Thin Solid Films 515 (2007) 7264 – 7268 www.elsevier.com/locate/tsf
Modeling and numerical study of electrical characteristics of polymer light-emitting diodes containing an insulating buffer layer Dexi Zhu ⁎, Hui Ye, Jun Gao, Xu Liu State key laboratory of Modern Optical Instrumentation, Zhejiang University, Hangzhou 310027, PR China Received 30 May 2006; received in revised form 1 February 2007; accepted 1 March 2007 Available online 12 March 2007
Abstract In order to improve the electrical characteristic of polymer light-emitting diodes, a simple model for the device characteristic with an insulating buffer layer at cathode is proposed. This model is based on Fowleer–Nordhein tunneling mechanism and Poission's equation. An additional tunneling factor which characterises the tunneling effect of buffer layer is introduced. The simulated current–voltage characteristic indicates how an insulating buffer layer with suitable thickness decreases the barrier height at the cathode and therefore increases the electron injection. The model is validated by experimental results of devices with BaO as the buffer material and poly[2-methoxy-5-(2′-ethyl-hexyloxy)-1,4phenylenevinylene] as the emission material. An optimum thickness of the buffer layer is also obtained from the model, which provides a guide to device design. © 2007 Elsevier B.V. All rights reserved. Keywords: Polymer light-emitting diodes; Model; Insulating buffer layer; Electrical characteristics
1. Introduction Polymer light-emitting diodes (PLEDs) have been attracting much attention recently as a technology for full-color displays. The electroluminescence from PLEDs is caused through relaxing of excitons which are formed by carrier injection from the anode and cathode. By the study of the basic principles of electroluminescence of PLEDs, it is found that the balance between the hole and electron concentrations has a crucial influence upon the electrical characteristics, such as drive voltage and efficiency of luminescence. In addition, most emissive polymeric materials act as hole transporters, and are found poor in electron transport. For example, hole mobility of poly[2-methoxy-5-(2′-ethyl-hexyloxy)-1,4-phenylenevinylene] (MEH-PPV) is about ten times larger than that of electron. For this reason, it's helpful to improve the electron injection from the cathode for the purpose of achieving better electrical and optical properties for PLEDs [1].
⁎ Corresponding author. Tel.: +86 571 87951190; fax: +86 571 87951758. E-mail address:
[email protected] (D. Zhu). 0040-6090/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2007.03.001
Many efforts have been made to improve the electron injection. Among them proper insertion of a thin buffer layer between the cathode and the polymer layer has been reported [2]. Usually, the materials used as the buffer layer are insulators, such as alkali–halide [3] or oxide [4]. In contrast to the conventional view that an extra insulating layer might increase the turn-on voltage, the fact is that the turn-on voltage decreases and quantum efficiency increases when a buffer layer is introduced [5]. Park et al. proposed that the mechanism of the enhanced electron injection originated from a lowering of the barrier height between the cathode and emission layer [6]. The energy level diagram is shown in Fig. 1, where 1, 2, 3 and 4 are the anode (or hole-injecting layer), the polymer layer, the cathode and the buffer layer, respectively. Ef is the cathode Fermi level. LUMO means polymer lowest unoccupied molecular orbital while HOMO means highest occupied molecular orbital. If no buffer layer is inserted, the electron must tunnel through the barrier from the Ef to LUMO as shown in the left diagram. When a buffer layer with proper thickness is introduced as shown in the right diagram, the forward voltage drops across the insulating layer is ΔV. So the barrier will be eΔV smaller than that in the buffer-free case, resulting in an increased injection current.
D. Zhu et al. / Thin Solid Films 515 (2007) 7264–7268
7265
exponentially inside. This is analogous to the diffusion of minority carriers in the active layer of semiconductor. Therefore, the densities of holes and electrons can be expressed as following [9]:
Fig. 1. A sample model of energy level diagram of PLEDs with different cathodes.
In general, device model provides a useful tool for facilitating device design. Some numerical models have been reported. Y. Kawabe et al. [7] proposed a model for single layer devices in which the current density–voltage (J–V) characteristics were obtained numerically. It showed that the current depends on the barrier height at the metal-organic interface. However, the additional insulating layer was ignored in this model, which was therefore unsuitable for bilayer-cathode devices. In this paper, we propose a simple model for simulating the device characteristics of PLEDs containing insulating buffer layers with different thickness. This model is based on Poisson's equation and Fowleer–Nordhein (FN) tunneling mechanism. A tunneling factor is introduced to take the effect of the insulating buffer layer into consideration. The current densities are simulated for different thicknesses of buffer layers and the optimum thickness is obtained for a certain material, which is in agreement with the experimental result. 2. The model For the research of the effect of a buffer layer on the carrier injection, the interfaces between emission layer and two electrodes are significant. So double-carrier injection is considered in our model. Meanwhile, we assume that the injection is driven by FN tunneling mechanism [8] which shows the relationship between current and potential barrier height at two interfaces. We define the coordinate that the anode position is at x = 0, and the cathode one is at x = d. The current density can be expressed by the following equation: E2 U3=2 J0 ¼ FNðEÞ ¼ k1 exp k2 U E k1 ¼
2:2e2 8kh
k2 ¼
8k ð2em⁎ Þ1=2 3h
ð1Þ
where e is the elementary charge, h is Planck's constant, m⁎ is the effective mass of an electron or a hole (for simplicity, m⁎ = m0 herewith), Φ is the potential barrier height, and E is the electric field at the interface where the carriers are injected. We assume that the carrier density in emission layer is not uniform. That means the carrier density is the largest at the interface between electrode and emission layer and declines
pðxÞ ¼ p0 expðx=r0 Þ
ð2:aÞ
nðxÞ ¼ n0 expððx dÞ=r0 Þ
ð2:bÞ
where x is the coordinate normal to the device. x = 0 corresponds to the interface position of anode and emission layer. p and n are the densities of holes and electrons, respectively. r0 is the characteristic length characterising the distribution of carriers in the emission layer. p0 and n0 are the largest densities in the device. We assume that the current densities caused by holes and electrons are constant throughout the device, given by: Jh ðxÞ ¼ Jh ¼ E0 p0 Ah e
ð3:aÞ
Je ðxÞ ¼ Je ¼ Ed n0 Ae e
ð3:bÞ
where subscripts h and e refer to holes and electrons, respectively. μ is the carrier mobility, Ed is the electric field at x = d. The electric field and carrier density are related through Poisson's equation: dEðxÞ e ¼ fpðxÞ nðxÞg: dx e
ð4Þ
Where ε is the dielectric constant of the polymer layer. By integrating Eq. (4) and combining it with Eqs. (2.a) and (2.b), we get the electric field distribution: ep0 r0 x EðxÞ ¼ E0 þ 1 exp r0 e en0 r0 xd d exp exp : ð5Þ r0 r0 e FNðE0 Þ FNðEd Þ ; n0 ¼ according to Eqs. (1), (3.a), E0 Ah e Ed Ae e and (3.b). So we can get Ed at x = d. The voltage can be derived by integrating Eq. (5): ep0 r0 d r0 exp V ¼ E0 d þ 1 þd r0 e en0 r0 d d r0 1 exp dexp : ð6Þ r0 r0 e Where p0 ¼
The total current density is the sum of current densities of electrons and holes: Jtotal ðE0 Þ ¼ Jh þ Je :
ð7Þ
Where Jh and Je are both in the form of Eq. (1), which is a function of electric field E0 when other parameters are known. Eq. (6) shows the voltage V is also a function of E0. It means all equations can be determined when E0 is chosen. Therefore, the J–V characteristic can be plotted according to Eqs. (6) and (7). Fig. 2 shows the simulated J–V curve based on the buffer-free
7266
D. Zhu et al. / Thin Solid Films 515 (2007) 7264–7268
Fig. 2. Calculated J–V plot of buffer-free device.
device structure discussed in Section (3). All parameters mentioned in the equations are listed in Table 1. We have discussed the current injection for the simplest device structure of anode/polymer/cathode. If an insulating buffer is introduced, the condition at the cathode is different. When the potential V is applied between two electrodes, the potential drop across polymer layer and insulating layer can be expressed as following [2,6]: Vp ¼
ðep =dp ÞV þ r ðeb =db ÞV r ; Vb ¼ ; Vp þ Vb ¼ V ep =dp þ eb =db ep =dp þ eb =db
ð8Þ
where subscripts p and b refer to polymer and buffer, respectively. d is the thickness of each layer and σ is the interface charge density caused by charge confinement at the interface between the emission layer and the insulating layer. The potential drop across the emission layer is related to the layer thickness and the dielectric constant. By inserting a thin buffer layer, the slop of band bending in the emission layer will be lowered. So the probability of electron tunneling will increase due to reduced energy barrier (Φ′e = Φe − Vb) of the device. On the other hand, the insulating layer may cause an additional barrier for electron injection. Here we introduce an additional concept to this model. Based on quantum theory, the transmission factor of square potential barrier can be described as following [10]: 2a pffiffiffiffiffiffiffiffiffiffiffiffiffiffi T ~Ub exp 2m0 Ub ð9Þ J
as one dimension square potential barrier. Therefore, the parameter a in Eq. (9) is replaced by the thickness of buffer layer. T is called the tunneling factor in our model. The thickness and barrier height of the buffer layer have important effects on the tunneling factor. BaO and LiF are used as buffer layers for comparison. Their work functions are 2.9 eV and 3.5 eV [11], respectively. Fig. 3 shows the calculated dependence of the tunneling factor on the thickness of the buffer layer according to Eq. (9). The tunneling factor T decreases exponentially as the thickness of the buffer layer increases. T is nearly zero when the thickness of buffer is more than 0.5 nm. This is not in agreement with the experimental results. We suppose that the atoms of the buffer material diffuse into the near-surface region. This interfacial region has an approximate scale of about several angstroms. Therefore the thickness of the buffer layer must be smaller than the expected value. From Fig. 3 we can also see that at the same nominal thickness, the tunneling factor of LiF is larger than that of BaO, that is, LiF buffer layer can be thicker than that of BaO at the optimum value of electron injecting current density. As discussed above, there is an analogy between the meanings of the tunneling factor T and the transmission factor of square potential barrier. The tunneling factor means the probability of electron injection, just as the meaning of transmission. Based on this ideal, we multiply Fne by T to get the modified current density of electrons injected from cathode. The modified current density J′e can be expressed as following: ! Ed2 ðA Vb ðdb ÞÞ3=2 exp k2 Je V¼ FN e T ~k1 A Vb ðdb Þ Ed 2db pffiffiffiffiffiffiffiffiffiffiffiffiffiffi Ab exp 2m0 Ab : (10) J Fig. 4 shows the dependence of current density of electrons on the thickness of the buffer layer (BaO) at the voltage of 12 V according to Eq. (10). As db increases, FNe increases, and T reduces synchronously. That means J′e reaches an optimum value when the thickness of the buffer layer is proper, as shown
where Φb is the barrier height of the buffer layer and a is the width of the barrier. For simplicity, we consider the buffer layer
Table 1 Parameters of the materials used in the experiment Parameter Hole mobility, μh Electron mobility, μe Characteristic length, r0 LUMO HOMO PEDOT:PSS Work function Relative dielectric constant of polymer, εp Relative dielectric constant of BaO, εb
Magnitude −8
0.5 × 10 0.5 × 10− 9 3 × 10− 8 3.4 4.9 5.2 3 34
Units m2V− 1s− 1 m2V− 1s− 1 m eV eV eV Unitless Unitless
Fig. 3. Calculated dependence of tunneling factor on the thickness of buffer layer when BaO and LiF are used.
D. Zhu et al. / Thin Solid Films 515 (2007) 7264–7268
Fig. 4. Calculated dependence of current density of electrons on the thickness of buffer layer (BaO) at the voltage of 12 V.
in Fig. 4. The modified Je − V characteristic at a proper thickness of the buffer layer is shown in Fig. 5. Here the current density of holes is not considered. When a buffer layer with proper thickness is introduced, the current density of electrons is larger than that of a buffer-free device at the same voltage, which attributes to the improved electron injection.
7267
Fig. 6. Experimental J–V characteristic of devices using BaO/Al or bare Al as cathode.
spectrum was measured by Ocean Optics OOIbase32 fiber spectrum meter. 4. Results and discussion
MEH-PPV purchased from Aldrich was used as emission material for the devices discussed in this paper. 80 nm thick poly (3,4-ethylene dioxythiophene):poly(styrenesulfonate) (PEDOT: PSS) was spin-coated on oxygen plasma treated Indium-TinOxide (ITO) substrates as hole-injecting layer. A drying process in vacuum drying oven at the temperature of 100 °C for 1 h was necessary for removing the water solvent. A 100 nm thick MEHPPV layer was spin-coated on PEDOT:PSS layer. Bare Al or bilayer-cathode with BaO as the buffer layer was evaporated as the cathode. The pressure during evaporation was kept at 10− 4 Pa. The rate of evaporation of BaO was kept at 0.2 nm/s then Al was evaporated from a second boat at the rate of 10 nm/s. All the prepared devices were encapsulated before the test. The J–V characteristic was measured by Agilent 4155C Semiconductor Parameter Analyzer. The electroluminescence (EL)
In order to clarify the effect of the inserted insulating buffer layer at the cathode, two device structures were fabricated for comparison: ITO/PEDOT:PSS/MEH-PPV/Al and ITO/PEDOT: PSS/MEH-PPV/BaO/Al. The experimental J–V characteristics of these devices are plotted in Fig. 6. It shows that the current density of the device with a buffer layer of proper thickness is higher than that of device with bare Al because of the lowering of the barrier height when a buffer layer exists, as discussed in Section (3). The optimal thickness was about 2 nm for our devices. Fig. 7 shows the EL spectrum of the devices with different cathodes. The lines from top to bottom stand for devices with 2 nm thick buffer layer, 1 nm thick buffer layer and no buffer layer, respectively. If the buffer layer is too thick, the electron tunneling across this layer will be difficult as the tunneling factor T decreases. Therefore, the intension of light will be lowered as expected. It indicates that a buffer layer with proper thickness will increase the output efficiency. These results prove our model reasonable.
Fig. 5. Calculated current density of electrons vs. voltage plot of devices using BaO/Al or bare Al as cathode.
Fig. 7. EL spectra of devices with buffer layers (BaO) of different thickness and without buffer layer.
3. Experimental details
7268
D. Zhu et al. / Thin Solid Films 515 (2007) 7264–7268
5. Conclusion
References
The model developed in this paper suggests that the electric characteristic of PLEDs can be affected by cathode obviously. The insertion of an ultra thin insulating buffer layer into metal cathode is helpful in lowering the potential barrier height at the cathode/polymer interface. Hence the electron injecting current increases. However, the buffer layer also introduces an additional barrier for electron injection, which can be expressed by the tunneling factor. When the thickness of the buffer layer increases, the barrier height increases and thus the efficiency decreases. The experimental characteristic of the devices with BaO as the buffer material indicates that the model proposed in this paper is reasonable.
[1] R.H. Friend, R.W. Gymer, A.B. Holmes, J.H. Burroughes, R.N. Marks, C. Taliani, D.D.C. Bradley, D.A. Dos Santos, J.L. BreÂdas, M. LoÈgdlund, W.R. Salaneck, Nature 397 (1999) 121. [2] Y.E. Kim, H. Park, J.J. Kim, Appl. Phys. Lett. 69 (1996) 599. [3] L.S. Hung, C.W. Tang, M.G. Mason, Appl. Phys. Lett. 70 (1997) 152. [4] F. Li, H. Tang, J. Anderegg, J. Shinar, Appl. Phys. Lett. 70 (1997) 1233. [5] S.T. Zhang, X.M. Ding, J.M. Zhao, H.Z. Shi, J. He, Z.H. Xiong, H.J. Ding, E.G. Obbard, Y.Q. Zhan, W. Huang, X.Y. Hou, Appl. Phys. Lett. 84 (2004) 425. [6] J.h. Park, O.O. Park, J.W. Yu, J.K. Kim, Y.Ch. Kim, Appl. Phys. Lett. 84 (2004) 1783. [7] Y. Kawabe, M.M. Morrell, G.E. Jabbour, S.E. Shaheen, B. Kippelen, J. Appl. Phys. 84 (1998) 5306. [8] I.D. Parker, J. Appl. Phys. 75 (1994) 1656. [9] H.L. Kwok, J.B. Xu, Solid State Electron 46 (2002) 645. [10] J.Y. Zeng, Quantum Mechanics, vol. 1, Science Publishing Company, Beijing, 2000, p. 98. [11] P.A. Zeijlmans van Evmmichoven, A. Niehaus, P. Stracke, F. Wiegershaus, S. Krischok, V. Kempter, A. Arnau, F.J. García de Abajo, M. Peňalba, Phys. Rev., B 59 (1999) 10950.
Acknowledgements We are grateful to H. Yang, W.D. Shen, X.Z. Sun, D. Yao and D.H. Yin for their support to our work in this paper.