Efficient Trilayer Phosphorescent Organic Light-Emitting Devices Without Electrode Modification Layer and Its Working Mechanism

At present, numerous functional layers are introduced to improve the carrier injection and balance the carrier transport in organic light-emitting devices (OLEDs). Although it may be a good way to enhance the efficiency of devices, the introduction of functional layers would also result in extra process and long manufacture period. Actually, with the enrichment of material system, many appropriate materials could be chosen to share two or even more functions in OLEDs. Here, via impedance spectroscopy and transient electroluminescence analysis, di-[4-(N,N-ditolyl-amino)-phenyl] cyclohexane (TAPC) and 4,7-diphenyl-1,10-phenanthroline (Bphen) are demonstrated to serve as carrier injection and transport layers simultaneously. As a result, efficient trilayer OLEDs are achieved with comparable performances to conventional multilayer devices. Further studies have also been carried out to analyze the recombination and quenching mechanisms in devices. TAPC can block electrons effectively, while Bphen avoids the accumulation of holes. It makes carriers in emitting layer become more balanced, resulting in the reduction of efficiency roll-off.


Background
It is known to all that organic light-emitting devices (OLEDs) have attracted considerable attention for solid-state lighting, full color displays, and so on. A good deal of functional layers, such as the anode modification layer (AML), cathode modification layer (CML), holeblocking layer (HBL), and electron-blocking layer (EBL), have been introduced in the OLEDs to achieve high-efficiency and low turn-on voltage. The AML and CML are used to enhance the hole or electron injection, respectively [1,2]. While the HBL and EBL can efficiently block the diffusion of the exciton from the luminescent layer into the transport layer [3]. Obviously, the multilayer structure becomes a frequently used way to improve device performance. However, since one more layer means an extra preparation process, excess function layers would also cause the long period and high cost that limit the development of their industrialization. With the improvement of the organic material system, some materials could play multiple roles in OLEDs due to their prominent properties. For example, deoxyribonucleic acid-cetyltrimetylammonium complex can act as hole-transporting layers (HTL) because of high hole mobility, meanwhile the low lowest unoccupied molecular orbital (LUMO) energy level makes it suit for the EBL [4]. 4,4′,4″-Tris (carbazol-9-yl)-triphenylamine (TCTA) is usually used to be HTL; besides, it can also serve as the host in emitting layer (EML) because of its high triplet energy [5,6]. Hence, it is possible to simplify the structure without sacrificing the device performance by choosing appropriate material. However, few studies have been carried out on phosphorescent white OLEDs (PHWO-LEDs) with simple structure [7,8].
More recently, capacitance characteristics based on impedance spectroscopy (IS) measurement has been a widely used tool to investigate the physical mechanisms of OLEDs. The inflection point of the first peak in capacitance-voltage (C-V) curves has been reported to be corresponded to the turn-on voltage of OLEDs. It is also a very sensitive probe of carrier accumulation caused by the barrier in the interface of organic layers or the imbalance of charge injection and transport in devices [9][10][11][12][13][14][15][16][17]. Meanwhile, transient electroluminescence (EL) has also been the subject of intense technological as well as fundamental research, because transient EL studies have generated insight into the internal working mechanism in OLEDs. Transient EL is investigated by driving the devices with short, rectangular voltage pulses. The response times obtained from transient EL characteristics of devices provides an essential criterion for their application [18][19][20][21][22][23][24][25][26][27][28].
All devices are prepared on glass substrates covered with patterned indium tin oxide (ITO) stripes. Before film deposition, the ITO glass substrates are subjected to a routine cleaning process with rinsing in Decon 90, deionized water, drying in an oven, and finally treated in a plasma cleaner chamber for about 5 min. The PEDOT:PSS films are fabricated by spin coating from aqueous solution before depositing with the thickness to be approximately 40 nm, and then the PEDOT:PSS films are all annealed at 120°C for 10 min.
All organic layers and cathode are evaporated by thermal vapor deposition using resistively heated tungsten filament and metal boats under high vacuum (~5 × 10 −4 Pa) at a rate of 1-2 Å s −1 monitored in situ with a quartz oscillator. The cathode we used in our experiments is Mg:Ag (15:1) alloy, which is controlled independently by separate thin-film deposition monitors, so does the doping process in EML. Finally, four active areas of the devices on each substrate were 10 mm 2 , which is decided by the overlap between the anode and cathode via using a shadow mask [24,25].

Characterizations
Luminance-current density-voltage characteristics and spectra of unpackaged devices are measured simultaneously using Goniophotometric Measurement System based on spectrometer (GP-500, Otsuka Electronics Co. Osaka, Japan) in air at room temperature.
For the transient voltage decay measurement, high-speed switching diode (Philips, 1N4531) and arbitrary waveform generator (Rigol, DG5102) are connected with our devices in series orderly, and the transient voltage of the devices is recorded by a digital oscilloscope (Rigol, DS4054) after a consecutive signal averaging. In the transient EL measurement, the tested devices are driven by pulsed voltage with a pulse width of 1 ms using arbitrary waveform generator (Rigol, DG5102) as an electrical switch for driving tested devices and a trigger signal for starting the collection of EL signals. The transient EL response was detected and collected by using an avalanche photodiode (C30902) and time-correlated single-photon counting system.
The capacitance-voltage (C-V) characteristics are measured with an Impedance Analyzer (TH2829C, Changzhou Tonghui Electronic Co., Ltd., China) with oscillating amplitude of 100 mV and the repetition rate of 1 kHz. The range of dc bias applied by this setup allows sweeping from 0 to + 10 V [26].

Efficient OLEDs Simplified Without AML
To get rid of AML, we choose TAPC as HTL in green phosphorescent OLEDs, because the highest occupied molecular orbital (HOMO) energy level is similar to the working function of ITO [5]. We perform contrastive experiments on an ITO/x/CBP:10 wt% Ir(ppy) 3 (30 nm)/ TmPyPB (50 nm)/LiF (0.5 nm)/Mg:Ag (120 nm) OLEDs, while the structure of x is TAPC (50 nm), MoO 3 (3 nm)/ TAPC (50 nm), and PEDOT:PSS (50 nm)/TAPC (50 nm), respectively. In order to differentiate the three devices, we mark them as D 1 , D 2 , and D 3 in turns. Firstly, we investigate hole injection ability of these devices by analyzing their capacitance-voltage and current density-voltage-luminance characteristics. As we can see in Fig. 1a, the turn-on voltage of the three devices is about 3 V. It is relevant to the maximum of the first peak in their capacitance-voltage characteristics, indicating that it makes no difference to the turn-on voltage without AML in D 1 [9][10][11]. Figure 1b shows the current density-voltage (J-V) characteristics of the three devices in log-log scale, we divide the J-V curves into three regions, (I) leakage or diffusion-limited current caused by Ohmic contact, (II) volume-controlled current with an exponential distribution of traps, and (III) volume-controlled current with partly filled traps [20]. The higher current density of device D 3 at low applied voltage in region I may be attributed to the leakage current caused by the rough film morphology of solution-processed PED-OT:PSS films. In addition, the right shift of the turning point between region I and region II (from A to A") presents the strongest carrier injection in D 1 , while the highest capacitance value of D 1 indicates that more holes inject in the device and then accumulate in the interface or bulk [29]. Obviously, the interface of ITO/TAPC shows better hole injection ability. We can also find that the current density of D 1 is larger than the values of the other two devices with the increase of the applied voltage. It may be attributed to the dipole layer generated between the ITO/ TAPC interfaces. After introducing an extra AML, the intrinsic dipole layer is broken, resulting in the weaker injection ability among the two devices [10,30]. In the reported references, the AML may be used to reduce the trap density which may have an impact on the stability of device [31]. For D 1 , the slope of the J-V curve in region III (m = 11) is larger than the values of D 2 and D 3 (m = 7, 8), the higher value of m always means higher trapping density [18]. The higher trapping density of device D 1 may be attributed to the morphology change of TAPC film because of the lack of wetting layer, such as MoO 3 or PEDOT:PSS. Moreover, the turning points C and C' shown in Fig. 1 are relevant to the rapid increase of the injection of electron with the increase in bias voltage.
A further study is carried out to research the carrier injection of the above devices by the transient voltage discharge characteristics. The test circuit is shown in Fig. 2a. Two response times are observed in Fig. 2b under the applied voltage of 5 V. The fast decay time τ 1 is about 100 μs in the inset of Fig. 2b. Then, a followed slower decay τ 2 is higher than one order of magnitude (τ 2 is in the millisecond scale) [7]. The diode is regarded as a wire when the generator provides positive voltage. Charge carriers can transfer into the device easily, and then with the carrier injection barrier, there are a certain number of holes and electrons accumulated at the interface between organic layers, anode and cathode, respectively. The diode becomes infinite resistance inversely when the applied voltage turns to be negative. Charge carriers cannot reach the device, so the residual holes in the interface of ITO/organic layer can flow through organic layers and neutralize the remanent electrons diffused or drifted by space charges from cathode interface. Therefore, the downtrend of two response times, especially the τ 1 are determined by the hole injection and transport ability of the organic layers in our contrast devices. It is obvious that the voltage of D 1 falls at fastest rates, representing an excellent hole injection ability with the structure of ITO/TAPC merely. As the resistances of internal resistances in our samples reach to the magnitude of MΩ, the influence of the oscilloscope with 1 MΩ resistance cannot be ignored. That is why, only a little distinction can be seen in the three downtrends of τ 2 [21,22].

Efficient OLEDs Simplified Without CML
Afterwards, we design a new contrastive experiment with the further simplification of ETL. As described in the reference reported by Scholz et al. [32], the metal-organic donor-acceptor adducts [Bphen+Ag] + and [2Bphen+Ag] + will form at the Ag-on-BPhen interface due to a self-doping effect. Our previous experimental results also indicated that these metal-organic adducts will improve the injection of electrons from Mg:Ag (15:1) to Bphen. Therefore, Bphen is chosen to be the appropriate experimental electron-transporting material here. The structure is ITO/TAPC (50 nm)/CBP:10 wt% Ir(ppy) 3 (30 nm)/y/Mg:Ag (120 nm). The y is TmPyPB (50 nm)/LiF (0.5 nm), TmPyPB (50 nm), and Bphen (50 nm). S 1 , S 2 , and S 3 are defined as the three samples, respectively. Figure 2c shows the turn-on characteristics of these three samples. It can be seen that S 3 has the same turn-on voltage (V on = 3 V) with S 1 , the luminance-voltage characteristics of S 3 are also similar to those of S 1 in the inset of Fig. 2c. So, we conclude that the simple structure in S 3 owns great electron injection ability, which  Fig. 2d show that the EL onset times of devices S 1 , S 2 , and S 3 are about 0.32 μs, 1.05 μs, and 0.48 μs, respectively. The EL onset time is also called delay time (t d ). It is composed of the injection time t inj and transport time t trans . The larger threshold voltage V th results directly in the longer t inj . Therefore, it is straightforward to prove that S 3 can also possess excellent electron injection ability [23][24][25].

Performance Comparison Between Simple Trilayer and Multilayer OLEDs
Finally, simple green PHOLED with a trilayer structure is obtained as shown in Fig. 3a,  Although the current density and luminance of device 3 are lower than those of the other two devices as shown in Fig. 3b, the same turn-on voltage could also be observed. It indicates that the carrier injection has not been influenced by simplifying the electrode modification layers. Nevertheless, it is confused that efficiency of device 3 shows a lowest roll-off in Fig. 3c.
To explain the existence of better efficiency roll-off in device 3, we then simulate exciton-quenching mechanism via the mathematical model with the function between the EQE and J. There are two exciton-quenching mechanisms existing in PHOLEDs, i.e., triplet-triplet annihilation (TTA) and triplet-polaron annihilation (TPA). The rate equation in the model is shown as follows: For Eq. (4), we deem that charge carriers recombine via Langevin recombination with the rate K L , where q is the elementary charge, μ e/h is the mobility, ε r is the relative permittivity, and ε 0 is the permittivity of free space. The triplet and polaron densities, n T and n P , were calculated by Eqs. (5) and (6), where K TT and K TP are the rate constants describing the kinetics of the TTA and TPA process. Actually, the internal quantum efficiency (IQE) is the ratio of radiative decaying triplets over the number of injected electrons from Eq. (7). For simplification, we do not consider light outcoupling. Moreover, the electric efficiency and the PL quantum efficiency at low current density are set to 1. Hence, the calculated IQE is used to compare with experimental EQE [33].
As we can see from Fig. 4b-d, serious exiton-quenching effect existed in device 1 and device 2, especially TPA. CBP is bipolar transport material, but the hole mobility is an order of magnitude higher than the electron mobility. Combined with the schematic energy-level diagrams in Fig. 4a, the recombination zone should be adjacent to the interface of EML/ETL. Besides, we find that the HOMO and LUMO energy levels of Bphen are similar to those of CBP; therefore, it is easier for holes to traverse CBP layer into Bphen and few holes are accumulated at the interface between CBP and Bphen. As to device 1 and device 2, a larger energy gap between TmPyPB and CBP can also be seen in Fig. 4a, resulting in an extra hole accumulation at the interface of CBP/TmPyPB. The different hole accumulation at the interface of CBP/TmPyPB would make different influences on the excitons formed at the same interface, resulting in different TPA of devices finally.

Analysis to the Mechanism of Exciton Recombination in Monochrome PHOLEDs
As we all know that the low concentration of phosphorescent dopant molecules leads to the long intermolecular distance, it is generally believed that phosphorescent materials act as trapping for the charge carrier. Therefore, there are two recombination mechanisms in EML of PHOLEDs, Langevin recombination I and trapassisted recombination II. For the former, when the device is driven by applied voltage, a mass of carriers inject continuously into EML. The holes transfer through the host material, followed by an accumulation in the interface of EML/ETL. On account of a good matching to energy levels between ETL and cathode, most electrons flow through ETL up to EML and then recombine with the stored charge. In this case, excitons generated in the host material transfer to the dopant by the Förster and/ or Dexter mechanisms; therefore, it belongs to the bimolecular recombination. The latter recombination zone is located in dopant due to the shallow-energy-level trapping formed by phosphorescent guest [27].
It is necessary to investigate the mechanisms mentioned above. As different recombination types playing a leading role in EML, it will have different impact on device performance. Structure of devices with the different dopant in EML is shown in Fig. 5a.    Fig. 5b, c is tested by changing the reverse bias (0 V, − 1 V, − 3 V, and 5 V) after the applied voltage turning off, while the voltage pulse height corresponds to a current density of 90 mA cm −2 . The voltage pulse width is 1 ms, and the pulse frequency is 100 Hz. As shown in Fig. 5b, c, the rise time of green and red devices slow down with the increase of reverse bias. However, this phenomenon does not occur in the other two devices. The reverse bias would take the captured carriers out of the trapping sites, and then the trapped carriers will make less contribution to EL intensity. So, we infer that trap-assisted recombination probably consists in devices fabricated by CBP-doped Ir(MDQ) 2 (acac) or Ir(ppy) 3 due to the existence of the trapped charges [27].
Further study of the existence of trapped charges is developed by impedance spectroscopy measurement with the result of capacitance-voltage curves shown in Fig. 6a. Two strong peaks could be observed in the C-V characteristics of green and red devices. Moreover, there is only one apparent peak in the blue device. The bias voltage corresponding to the first peak of the three devices is almost identical to the turn-on voltage. It can be interpreted that charge carriers inject constantly into devices when devices start to be driven by the applied voltage, resulting in the increase of capacitance at low voltage. And then for the green device, we deem that a small amount of the injected holes are captured by trapping via phosphorescent dye. Subsequently, they are recombined with electrons from cathode causing the trap-assisted recombination. Hence, parts of these accumulated charges begin to reduce at approximately 3 V. Similar phenomenon can be seen in the C-V curve of the red device, the falling of the first peak at 3.5 V is caused by trap-assisted recombination. In addition, the higher peak of C-V curve from 2.5 to 5 V can attribute to the stronger trapping effect in the red device.
More holes inject with the augment of applied voltage; besides the trapped ones, most of them get to store at the interface of EML/Bphen. Therefore, both of the C-V curves of green and red devices rise again. At this point, the Langevin recombination has happened in the EML causing the reduction of internal stored carriers. When the dissipative rate of charges exceeds their injection rates, the accumulated charges reduce rapidly and the C-V curve exhibits a sharp drop. The recombination process is shown in Fig. 6b, c. For comparison, only one strong peak appears in the capacitance characterize of the blue device, indicating that only the Langevin recombination occurs in the EML. Schematic energy-level diagrams with the recombination mechanism are shown in Fig. 6d.
We can also verify our results via the mathematical model mentioned above. It is well known that TTA is caused by high triplet density, while the high Langevin recombination rate would reduce the triplet density. So, the TTA can be associated with the Langevin recombination. TPA depends on the charge trapping characteristics of the host-guest system: when the emitter molecules constitute a trapping site for polarons within the host, accelerated TPA can be expected [33].
The corresponding contribution of TTA and TPA to the overall annihilation for the two devices with the EML of CBP:Ir(ppy) 3 and CBP:Ir(MDQ) 2 (acac) is shown in Fig. 6e, f. The calculated IQE is coincident to the measured EQE; moreover, the distinction between IQE and EQE curves at low bias voltage is caused by leak current. For the two devices, the polaron density is larger than the triplet density when the current density is below 5 mA cm −2 . Therefore, we believe that there are two quenching processes on operation condition, meaning that two recombination types occur in the EML. A higher percentage of TPA occurs in the red device, reflecting the stronger trap-assisted recombination [33,34].
In terms of the quenching process discussed above, it is obvious that TTA and TPA may dramatically decrease the efficiency of phosphorescent OLEDs. Therefore, in order to research the effect on device performance by changing host material, we prepare red devices with different hosts, i.e., CBP, TCTA, 2,6-bis(3-(carbazol 9,9′-[4′-(2-ethyl-1H-benzimidazol-1-yl)-9-yl) phenyl)pyridine [26DCzPPy] and 2,2′[2″-1,3,5-benzinetriyl)-tris(1-phenyl-1-H-benzimidazole) [TPBi]. When CBP is used as the host, the TTA and TPA are efficiently limited. Therefore, the CBP is chose to act as the host in this work.  curves. c Normalized EL spectra of the white OLEDs. The orange arrow shows the weakened spectra versus applied voltage single-EML WOLEDs possess a low turn-on voltage below 3 V. Moreover, we achieve a high current efficiency of 21 cd A −1 . Normalized EL spectra of the device in Fig. 7c show that the red intensity tends to be weakened when the bias voltage increases from 5 to 9 V. It should be attributed to that the trapping effect of the red dye molecule merely plays a major role under low bias voltage. At a practical luminance of 5840 cd m −2 , the CIE coordinates of devices are (0.39, 0.39), corresponding to warmish-white emission.

Conclusions
In summary, efficient phosphorescent OLEDs have been prepared based on a simple trilayer structure (TAPC/ EML/Bphen). We simplify the devices gradually via impedance spectroscopy and transient measurement. The EL performances of trilayer devices could be still comparable to the conventional devices with modification layers. Langevin recombination and trap-assisted recombination are certified to be existed in red and green phosphorescent devices by capacitance-voltage measurement. In addition, mathematical model is used to describe the TTA and TPA quenching processes, which are relevant to the two recombination types mentioned above. Based on the above analysis, we obtain the efficient WOLEDs with low roll-off. These results demonstrate an effective approach towards simplified OLED with high efficient and low cost.

Availability of Data and Materials
The datasets supporting the conclusions of this article are included within the article.
Authors' Contributions XMP designed and conducted most of the experiments, analyzed the data, and prepared the manuscript. HWF contributed to transient EL and transient voltage decay measurement. JXZ helped with the white OLED fabrication. SHL helped with the mathematical simulation and prepared the manuscript. LTZ contributed to capacitance-voltage characteristics and data analysis. WFX initiated the study, designed all the experiments, analyzed the data, and prepared the manuscript. All authors discussed the results and commented on the manuscript. All authors read and approved the final manuscript.