STEADY-STATE TRANSIENT CURRENT-TRANSIENT VOLTAGE CHARACTERIZATION OF OLEDS B. J. Norris, J. P. Bender, and J. F. Wager Department of Electrical and Computer Engineering Center for Advanced Materials Research Oregon State University Corvallis, OR 97331-3211 Phone: (541)737-4862, (541)737-4862, (541)737-2994 Fax: (541)737-1300 Email: norris@ece.orst.edu, bender@ece.orst.edu, jfw@ece.orst.edu Keywords: electroluminescence, OLED, organic light emitting device Abstract Steady-state transient current-transient voltage [i(t)-v(t)] analysis is introduced as a novel method for the electrical characterization of organic light-emitting devices (OLEDs). i(t)-v(t) analysis involves measurement of the instantaneous voltage across and the instantaneous current through an OLED when it is subjected to a bipolar, piecewise-linear applied voltage waveform. The i(t)-v(t) method provides information related to the conduction and displacement currents and to charge trapping and detrapping properties of an OLED. Charge trapping, detrapping and hole accumulation on the hole transport layer (HTL) side of electron transport layer (ETL)/HTL interface are shown to play a signicant role in establishing the conduction and radiative recombination characteristics of OLEDs. As an extension of the i(t)-v(t) method, the total charge released from the OLED versus total charge transferred [Q detrapped, Q transf erred ] is plotted as a means of assessing how charge transferred during the forward bias portion of the waveform is related to charge detrapping and hole accumulation on the ETL/HTL interface. Additionally, 1
brightness transient-voltage transient [b(t)-v(t)] analysis is shown to be useful for assessing how charge trapping and accumulation eects the radiative recombination rate of OLEDs. 1 INTRODUCTION Organic light-emitting devices (OLEDs) oer an inexpensive and ecient alternative for full-color displays.[1, 2] They can be deposited onto a variety of substrates including exible plastics, making OLEDs attractive to the display industry.[3] However the details of the physics of OLED operation are still a matter of debate and it is likely that a better understanding of OLED physics will lead to device improvements. Recent papers have addressed the DC steady-state current-voltage and brightness-current characteristics but few have studied the transient conduction characteristics of OLEDs.[4, 5, 6, 7] The most relevant papers are by Egusa et al. [6, 7] They measured the transient characteristics of OLEDs at extremely low frequencies (1 Hz) and obtained important results related to device capacitance. The characterization method described in this paper is performed at high frequency (1-5 khz) as opposed to the very low frequency method of Egusa et al. The results presented in this paper are interpreted in terms of a model similar to the model for OLED operation introduced by Khramtchenkov et al. in which hole accumulation at the ETL/HTL interface is shown to play an important role in OLED operation. [8, 9] However, Khramtchenkov et al. consider the DC mode of OLED operation only and do not consider the possibility of charge trapping in the OLED. In this paper we propose steady-state transient current-transient voltage [i(t)-v(t)] analysis as a novel method for the electrical characterization of OLEDs. This method involves plotting the instantaneous i(t) versus v(t) characteristics of an OLED when it is driven by a bipolar, piecewiselinear applied voltage waveform. i(t)-v(t) analysis is found to be useful as a means of monitoring 2
charge trapping, detrapping and hole accumulation in OLEDs. Additionally, i(t)-v(t) analysis allows assessment of both conduction and displacement currents owing in an OLED, from which information related to the capacitance and conductance properties of the OLED may be extracted. As an extension of i(t)-v(t) analysis, the total transferred charge versus the total detrapped charge and charge release from the ETL/HTL interface [Q detrapped, Q transferred ] is plotted. These curves provide information related to how traps ll and charge accumulates during forward bias. Finally, transient brightness-transient voltage [b(t) - v(t)] analysis is introduced as a method of viewing changes in the radiative recombination rate as charge is accumulated and released from the OLED. 2 EXPERIMENTAL PROCEDURE The OLEDs employed in this study are fabricated by the Eastman-Kodak Company.[10] The devices are constructed on glass substrates and consist of a thin-lm stack comprised of a transparent indium-tin-oxide (ITO) anode, a CuPc hole injection layer, a naphthyl-substituted benzidine derivative (NPB) hole transport layer (HTL), an 8 hydroxyquinoline aluminum (Alq 3 ) electron transport layer (ETL), and a Mg:Ag cathode. The i(t)-v(t) experimental setup is shown in Fig. 1. An arbitrary waveform generator (Wavetek model 395) is used to provide the voltage waveform which is amplied and applied to a series combination of the OLED under test and a sense resistor, typically 19. The instantaneous current, i(t), is assessed by monitoring the voltage across the sense resistor while the instantaneous voltage, v(t), is evaluated as the instantaneous dierence in voltage across the OLED. i(t) and v(t) are assessed only after the OLED has reached steady-state with the applied voltage waveform. Luminance is measured via a photomultiplier tube (PMT). All of the voltage signals related to i(t), v(t), and the luminance are monitored via a digitizing oscilloscope (Tektronix model TDS 420A). Experimental control, data acquisition, and data analysis are accomplished using a personal 3
computer. The bipolar, piecewise-linear applied voltage waveform usually employed for i(t)-v(t) analysis is shown in Fig. 2. Typical waveform parameters are positive and negative maximum applied voltage amplitudes (V + max, V, max) of 5-15 V, rise and fall times (RT, FT) of 15-150 s, positive and negative pulse widths (PW +, PW, )of50-500 s, an intermediate transition time (ITT) of 30-300 s, and repetition frequency of 1 khz to 5 khz. Note that positive and negative refer to applied voltage polarities in which the OLED is forward and reverse biased, respectively. The intermediate fall time is typically chosen to continuously transition from positive to negative voltages so that detrapping phenomena may be more accurately assessed. 3 EXPERIMENTAL RESULTS AND DISCUSSION A representative i(t)-v(t) curve is shown in Fig. 3. Positive and negative voltages correspond to forward- and reverse-bias, respectively. The measured current corresponds to a superposition of the normal conduction current owing through the OLED device and of the displacement current arising from the C dv/dt current associated with the capacitive nature of the OLED. Below the threshold voltage there is a negligible amount of conduction current so that the measured current is exclusively displacement current. Thus, for these below-threshold, horizontal, baseline portions of the i(t)-v(t) curve, the physical capacitance of the entire OLED stack, C t may be estimated as the current divided by the voltage slew rate, C t = i(t) : (1) @v(t) @t The baseline current shown in Fig. 3 is 0.25 ma and the voltage slew rate is 0.1 V/s so that the total physical capacitance of the OLED stack is estimated from Eq. 1 to be 2.5 nf. This is in good agreement with that expected from the thicknesses and dielectric constants of the constituent 4
layers comprising the OLED. During aging experiments, we have sometimes noted a decrease in the measured C t with aging and have correlated this to burnout of portions of the OLED, which decreases the active area of the device. Thus, monitoring changes in the physical capacitance of an OLED with aging is a quantitative way of measuring burnout failure of an OLED. The most interesting aspect of the i(t)-v(t) curves measured to date is the small bump found at the bottom center portion of the i(t)-v(t) curve which occurs during the ITT portion of the applied voltage waveform. We attribute this bump primarily to current caused by holes that have accumulated at the ETL/HTL interface during the positive bias, transport across the HTL, and exit through the anode. A secondary mechanism that may also be responsible for this bump is the detrapping of trapped electrons in the ETL that are exiting through the cathode. Notice that this bump is initiated while the applied voltage to the OLED is still positive at 1-2 V. This is partially associated with the 1.1 V at-band voltage of the OLED and partially due to the voltage drop across the ETL due to the charge stored at the ETL/HTL interface. In addition, space charge due to trapped charge could create an internal electric eld which initiates charge detrapping at a voltage greater than the at-band voltage. The area of this bump (which is related to the total charge stored in the OLED) increases and then approximately saturates with increasing forward bias, as shown in Fig. 4. It is expected that the charge stored at the ETL/HTL interface increases with increasing applied voltage, just as the charge stored on a capacitor increases as the voltage applied across it is increased. The saturation of the bump is most likely related to the increasing ability of the ETL to transport charge at large applied voltages (e.g. the ETL capacitance is shunted). By integrating the current transient curve over the bump portion of the applied voltage waveform, the total detrapped charge (Q detrapped ) may be estimated. Here detrapped charge refers to both charge from hole accumulation at the ETL/HTL interface and charge from detrapping elec- 5
trons in the ETL. The resulting charge per unit area when the bump is saturated is approximately 1:3 10,7 C=cm 2 (i.e. 8:1 10 11 charges=cm 2 ). However, this may be larger than the actual hole density at the ETL/HTL interface as this charge density also includes charge due to electron detrapping in the ETL. By plotting Q detrapped versus the transferred charge (Q transferred ), several trends become evident. A Q detrapped - Q transferred curve is obtained by varying PW + for a given V + max and calculating Q detrapped and Q transferred for each applied voltage waveform. Figure 5 shows Q detrapped, Q transferred curves obtained with various V + max's. The most important feature of this plot is that a larger V + max yields a larger saturation Q detrapped magnitude but that before saturation Q detrapped is approximately equal to Q transferred. Our interpretation of this trend is that the initial charge injected into the OLED during forward bias is either trapped in the ETL or accumulates at the ETL/HTL interface. Thus, little of the charge initially injected into the OLED during forward bias contributes to conduction of charge through the OLED. Notice that if the ETL is thought ofasa capacitor and the HTL a resistor, the current due to the charging up of the ETL would be expected to decrease as time passed. Saturation is thought to be caused by the increasing conductance of the ETL at high applied biases. Thus, at large positive biases holes injected into the HTL are likely to be injected into the ETL as opposed to accumulating at the ETL/HTL interface. In addition to the area of the bump increasing with increasing V + max, there is a noticeable change in the forward bias i(t)-v(t) characteristics, as shown in Fig. 6. The OLED turns on harder with increasing V + max. As stated above, the amount ofcharge accumulated across the ETL increases with larger V + max. A consequence of charge accumulation across the ETL is an increased voltage drop across the ETL as compared to the voltage drop across the HTL. This increased voltage drop across the ETL will aid the injection of electrons and holes into the ETL (from the cathode and the HTL, respectively) and should also enhance the conduction across the ETL. Thus, if the negative bias 6
portion of the waveform was insucient toremove all of the holes accumulated at the ETL/HTL interface the OLED would turn on harder than if there were no holes at the ETL/HTL interface. When the period of the applied voltage waveform is increased, the OLED turns on softer, as shown in Fig. 7. Increasing the period allows more time for holes to be removed from the ETL/HTL interface during the delay portion of the waveform. Thus, fewer holes remain at the ETL/HTL interface at the beginning of the positive pulse and the OLED turns on softer. Although the amount charge build up within the OLED is primarily controlled by the extent of the forward biasing of the OLED (via V + max), the reverse bias characteristics of the applied voltage waveform have a small eect on the i(t)-v(t) characteristics. Evidence that the reverse bias portion of the applied voltage waveform inuences the i(t)-v(t) curve arises from the fact that even though the detrapping bump does not change appreciably with changes in V, max or PW,, increasing the magnitudes of these reverse bias parameters slightly softens the i(t)-v(t) turn on. This suggests that a more aggressive reverse bias (increasing V, max and PW, )may enhance the remove ofcharge from the OLED. The steady-state transient brightness-transient voltage [b(t)-v(t)] curve exhibits a similar trend to that of the i(t)-v(t) curveasv + max is increased, as shown in Fig. 8. The harder turn on of the b(t)- v(t) curve with increasing V + max is due to a harder turn on of the i(t)-v(t) curve. The linkage between b(t)-v(t) and i(t)-v(t) curves is more evident when a steady-state transient brightness-transient current [b(t)-i(t)] curve (not shown) is obtained, since it is found that the brightness increases linearly with current. Therefore, a harder turn on of a i(t)-v(t) curve leads to a concomitant harder turn on of the b(t)-v(t) curve. Returning to Fig. 8, note that these b(t)-v(t) curves exhibit counterclockwise hysteresis. With counterclockwise hysteresis, the brightness is higher for a given voltage during the ITT portion of the forward bias pulse than it is during the RT portion. We attribute this counterclockwise 7
hysteresis of the b(t)-v(t) curve to lling of traps and the accumulation of holes during the RT and plateau portions of the forward bias pulse. Trapping during the RT portion of the applied voltage waveform causes fewer carriers to be lost to traps during the ITT portion of the forward bias pulse. Build up of holes at the ETL/HTL interface during the RT portion of the waveform causes a greater voltage drop across the ETL. Thus, conduction and radiative recombination is more ecient during the ITT portion of the forward bias pulse due to reduced trapping and a greater voltage drop across the ETL. It is likely that the build up of holes at the ETL/HTL interface is more important for the hysteresis of the b(t)-v(t) curve than the trapping of electrons. [11] Note that the preceding discussion of the hysteresis and its consequences is undertaken in the context of b(t)-v(t) curves. Originally, we attempted to investigate OLED memory eects associated with hysteresis using i(t)-v(t) curves. However, in order to meaningfully assess hysteresis in this manner, it is necessary to subtract the displacement current, i displacement (t), from the total measured current, i(t), in order to estimate the conduction current, i cond (t); subsequently, the hysteretic nature of the i cond (t)-v(t) is assessed. From SPICE simulation, we have discovered that it is exceedingly dicult to unambiguously extract i cond (t) from i(t) in the above-threshold portion of the i(t)-v(t) curve. Therefore, since b(t)-v(t) curves have no dependence upon the displacement current, we now prefer to evaluate the hysteretic nature of an OLED from b(t)-v(t) curves rather than i cond (t)-v(t) curves since b(t)-v(t) curves can be interpreted in a much less ambiguous manner. 4 Conclusions Steady-state transient current-transient voltage [i(t)-v(t)] analysis is proposed as a useful method for the characterization of OLEDs. With the i(t)-v(t) method, the removal of charge from the OLED is observed directly as a bump. Additionally, the forward bias i(t)-v(t) characteristics are modied by trapping and hole accumulation at the ETL/HTL interface. Also, the physical 8
capacitance of the entire OLED stack is obtained from the i(t)-v(t) curve by measuring the displacement current below threshold. By varying the applied voltage waveform used in i(t)-v(t) analysis, hole accumulation at the ETL/HTL interface is shown to enhance conduction and radiative recombination. As an extension of i(t)-v(t) analysis, detrapped charge versus transferred charge (Q detrapped,q transferred ) curves are plotted as means of viewing the relation between transferred charge and detrapped charge. Q detrapped, Q transferred analysis reveals that injected charge is initially trapped and accumulated at the ETL/HTL interface during the forward bias pulse until the traps ll and the hole accumulation stops (the ETL capacitance charges up). The magnitude of the charge at which Q detrapped saturates increases with increasing V + max, indicating that a larger forward bias increases the amount ofcharge accumulated at the ETL/HTL interface. These observations suggest that under moderate forward bias the HTL easily conducts charge while the ETL acts primarily as a capacitor. These conclusions imply that conduction and brightness would be improved by improving electron transport across and/or injection into the ETL. 9
ACKNOWLEDGMENTS We wish to thank Pat Green and Bill Barrow for helpful discussions and Ching Tang for providing OLEDs. This work was supported by ARO under Contract No. DAAG55-9710226. 10
References [1] C. W. Tang and S. A. VanSlyke, Appl. Phys. Lett. 51, 913 (1987). [2] C. W. Tang, S. A. VanSlyke, and C. H. Chen, J. Appl. Phys. 65, 3610 (1989). [3] P. E. Burrows, G. Gu, V. Bulovic, Z. Shen, S. R. Forrest, and M. E. Thompson, IEEE Transactions on Electron Devices 44, 1188 (1997). [4] P. E. Burrows, Z. Shen, V. Bulovic, D. M. McCarty, and S. R. Forrest, J. Appl. Phys. 79, 7991 (1996). [5] M. Matsumara, Y. Jinde, T. Akai, and T. Kimura, Jpn. J. Appl. Phys. 35, 5735 (1996). [6] S. Eugusa, N. Germma, A. Miura, K. Mizushima, and M. Azuma, J. Appl. Phys. 71, 2042 (1992). [7] S. Eugusa, A. Miura, N. Germma, and M. Azuma, Jpn. J. Appl. Phys. 33, 2741 (1994). [8] D. V. Khramtchenkov, H. Bassler, and V. I. Arkhipov, J. Appl. Phys. 79, 9283 (1996). [9] D. V. Khramtchenkov, V. I. Arkhipov, and H. Bassler, J. Appl. Phys. 81, 6954 (1997). [10] S. A. Van Slyke, C. H. Chan, and C. W. Tang, Appl. Phys. Lett. 59, 2160 (1996). [11] B. J. Norris, "Characterization of Organic Light-Emitting Devices," MS Thesis, Oregon State University, 1999. 11
FIGURE CAPTIONS Figure 1. The circuit used to test OLEDs. The voltages V1(t), V3(t), and V4(t) go to the oscilloscope and V2(t) (not shown) goes from the arbitrary waveform generator (AWG) to the oscilloscope for synchronization. The sense element is a resistor, typically 19.1. Figure 2. Representative applied voltage waveform for OLED testing. Waveform parameters identied are positive and negative maximum applied voltages (V + max ;V, max), rise and fall times (RT, FT), and intermediate transition time (ITT). Figure 3. A representative i(t)-v(t) curve for an OLED. Positive and negative voltage corresponds to forward- and reverse-bias, respectively. The bump is found near the zero voltage portion of the negative going trace. Figure 4. The bump portion of a set of i(t)-v(t) curves obtained using a constant V, max = -12 V and variable V + max = 5, 5.5, 6.5, 7.5, 8.5, and 9 V, respectively. The bump increases and then approximately saturates with increasing V + max. Figure 5. Q detrapped versus Q transferred with V + max = 6, 7, 8, and 9 V, respectively. Note that decreasing Q detrapped indicates that more trapped charge is released; the bump current is negative, so the detrapped charge is negative. The Q detrapped saturation magnitude increases with increasing V + max. In the initial presaturation portion of the curve, Q detrapped is approximately linear with Q transferred. Figure 6. A blow up of the forward bias portion of i(t)-v(t) curves with V + max = 11, 13, and 15 V, respectively. The OLED turns on harder with increasing V + max. Figure 7. A blow upof the positive portion of i(t)-v(t) curves with the period equal to 320, 470, and 1320 s, respectively. Other relevant waveform parameters are RT and FT = 30 s, ITT = 60 s, PW + and PW, = 100 s, V + max =15V,and V, max = -15 V. The device turns on softer with larger periods. 12
Figure 8. Transient brightness versus transient voltage with V + max = 11, 13, and 15 V, respectively. The brightness has a counterclockwise hysteresis and turns on harder with increasing V + max. 13
V1(t) V4(t) AMP AWG sense OLED PMT V3(t) Figure 1: Norris, et al. t RT PW + ITT PW - FT V + max - V max Figure 2: Norris, et al. 14
7 Current (ma) 5 3 1-1 -12-8 -4 0 4 8 12 Applied Voltage (V) Figure 3: Norris, et al. Current (ma) -0.3-0.4 V + max Increasing -0.5-12 -8-4 0 4 Applied Voltage (V) Figure 4: Norris, et al. 15
Detrapped Charge (nc) 0-4 -8-12 0 50 100 Increasing V + max Transferred Charge (nc) Figure 5: Norris, et al. 8 Current (ma) 6 4 2 Increasing V + max 0 9 10 11 12 Applied Voltage (V) Figure 6: Norris, et al. 16
Current (ma) 40 30 20 10 Increasing Period 0 9 11 13 15 Applied Voltage (V) Figure 7: Norris, et al. Brightness (au) 0.02 0.01 Increasing V + max 0 9 11 13 15 Applied Voltage (V) Figure 8: Norris, et al.
BIOGRAPHY Benjamin J. Norris received the B.A. degree in Physics and Math from Whitman College, Walla Walla, in 1997. He is currently working towards the M. S. degree in electrical engineering at Oregon State University. His current research interest is the characterization of organic light emitting devices. Jerey P. Bender received the B.S. degree in electrical engineering from Oregon State University, Corvallis, in 1998. He is currently working towards the M. S. degree in electrical engineering at Oregon State University. John F. Wager received the B.S. degree in engineering physics from Oregon State University, Corvallis, in 1977 and the M.S. and Ph.D. degrees in electrical engineering from Colorado State University, Fort Collins, in 1978 and 1981, respectively. From 1982-84 he was at Hughes Research Laboratories, Malibu. In 1984, he joined the Department of Electrical and Computer Engineering at Oregon State University where he is presently a Professor. His current research focus involves inorganic and organic electroluminescence for at-panel display applications.