1. Introduction

The location of the emission zone (EZ) significantly influences the performance of organic light emitting diodes (OLEDs), such as their emission spectrum, internal and outcoupling efficiencies and stability [1–3]. Therefore, the direct observation or estimation of the position of the EZ is of great importance in terms of realizing high efficiency and high stability OLEDs. Various approaches have been made to estimate the EZ up to now. These include the analysis of the polarization of the electroluminescence (EL) spectra emitted from emitting polymer chains aligned vertically and in parallel [4], and the comparison of the experimental and calculated EL spectra by assuming the distribution of EZ [5,6], or by a linear superposition of the calculated spectra at various positions inside the emitting polymer [7]. Most analyses were based on the principle of spectral change with the position of the emission zone in OLEDs by the interference effect and used polymeric LEDs with a thick emitting layer (EML) in the range of hundreds of nanometers where large EL spectral changes are observed due to the large shift of EZ.

Recent high efficiency and low driving voltage OLEDs adopt thin EMLs in the range of 10-30 nm by means of optimized material systems [1,8,9], p-i-n structures [10–12], and microcavity structures [13]. Estimating EZs in these devices is still important to analyze and optimize the device performance. Unfortunately, however, determination of EZs based on the change of emission spectra is difficult in these devices since the shapes of the EL spectra are almost identical. The EZ shifts throughout the EML result in no significant difference in the interference effect due to the thin EML. Therefore, EZs have been determined experimentally in most cases by inserting sensing layers of a fluorescent dye at various positions in OLEDs [3,14,15], but the additional layer modifies the charge transport within the device, resulting in an uncertainty of the EZs. Very recently, Young et al. [16] reported a method by using the linear superposition of the experimental EL spectra of the reference OLEDs with very thin EMLs located at different distances from the cathode to estimate the relevant EZ positions. However, this method still requires many experimental EL spectra of the reference OLEDs. Thus, a simple new simulation method without either significant electrical disturbance or vast experimental data is strongly required in order to determine the position of the EZs in OLEDs with a thin EML.

In this letter, we present an approach to estimate the EZ positions in high efficiency phosphorescent OLEDs with thin EMLs. We used two devices with different distances between the EML and the cathode (i.e. they are optically different) but exhibiting the same current density-voltage (J-V) characteristics (i.e. they are electrically the same). Therefore, the optical interference effect can be independently monitored from the electrical effect and considered as a main parameter for the difference in device performances. The mean EZ positions in the OLEDs were directly extracted from the comparison of the experimental luminous intensity ratio vs. the current density with the calculated intensity ratio vs. EZ position of the devices.

2. Experiment and brief description about theoretical approach

The OLEDs were fabricated on UV ozone-treated ITO substrates with the layer sequences of a 4 wt% rhenium oxide-doped N,N’-diphenyl-N,N’-bis(1,1’-biphenyl)-4,4’-diamine (NPB) p-hole transporting layer (p-HTL) (80 nm), undoped NPB HTL (20 nm), a double EML of 8 wt% Ir(ppy)₃-doped 4-4’-N,N’-dicarbazolylbiphenyl (CBP) (10 nm) and 8wt % Ir(ppy)₃-doped 4,7-diphenyl-1,10-phenanthroline (Bphen) (20 nm), undoped Bphen ETL (30 nm), 15wt % rubidium carbonate-doped Bphen n-ETL (15 or 25 nm), and an Al cathode. Further detailed device structures with a possible energy level diagram are depicted in the inset of Fig. 1 . The current density-voltage-luminance characteristics of the devices were measured by a Keithley 2400 semiconductor parameter analyzer and a Photo Research PR-650 spectrophotometer. All devices were encapsulated prior to the measurement. Classical electromagnetic theory has been applied for the optical modeling of the OLEDs [1,17–21]. Changes of the EL spectra and intensities with the position of the EZ within the devices were simulated under the assumptions of randomly oriented sheet dipoles in the layered structures. The position of the sheet dipole can be interpreted as the mean position of the emission zone. The assumption of the sheet dipoles will be discussed later. The intrinsic quantum yield of the emissive material (q) was varied in the calculation. Even though the photoluminescence quantum yield of Ir(ppy)₃ in CBP is close to 100% [20,22], the q can be reduced at high current density [23]. The optical constants of all organic layers for optical modeling were measured by spectroscopic ellipsometry (Korea Research Institute of Standards and Science).

 

Fig. 1 The current efficiency-voltage-current density characteristics of the OLEDs with different n-ETL thicknesses. The inset shows the schematic energy level diagrams of the OLEDs used.

Download Full Size | PDF

3. Results and discussion

Figure 1 shows the luminous efficiency-current density-voltage (L-J-V) characteristics of the OLEDs with two different n-ETL thicknesses. The J-V characteristics of the devices are almost identical between the devices in spite of the different thicknesses of the n-ETL, indicating that carrier movements (charge injection and transport) are very similar in both OLEDs independent of the n-ETL thickness due to the negligible Ohmic loss for electron injection and transport in the layer by n-doping [24] and thus, the exciton formation and the position of EZs within the EML must be the same in both devices. In contrast, the L-J characteristics are apparently different for the two OLEDs. At the low current density region, the device with the 25 nm-thick-n-ETL produces higher peak luminance efficiency than that of the device with the 15 nm-thick-n-ETL, whereas the efficiency is reversed in the high current density region. Since the electrical characteristics of the devices are identical, the difference in efficiencies between the devices originates from purely optical effects.

Figure 2 exhibits the EL spectra of the OLEDs with two different n-ETL thicknesses at four different current densities. The EL spectra of the devices are almost the same and do not change significantly with current densities, indicating that determination of the location of the EZ is not possible based on the variation of the emission spectra. However EL intensity ratio varies significantly with increasing current density as expected from the efficiency-current density plots of the OLEDs in Fig. 1. It is the intensity ratio that we are using to determine the mean position of the EZ. It is possible because the only difference between the two devices is the distance between the EZ and the cathode, due to the different ETL thicknesses.

 

Fig. 2 Experimental (symbol) and calculated (line) EL spectra of OLEDs with the different n-ETL thicknesses at the current density of (a) 0.2 mA/cm², (b) 2.46 mA/cm², (c) 9.5 mA/cm², and (d) 22 mA/cm².

Download Full Size | PDF

The luminous intensities of the OLEDs are calculated for different EZ locations using classical electromagnetic theory. The relative luminous intensities and their intensity ratios are displayed in Fig. 3 and in the inset of Fig. 3, respectively, when q = 1. The EZ positions of “0” and “30” in the figure correspond to the interface of the EML/ETL and the HTL/EML, respectively. If the EZ position is located below 16 nm, the calculated luminous intensity of the 25 nm-thick n-ETL device is higher than the 15 nm-thick n-ETL device, whereas the intensity is reversed for the EZ positions beyond 16 nm, similar to the L-J plots of the OLEDs in Fig. 1. The similarity between them is apparent in the intensity ratios shown in the inset of Fig. 3. The efficiency roll-off observed in Fig. 1 does not appear in the calculated luminous intensity in Fig. 3 because the annihilation of excitons was not considered in the calculation. By relating the experimentally obtained EL intensity ratio vs. the current density with the calculated EL intensity ratio vs. the EZ location, we can now extract the EZ location vs. the current density, allowing the estimation of the movement of the EZ in the OLEDs with the current density.

 

Fig. 3 The calculated luminous intensity of the OLEDs with 25 nm and 15 nm-thick n-ETLs depending on the position of the emission zone when q = 1. The inset shows the relative ratios of the calculated luminous intensity (line) of two devices depending on the emission zone positions. For comparison, the relative ratios of the current efficiency (symbol) of two devices are also plotted. The values in the inset are the current densities corresponding to the relative ratios of current efficiencies of the two devices.

Download Full Size | PDF

Figure 4 shows the variation of the obtained mean location of the EZ with the current density in the OLEDs with 30 nm-thick EMLs for different q values. We repeated theprocedure displayed in Fig. 3 for different values of q. The ratio of current efficiency could be fitted only when q is greater than 0.65. When q is below 0.65, the average emission zone at high current density is positioned in the HTL, which is contradictory to the experimental results. The EL spectra show little emission from NPB even at the high current density of 30 mA/cm² (not shown), indicating that the excitons are well confined in the EML even at high current densities. We believe this is due to our device geometry with the double EMLs and the p-i-n structure. The maximum deviation of the average emission zone is 3.3 nm at 30 mA/cm² while q varies from 1.0 to 0.65. These results indicate that our approach is adequate with an accuracy of 3 nm.

 

Fig. 4 The mean position of the emission zone of the OLEDs with applied current density at various internal quantum efficiencies (square: 1.0, circle: 0.75, and triangle: 0.65). (The lines are for a guide to the eye only).

Download Full Size | PDF

The mean position of the EZ in the devices is located near the EML/ETL interface at low current densities and moves toward the HML/EML interface with increasing current density. The result is consistent with the experimental reports in a double EML OLED with a similar device structure [3]. It is interesting to note that the EZ at low current density is located within the Ir(ppy)₃ doped BPhen EML rather than at the 1st EML/2nd EML interface. Even with the energy barrier for hole injection from the CBP to the BPhen host, holes must be injected to the 2nd EML probably through the Ir(ppy)₃ dopants even at low electric field [25]. The EZ moves rather rapidly with increasing the current density in the range of 1-5 mA/cm². Further increment of the current density shifts the EZ further toward the HTL/EML interface but with a rather slower rate.

The validity of our calculation on the profiles of the EZ in OLEDs with a thin EML can be obtained from the comparison of the calculated emission spectra with the experimental spectra shown in Fig. 2. Excellent agreement in the EL spectra and the relative intensities between the simulated and measured EL data are observed. Even though the EZ is located within the EML up to 10,000 cd/m², it does not necessarily mean that all the excitons are confined in the EML. Excitons in real OLEDs must be distributed with a certain dispersion around an mean position rather than a delta function assumed in our calculation [3,16]. Therefore the EZ position in Fig. 4 must be considered as the mean position in the distribution, even though the excellent agreement of the calculated spectra with the experimental spectra indicates that the distribution is narrow in these devices. The fact that the roll-off of the efficiency begins at a rather low current density of 1 mA/cm² supports the narrow distribution of excitons in positions resulting in a high likelihood of triplet-triplet annihilation [3]. A recent paper revealed that the width of the exciton recombination zone of a PhOLED is in the range of 5 nm [26].

4. Conclusion

In summary, we presented an approach to estimate the EZ positions in high efficiency phosphorescent OLEDs with thin EMLs. Two devices with different distances between the emitting layer and the cathode (i.e. they were optically different), but exhibiting same current density-voltage characteristics (i.e. they were electrically the same) were used for the purpose. EZ positions in the OLEDs were extracted from the comparison of the experimental luminous intensity ratio vs. the current density with the calculated intensity ratio vs. the EZ position. The calculation showed that the EZ moves from the EML/ETL interface to the HTL/EML interface with increasing current densities in the dual EML OLEDs, but the entire EZ positions still exists inside the 30 nm-thick EML even at 10000 cd/m² mainly due to the exciton confinement effects. We believe that this simple approach can contribute to further understanding of exciton behavior related to the realization of high performance OLEDs and greater understanding of the internal efficiency.