- Email: [email protected]
Reliability comparison of BTEL and bilayer organic LEDs
Synthetic Metals 107 Ž1999. 53–56 www.elsevier.comrlocatersynmet
Reliability comparison of BTEL and bilayer organic LEDs J. Curless ) , S. Rogers, M. Kim, M. Lent, S. Shi, V.-E. Choong, C. Briscoe, F. So Physical Sciences Research Laboratories, Motorola Labs, Motorola, 2100 E. Elliot, Tempe, AZ 85284, USA Received 22 April 1999; received in revised form 7 July 1999; accepted 8 July 1999
Abstract Organic structures consisting of two distinct layers Žbilayer., a hole transport layer and an electron transportremitter layer are compared to structures using a bipolar transport and emitting layer ŽBTEL.. Reverse bias and different duty cycle did not significantly affect reliability. The BTEL structure had significantly improved reliability compared to the bilayer structure. The relative change in device voltage was found to be linearly proportional to the relative change in luminance and the constant of proportionality was a function of the contact. This constant of proportionality can be used as a figure of merit for voltage increase during operation. The BTEL structure also gives improved reliability at elevated temperature. q 1999 Elsevier Science S.A. All rights reserved. Keywords: Electroluminescence; Light emission; Reliability; Organic light-emitting diode; Organic light-emitting device
1. Introduction In the 1980s, Tang and VanSlyke w1x and Tang et al. w2x developed an organic light-emitting diode ŽOLED. structure which gave high luminance while operating at low voltages. Their structure is called a bilayer structure because it contains two distinct transport layers: the hole transport layer ŽHTL. and the emitterrelectron transport layer. Recently, an OLED structure has been developed which uses a single bipolar transport and emitting layer ŽBTEL. in place of the bilayer structure w3x. This single BTEL is constructed by mixing the hole transport and electron transport in an appropriate ratio. Reliability of OLEDs has always been a concern. Many degradation mechanisms have been proposed and observed for OLEDs. McElvain et al. w4x demonstrated that breaks in the metalrorganic contact caused by particles can nucleate sites for delamination of the electrode and for oxidizing species to enter thus reducing or stopping the luminescence process. These processes result in black spots forming in the device. At the other electrode, Adachi et al. w5x
) Corresponding author. Tel.: q1-480-413-5957; fax: q1-480-4135934; E-mail: [email protected]
found that the ionization potential Ž Ip . of the HTL was critical to reliability. Lower Ip results in a lower rate of degradation. Further, Tang and VanSlyke found that the drive voltage increases as the device degrades during constant current lifetest. In this work, we observed degradation in contact resistance to both bilayer and BTEL devices unrelated to the usual black spot formation. Joswick et al. w6x found that the crystallization of the emissive layer caused a reduction in the electroluminescence efficiency. Thus, the use of high glass temperature materials is prudent to avoid recrystallization. Also chemical reactions in the bulk of the semiconductor can affect reliability. Papadimitrakopoulos et al. w7x proposed that water and oxygen can react with Alq 3 Ž8-hydroxyquinoline aluminum. to form a non emissive polymer. We have proposed that charge accumulation at the interface can adversely affect reliability. Fields associated with the charge buildup result in higher voltage drop at the interface and Joule heating which results in device degradation w3x. The BTEL structure replaces a well defined 2 dimensional heterojunction with a three-dimensional BTEL layer. This has strong implications for reliability. In this work, we report our lifetest results on OLED devices of both the bilayer and BTEL structures. Two cathode contact schemes, and various temperatures and duty cycles were examined. Both relative change in device operating voltage and luminance were monitored. A corre-
0379-6779r99r$ - see front matter q 1999 Elsevier Science S.A. All rights reserved. PII: S 0 3 7 9 - 6 7 7 9 Ž 9 9 . 0 0 1 4 2 - 3
54
J. Curless et al.r Synthetic Metals 107 (1999) 53–56
values represent the values obtained at a driving current of 40 mArcm2 .
3. Results and discussion
Fig. 1. The three typical device structures discussed in this work: Ža. conventional bilayer, Žb. bilayer with LiFrAl cathode, and Žc. BTEL device.
lation between the two values during constant current lifetest was found.
2. Experimental The devices used in this study fall into three categories as illustrated in Fig. 1: Ža. is conventional bilayer, Žb. is bilayer with a LiFrAl cathode, and Žc. is BTEL with LiFrAl cathode. The method of manufacture was briefly described previously w3x. The thicknesses and doping are given in Fig. 1 except where noted. MQA Ž N, N-dimethylquinacridone. is a dopant to increase luminescence. NPB X .1,4-biphenyl. is a hole ŽbisŽ N, N-phenyl-1-naphthylamino transporting material and CuPc Žcopper phthalocyanine. improves the ITOrorganic interface. All accelerated lifetests were conducted at a constant current density of 40 mArcm2 in dry nitrogen. In order to fairly compare the accelerated lifetest data, luminances were normalized to 1000 Cdrm2 . The normalization process keeps the same luminance time product for each point on the plot and allows useful comparison between devices. The normalization was accomplished by multiplying all luminances by 1000 Cdrm2 and dividing by the initial luminance. The time scale is likewise adjusted by multiplying the actual time by the initial luminance and dividing by 1000 Cdrm2 . Further justification for this normalization process comes from the experimentally observed fact that luminance is linearly proportional to current. Essentially, the lifetest data is that which would have occurred if the devices had been set to 1000 Cdrm2 initially. Test devices yielded luminances from 700 to 1400 Cdrm2 at 40 mArcm2 ; thus, 1000 Cdrm2 was chosen as a convenient level for comparison of the test samples. Because operating voltage is not linearly proportional to current, data involving voltage change during lifetest is not normalized in this paper. In the case where operating voltage is examined in this paper, luminance and lifetime
Fig. 2 is a lifetest plot for four duty cycle conditions for each of three device structures, conventional bilayer Ždevice type a., AlrLiF cathode bilayer Ždevice type b. and BTEL device Ždevice type c.. The lifetests were conducted at room temperature. The average forward current for all devices was 40 mArcm2 . The duty cycles examined were: DC Žcircle., 50% time on Žsquare., 50% time on with 5 V reverse bias when off Ždiamond. and 10% time on Žtriangle.. Only DC bias was examined for device type b. The devices in this test were all manufactured consecutively from the same batches of materials in the same tool, except for the device of type b, which was manufactured at an earlier time from different batches of materials. Duty cycle has some effect on reliability. The duty cycle with the best reliability is the case of 50% duty cycle with 5 V reverse bias when the device is off. It is also apparent that switching to the BTEL structure greatly improves reliability. Tang and VanSlyke w1x noted that operating voltage increased during lifetest at constant current. We have found that the relative change in voltage Žoverall increase in voltage divided by the initial voltage. is linearly proportional to the relative change in luminance Žoverall drop in luminance divided by the initial luminance.. Fig. 3 shows typical plots of relative change in voltage as a function of relative change in luminance for the three device types examined in this work. The data represents the results of 40 mArcm2 lifetests with no normalization. The proportionality often holds over most of the half luminance lifetime, but at some point will change to a steeper slope. The slope from the linear portion of the graph is a figure of merit for the voltage increase over the device life and
Fig. 2. Room temperature lifetest data comparing device types a, b and c Žcircle is DC, square is 50% duty cycle, diamond is 50% duty cycle with 5 V reverse bias when device is off, and triangle is 10% duty cycle..
J. Curless et al.r Synthetic Metals 107 (1999) 53–56
Fig. 3. Relative change in voltage during room temperature lifetest as a function of relative change in luminance during a 40 mArcm2 lifetest.
will be designated here as ŽdVrd L.Ž LrV .. This indicates that the same mechanism may be responsible for the degradation of luminance and the increase in voltage. Fig. 4 is a plot of ŽdVrd L.Ž LrV . vs. the half luminance lifetime Žnot normalized. at room temperature. The data points fall into three clusters corresponding to our three classifications: bilayer with Mg:Ag, bilayer with LiFrAl contact, and BTEL devices. It should be noted that in the case of the BTEL devices, while having the same layered structure, different layer thicknesses were employed on the various devices. Also the initial luminances are different due to concentration differences. Thus, there is a wide scatter of lifetimes in the data. It is apparent from the plot that ŽdVrd L.Ž LrV . is determined primarily by the contact type. Average values for ŽdVrd L.Ž LrV . are indicated by the horizontal lines on the graph. MgrAg contacts have a ŽdVrd L.Ž LrV . of about 0.57, while LiFrAl contacts have a ŽdVrd L.Ž LrV . of about 0.23. Since bilayer with LiFrAl and BTEL structures both have essentially the same ŽdVrd L.Ž LrV . values, ŽdVrd L.Ž LrV . is independent of the organic material structure. In addition, ŽdVrd L.Ž LrV . is independent of the device half luminance lifetime. The fact that voltage
Fig. 4. Plot of the figure of merit for voltage change vs. unnormalized half luminance lifetimes for three device structures.
55
increase is linked to luminance degradation indicates that the same mechanism may be responsible for both. It is apparent that the cathode material is also crucial to the ŽdVrd L.Ž LrV . observed. It is possible that the differences in injection efficiency for the different cathodes results in different Joule heating which accelerates the degradation in the organic material where the degradation occurs. One conclusion is apparent. The figure of merit ŽdVrd L.Ž LrV . indicates that the devices with LiFrAl cathode organic interface exhibit less degradation than those with MgrAg cathode organic interface. Fig. 5 shows some results from an accelerated lifetest performed at elevated temperatures. It is seen that at 458C the BTEL device takes approximately 10 = more time to fall to the same luminance level as a conventional bilayer device. The reason for the improvement in reliability is believed to be the reduction or the elimination of the organic heterojunction. The carriers do not build up at the heterojunction, which results in a reduction of fields within the organic materials. The reduced fields slow the degradation caused by Joule heating as the carriers transverse the fields in the organic materials. At 858C, the BTEL device starts out requiring 10 = as much time to reach 800 Cdrm2 as does the bilayer device. However, the luminance values for the BTEL device falls off more rapidly after reaching 800 Cdrm2 and converges with that of the bilayer device. Examination of both type a and c devices from the 858C test revealed that the devices were darkening except at the periphery and around the initial dark spot defects. New devices of both types were placed at 858C without any current applied. These devices showed the same pattern of degradation, a darkening center with bright periphery and bright regions about the initial dark spots. Fig. 6 is an example of a device after storage at 858C. When the device was operated at 40 mArcm2 , it was revealed that the bright areas were actually brighter than they were initially. While the dark areas were dimmer. The device was thus becoming resistive in the dark regions and shunting current to the less resistive bright regions. The fact that the periphery and regions
Fig. 5. Lifetest data at elevated temperatures.
J. Curless et al.r Synthetic Metals 107 (1999) 53–56
56
Fig. 6. Bilayer device with LiFrAl contact after 858C storage.
around the dark spots remained very luminescent indicates that this particular reliability problem was not due to the in-diffusion of oxidizing species as observed by McElvain et al. w4x. Any species diffusing in would undoubtedly effect the periphery and black spot regions first. However, it is likely due to a contact problem as the organic materials in the bright regions seem unaffected. The change in the interface occurs whether or not the device is turned on. It is apparent that this contact reliability problem limits the ultimate lifetime for devices tested in this work at 858C.
4. Conclusions The above results show that the BTEL device is significantly more reliable than bilayer devices at room tempera-
ture. The degradation of the device is determined by the amount of charge passing in the forward direction through the device. Duty cycle is not of primary importance for the devices in this study. Operating temperature is a critical parameter to the devices we fabricated, and BTEL devices have a longer lifetime than bilayer devices at elevated temperatures. A contact resistance problem developed for all our devices tested at 858C. This problem manifested itself as a darkening region where the contact resistance increased. This problem appeared in both bilayer and BTEL devices. This problem imposed an ultimate limit to lifetime tested at 858C for the devices tested. A figure of merit was proposed for the change in voltage during lifetest. This figure of merit is the proportionality constant between relative change in voltage and relative change in luminance. The figure of merit was found to be determined by the cathode contact scheme. The AlrLiF cathode gave a much reduced relative change in voltage for the same relative change in luminance than did the AgrMg:Ag cathode.
References w1x C.W. Tang, S.A. VanSlyke, Appl. Phys. Lett. 51 Ž1987. 913. w2x C.W. Tang, S.A. VanSlyke, C.H. Chen, J. Appl. Phys. 65 Ž1989. 3610. w3x V.-E. Choong, S. Shi, J. Curless, F. So, J. Shen, J. Yang, Appl. Phys. Lett. 75 Ž1999. 172. w4x J. McElvain, H. Antoniadis, M.R. Hueschen, J.N. Miller, D.M. Roitman, J.R. Sheats, R.L. Moon, J. Appl. Phys. 80 Ž1996. 6002. w5x C. Adachi, K. Nagai, N. Tamoto, Appl. Phys. Lett. 66 Ž1995. 2679. w6x M.D. Joswick, I.H. Campbell, N.N. Barashkov, J.P. Ferraris, J. Appl. Phys. 80 Ž1996. 2883. w7x F. Papadimitrakopoulos, X-M. Zhang, D.L. Thomsen III, K.A. Higginson, Chem. Mater. 8 Ž1996. 1363.
Reliability comparison of BTEL and bilayer organic LEDs J. Curless ) , S. Rogers, M. Kim, M. Lent, S. Shi, V.-E. Choong, C. Briscoe, F. So Physical Sciences Research Laboratories, Motorola Labs, Motorola, 2100 E. Elliot, Tempe, AZ 85284, USA Received 22 April 1999; received in revised form 7 July 1999; accepted 8 July 1999
Abstract Organic structures consisting of two distinct layers Žbilayer., a hole transport layer and an electron transportremitter layer are compared to structures using a bipolar transport and emitting layer ŽBTEL.. Reverse bias and different duty cycle did not significantly affect reliability. The BTEL structure had significantly improved reliability compared to the bilayer structure. The relative change in device voltage was found to be linearly proportional to the relative change in luminance and the constant of proportionality was a function of the contact. This constant of proportionality can be used as a figure of merit for voltage increase during operation. The BTEL structure also gives improved reliability at elevated temperature. q 1999 Elsevier Science S.A. All rights reserved. Keywords: Electroluminescence; Light emission; Reliability; Organic light-emitting diode; Organic light-emitting device
1. Introduction In the 1980s, Tang and VanSlyke w1x and Tang et al. w2x developed an organic light-emitting diode ŽOLED. structure which gave high luminance while operating at low voltages. Their structure is called a bilayer structure because it contains two distinct transport layers: the hole transport layer ŽHTL. and the emitterrelectron transport layer. Recently, an OLED structure has been developed which uses a single bipolar transport and emitting layer ŽBTEL. in place of the bilayer structure w3x. This single BTEL is constructed by mixing the hole transport and electron transport in an appropriate ratio. Reliability of OLEDs has always been a concern. Many degradation mechanisms have been proposed and observed for OLEDs. McElvain et al. w4x demonstrated that breaks in the metalrorganic contact caused by particles can nucleate sites for delamination of the electrode and for oxidizing species to enter thus reducing or stopping the luminescence process. These processes result in black spots forming in the device. At the other electrode, Adachi et al. w5x
) Corresponding author. Tel.: q1-480-413-5957; fax: q1-480-4135934; E-mail: [email protected]
found that the ionization potential Ž Ip . of the HTL was critical to reliability. Lower Ip results in a lower rate of degradation. Further, Tang and VanSlyke found that the drive voltage increases as the device degrades during constant current lifetest. In this work, we observed degradation in contact resistance to both bilayer and BTEL devices unrelated to the usual black spot formation. Joswick et al. w6x found that the crystallization of the emissive layer caused a reduction in the electroluminescence efficiency. Thus, the use of high glass temperature materials is prudent to avoid recrystallization. Also chemical reactions in the bulk of the semiconductor can affect reliability. Papadimitrakopoulos et al. w7x proposed that water and oxygen can react with Alq 3 Ž8-hydroxyquinoline aluminum. to form a non emissive polymer. We have proposed that charge accumulation at the interface can adversely affect reliability. Fields associated with the charge buildup result in higher voltage drop at the interface and Joule heating which results in device degradation w3x. The BTEL structure replaces a well defined 2 dimensional heterojunction with a three-dimensional BTEL layer. This has strong implications for reliability. In this work, we report our lifetest results on OLED devices of both the bilayer and BTEL structures. Two cathode contact schemes, and various temperatures and duty cycles were examined. Both relative change in device operating voltage and luminance were monitored. A corre-
0379-6779r99r$ - see front matter q 1999 Elsevier Science S.A. All rights reserved. PII: S 0 3 7 9 - 6 7 7 9 Ž 9 9 . 0 0 1 4 2 - 3
54
J. Curless et al.r Synthetic Metals 107 (1999) 53–56
values represent the values obtained at a driving current of 40 mArcm2 .
3. Results and discussion
Fig. 1. The three typical device structures discussed in this work: Ža. conventional bilayer, Žb. bilayer with LiFrAl cathode, and Žc. BTEL device.
lation between the two values during constant current lifetest was found.
2. Experimental The devices used in this study fall into three categories as illustrated in Fig. 1: Ža. is conventional bilayer, Žb. is bilayer with a LiFrAl cathode, and Žc. is BTEL with LiFrAl cathode. The method of manufacture was briefly described previously w3x. The thicknesses and doping are given in Fig. 1 except where noted. MQA Ž N, N-dimethylquinacridone. is a dopant to increase luminescence. NPB X .1,4-biphenyl. is a hole ŽbisŽ N, N-phenyl-1-naphthylamino transporting material and CuPc Žcopper phthalocyanine. improves the ITOrorganic interface. All accelerated lifetests were conducted at a constant current density of 40 mArcm2 in dry nitrogen. In order to fairly compare the accelerated lifetest data, luminances were normalized to 1000 Cdrm2 . The normalization process keeps the same luminance time product for each point on the plot and allows useful comparison between devices. The normalization was accomplished by multiplying all luminances by 1000 Cdrm2 and dividing by the initial luminance. The time scale is likewise adjusted by multiplying the actual time by the initial luminance and dividing by 1000 Cdrm2 . Further justification for this normalization process comes from the experimentally observed fact that luminance is linearly proportional to current. Essentially, the lifetest data is that which would have occurred if the devices had been set to 1000 Cdrm2 initially. Test devices yielded luminances from 700 to 1400 Cdrm2 at 40 mArcm2 ; thus, 1000 Cdrm2 was chosen as a convenient level for comparison of the test samples. Because operating voltage is not linearly proportional to current, data involving voltage change during lifetest is not normalized in this paper. In the case where operating voltage is examined in this paper, luminance and lifetime
Fig. 2 is a lifetest plot for four duty cycle conditions for each of three device structures, conventional bilayer Ždevice type a., AlrLiF cathode bilayer Ždevice type b. and BTEL device Ždevice type c.. The lifetests were conducted at room temperature. The average forward current for all devices was 40 mArcm2 . The duty cycles examined were: DC Žcircle., 50% time on Žsquare., 50% time on with 5 V reverse bias when off Ždiamond. and 10% time on Žtriangle.. Only DC bias was examined for device type b. The devices in this test were all manufactured consecutively from the same batches of materials in the same tool, except for the device of type b, which was manufactured at an earlier time from different batches of materials. Duty cycle has some effect on reliability. The duty cycle with the best reliability is the case of 50% duty cycle with 5 V reverse bias when the device is off. It is also apparent that switching to the BTEL structure greatly improves reliability. Tang and VanSlyke w1x noted that operating voltage increased during lifetest at constant current. We have found that the relative change in voltage Žoverall increase in voltage divided by the initial voltage. is linearly proportional to the relative change in luminance Žoverall drop in luminance divided by the initial luminance.. Fig. 3 shows typical plots of relative change in voltage as a function of relative change in luminance for the three device types examined in this work. The data represents the results of 40 mArcm2 lifetests with no normalization. The proportionality often holds over most of the half luminance lifetime, but at some point will change to a steeper slope. The slope from the linear portion of the graph is a figure of merit for the voltage increase over the device life and
Fig. 2. Room temperature lifetest data comparing device types a, b and c Žcircle is DC, square is 50% duty cycle, diamond is 50% duty cycle with 5 V reverse bias when device is off, and triangle is 10% duty cycle..
J. Curless et al.r Synthetic Metals 107 (1999) 53–56
Fig. 3. Relative change in voltage during room temperature lifetest as a function of relative change in luminance during a 40 mArcm2 lifetest.
will be designated here as ŽdVrd L.Ž LrV .. This indicates that the same mechanism may be responsible for the degradation of luminance and the increase in voltage. Fig. 4 is a plot of ŽdVrd L.Ž LrV . vs. the half luminance lifetime Žnot normalized. at room temperature. The data points fall into three clusters corresponding to our three classifications: bilayer with Mg:Ag, bilayer with LiFrAl contact, and BTEL devices. It should be noted that in the case of the BTEL devices, while having the same layered structure, different layer thicknesses were employed on the various devices. Also the initial luminances are different due to concentration differences. Thus, there is a wide scatter of lifetimes in the data. It is apparent from the plot that ŽdVrd L.Ž LrV . is determined primarily by the contact type. Average values for ŽdVrd L.Ž LrV . are indicated by the horizontal lines on the graph. MgrAg contacts have a ŽdVrd L.Ž LrV . of about 0.57, while LiFrAl contacts have a ŽdVrd L.Ž LrV . of about 0.23. Since bilayer with LiFrAl and BTEL structures both have essentially the same ŽdVrd L.Ž LrV . values, ŽdVrd L.Ž LrV . is independent of the organic material structure. In addition, ŽdVrd L.Ž LrV . is independent of the device half luminance lifetime. The fact that voltage
Fig. 4. Plot of the figure of merit for voltage change vs. unnormalized half luminance lifetimes for three device structures.
55
increase is linked to luminance degradation indicates that the same mechanism may be responsible for both. It is apparent that the cathode material is also crucial to the ŽdVrd L.Ž LrV . observed. It is possible that the differences in injection efficiency for the different cathodes results in different Joule heating which accelerates the degradation in the organic material where the degradation occurs. One conclusion is apparent. The figure of merit ŽdVrd L.Ž LrV . indicates that the devices with LiFrAl cathode organic interface exhibit less degradation than those with MgrAg cathode organic interface. Fig. 5 shows some results from an accelerated lifetest performed at elevated temperatures. It is seen that at 458C the BTEL device takes approximately 10 = more time to fall to the same luminance level as a conventional bilayer device. The reason for the improvement in reliability is believed to be the reduction or the elimination of the organic heterojunction. The carriers do not build up at the heterojunction, which results in a reduction of fields within the organic materials. The reduced fields slow the degradation caused by Joule heating as the carriers transverse the fields in the organic materials. At 858C, the BTEL device starts out requiring 10 = as much time to reach 800 Cdrm2 as does the bilayer device. However, the luminance values for the BTEL device falls off more rapidly after reaching 800 Cdrm2 and converges with that of the bilayer device. Examination of both type a and c devices from the 858C test revealed that the devices were darkening except at the periphery and around the initial dark spot defects. New devices of both types were placed at 858C without any current applied. These devices showed the same pattern of degradation, a darkening center with bright periphery and bright regions about the initial dark spots. Fig. 6 is an example of a device after storage at 858C. When the device was operated at 40 mArcm2 , it was revealed that the bright areas were actually brighter than they were initially. While the dark areas were dimmer. The device was thus becoming resistive in the dark regions and shunting current to the less resistive bright regions. The fact that the periphery and regions
Fig. 5. Lifetest data at elevated temperatures.
J. Curless et al.r Synthetic Metals 107 (1999) 53–56
56
Fig. 6. Bilayer device with LiFrAl contact after 858C storage.
around the dark spots remained very luminescent indicates that this particular reliability problem was not due to the in-diffusion of oxidizing species as observed by McElvain et al. w4x. Any species diffusing in would undoubtedly effect the periphery and black spot regions first. However, it is likely due to a contact problem as the organic materials in the bright regions seem unaffected. The change in the interface occurs whether or not the device is turned on. It is apparent that this contact reliability problem limits the ultimate lifetime for devices tested in this work at 858C.
4. Conclusions The above results show that the BTEL device is significantly more reliable than bilayer devices at room tempera-
ture. The degradation of the device is determined by the amount of charge passing in the forward direction through the device. Duty cycle is not of primary importance for the devices in this study. Operating temperature is a critical parameter to the devices we fabricated, and BTEL devices have a longer lifetime than bilayer devices at elevated temperatures. A contact resistance problem developed for all our devices tested at 858C. This problem manifested itself as a darkening region where the contact resistance increased. This problem appeared in both bilayer and BTEL devices. This problem imposed an ultimate limit to lifetime tested at 858C for the devices tested. A figure of merit was proposed for the change in voltage during lifetest. This figure of merit is the proportionality constant between relative change in voltage and relative change in luminance. The figure of merit was found to be determined by the cathode contact scheme. The AlrLiF cathode gave a much reduced relative change in voltage for the same relative change in luminance than did the AgrMg:Ag cathode.
References w1x C.W. Tang, S.A. VanSlyke, Appl. Phys. Lett. 51 Ž1987. 913. w2x C.W. Tang, S.A. VanSlyke, C.H. Chen, J. Appl. Phys. 65 Ž1989. 3610. w3x V.-E. Choong, S. Shi, J. Curless, F. So, J. Shen, J. Yang, Appl. Phys. Lett. 75 Ž1999. 172. w4x J. McElvain, H. Antoniadis, M.R. Hueschen, J.N. Miller, D.M. Roitman, J.R. Sheats, R.L. Moon, J. Appl. Phys. 80 Ž1996. 6002. w5x C. Adachi, K. Nagai, N. Tamoto, Appl. Phys. Lett. 66 Ž1995. 2679. w6x M.D. Joswick, I.H. Campbell, N.N. Barashkov, J.P. Ferraris, J. Appl. Phys. 80 Ž1996. 2883. w7x F. Papadimitrakopoulos, X-M. Zhang, D.L. Thomsen III, K.A. Higginson, Chem. Mater. 8 Ž1996. 1363.