Investigation of triplet harvesting and outcoupling efficiency in highly efficient two-color hybrid white organic light-emitting diodes


Corresponding author: e-mail, Phone: +49-(0)351-463-35117, Fax: +49-(0)351-463-37065, Web:


We investigate singlet and triplet transfer processes in white triplet harvesting (TH) organic light-emitting diodes (OLEDs), comprising fluorescent and phosphorescent emitter molecules. By analyzing electroluminescence spectra and performing time-resolved measurements, we prove direct TH from the blue fluorescent emitter N,N′-di-1-naphthalenyl-N,N′-diphenyl-[1,1′:4′,1″:4″,1′′′-quaterphenyl]-4,4′′′-diamine (4P-NPD) to the yellow phosphorescent emitter bis(2-(9,9-dihexyluorenyl)-1-pyridine)(acetylacetonate)iridium(III) (Ir(dhfpy)2(acac)). Singlet transfer is identified as a second exciton transfer mechanism in the TH OLEDs under investigation. The CIE coordinates of these devices can be adjusted over a wide range by varying the distance between exciton generation and TH zone. For an OLED with CIE color coordinates of (0.42/0.40), close to the warm white color point A, we obtain nearly perfect Lambertian emission characteristics. This device achieves an external quantum efficiency (EQE) of 9.4% and a luminous efficacy (LE) of 27.1 lm W−1 (at 1000 cd m−2) which can be further increased to 46 lm W−1 using outcoupling enhancement techniques. Optical modeling of the electromagnetic field inside the TH OLED and a comparison to the position of the emitters within the device confirms that the devices operate close to the optical optimum and yields an estimated internal quantum efficiency (IQE) of 48.7% and a singlet/(harvested) triplet ratio of 0.28/0.72 = 0.39.

1 Introduction

Organic light-emitting diodes (OLEDs) are promising candidates for general illumination due to their unique properties such as area emission, wide viewing angle, and the possibility to be processed on flexible substrates by roll-to-roll production. White OLEDs with high color rendering index (CRI) and efficiencies comparable to fluorescent tubes have already been demonstrated [1-5]. White light from OLEDs is usually obtained by combining the emission of several emitters in a multilayer structure. To achieve very high efficiencies, two main approaches, which differ by the used emitters, are currently discussed: full phosphorescent (PH) and triplet harvesting (TH) white OLEDs [6, 7]. While PH OLEDs usually employ two or more phosphorescent emitters, TH OLEDs are a hybrid system of fluorescent and phosphorescent emitters. Recently, Uoyama et al. [8] developed fluorescent emitters with a very low singlet–triplet energy splitting, having efficiencies comparable to that of phosphorescent emitters due to thermally activated delayed fluorescence. So far, this approach has only been shown for monochrome OLEDs, but might be used in future to avoid the need of phosphorescent emitters in highly efficient white OLEDs.

Due to spin statistics, about 25% singlets and 75% triplets are generated by the recombination of electrons and holes inside the OLED [9]. Phosphorescent emitters have high intersystem crossing rates, and emit efficiently from the triplet state. Therefore, the internal quantum yield, i.e., the conversion efficiency of electrons into photons, can be very high in PH OLEDs, in the range between 70 and 100% [10-13]. However, developing blue phosphorescent emitters with saturated blue emission color and long lifetimes has turned out to be challenging. This limits the CRI and leads to color changes during aging of white PH OLEDs.

In 2006, Sun et al. [3] and Schwartz et al. [14] introduced the TH concept, which overcomes these issues, since it allows the use of a fluorescent blue emitter while at the same time providing similarly high internal quantum yields. Fluorescent emitters yield light only from the singlet state, which means that in completely fluorescent OLEDs the triplet excitons are non-radiative, and thus 75% of the injected charge carriers are lost. However, if a phosphorescent emitter is inserted at an appropriate position, these triplets can diffuse to the phosphorescent emitter, are harvested, and lead to additional emission at longer wavelength. In the ideal white TH OLED, all singlets will recombine at the blue fluorescent emitter and all triplets will be transferred to a phosphorescent emitter, which would again allow an internal quantum yield close to 100%. However, this requires careful design of the emission layer (EML) since singlet transfer and direct recombination of charge carriers on the phosphorescent emitter need to be avoided. The first requirement is fulfilled either by using a very low doping concentration of the phosphorescent emitter [15, 16], or by placing the phosphorescent emitter at an appropriate distance from the exciton generation zone [2, 3, 15, 17-21]. In this way, singlets, which usually have a smaller diffusion length than triplets due to their shorter lifetime, are not able to reach the phosphorescent emitter. Direct recombination of charges on the phosphorescent emitter can be prevented by using materials with suitable highest occupied (HOMO) and lowest unoccupied molecular orbital (LUMO) energies.

Known methods to verify TH include comparing the spectral emission to a reference OLED without the phosphorescent emitter [3], varying the concentration of the phosphorescent emitter [15, 16], or a distance variation between recombination zone and TH layer [2]. Furthermore, Kondakova et al. [17] investigated the harvesting process by time-resolved measurements and the dependence of emission intensity on an external magnetic field.

White TH OLEDs exceeding a luminous efficacy (LE) of 30 lm W−1 (at 1000 cd m−2) were reported by several groups: in 2008, Schwartz et al. [22] demonstrated 31.6 lm W−1 and an external quantum efficiency (EQE) of 15.2% [at 1000 cd m−2, CIE (0.49/0.41)] based on an OLED with three emitters. However, in these devices TH takes place only between the deep blue fluorescent emitter N,N′-di-1-naphthalenyl-N,N′-diphenyl-[1,1′:4′,1″:4″,1′′′-quaterphenyl]-4,4′′′-diamine (4P-NPD) which is also used as matrix, and the red phosphorescent emitter iridium(III)bis(2-methyldibenzo-[f,h]quinoxaline) (acetylacetonate) (Ir(MDQ)2(acac)). Triplets on the green emitter fac-tris(2-phenylpyridine)iridium (Ir(ppy)3) are generated either by direct charge recombination or by singlet transfer from 4P-NPD. Due to the low triplet energy of 4P-NPD (T1 = 2.3 eV), harvesting by Ir(ppy)3 (T1 = 2.4 eV) is not feasible.

A possibility to overcome this problem is to stack a blue–red TH OLED and a green–yellow PH OLED as demonstrated by Rosenow et al. [2]. With this four-color tandem OLED an LE of 33.0 lm W−1 [at 1000 cd m−2, CIE (0.51/0.42)] was obtained. Using high index glass and light outcoupling structures even 90.5 lm W−1 [at 1000 cd m−2, CIE (0.46/0.43)] were achieved. However, compared to single unit OLEDs the stacked design tends to increase the driving voltage, thus potentially impacting LE. Furthermore, the large cavity complicates light extraction since all emitters have to be placed into the corresponding maximum of the electromagnetic field simultaneously [13, 23].

A two-color white TH OLED with an LE of 30.1 lm W−1 and an EQE of 13.6% [at 1000 cd m−2, CIE (0.32/0.36)] based on the sky-blue fluorescent emitter difluoro[6-mesityl-N-(2-(1H)-quinolinylidene-κN)-(6-mesityl-2-quinolinaminato-κN1)]boron (MQAB) and the yellow emitter fac-bis(2-phenylpyridyl)(2-pyridylcoumarin)-iridium(III) (Ir(ppy)2(pc)) was shown by Kondakova et al. [17] in 2010. However, the emitter MQAB was inserted into a matrix material with a high band gap leading to higher driving voltage in comparison to OLEDs where blue emission originates from the matrix material.

Recently, Ye et al. [16] demonstrated high efficiencies of 33.7 lm W−1 and 22.2% [at 1000 cd m−2, CIE (0.46/0.44) at 100 cd m−2] by doping the orange phosphor tris(2-phenylquino-line)iridium(III) (Ir(2-phq)3) into the ambipolar sky-blue fluorescent matrix emitter 2,8-di[4-(diphenylamino)phenyl] dibenzothiophene-S,S-dioxide (DADBT) at a very low concentration of 0.1 wt.%. While using a low concentration can indeed suppress singlet transfer to the phosphorescent emitter in these structures, reproducible fabrication of mixtures with such low concentrations is experimentally challenging.

In this paper, we investigate TH in a simple two-color OLED based on 4P-NPD, which is used as matrix and blue fluorescent emitter, and the yellow phosphorescent emitter bis(2-(9,9-dihexyluorenyl)-1-pyridine) (acetylacetonate) iridium(III) (Ir(dhfpy)2(acac)). In contrast to the work of Ye et al., we use a preferentially hole transporting blue emitter. The singlet and triplet excitons are therefore generated close to the hole blocking layer (HBL). Furthermore, only a thin section of the 4P-NPD layer next to the electron blocking layer (EBL) is doped with Ir(dhfpy)2(acac). This allows us to use a doping concentration of 5 wt.% for Ir(dhfpy)2(acac) and to have triplets diffuse from the generation zone toward this doped layer where they are harvested by the phosphorescent emitter.

TH is studied by using a reference device without the phosphorescent emitter in conjunction with spectrally and time-resolved electroluminescence measurements by a streak camera. A variation in distance between the exciton generation zone and TH layer is used to adjust the color coordinates toward the warm white color point A. Furthermore, we study how the emission spectrum and the color coordinates depend on current density. Using optical simulation, we are able to fit the angular dependent spectral radiant intensity and model the electromagnetic field distribution inside the device. Furthermore, the simulation enables us to estimate the internal quantum efficiency (IQE), for which a value of 48.7% is obtained.

As a result of targeted optimization, we obtain highly efficient two-color white TH OLEDs with Lambertian emission characteristics and a very small color shift with viewing angle (Δxy) = (0.040/0.012) from 0° to 70°. At 1000 cd m−2, the device has CIE coordinates of (0.42/0.40) and reaches an LE and EQE of 27.1 lm W−1 and 9.4%, respectively, which can be further improved to 46.1 lm W−1 and 15.2% when using outcoupling enhancement techniques. We believe that these results are a significant step toward the successful application of TH for highly efficient commercial white OLEDs.

2 Experimental

2.1 Absorbance

Absorbance is calculated from transmission of Ir(dhfpy)2(acac) which was measured in toluol solution (concentration <10−5 M) with an MPC 3100 spectrometer (Shimadzu GmbH).

2.2 OLED processing

Organic materials are purchased and further purified by vacuum gradient sublimation. Thermal evaporation in a UHV chamber (Kurt J. Lesker Co.) at a base pressure of about 10−7 mbar is used to deposit the organic layers and the aluminum (Al) cathode without breaking the vacuum. The thickness of all layers is measured in situ via quartz crystals and doping is achieved by co-evaporation.

Devices are processed on commercial, pre-cleaned, and pre-structured indium tin oxide (ITO) coated glass substrates (Thin Film Devices Inc.). After fabrication, the OLEDs are immediately encapsulated under nitrogen atmosphere using glass lids which include a getter material. The active area of all devices is 6.49 mm2. All OLEDs are processed in one run on a single substrate eliminating possible run-to-run variability. The variation of layer thicknesses is achieved using mask systems. We employ blocking and doped charge transport layers [33, 34]. As p-type hole injection and transport material (HTL) 55 nm of N,N,N′,N′-tetrakis(4-methoxyphenyl)-benzidine (MeO-TPD) doped with 2 wt.% 2,2′-(perfluoronaphthalene-2,6-diylidene)dimalononitrile (F6-TCNNQ) is used. The n-type electron injection and transport layer (ETL) is 55 nm 4,7-diphenyl-1,10-phenanthroline (BPhen) doped with cesium (Cs). Cs doping was adjusted such that ETL exhibits a conductivity of 10−5 S cm−1. The EBL and HBL consist of 10 nm of 2,2′,7,7′-tetrakis-(N,N-diphenylamino)-9,9′-spirobifluorene (Spiro-TAD) and BPhen, respectively. These blocking layers confine charge carriers and excitons in the EML. The EML consists of an intrinsic 4P-NPD layer [15, 18] and 5 nm of 4P-NPD doped with Ir(dhfpy)2(acac).

2.3 OLED characterization and measurements

OLED characterization is entirely performed in air and at ambient temperature. Current–Voltage–Luminance (IVL) measurements are performed with a Keithley SM2400 source-measure unit and a calibrated photo diode. The spectral radiance in forward direction is recorded by a calibrated Instrument Systems GmbH CAS140 spectrometer. Time- and spectrally resolved measurements are done using a self-calibrated HPD-TA C5680 (Hamamatsu) streak camera with a C4792 (Hamamatsu) trigger unit. A 8114A (Hewlett Packard) pulse generator is used to drive the OLED in pulsed operation and to trigger the streak camera at the same time. Angular-dependent spectra are measured with a self-made spectrogoniometer including a calibrated Ocean Optics USB4000 miniature fiber optic spectrometer. Emission perpendicular to the substrate is referred to as forward emission or emission under 0°. EQE and LE are measured in a calibrated Ulbricht sphere. To restrict the measurement to light emitted into the forward hemisphere, the edges of the sample are covered. Light-outcoupling enhancement is realized by attaching a glass half-sphere lens (Biomedical Optics, 18 mm) to the substrate using index matching oil (Zeiss 518F, refractive index of 1.518). Singlet and triplet energy levels are taken from photoluminescence experiments of previous studies [15, 24, 26].

3 Working principle of the device

The layer architecture of the OLEDs under investigation is shown in Fig. 1(a). Electrical charges are injected via the Al and the ITO contact and transported through the transport and blocking layers to the EML. Since the hole mobility in 4P-NPD is significantly higher than the electron mobility (µh = 6.6 × 10−4 cm2/Vs and µe = 3.6 × 10−8 cm2/Vs [15]), the exciton generation zone is assumed to be located at the 4P-NPD/BPhen interface. A scheme of the energy levels and the suggested excitonic processes is shown in Fig. 1(b). The high singlet and triplet energy levels of BPhen (3.2 eV and 2.6 eV, respectively [24]) provide an efficient energy barrier for both exciton species. Hence, excitons can only diffuse in the direction of the phosphorescent dopant. In previous studies, we obtained a diffusion length of 4.6 ± 0.5 nm for singlets and 11.3 ± 3 nm for triplets in 4P-NPD [24, 25]. This means that for sufficient separation between the recombination zone and the phosphorescent emitter, the majority of singlet excitons will decay on the 4P-NPD molecules, resulting in blue emission.

Figure 1.

(a) Layer architecture of the OLED series under investigation. (b) Energy level scheme and proposed energy transfer processes. Due to their short lifetime, singlet excitons decay close to the BPhen HBL while triplet excitons diffuse to the phosphorescent emitter Ir(dhfpy)2(acac), are transferred to the lower-lying triplet level and emit additional light at longer wavelength.

The singlets which reach Ir(dhfpy)2(acac) are transferred to the singlet state of Ir(dhfpy)2(acac) and are converted to its emissive triplet state via intersystem crossing. Triplet excitons reaching Ir(dhfpy)2(acac) will also be transferred to its lower lying triplet level and lead to emission of the yellow phosphor. Since the emission of Ir(dhfpy)2(acac) depends on the amount of harvested triplets, i.e., on a diffusion-based process, the amount of yellow emission is expected to be sensitive to changes in distance between the generation and the harvesting zone.

It has been shown that despite the fact that the electron mobility in 4P-NPD is much lower than the hole mobility a non-negligible electron current is present in the 4P-NPD EML [24, 25]. Furthermore, the HOMO energy of Ir(dhfpy)2(acac) (−5.1 eV [26]) is higher than that of 4P-NPD (−5.7 eV [24]), which is expected to result in hole trapping on the Ir(dhfpy)2(acac) molecules. Hence, emission from Ir(dhfpy)2(acac) that results from direct recombination cannot be fully excluded. The interplay between the different excitation processes will be investigated in the following sections.

4 Results and discussion

4.1 Proof of triplet harvesting

In order to provide clear evidence of TH, we compare the emission spectra of a TH OLED and a reference OLED. The TH OLED has a 4P-NPD layer thickness of x = 9 nm [cf. Fig. 1(a)]. The reference OLED is exactly the same device with the exception that the doping with the yellow phosphor Ir(dhfpy)2(acac) is omitted. Using a similar device design has the advantage that the outcoupling efficiency (OE) will be identical in the TH and the reference OLED.

The IV characteristics of both OLEDs are shown in Fig. 2(a). The curves are perfectly overlapping which indicates that charge carrier transport or charge trapping on Ir(dhfpy)2(acac) is negligible, thus providing evidence that direct recombination on the Ir(dhfpy)2(acac) molecules does not significantly contribute to the overall emission.

Figure 2.

(a) Current density–voltage characteristics for OLEDs with and without the harvesting phosphorescent emitter at x = 9 nm. Inset: EQE versus current density. (b) Spectral radiance at a constant current density of 1.54 mA cm−2. Inserting Ir(dhfpy)2(acac) leads to a slightly reduced blue emission from 4P-NPD accompanied by significant additional emission in the yellow wavelength regime and a strong increase in EQE at low current density. The two latter effects strongly indicate TH from 4P-NPD to Ir(dhfpy)2(acac) as explained in the text. Also shown is the absorbance spectrum of Ir(dhfpy)2(acac) (gray line).

Figure 2(b) shows the emission spectrum of both devices for a constant current density of 1.54 mA cm−2. For a constant current density, we expect the charge injection to be the same in both devices. Relative to the reference device (Ir(dhfpy)2(acac) free), the blue emission of the device containing the yellow phosphor is slightly decreased, while a strong peak is observed in the yellow part of the spectrum. The reduction of the blue emission can be explained by diffusion and transfer of a certain amount of singlets from 4P-NPD toward Ir(dhfpy)2(acac). As shown in Fig. 2(b), 4P-NPD emission and Ir(dhfpy)2(acac) absorption have a considerable overlap which favors Förster transfer. The main contribution to the emission in the yellow part of the spectrum, however, is expected to be a result of triplets harvested by the Ir(dhfpy)2(acac) molecules. As a result, the EQE is more than doubled [cf. inset in Fig. 2(a)] for the Ir(dhfpy)2(acac) device at low current densities.

It is worth noting that the reference device-based solely on 4P-NPD as a fluorescent emitter reaches an EQE of more than 4%, which is close to the theoretical maximum for fluorescent bottom emitting OLEDs without outcoupling enhancement structures [27]. This indicates that the emitter 4P-NPD itself has a high IQE. The strong enhancement in EQE for the TH OLED and the nearly unchanged emission in the blue spectral region with respect to the reference device strongly suggest the presence of TH. At high current densities, the EQE of the TH OLED reduces significantly (roll-off). This finding is attributed to the high triplet density reached at these current levels which strongly increases the probability for triplet annihilation processes, such as triplet–triplet annihilation (TTA) or triplet-polaron annihilation (TPA) [28].

To further analyze TH in these devices, we investigate changes in the emission spectrum of the OLED in response to short electrical pulses using time-resolved measurements. During electrical excitation, singlet and triplet excitons are generated at the same time. Due to the short lifetime of singlets, fluorescence of 4P-NPD is expected to decay relatively fast, and will be followed by the delayed and extended Ir(dhfpy)2(acac) phosphorescence signal. The delay of the second signal is related to the triplet diffusion to the phosphor, i.e., it should depend on the distance x between the exciton generation zone and the TH layer. In Fig. 3(a) the time- and spectrally resolved emission following a voltage pulse (3.6 V for 2.5 µs) is shown for a device containing 4P-NPD and Ir(dhfpy)2(acac). Both emitters can be distinguished not only by wavelength, but also by their different transient behavior after the pulse is switched off. As expected, the emission at small wavelength shows a short decay and can be attributed to 4P-NPD; the signal at longer wavelength is characterized by a slow decay and is associated with Ir(dhfpy)2(acac) emission. Integrating the measured intensity in the two different wavelength regimes (from 390 to 540 nm for 4P-NPD and 540 to 610 nm for Ir(dhfpy)2(acac)), we obtain temporal profiles for each emitter [Fig. 3(b)]. We define τ as the delay time between the maximum of the emission intensity from the fluorescent and the phosphorescent emitter. For x = 9 nm, we find τ = 1.5 µs. As expected, the delay between fluorescent and phosphorescent signal increases with increasing distance x [inset in Fig. 3(a)], which is further evidence for the presence of TH.

Figure 3.

(a) Time- and spectrally resolved emission of the TH OLED with a 4P-NPD layer thickness of x = 9 nm after application of short electrical pulses (red arrow) of 2.5 µs and 3.6 V. (b) Normalized integrated intensity for both emitters. The delay time is defined as the time between the emission maxima. Inset in (a): delay time versus 4P-NPD layer thickness.

Interestingly, the emission of Ir(dhfpy)2(acac) shows an intermediate maximum at 2.4 µs. It is difficult to clearly assign this peak to one of the emitters: on the one hand, emission from 4P-NPD is not zero at wavelengths larger than 540 nm so that this signal could result from 4P-NPD emission. On the other hand, singlet transfer may lead to prompt emission from Ir(dhfpy)2(acac).

4.2 IVL and spectral emission

The IVL characteristics of the OLEDs under investigation are shown in Fig. 4(a). For a constant voltage, increasing the intrinsic layer thickness of 4P-NPD leads to a continuous decrease in current density and, even more dramatically, in luminance. The reduced current is attributed to a reduction in electric field within the thicker EML. The strong decrease in luminance can be explained by considering the change in emission spectrum between devices of this series [Fig. 4(b)].

Figure 4.

(a) Current density–voltage–luminance characteristics of OLEDs with 4P-NPD layers of different thickness. (b) Spectral radiance in forward emission direction at a constant current of 15.4 mA cm−2. The corresponding luminance is given in brackets. Inset: CIE color coordinates of the light emitted by the different devices; also shown is the warm white color point A and the CIE coordinates of the photoluminescence emission from 4P-NPD and Ir(dhfpy)2(acac).

The luminance is calculated using the luminosity function, i.e., the V-lambda curve, describing the sensitivity of the eye to light of different wavelengths. The maximum of this curve is at 555 nm which almost matches the emission peak of Ir(dhfpy)2(acac) (557 nm). With increasing 4P-NPD layer thickness, the emission of Ir(dhfpy)2(acac) and, as a result, the luminance decrease significantly. At the same time, blue emission from 4P-NPD increases but this affects the measured luminance to a lesser extent due to the small overlap of the blue emission with the V-lambda curve.

The increase in 4P-NPD emission and decrease in Ir(dhfpy)2(acac) emission with increasing 4P-NPD layer thickness can be explained when considering the singlet and TH processes (cf. Section 'Experimental'). While the blue singlet emission benefits from a larger distance, since fewer singlets are transferred to Ir(dhfpy)2(acac), yellow triplet emission suffers from a large distance, since triplets will then decay non-radiatively before they can reach the phosphorescent emitter. Hence, an optimized TH OLED structure is a trade-off between singlet loss and triplet gain. Using this approach, the CIE color coordinates can be conveniently adjusted along a line between the CIE coordinates of the 4P-NPD and the Ir(dhfpy)2(acac) emission [inset in Fig. 4(b)]. Emission close to the warm white color point A can be obtained for a 4P-NPD layer thickness of 3 and 5 nm.

A detailed analysis of the spectral emission and the CIE color coordinates for different applied current densities is given in Fig. 5 for a device with a 5 nm thick 4P-NPD layer. The inset shows the spectral radiance normalized to the applied current density. Due to the high triplet density, triplet quenching mechanisms, including TTA and TPA, reduce the emission of the Ir(dhfpy)2(acac) above a certain current density. The CIE color coordinates remain within one 7-step chromaticity quadrangle, thus meeting the Energy Star requirements for white light sources [29].

Figure 5.

CIE1931 color coordinates of the emission from a device with 4P-NPD thickness of 5 nm for different current densities. The corresponding luminance and CRI values are given in brackets. Inset: spectral radiance in forward direction for different current densities. While the blue emission is almost constant, the yellow emission decreases with increasing current (black arrow).

A CRI in the range of 40–50 can be obtained with this simple two-emitter white OLED. This value is too low for indoor lighting applications which require a CRI of >70, but in good agreement with other two-color white OLEDs. Ho et al. [30], e.g., obtained CRI values from 50 to 60 for blue/orange TH OLEDs. Using emitters with a broader photoluminescence spectrum or the insertion of additional green or red emitters would further enhance the CRI.

4.3 Angular dependence

The spectral radiant intensity of the OLED with 5 nm 4P-NPD layer is shown in Fig. 6(a) for viewing angles between 0° and 80°. The spectra are all measured at a current density of 15.4 mA cm−2. With increasing viewing angle, the blue and yellow emission decrease by similar amounts, which results in a negligible color shift with viewing angle. From 0° to 70°, the color shift of the CIE coordinates is only (Δxy) = (0.040/0.012). This finding is in good agreement with other studies of two-color white OLEDs. For example, Chang et al. [31] reported a CIE shift of (Δxy) = (0.021/0.012) from 0° to 75°.

Figure 6.

(a) Measured (lines) angular dependent spectral emission of the OLED with a 5 nm thick 4P-NPD layer at a current density of 15.4 mA cm−2. Open circles are fits based on a dipole model. The spectra are normalized to the peak of the 0° spectrum. Inset: polar diagram of the normalized radiant intensity as a function of viewing angle (red line) in comparison to an ideal Lambertian emitter (gray line). (b) Calculated 0° field profiles at peak emitter intensities 428 and 557 nm, respectively. The position of the emitters (indicated by a blue and orange box) is located near the maximum of the corresponding field.

Using optical simulations based on a semi-classical dipole model [13, 32], we fit the experimental data (open circles) by adjusting transport layer thicknesses within the uncertainties of our fabrication process. The agreement between simulation and experiment is remarkably good.

Furthermore, the device exhibits Lambertian like emission characteristics [Fig. 6(a) inset], which usually indicates that the OLED structure is close to the optical efficiency maximum. It is interesting to analyze the optical field profile calculated at the peak emission wavelengths of the emitters (428 nm for 4P-NPD, 557 nm for Ir(dhfpy)2(acac)) at 0° [Fig. 6(b)]. If the emitter is placed in the corresponding field maximum, the radiative efficiency can be expected to be high due to efficient coupling to radiative modes. On the other hand, emission is strongly suppressed for emitters placed in antinodes of the optical field. As shown in Fig. 6(b), the emitters are indeed in very close proximity to their respective field maximum, indicating the empirical device optimization indeed converges toward the optical optimum.

By using the model described in Refs. [13, 23] and fitting the 0° emission spectrum at a current density of 1.54 mA cm−2, we find a total OE of 22.4% for this OLED. The OEs of the two individual emitters are 16.4% for blue and 24.7% for yellow, respectively. By calculating the photon contribution for each emitter, we obtain a singlet to (harvested) triplet ratio of 0.28/0.72 = 0.39. This value is only slightly higher than the theoretical optimum which is given by the spin statistics of exciton formation, i.e., a singlet/triplet ratio of 0.33. Assuming a random orientation of the emitting dipoles and taking the measured EQE of 10.9% at 1.54 mA cm−2 (420 cd m−2) into account, an IQE of 48.7% is calculated, which is considerably lower than 73% obtained by Rosenow et al. [2] where TH was investigated using a similar OLED structure with 4P-NPD as blue emitter doped partially with the red phosphorescent emitter Ir(MDQ)2(acac). This result strongly suggests that the number of harvested triplets and/or the internal quantum yield of Ir(dhfpy)2(acac) are lower than that of Ir(MDQ)2(acac). It is important to mention that the calculated IQE includes losses due to non-perfect charge balance, the non-unity internal quantum yield of both emitters, and exciton quenching (TTA, TPA).

4.4 Efficiencies

Figure 7 shows the EQE and LE of the device series with varying 4P-NPD layer thickness as function of the luminance. Both efficiency measures tend to decrease with increasing 4P-NPD thickness. This behavior can again be understood by considering the properties of TH. At small thickness, a larger number of triplets (and singlets) can diffuse toward the phosphor and lead to additional yellow emission. When increasing the distance, triplets will undergo annihilation processes or decay non-radiatively before they reach the phosphor. Consequently, the efficiency will drop with increasing 4P-NPD thickness.

Figure 7.

(a) EQE and (b) LE versus forward luminance for the four devices under investigation. The efficiency decreases with increasing 4P-NPD thickness of the device due to lower number of harvested triplets. The inset shows a photograph of the OLEDs for a 4P-NPD layer thickness of 5 nm operated at 1000 cd m−2.

This fact is also reflected in the roll-off of the different devices: The roll-off for the OLED with a 3 nm thick 4P-NPD layer is lower than for the OLED with 9 nm. We attribute this to the larger interaction volume for 4P-NPD triplets in the intrinsic 4P-NPD layer which is expected to result in high annihilation rates. It is worth noting that the simple two color white OLED shows efficacies in the range of 30 lm W−1 at 1000 cd m−2 which is comparable to other state-of-the-art white OLEDs without additional outcoupling structures [6, 7].

The key data for the devices with 3 nm and 5 nm thick 4P-NPD layers are summarized in Table 1. The values are obtained at a forward luminance of 1000 cd m−2 for the bare device without outcoupling enhancement and exclude edge emission from the substrate. For x = 3 nm, an LE of 32.6 lm W−1 and EQE of 10.2% with CIE coordinates of (0.46/0.45) are achieved, whereas an LE of 27.1 lm W−1, EQE of 9.4%, and CIE of (0.42/0.40) are reached for x = 5 nm. By attaching a glass half-sphere, light outcoupling is enhanced and an LE of 46.1 lm W−1 (55.4 lm W−1) is obtained for the 5 nm (3 nm) device.

Table 1. Summary of the characteristics of the two most efficient devices at a luminance of 1000 cd m−2. Furthermore, the onset voltage, the efficiency values using outcoupling enhancement, and the maximum efficiency values are given
x (nm)U (V)j (mA  cm−2)CIE (x,y)CRICE (cd A−1)EQE (%)LE (lm W−1)Uon (V)EQEmax (%)LEmax (lm W−1)
  1. aUsing attached half-sphere for light outcoupling.

Finally, both OLEDs operate at low driving voltages, 1000 cd m−2 are reached at 2.9 and 3.0 V, respectively, the turn-on voltage is as low as 2.6 V.

5 Conclusions

We investigated excitonic processes in two-color hybrid white TH OLEDs and the optical optimization of these devices. The presence of TH in the analyzed OLEDs was proven by comparing emission intensities and EQEs of devices with and without the phosphorescent emitter. Spectrally and time-resolved measurements supported our finding that the emission of the phosphorescent emitter is based on triplet diffusion rather than on direct recombination. Additionally, singlet transfer was found to take place in our OLEDs, which was strongly increased if the phosphorescent emitter was in close proximity to the exciton generation zone. In addition, more triplets were harvested at a small distance, which had a positive influence on the efficiency roll-off. Using optical simulation and angular dependent emission spectra, we found that both emitters were located close to the electromagnetic field maximum of the cavity mode at the corresponding peak wavelength. We calculated an IQE of 48.7% and a singlet/(harvested) triplet ratio of 0.39 for a TH OLED, which had color coordinates close to the warm white color point A. This device showed an LE of 27.1 lm W−1 and an EQE of 9.4% at 1000 cd m−2 (CIE (0.42/0.40), CRI 46).

In future, to enhance the CRI, integration of green and red emitters would be required. Therefore, an efficient blue matrix emitter with a high triplet energy (T1 > 2.4 eV) allowing the harvesting by a green phosphorescent emitter could be beneficial. The achieved efficiencies are comparable to white state-of-the-art OLEDs and the calculations reveal great potential of efficiency enhancement, if internal losses can be further reduced.


Novaled AG, Dresden is acknowledged for materials support. Dr. Thomas C. Rosenow and Caroline Murawski are acknowledged for fruitful discussions. Dr. Susanne Hintschich, Jens Ludwig, and Paul-Anton Will are acknowledged for their help with streak camera measurements. The work leading to these results has received funding from the European Community's Seventh Framework Programme under grant agreement no FP7-224122 (


  • Image of creator

    Simone Hofmann is currently working as a Ph.D. candidate under the supervision of Prof. Karl Leo at the Institut für Angewandte Photophysik, Dresden (Germany). In 2009, Simone obtained the Diplomphysiker degree from the TU Dresden (Germany), where she was involved in the work of highly efficient monochrome top-emitting OLEDs. Her main research interests are singlet and triplet exciton dynamics in organic semiconductors and the subsequent development of highly efficient white OLEDs.

  • Image of creator

    Karl Leo obtained the Diplomphysiker degree from the University of Freiburg in 1985, and a Ph.D. degree from the University of Stuttgart in 1988. From 1989 to 1991, he was postdoc at AT&T Bell Laboratories in Holmdel, NJ, USA. From 1991 to 1993, he was with the Rheinisch-Westfälische Technische Hochschule (RWTH) in Aachen (Germany). Since 1993, he is full professor of optoelectronics at the TU Dresden. Since 2002, he is also working at the Fraunhofer-Institution for Organics, Materials and Electronic Devices COMEDD, presently as director. His main interests are novel semiconductor systems like semiconducting organic thin films, with special emphasis to understand basics device principles and the optical response.

  • Image of creator

    Malte C. Gather studied physics, materials sciences and photonics at RWTH Aachen University (Germany) and Imperial College London (UK). He obtained his Ph.D. from the University of Cologne (Germany) in 2008 under the supervision of Prof. Klaus Meerholz with a thesis on direct photolithography for OLED displays and color stability of white OLEDs. He was a Postdoctoral Fellow at the University of Iceland in 2008/2009 and Bullock Fellow at the Wellman Center, Harvard University (USA) from 2009 to 2011. Since 2011, he holds an Assistant Professorship at TU Dresden (Germany). His research revolves around soft-matter photonics, including organic electronics, biophotonics and plasmonics.