Challenges for Small- and Medium-Sized AMOLED Displays

Many issues stand in the way of the mass production of active-matrix organic-light-emitting-diode (AMOLED) displays, which includes front-of-screen (FOS) performance, infrastructure cost, and production-yield control. Here, solutions to several of these problems are proposed, including methods for realizing higher performance with excellent FOS performance, how to reduce fabrication cost, and how to increase yield.

by Du-Zen Peng, Hsiang-Lun Hsu, and Ryuji Nishikawa

ORGANIC ELECTROLUMINESCENCE (OEL) continues to draw attention for the next generation of display technology.Active-matrix organic light-emitting-diode (AMOLED) displays promise myriad advantages compared to other displays, which include high picture quality, fast response time suitable for displaying moving images, ultra-thin panel structure, low-power consumption, and wide viewing angle. However, the commercialization of AMOLED displays has met repeated delays, as the move to full color, good uniformity, higher resolution, and larger screen size with reasonable fabrication cost has proven challenging. Continued advances and improvements in liquid-crystal-display (LCD) technology are also acting as a barrier to the introduction of AMOLED displays.

In 2002, Sanyo and Kodak formed a joint venture – SK Display – with the intent of developing AMOLED displays for digital still cameras (DSCs). In 2004, Sony began mass production of AMOLED displays for personal digital assistants (PDAs). Then in 2006, AU Optronics Corp. used AMOLED displays for the first time in mobile-phone applications. However, all of these efforts have met with delays to or termination of the mass-production schedule.2,3 Despite the superior performance of AMOLED displays, they still have many issues that need to be overcome in order to achieve the full capabilities required for mass production. These include:

Du-Zen Peng, Hsiang-Lun Hsu, and Ryuji Nishikawa are with TPO Displays Corp., No. 12 Ke Jung Rd., Science-Based Industrial Park, Chu-Nan 350, Miao-Li County, Taiwan; telephone +886-37-586393 ext. 11039, e-mail:


Fig. 1: Comparison of luminance efficiencies for structures A, B, and C.

•   Front-of-Screen (FOS) Performance and Device Stability: The FOS performance, which includes brightness, resolution, and power consumption, should be comparable to that of LCDs and meet customer specifications. Although a typical brightness specification for an AMOLED display is lower than that for an LCD, the contrast of an AMOLED in ambient lighting conditions is much higher, which leads to better image quality. Nevertheless, it is not easy to produce AMOLED displays with higher resolution and lower power consumption. Solving these issues within reasonable costs is therefore critical.

•   Infrastructure and Fabrication Cost: The cost of infrastructure includes the equipment and material costs. Because the manufacturing equipment required to fabricate AMOLED displays is not mature, higher fixed costs are expected at the initial stage compared to that of the mature LCD industry. Materials costs as well as fabrication costs must be reduced to those of competing technologies before significant market share can be won, which means that the production yield as well as the EL material utilization efficiency must be improved.

•   Production-Yield Control: The main yield loss in the fabrication of AMOLED displays is attributed to the intrinsic characteristics of the thin-film transistors (TFTs) as well as to variations caused during the TFT process. The resultant luminance non-uniformity on the panel reduces production yield. Although many solutions have been proposed to tackle this problem, few of them have been effective and low in cost. 

Analyses of these issues have led the majority of the OLED community to conclude that the advantages of OLEDs are seriously compromised in flat-panel-display applications. However, the development of new EL materials, equipment, and a manufacturing processes are seen as an essential first step in the establishment of an AMOLED business. The schedule for this development should be coordinated with improvements in the fabrication of the high-performance array substrates that will be needed to drive uniform AMOLED displays.

High Performance with Excellent FOS

Low Power Consumption (Device/Display Method/Management): Power consumption is one of the most important parameters in mobile-display applications. Although the power consumption required by an AMOLED device for the display of a typical image is about one-quarter of that for a fully white image, it still needs further improvement to be comparable to that of a TFT-LCD when a highly efficient LED backlight is used. The fundamental breakthrough of power savings should be the lower turn-on voltage of an EL device and the higher efficiency of the EL material itself. Moreover, a modified EL structure, such as tandem and PIN devices, has also been proposed to further increase the efficiency of the material.4,5 Because the improvement of the EL material is unpredictable, we therefore reported an effective connecting unit for the tandem white OLED (WOLED).a

The four-color RGBW (red, green, blue, and white) format within a single pixel has been widely studied in the industry for both AMLCDs and AMOLED displays. Although an RGBW format for LCD applications was proposed to increase the panel brightness, the purpose for OLED applications focused on reducing the power consumption of the AMOLED display. The power consumption for a RGBW format could be reduced to one-half that for a RGB format6 depending on the image itself, the white point of the material, and the EL efficiency. Because the white point of the WOLED itself is not exactly D65 (customer defined), color adjustment needed to meet the desired white point is required. The closer to D65 the white point is, the more power can be saved for most images.

The power consumption can be lower for higher-efficiency AMOLED displays and for AMOLED displays with a white point closer to the customer-defined point, as shown in Figs. 2 and 3. Figure 2 depicts the power-consumption ratio for three different types of white materials – WOLED 1 (20 cd/A with a white point close to D65), WOLED 2 (20 cd/A with white point of D65), and WOLED 3 (30 cd/A with a white point of D65). Figure 3 shows the power-consumption ratio of the RGBW and RGB formats for a white image for different EL-material white points. The other feature of the RGBW format is that the color saturation can be made as high as possible without greatly increasing the power consumption for a white image.

The power consumption can be further reduced through better light management, such as luminance adjustment according to the ambient environment. To meet this requirement for AMOLED display applications, a system with a self-adjusting luminance was developed via the fully integrated low-temperature polysilicon (LTPS) technique.

High Resolution: Color patterning is another key issue for high-resolution AMOLED display applications. A conventional fine metal mask (FMM) limits the ability to mass produce high-resolution AMOLED panels due to the FMM capability and yield loss caused by pixel defects. Several techniques such as laser-induced thermal imaging (LITI) and radiation-induced sublimation transfer (RIST) have been developed to solve this issue. However, these two techniques are still not mature.



Fig. 2: Power-consumption ratio for three different WOLED materials.

We have applied color-filter-on-array (COA) technology together with a white OLED (WOLED) to fulfill the need for a high-resolution large panel without using the FMM process. Table 1 indicates the specification of the 7-in. AMOLED panel with an WOLED and COA architecture. The restrictions of this approach are the increase in power consumption and reduction in NTSC ratio if a conventional LCD-type color filter is used. The power-consumption issue can be improved by using a high-transmittance color filter in addition to an increase in WOLED efficiency. Low-power approaches, such as the use of a RGBW four-color format7 or the luminance-adjustment method, can also be applied to further reduce the power consumption as described in the previous section.

The color saturation can be much improved with better optimization of the color filter and white OLED spectrum. Figure 4 shows the white OLED spectrum and the white OLED spectrum through a color filter, as well as the R, G, and B color-filter transmittance. The color saturation of OLED devices is typically 60% NTSC using 1.75 μm for each color-filter thickness. Higher color saturation, for example, >100% of NTSC, can be achieved if thicker color filters are used, along with an optimized spectrum for the white material. Therefore, the use of a white OLED with a COA structure can realize high resolution as well as high color saturation.

Table 1: Specifications of a 7-in. AMOLED display using WOLED and COA technology.

Display Size

7-in. Diagonal
Driving Method Voltage-Driven
Resolution 800 ´ RGB ´ 480
Pixel Pitch 64 μm ´ 192 μm
Pixel Arrangement Stripe
Driver Integrated/COG
Number of Colors 262,144 (6 bits)


Fig. 3: Power-consumption ratio for RGBW and RGB pixel formats for a white image using EL materials with different white points.


Fig. 4: Color spectra for white OLEDs with and without the use of color filters and the R, G, and B color-filter transmittance.

Figure 5 depicts how several technologies are being used to increase resolution. Compared to the normal LTPS design rule, the advanced LTPS design rule allows for the fabrication of panels with higher resolutions. By utilizing subpixel-rendering technology, the resolution can be even higher. For those displays with an aperture ratio smaller than 30%, it is suggested that a top-emission structure be applied. For top emission, the aperture ratio is not determined by the LTPS layout area and is usually higher than that for bottom emission, which is suitable for high-resolution applications.

Longer Lifetime: The lifetime of OLED material is usually determined by the duration of operating time until the luminance reaches one-half of its initial value. The material lifetime is the most important factor for AMOLED product lifetime and, therefore, it is very important to increase the material lifetime for mass production. Moreover, good control of the EL process is also a very important item for longer lifetime because of the higher stability of an organic interface. As an example, for evaporation, a common approach is the use of a cluster-type machine, which always requires queuing time during organic-layer evaporation. This leads to contamination and a poor interface on the EL material.

The impact on the lifetime of an EL device after queuing is shown in Fig. 6; the longer the queuing time, the poorer the lifetime. It is therefore important to shorten or even eliminate the queuing time between organic-layer depositions for longer lifetimes. Materials lifetimes have improved substantially in the past few years and are now adequate for many mobile applications. However, if one color degrades faster than the others, it leads to the phenomenon known as image sticking (burn-in of commonly displayed patterns) – a concern of many customers. Therefore, it is important that technologies be applied to further extend the product lifetime. The lifetime and current density has the relationship as shown in Eq. (1); it is essential that the current density in one subpixel can be reduced to increase the product lifetime.

Lifetime µ 1/(current density)n, (1)

where n > 0.

As the emission aperture increases, the current density decreases, and according to Eq. (1), the lifetime is expected to increase. To increase the aperture ratio, several approaches can be applied, including the use of an advanced LTPS design rule, a top emission structure, and subpixel-rendering technology for high-resolution AMOLED displays. For an advanced LTPS design rule, the layout areas for the TFTs as well as the storage capacitor are reduced and, therefore, the emission aperture (25–40%) is larger when bottom emission is considered. For the top-emission structure, the typical aperture ratio is around 50–75%, which is supposed to be most beneficial to longer lifetimes.


Fig. 5: Aperture ratio versus panel resolution for different LTPS technologies.


Fig. 6: Lifetime comparisons with and without queuing.

By combining all the technologies as described above, AMOLED displays that feature high resolution, low power consumption, and longer lifetime (>50 khours) can be realized for high-performance mobile displays.

Low Cost, High Yield

One method for reducing the production costs of AMOLED displays is to improve the yield of the backplane. An LTPS backplane is considered mandatory for an AMOLED display because of its higher driving capability, superior reliability, and higher thermal endurance compared to an amorphous-silicon (a-Si) TFT backplane.8 However, the spatial non-uniformity in output characteristics caused by LTPS process variations9 make it difficult to generate uniform current through each pixel for LTPS TFTs. The non-uniformity of the TFT characteristics leads to what is called "mura." To overcome this non-uniformity issue, pixel compensation circuits have been proposed.10-12 The pixel circuit contains six transistors (6T) and one storage capacitor, which compensates not only the threshold voltage of the driving TFT but also the OLED power voltage drop on the power lines (Fig. 7).

The operation of the pixel is divided into three periods. The first period is to discharge the Cst in preparation for data loading and programming. The second period loads the Vdata into node A, while the driving TFTs M2 and M4 form a diode-connection structure, and the voltage at node B will be charged to approximately PVdd – Vth, where Vth is the threshold voltage of M2. Therefore, the voltage across the Cst now becomes VAB = Vdata – PVdd + Vth. In the final period, M5 turns on and the voltage at node A is equal toVref, and the voltage at node B now becomes Vref – VAB = Vref – Vdata + PVdd – Vth. The current I flowing into the EL material has the following relationship

I = (1/2)Cox(μW/L)(Vdata – Vref)2.   (2)

Note that current I is independent of the Vth of M2 and the OLED power voltage PVdd. The simulation results are similar to that of the previous 5T pixel circuit13 and can be found in Fig. 8, in which the current variation is small (< 5%) for Vth changes from –1.1 to –1.8 V (calculation normalized to Vth = -1.4 V).

It is advantageous that the OLED current be independent of PVdd, especially for large-panel AMOLED display applications since, in large panels, the current in the power line is large, which causes a voltage drop due to the finite resistance on the power line. For a conventional 2T structure, the luminance and hence uniformity will be influenced by the OLED power voltage drop. Figure 9 compares the current variation for 2T and 6T pixel circuits, assuming PVdd drops from 10 to 9 V. The current variation is within 10% even if there is a 1-V drop in PVdd for the proposed 6T structure.


Fig. 7: The pixel circuit contains six transistors and one storage capacitor, which compensates not only the threshold voltage of the driving TFT but also the OLED power voltage drop on the power lines.


Fig. 8: Variations in current for threshold voltages from –1.1 to –1.8 V.

Another way to lower the production cost is to reduce the fixed cost. Figure 10 compares the cost structure for AMLCDs and AMOLED displays for different glass sizes and TACT times. As can be seen in the figure, the fixed cost for AMOLED displays is lower for larger motherglass sizes because the average cost per unit panel is lower on larger-dimension glass. Additionally, if the general TACT time for OLEDs can be shortened from the current standard of about 4 minutes (with cluster evaporation and encapsulation) to less than 2 minutes, then the fixed cost can be further reduced. Therefore, enhancing the productivity, improving the yield, and increasing the EL-material utilization efficiency are important factors in lowering the production cost of AMOLED displays.14-19


In 2007, TPO Displays Corp. will explore new processes and improved substrate technologies. Our target is to match the performance of LCDs within 2 years, mainly in terms of FOS performance and unit price per panel. Currently, AMOLED displays are almost competitive with LCDs and are expected to exceed LCD performance in terms of panel resolution, lifetime, power consumption, and production cost in 2009.


aThe white-emitting two-unit tandem devices were fabricated with a structure of ITO/HI-01 (60 nm)/HT-01 (20 nm)/BH-01: BD-04 (10 nm)/BH-01: RD-01 (25 nm)/Alq3 (10 nm)/ Alq3:Cs2CO3 (20 nm)/Al (1 nm)/MoO3 (5 nm) /HI-01 (device A: 50 nm, device B: 55 nm, device C: 60 nm)/HT-01 (20 nm)/BH-01: BD-04 (10 nm)/BH-01: RD-01 (25 nm)/Alq3 (25 nm)/Cs2CO3 (1 nm)/Al. The structure consists of a thin metal layer (Al) as the common electrode, a hole-injection layer (MoO3) providing hole injection into the upper unit, and an electron-injection layer (Alq3:Cs2CO3) providing electron injection into the lower unit. The efficiency curves shown in Fig. 1 indicate A: 16.9 cd/A, B: 16.6 cd/A, C: 17.5 cd/A at 20 mA/cm2, which is more than double that of a single-unit device (8.3 cd/A at 20 mA/cm2).

1C. W. Tang and S. Van Slyke, Appl. Phys. Lett. 51, 913 (1987).

2Nikkei Microdevices, 35-49 (July 2006).

3Monthly Display, 5-10 (September 2006).

4L. S. Liao, et. al.Appl. Phys. Lett. 84, No. 2 (January 2004).

5G. He et. al.Appl. Phys. Lett. 85, No. 17 (October 2004).

6J. P. Spindler, et. al.SID Symposium Digest Tech Papers 36, 36-39 (2005).

7J. P. Spindler, et. al.SID Symposium Digest Tech Papers 36, 36-39 (2005).

8D-Z. Peng, et al.Proc. IDW/Asia Display '05, 629 (2005).


Fig. 9: Variations in current for EL power voltages from 10 to 9 V.


Fig. 10: Cost structure comparisons for different display technologies.

9M. Hack, et al.Proc IDMC, 57 (2002).

10D-Z. Peng, et al., SID Symposium Digest Tech Papers 36, 814 (2005).

11D-Z. Peng, et al.Proc. IDW/Asia Display '05, 629 (2005).

12Y-M. Tsai, et al.AM-FPD Digest '06, 289 (2006).

13D-Z. Peng., et al., SID Symposium Digest Tech Papers 36, 814 (2005).

14M. Long, J. M. Grace, D. R. Freeman, N. P. Redden, B. E. Koppe, and R. C. Brostl, SID Symposium Digest Tech Papers 37, 1474-1476 (2006).

15S. Hartmann, H.-H. Johannes, R. Caspary, W. Kowalsky, H. Schwab, D. Bertram, H. Blei, N. Meyer, B. Marheineke, and M. Heuken,SID Symposium Digest Tech Papers 37, 1563-1566 (2006).

16K. Chung; N. Kim, J-H. Choi, C-W. Chu, and J-M. Huh, SID Symposium Digest Tech Papers 37, 1958-1963 (2006).

17C-W. Kim, H-G. Kwon, J-S. Rho, J-S. Yoon, and K-B. Bae, SID Symposium Digest Tech Papers 37, 432-435 (2006).

18M. Schreil, M. Eritt, J. Amelung, K. Leo, U. Hoffmann, and M. Bender, SID Symposium Digest Tech Papers 35, 1376-1379 (2004).

19L. M. Moro, X. Chu, H. Hirayama, T. Krajewski, and R. J. Visser, Proc. IMID/IDMC (2006). •