Enhanced Emission by Accumulated Charges at Organic/Metal Interfaces Generated during the Reverse Bias of Organic Light Emitting Diodes

A high frequency rectangular alternating voltage was applied to organic light emitting diodes (OLEDs) with the structure ITO/TPD/Alq3/Al and ITO/CoPc/Alq3/Al, where ITO is indium-tin-oxide, TPD is 4,4′-bis[N-phenyl-N-(m-tolyl)amino]biphenyl, CoPc is cobalt phthalocyanine, and Alq3 is Tris(8-quinolinolato)aluminum, and the effect on emission of the reverse bias was examined. The results reveal that the emission intensity under an alternating reverse-forward bias is greater than that under an alternating zero-forward bias. The difference in the emission intensity (∆I) increased both for decreasing frequency and increasing voltage level of the reverse bias. In particular, the change in emission intensity was proportional to the voltage level of the reverse bias given the same frequency. To understand ∆I, this paper proposes a model in which an OLED works as a capacitor under reverse bias, where positive and negative charges accumulate on the metal/organic interfaces. In this model, the emission enhancement that occurs during the alternating reverse-forward bias is rationalized as a result of the charge accumulation at the organic/metal interfaces during the reverse bias, which possibly modulates the vacuum level shifts at the organic/metal interfaces to reduce both the hole injection barrier at the organic/ITO interface and the electron injection barrier at the organic/Al interface under forward bias.


Introduction
About 25 years after the first report of electroluminescence from a single crystal of anthracene under a high applied voltage [1], Tang proposed an organic light emitting diode (OLED) composed of a double-layer structure of organic thin films [2].Since then, various OLEDs with different structures have been developed.The device structures are designed based on the work functions of the electrodes and molecular orbital energy levels of the organic materials, which are key factors of the OLEDs' performance.Meanwhile, using ultraviolet photoemission spectroscopy and the Kelvin probe method, Ishii et al. discovered the vacuum level shift at various organic/metal interfaces, where band bending does not occur [3,4], meaning that the Mott-Schottky model is not applicable to OLEDs.Furthermore, by using a time-resolved optical second harmonic generation measurement while applying a rectangular alternating voltage to typical OLEDs, Iwamoto et al. reported that charges accumulate at the organic/organic interface under forward bias with high frequency, and they proposed a model of emission induced by the accumulated charges at the organic/organic interface [5][6][7][8].Moreover, these studies signify that, to understand the OLED emission mechanism and control its performance, we must focus on the electronic states at the interfaces as well as the work functions and molecular orbital energy levels.
We have been attracted by the report that the emission intensity for an ITO/α-NPD/Alq 3 /Al device under an alternating reverse-forward bias becomes higher than that under an alternating zero-forward bias in the frequency region between 8 × 10 5 and 10 Hz [7].The change in the emission intensity is reported to be due to suppression of the charge injection caused by the charge accumulation at the organic/organic interface under the alternating zero-forward bias.
In this study, we focused on the effect of reverse bias on the device properties.Zou et al. reported that a reverse bias with a long application time (10 1 -10 4 s) induces the movement of ionic impurities in the active layer, leading to emission enhancement and lifetime improvement [9,10]; however, this mechanism is not acceptable in the high frequency region because the ionic impurities could not follow the frequency.To investigate the generality of the higher emission intensity under an alternating reverse-forward bias than that under an alternating zero-forward bias, OLEDs with the structure ITO/TPD/Alq 3 /Al and ITO/CoPc/Alq 3 /Al were fabricated, where TPD is 4,4 -bis[N-phenyl-N-(m-tolyl)amino]biphenyl, Alq 3 is Tris(8-quinolinolato)aluminum, and CoPc is cobalt phthalocyanine.We propose another model in which an OLED works as a capacitor under reverse bias, and the accumulated charges at the organic/metal interfaces induce an enhancement of emission under the alternating reverse-forward bias.

Materials and Methods
CoPc was synthesized following the procedure reported in [11], and used as a hole injection and transport material.TPD was used as a hole injection and transport material and Alq 3 was used as an electron injection, transport, and emitting material.Both are commercially available from Tokyo Kasei and were used without further purification.
The OLEDs were fabricated by the vacuum deposition of TPD or CoPc and Alq 3 films (30 nm), followed by the Al cathode (30 nm) on an ITO (indium-tin-oxide) electrode.The base pressure for the organic materials and Al was less than 1.5 × 10 −5 torr and 5.0 × 10 −6 torr, respectively.The ITO substrate was cleaned by ultrasonic agitation for 15 min in acetone, 2-propanol, and methanol prior to film fabrications.Energy diagrams of the fabricated devices are shown in Figure 1.
We have been attracted by the report that the emission intensity for an ITO/α-NPD/Alq3/Al device under an alternating reverse-forward bias becomes higher than that under an alternating zero-forward bias in the frequency region between 8 × 10 5 and 10 Hz [7].The change in the emission intensity is reported to be due to suppression of the charge injection caused by the charge accumulation at the organic/organic interface under the alternating zero-forward bias.
In this study, we focused on the effect of reverse bias on the device properties.Zou et al. reported that a reverse bias with a long application time (10 1 -10 4 s) induces the movement of ionic impurities in the active layer, leading to emission enhancement and lifetime improvement [9,10]; however, this mechanism is not acceptable in the high frequency region because the ionic impurities could not follow the frequency.To investigate the generality of the higher emission intensity under an alternating reverse-forward bias than that under an alternating zero-forward bias, OLEDs with the structure ITO/TPD/Alq3/Al and ITO/CoPc/Alq3/Al were fabricated, where TPD is 4,4′-bis[N-phenyl-N-(m-tolyl)amino]biphenyl, Alq3 is Tris(8-quinolinolato)aluminum, and CoPc is cobalt phthalocyanine.We propose another model in which an OLED works as a capacitor under reverse bias, and the accumulated charges at the organic/metal interfaces induce an enhancement of emission under the alternating reverse-forward bias.

Materials and Methods
CoPc was synthesized following the procedure reported in [11], and used as a hole injection and transport material.TPD was used as a hole injection and transport material and Alq3 was used as an electron injection, transport, and emitting material.Both are commercially available from Tokyo Kasei and were used without further purification.
The OLEDs were fabricated by the vacuum deposition of TPD or CoPc and Alq3 films (30 nm), followed by the Al cathode (30 nm) on an ITO (indium-tin-oxide) electrode.The base pressure for the organic materials and Al was less than 1.5 × 10 −5 torr and 5.0 × 10 −6 torr, respectively.The ITO substrate was cleaned by ultrasonic agitation for 15 min in acetone, 2-propanol, and methanol prior to film fabrications.Energy diagrams of the fabricated devices are shown in Figure 1.An alternating rectangular voltage of various frequencies was applied using a Keithley 3390 waveform generator.The duty ratio for the applied alternating rectangular voltage signal was always 50%.The emission spectra were measured using an Ocean Optics QE65000 spectrometer with a 10 s exposure time.All measurements were performed in a vacuum (less than 1.0 × 10 −4 torr) in an Oxford Optistat-CF cryostat.
Because of the deterioration of the devices, the measurements of the frequency dependence (Figure 2) and reverse-bias voltage dependence (Figure 3) were performed using different devices.However, the reproducibility of the tendency of the emission properties was ascertained several times using several devices for each measurement.The energy levels of organic materials and electrodesare cited from [12].
An alternating rectangular voltage of various frequencies was applied using a Keithley 3390 waveform generator.The duty ratio for the applied alternating rectangular voltage signal was always 50%.The emission spectra were measured using an Ocean Optics QE65000 spectrometer with a 10 s exposure time.All measurements were performed in a vacuum (less than 1.0 × 10 −4 torr) in an Oxford Optistat-CF cryostat.
Because of the deterioration of the devices, the measurements of the frequency dependence (Figure 2) and reverse-bias voltage dependence (Figure 3) were performed using different devices.However, the reproducibility of the tendency of the emission properties was ascertained several times using several devices for each measurement.

Results and Discussion
Figure 2 shows the emission intensity (I) as a function of the frequency (f), which alternated between −10 and 10 V (for reverse-forward bias) and 0 and 10 V (for zero-forward bias).In both devices, I decreased when the frequency increased under both reverse-forward and zero-forward biases.Under the reverse-forward bias, I is higher than under the zero-forward bias in the frequency region between 1 and 5 × 10 3 Hz and 1 and 1 × 10 5 Hz for the ITO/TPD/Alq3/Al and ITO/CoPc/Alq3/Al devices, respectively.These results resemble that reported in [7], suggesting that the higher emission intensity under an alternating reverse-forward bias than under an alternating zero-forward bias is generally observed in OLEDs.
We investigated how the reverse-bias voltage alters I. Figure 3 depicts I as a function of the reverse-bias voltage alternated with a 10 V forward bias with f = 1 × 10 3 Hz.In this figure, I linearly correlates with the reverse-bias voltage.Because the change in emission intensity (∆I) can be regarded as the difference of I under zero-forward bias and that under reverse-forward bias, ∆I is also proportional to the applied reverse-bias voltage, as shown on the right axis of Figure 3.This result signifies that ∆I depends on both the frequency and applied voltage of the reverse bias.
In [7], ∆I is rationalized by a model in which excess charges accumulate at the organic/organic interface: the accumulated charges under the alternating zero-forward bias suppress the charge injection, and induce the decrease in the emission intensity.The model is reasonable, however, the proportional dependence of the emission intensity change on the reverse-bias voltage, observed in this study is unclear in the model.Here, we focus on the proportional dependence, and propose another model in which an OLED acts as a capacitor during the reverse bias, where positive and negative charges respectively accumulate on the organic/Al and organic/ITO interfaces, and the

Results and Discussion
Figure 2 shows the emission intensity (I) as a function of the frequency (f), which alternated between −10 and 10 V (for reverse-forward bias) and 0 and 10 V (for zero-forward bias).In both devices, I decreased when the frequency increased under both reverse-forward and zero-forward biases.Under the reverse-forward bias, I is higher than under the zero-forward bias in the frequency region between 1 and 5 × 10 3 Hz and 1 and 1 × 10 5 Hz for the ITO/TPD/Alq3/Al and ITO/CoPc/Alq3/Al devices, respectively.These results resemble that reported in [7], suggesting that the higher emission intensity under an alternating reverse-forward bias than under an alternating zero-forward bias is generally observed in OLEDs.
We investigated how the reverse-bias voltage alters I. Figure 3 depicts I as a function of the reverse-bias voltage alternated with a 10 V forward bias with f = 1 × 10 3 Hz.In this figure, I linearly correlates with the reverse-bias voltage.Because the change in emission intensity (∆I) can be regarded as the difference of I under zero-forward bias and that under reverse-forward bias, ∆I is also proportional to the applied reverse-bias voltage, as shown on the right axis of Figure 3.This result signifies that ∆I depends on both the frequency and applied voltage of the reverse bias.
In [7], ∆I is rationalized by a model in which excess charges accumulate at the organic/organic interface: the accumulated charges under the alternating zero-forward bias suppress the charge injection, and induce the decrease in the emission intensity.The model is reasonable, however, the proportional dependence of the emission intensity change on the reverse-bias voltage, observed in this study is unclear in the model.Here, we focus on the proportional dependence, and propose another model in which an OLED acts as a capacitor during the reverse bias, where positive and negative charges respectively accumulate on the organic/Al and organic/ITO interfaces, and the

Results and Discussion
Figure 2 shows the emission intensity (I) as a function of the frequency (f ), which alternated between −10 and 10 V (for reverse-forward bias) and 0 and 10 V (for zero-forward bias).In both devices, I decreased when the frequency increased under both reverse-forward and zero-forward biases.Under the reverse-forward bias, I is higher than under the zero-forward bias in the frequency region between 1 and 5 × 10 3 Hz and 1 and 1 × 10 5 Hz for the ITO/TPD/Alq 3 /Al and ITO/CoPc/Alq 3 /Al devices, respectively.These results resemble that reported in [7], suggesting that the higher emission intensity under an alternating reverse-forward bias than under an alternating zero-forward bias is generally observed in OLEDs.
We investigated how the reverse-bias voltage alters I. Figure 3 depicts I as a function of the reverse-bias voltage alternated with a 10 V forward bias with f = 1 × 10 3 Hz.In this figure, I linearly correlates with the reverse-bias voltage.Because the change in emission intensity (∆I) can be regarded as the difference of I under zero-forward bias and that under reverse-forward bias, ∆I is also proportional to the applied reverse-bias voltage, as shown on the right axis of Figure 3.This result signifies that ∆I depends on both the frequency and applied voltage of the reverse bias.
In [7], ∆I is rationalized by a model in which excess charges accumulate at the organic/organic interface: the accumulated charges under the alternating zero-forward bias suppress the charge injection, and induce the decrease in the emission intensity.The model is reasonable, however, the proportional dependence of the emission intensity change on the reverse-bias voltage, observed in this study is unclear in the model.Here, we focus on the proportional dependence, and propose another model in which an OLED acts as a capacitor during the reverse bias, where positive and negative charges respectively accumulate on the organic/Al and organic/ITO interfaces, and the accumulated charges at the organic/metal interfaces induce an enhancement of emission under the alternating reverse-forward bias.A series circuit of two parallel circuits of a resistor and capacitor is described to understand the charge accumulation at the organic/organic interface in a double-layer OLED [7,13].In contrast, to estimate the charge accumulation at the organic/metal interfaces, the equivalent circuit shown in Figure 4, which is a series circuit of a resistor and a parallel circuit of a resistor and capacitor, is considered.The circuit equations are as follows where V is the absolute value of the applied reverse-bias voltage, i is current, C is capacitance, R 1 is the internal resistance, R 2 is the resistance of the OLED, and t is the time of the reverse-bias application, which is half the reciprocal of the frequency (t = 1 2 f −1 ).These equations lead to the equation Because i c can be denoted as i c = dQ c /dt using the quantity of electric charge (Q c ), Equation ( 1) can be expressed as which can be transformed into Considering an initial condition of Q c = 0, Equation (3) can be solved as and supposing the emission intensity change induced by the accumulated charges (∆I a ) is proportional to Q c , we obtain by introducing proportionality constant A. Here, because ∆I a is the change in emission intensity during t, ∆I for an exposure time of 10 s can be denoted as This equation adequately reproduces the observed proportional dependence of the emission intensity change on the reverse-bias voltage.With the parameter sets R 1 = 50 Ω, R 2 = 5.0 × 10 6 Ω, and C = 10 × 10 −9 F, where R 1 and R 2 are the estimated resistances of the ITO electrode and OLED, respectively, and C is a typical value [14], Equation ( 6) produces the solid line shown in Figure 3 with a variable parameter of A = 2.88 × 10 6 and 6.35 × 10 5 count C −1 for the ITO/TPD/Alq 3 /Al and ITO/CoPc/Alq 3 /Al devices, respectively.Equations ( 5) and ( 6) clearly indicate that ∆I also depends on the time of the reverse-bias application t. Figure 5 depicts the t −1 (i.e., 2f ) dependence of ∆I derived from Figure 2 together with the values calculated using the parameter sets in Figure 3. Clearly, they are not consistent.This might be because A is not a common constant for t but a function of t.At this time, we cannot evaluate A(t).To understand the mechanism of the reverse-bias emission enhancement, we consider the vacuum level shift (∆) at the organic/metal interface.When a reverse bias is applied to an OLED and it acts as a capacitor, the interfacial dipole at the organic/ITO interface with negative (positive) Δ decreases (increases), while that at the organic/Al interface with negative (positive) Δ increases (decreases), because of the accumulation of negative and positive charges at the organic/ITO and organic/Al interfaces, respectively.In any case, whether Δ at the organic/metal interfaces is negative or positive, this situation causes Δ at the organic/ITO interface to shift positively and Δ at the organic/Al interface to shift negatively.Therefore, the vacuum level shifts at the organic/metal interfaces were modulated to shift both the HOMO of a hole injection material and the LUMO of an electron injection material close to the Fermi levels of the ITO and Al electrodes, respectively, leading to a decrease in both the hole injection barrier at the organic/ITO interface and the electron injection barrier at the organic/Al interface.Consequently, we believe that both the hole and electron injections occur easily under forward bias after reverse bias has been applied, resulting in an increase of the number of carriers that can arrive at the organic/organic interface compared to ordinary DC or alternating zero-forward bias.
Based on this model, we can interpret the tendency of ΔI to depend on t by dividing Figure 5 into five regions, I, II, III, IV and V, according to t.
In region I, no emission was observed for both the alternating zero-forward and reverse-forward biases.In this study, only the ITO/CoPc/Alq3/Al device has a region I, meaning that no carriers can arrive at the CoPc/Alq3 interface under the high frequency bias.
In region II, emission was only observed under the alternating zero-forward bias.The carriers can arrive at the organic/organic interface during the forward-bias phase of the alternating zero-forward bias.In contrast, under the alternating reverse-forward bias, the carriers are consumed to cancel the accumulated charges.As a result, because few residual carriers occur during the   To understand the mechanism of the reverse-bias emission enhancement, we consider the vacuum level shift (∆) at the organic/metal interface.When a reverse bias is applied to an OLED and it acts as a capacitor, the interfacial dipole at the organic/ITO interface with negative (positive) Δ decreases (increases), while that at the organic/Al interface with negative (positive) Δ increases (decreases), because of the accumulation of negative and positive charges at the organic/ITO and organic/Al interfaces, respectively.In any case, whether Δ at the organic/metal interfaces is negative or positive, this situation causes Δ at the organic/ITO interface to shift positively and Δ at the organic/Al interface to shift negatively.Therefore, the vacuum level shifts at the organic/metal interfaces were modulated to shift both the HOMO of a hole injection material and the LUMO of an electron injection material close to the Fermi levels of the ITO and Al electrodes, respectively, leading to a decrease in both the hole injection barrier at the organic/ITO interface and the electron injection barrier at the organic/Al interface.Consequently, we believe that both the hole and electron injections occur easily under forward bias after reverse bias has been applied, resulting in an increase of the number of carriers that can arrive at the organic/organic interface compared to ordinary DC or alternating zero-forward bias.
Based on this model, we can interpret the tendency of ΔI to depend on t by dividing Figure 5 into five regions, I, II, III, IV and V, according to t.
In region I, no emission was observed for both the alternating zero-forward and reverse-forward biases.In this study, only the ITO/CoPc/Alq3/Al device has a region I, meaning that no carriers can arrive at the CoPc/Alq3 interface under the high frequency bias.
In region II, emission was only observed under the alternating zero-forward bias.The carriers can arrive at the organic/organic interface during the forward-bias phase of the alternating zero-forward bias.In contrast, under the alternating reverse-forward bias, the carriers are consumed to cancel the accumulated charges.As a result, because few residual carriers occur during the To understand the mechanism of the reverse-bias emission enhancement, we consider the vacuum level shift (∆) at the organic/metal interface.When a reverse bias is applied to an OLED and it acts as a capacitor, the interfacial dipole at the organic/ITO interface with negative (positive) ∆ decreases (increases), while that at the organic/Al interface with negative (positive) ∆ increases (decreases), because of the accumulation of negative and positive charges at the organic/ITO and organic/Al interfaces, respectively.In any case, whether ∆ at the organic/metal interfaces is negative or positive, this situation causes ∆ at the organic/ITO interface to shift positively and ∆ at the organic/Al interface to shift negatively.Therefore, the vacuum level shifts at the organic/metal interfaces were modulated to shift both the HOMO of a hole injection material and the LUMO of an electron injection material close to the Fermi levels of the ITO and Al electrodes, respectively, leading to a decrease in both the hole injection barrier at the organic/ITO interface and the electron injection barrier at the organic/Al interface.Consequently, we believe that both the hole and electron injections occur easily under forward bias after reverse bias has been applied, resulting in an increase of the number of carriers that can arrive at the organic/organic interface compared to ordinary DC or alternating zero-forward bias.
Based on this model, we can interpret the tendency of ∆I to depend on t by dividing Figure 5 into five regions, I, II, III, IV and V, according to t. that the accumulated charges at the organic/metal interfaces modulate the vacuum level shifts at the organic/metal interfaces to reduce both the hole injection barrier at the organic/ITO interface and the electron injection barrier at the organic/Al interface under forward bias.
The model proposed in this study might be applicable to any OLED, suggesting that the application of a rectangular alternating reverse-forward bias is a fruitful method for tuning the light emission efficiency of OLEDs, and there is a fair possibility of obtaining higher device performances than under the ordinary DC bias.

Figure 2 .
Figure 2. Emission intensity I versus the frequency f of the voltage for the ITO/TPD/Alq3/Al device (a) and ITO/CoPc/Alq3/Al device (b).The solid and dashed lines indicate the emission intensity under −10 and 10 V (reverse-forward) bias and 0 and 10 V (zero-forward) bias, respectively.

Figure 2 .
Figure 2. Emission intensity I versus the frequency f of the voltage for the ITO/TPD/Alq 3 /Al device (a) and ITO/CoPc/Alq 3 /Al device (b).The solid and dashed lines indicate the emission intensity under −10 and 10 V (reverse-forward) bias and 0 and 10 V (zero-forward) bias, respectively.

Figure 2 .
Figure 2. Emission intensity I versus the frequency f of the voltage for the ITO/TPD/Alq3/Al device (a) and ITO/CoPc/Alq3/Al device (b).The solid and dashed lines indicate the emission intensity under −10 and 10 V (reverse-forward) bias and 0 and 10 V (zero-forward) bias, respectively.

Figure 4 .
Figure 4. Equivalent circuit based on a model in which the OLED acts as a capacitor during the reverse bias.Here, V is the absolute value of the applied voltage, R1 is the internal resistance, R2 is the resistance of the OLED, and C is capacitance.

Figure 5 .
Figure 5. ∆I versus the reciprocal of t, i.e., 2f, for the ITO/TPD/Alq3/Al device (a) and ITO/CoPc/Alq3/Al device (b).The solid line is calculated by Equation (6) with the parameter sets used in Figure3.To interpret the t dependence of ∆I, we divided t into five regions, I, II, III, IV, and V.

Figure 4 . 8 Figure 4 .
Figure 4. Equivalent circuit based on a model in which the OLED acts as a capacitor during the reverse bias.Here, V is the absolute value of the applied voltage, R 1 is the internal resistance, R 2 is the resistance of the OLED, and C is capacitance.

Figure 5 .
Figure 5. ∆I versus the reciprocal of t, i.e., 2f, for the ITO/TPD/Alq3/Al device (a) and ITO/CoPc/Alq3/Al device (b).The solid line is calculated by Equation (6) with the parameter sets used in Figure3.To interpret the t dependence of ∆I, we divided t into five regions, I, II, III, IV, and V.

Figure 5 .
Figure 5. ∆I versus the reciprocal of t, i.e., 2f, for the ITO/TPD/Alq 3 /Al device (a) and ITO/CoPc/Alq 3 /Al device (b).The solid line is calculated by Equation (6) with the parameter sets used in Figure3.To interpret the t dependence of ∆I, we divided t into five regions, I, II, III, IV, and V.