept at the level of the standard device. The normalized current density j15V (normalized to the current of the reference device) is plotted for the highest simulated voltage, i.e., 15 V. tive work function of the cathode between 3.0 and 2.6 eV. The obtained results are summarized in Figure 11, where the current density at an applied voltage of 15 V is plotted versus the effective barrier height for hole and electron injection, respectively. As expected, the device current increases significantly upon reduction of either barrier. However, reducing the hole-injection barrier causes a considerably stronger increase in the current and by that reduces the bias needed to give a certain luminance. To clarify the impact of the effective barrier for hole or electron injection on the whole device performance, the power efficiency of the simulated devices was calculated at a luminance of 1000 cd/m2 and the results are shown in Figure 12. In accordance with the results shown previously, the power efficiency increases upon reduction of either barrier. An improvement of up to 50% on the device’s power FIGURE 12 The power efficiency at a luminance of 1000 cd/m2 normalized to the efficiency of the reference device as a function of the effective barrier for hole or electron injection. The barrier for the other contact was kept at the level of the standard device. JOURNAL OF POLYMER SCIENCE WWW.POLYMERPHYSICS.ORG efficiency can be achieved by reducing the effective barrier for hole injection. A reduction of the electron injection barrier, on the other hand, increases the device efficiency only by 15%. In summary, the barrier for hole injection is found to be limiting the device efficiency, as the influence of this barrier is much stronger on the current and efficiency when compared to an equal change of the electron injection barrier. To explore this behavior, the carrier density, the electric field, and the recombination rate distributions were investigated for different values of the injection barriers. The results of these simulations indicate that reducing the injection barrier for holes leads to a stronger increase of the hole density, the electric field, and the recombination rate inside the device, compared to a similar reduction of the effective barrier for electron injection at the anode side. FULL PAPER Research for supplying with the electron transporting polymer. They also thank Anna Hayer and her colleagues at Merck KGaA for fruitful discussions and supplying the phosphorescent dye, the hole-injecting interlayer, sensing layer materials, and the filler. For the measurement of optical constants, authors are grateful to Steve Albrecht (University of Potsdam). Authors thank Norbert Koch (Humboldt-Universit€at zu Berlin) for supporting UPS measurements and data analysis, and Michael Fl€ammich (Fraunhofer IOF Jena) for measurements of the spatial luminance distribution of the devices. This work was supported by the Bundesministerium für Bildung und Forschung (BMBF project ‘‘NEMO’’, FKZ 13N10622). REFERENCES AND NOTES 1 Tang, C. W.; VanSlyke, S. A. Appl. Phys. Lett. 1987, 51, 913–915. CONCLUSIONS 2 Gong, X.; Robinson, M. R.; Ostrowski, J. C.; Moses, D.; Bazan, G. C.; Heeger, A. J. Adv. Mater. 2002, 14, 581–585. In this article, we have successfully modeled the electrical and optical properties of a new highly efficient material system for phosphorescent OLEDs. The good agreement of simulated characteristics of the bipolar devices on the basis of a set of parameters gained from simulations of electronand hole-only devices verified the validity and accuracy of the models and parameters used. The simulation of the investigated devices allows us to extract relevant information such as position and extent of the recombination zone, carrier concentration profiles, and field distribution inside the device, which is difficult to gain otherwise. The simulation shows that the device performance is limited by the injection of holes, which we assign to a large hole-injection barrier between the PEDOT:PSS anode and HIL. As HIL serves as an efficient electron blocking layer, the device current and with this the device brightness at a given voltage is largely reduced compared to an ideal ohmic hole injection. We propose that the presence of the hole-injection barrier is the main reason for the strong reduction of the power conversion efficiency at higher bias. Also, due to the high barrier for hole injection, the hole concentration in the EML is quite low and rather constant throughout the layer. At the same time, the electron blocking property of the HIL leads to a high electron concentration at the EML/HIL interface. As a result, die recombination zone is very narrow and located directly at the EML/HIL interface. With this recombination zone as input for an optical device model, the luminance characteristics of the device can be well reproduced. Simulations with varying injection barriers show that even a slight reduction of the hole-injection barrier would improve the device efficiency by almost 50%, while a reduced electron injection barrier has only little effect on device characteristics. Clearly, the information gained from these simulations will not only allow us to develop new approaches for further device optimization, but also serve as an important input to understand device degradation. 3 Anthopoulos,T. D.; Markham, J. P. J.; Namdas, E. B.; Samuel, D. W.; Lo, S. C.; Burn, P. L. Appl. Phys. Lett. 2003, 82, 4824–4826. 4 van Dijken, A.; Bastiaansen, J. J. A. M.; Kiggen, N. M. M.; € ssel, Langeveld, B. M. W.; Rothe, C.; Monkman, A.; Bach, I.; Sto P.; Brunner, K. J. Am. Chem. Soc. 2004, 126, 7718–7727. 5 Yang, X. H.; Jaiser, F.; Klinger, S.; Neher, D. Appl. Phys. Lett. 2006, 88, 021107. 6 Wu, H.; Zhou, G.; Zou, J.; Ho, C. L.; Wong, W. Y.; Yang, W.; Peng, J.; Cao,Y. Adv. Mater. 2009, 21, 4181–4184. €ubler, T. K. Adv. Mater. 7 Yang, X.; Neher, D.; Hertel, D.; Da 2004, 16, 161–166. 8 Zhu, M.; Ye, T.; He, X.; Cao, X.; Zhong, C.; Ma, D.; Qin, J.; Yang, C. J. Mater. Chem. 2011, 21, 9326–9331. 9 Zhang, Y.; Zuniga, C.; Kim, S. J.; Cai, D.; Barlow, S.; Salman, S.; Coropceanu, V.; Bredas;J. L.; Kippelen, B.; Marder, S. Chem. Mater. 2011, 23, 4002–4015. 10 Bagnich, S. A.; Unger, Th.; Jaiser, F.; Neher, D.; Thesen, M. W.; Krueger, H. Appl. Phys. Lett. 2011, 110, 033724. 11 Kim, J. S.; Friend, R. H.; Grizzi, I.; Burroughes, J. H. Appl. Phys. Lett. 2005, 87, 023506. 12 Choulis, S. A.; Choong, V. E.; Mathai, M. K.; So, F. Appl. Phys. Lett. 2005, 87, 113503. 13 Yang, X. H.; Jaiser, F.; Stiller, B.; Neher, D.; Galbrecht, F.; Scherf, U. Adv. Funct. Mater. 2006, 16, 2156–2162. 14 Choulis, S. A.; Choong, V. E.; Patwardhan, A.; Mathai, M. K.; So, F. Adv. Funct. Mater. 2006, 16, 1075–1080. 15 Yang, X. H.; Wu, F. L.; Neher, D.; Chien, C. H.; Shu, C. F. Chem. Mater. 2008, 20, 1629–1635. 16 Silvaco, TCAD. http://www.silvaco.com/, August, 2012. 17 Salert, B. Ch. D.; Krueger, H.; Bagnich, S. A.; Unger, Th.; Jaiser, F.; Al-Sa’di, M.; Neher, D. In Preparation. 18 Wang, B. Z.; Liu, J.; Wu, H. B.; Zhang, B.; Wen, S. S.; Yang, W. Chin. Phys. B 2011, 20, 088502. 19 Shin, J.; Um, H. A.; Cho, M. J.; Lee, T. W.; Kim, K. H.; Jin, J. I.; Kang, S.; Park, T.; Joo, S. H.; Yang, J. H.; Choi, D. H. J. Polym. Sci. Part A: Polym. Chem. 2012, 50, 388–399. ACKNOWLEDGMENTS 20 Zhu, M.; Li, Y.; Li, C. G.; Zhong, C.; Yang, C.; Wu, H.; Qin, J.; Cao, Y. J. Mater. Chem. 2012, 22, 11128–11133. Authors gratefully acknowledge Beatrice Salert and Hartmut Krueger from Fraunhofer Institute of Applied Polymer 21 Bange, S.; Kuksov, A.; Neher, D. Appl. Phys. Lett. 2007, 91, 143516. WWW.MATERIALSVIEWS.COM JOURNAL OF POLYMER SCIENCE PART B: POLYMER PHYSICS 2012, 50, 1567–1576 1575 FULL PAPER WWW.POLYMERPHYSICS.ORG 22 Peter, K. Optische Charakterisierung von Halbleiterschichten. Diploma Thesis, University of Konstanz, Germany, 1993. €mmich, M. Optical Characterization of OLED Emitter 23 Fla Properties by Radiation Pattern Analyses. PhD Thesis, Frie€t Jena, Germany, 2011. drich-Schiller-Universita 24 Blom, P. W. M.; de Jong, M. J. M.; Breedijk, S. Appl. Phys. Lett. 1997, 71, 930–932. 25 Scott, J. C.; Karg, S.; Carter, S. A. J. Appl. Phys.1997, 82, 1454–1460. 26 Lee, C. C.; Chang, M. Y.; Huang, P. T.; Chen, Y. C.; Chang, Y.; Liu, S. W. J. Appl. Phys. 2007, 101, 114501. 27 Sze, S. M.; Ng, K. K. Physics of Semiconductor Devices; Wiley: New York, 2007. 28 Selberherr, S. Analysis and Simulation of Semiconductor Devices; Springer-Verlag: Wien, 1984. 29 Scott, J. C.; Malliaras, G. G. Chem. Phys. Lett. 1999, 299, 115–119. 32 Benisty, H.; Stanley, R.; Mayer, M. J. Opt. Soc. Am. A 1998, 15, 1192–1201. 30 Walker, A. B.; Kambili, A.; Martin, S. J. J. Phys.: Condens. Matter 2002, 14, 9825–9876. 41 Tang, K. C.; Liu, K. L.; Chen, I. C. Chem. Phys. Lett. 2004, 386, 437–441. € ssler, H. Phys. Status Solidi B 1993, 175, 15–56. 42 Ba 31 Wang, Z. B.; Helander, M. G.; Xu, X. F.; Puzzo, D. P.; Qiu, J.; Greiner, M. T.; Lu, Z. H. J. Appl. Phys. 2011, 109, 053107. 1576 JOURNAL OF POLYMER SCIENCE JOURNAL OF POLYMER SCIENCE PART B: POLYMER PHYSICS 2012, 50, 1567–1576 33 Lukosz, W. J. Opt. Soc. Am. A 1979, 69, 1495–1503. 34 Lukosz, W.; Kunz, R. E. J. Opt. Soc. Am. A 1977, 12, 1607–1615. 35 Neyts, K. A. J. Opt. Soc. Am. A 1998, 15, 962–971. 36 Sapp, S.; Luebben, S.; Losovyj, Y. B.; Jeppson, P.; Schulz, D. L.; Caruso, A. N. Appl. Phys. Lett. 2006, 88, 152107. 37 Li, J. H.; Huang, J.; Yang, Y. Appl. Phys. Lett. 2007, 90, 173505. 38 Koch, N.; Kahn, A.; Ghijsen, J.; Pireaux, J. J.; Schwartz, J.; Johnson, R. L.; Elschner, A. Appl. Phys. Lett. 2003, 82, 70–72. 39 Wei, M. K.; Lin, C. W.; Yang, C. C.; Kiang, Y. W.; Lee, J. H.; Lin, H. Y. Int. J. Mol. Sci. 2010, 11,1527–1545. 40 Reineke, S.; Walzer, K.; Leo, K. Phys. Rev. B 2007, 75, 125328. 43 Koch, N.; Vollmer, A.; Elschner, A. Appl. Phys. Lett. 2007, 90, 043512.