Amorphous-Silicon TFT AMOLEDs

If amorphous silicon is not the ideal material for making active-matrix backplanes for organic light-emitting-diode displays, why are so many trying to use it?

by Corbin Church and Arokia Nathan

DESIGNERS have been committed to the idea that an effective backplane for active-matrix organic light-emitting-diode (AMOLED) displays must be powered by low-temperature polysilicon (LTPS) to provide the currents required by the organic light-emitting material. But LTPS technology represents a departure from the more-conventional and proven technology of amorphous silicon (a-Si), which has been a key enabler of the emergence and present dominance of active-matrix LCDs (AMLCDs). And companies have had concerns about adopting not just one new technology for OLEDs, but a second one (LTPS) for implementing high-value active-matrix products.

In the announcements of plans for Gen 6, Gen 7, and even Gen 8 fabs that seem to be a monthly occurrence, every massive investment is in a plant based on a-Si thin-film-transistor (TFT) technology. A parallel has been drawn between a-Si in the flat-panel-display (FPD) industry and the incredible staying power of optical photolithography in the semiconductor industry.

Optical photolithography is the imaging technology supposedly doomed more than a decade ago, but today it is pushing into the regime of 45-nm-feature sizes, a reality that few would have predicted ten years ago. But is this really that surprising? When there are such enormous financial implications, it is only natural to push an existing technology to its apparent limits and beyond, with the pushing fueled by innovation and persistence. Similarly, is a-Si about to be pushed beyond its apparent limits into a new regime of successful applications? Some observers argue that we are indeed nearing this stage in the context of the growing OLED revolution.

Limitations of LTPS

LTPS technology became the presumed backplane of choice for AMOLEDs for three primary reasons: higher charge-carrier mobility, immunity to threshold-voltage shift, and the ability to integrate drivers directly on the panel because it can accommodate both n- and p-MOS structures. It was theorized that not only would LTPS technology enable AMOLEDs in the short term, its promise of integrated drivers would mean lower costs in the long term. These benefits are compelling, but the technology has had a difficult time delivering on its promises.

In practice, it has been extremely difficult to achieve good yields with LTPS, primarily because of the challenges associated with the re-crystallization step, which results in a lack of uniformity across the display. Moreover, LTPS requires nine mask steps compared to five for a-Si, making LTPS roughly 40% more expensive than a-Si on an array-to-array basis for small displays. Integrated drivers appear to be cost-effective only below panel sizes of 3 in. on the diagonal.

But perhaps the most problematical aspect of LTPS is the world's limited LTPS fabrication capacity. The largest LTPS fabs are only of Gen 3.5 size (600 x 720 mm2), compared to the Gen 5 fab size of 1100 x 1250 mm2 and over for a-Si, which limits the size of the final display to roughly 17 in. (Table 1). Of the nearly 50 million square meters of TFT glass projected to be produced in 2007, a mere 6.5% will be LTPS based, according to DisplaySearch, and all of that will be used to satisfy captive demand. There will be no capacity available for third parties.

 


Table 1: Comparison of Backplane Technologies
Attribute a-Si:H Poly-Si
Drive capacity (type) Low mobility High mobility
Large W/L (nMOS) Small W/L (nMOS/pMOS)
Manufacturability and accessibility Mature and accessible New and not yet accessible
Vth uniformity across array Good Poor
ΔVth Poor Good
Fabrication plants Gen 7 (1850 x 2100 mm2) Gen 3.5 (600 x 720 mm2)
Drop-in solution to existing infrastructure? Yes No
Normalized array-to-array cost for small displays 60% 100%

 

Advances Favoring a-Si

With a maturing base of OLED ink-jet-printing technology and recent advances in driver-IC packaging, the FPD industry is taking a bold new look at a-Si as an alternative for AMOLEDs. Ink-jet printing is clearly important for large-area-display manufacturing, while today's fine-pitch driver packages offer competitive footprints for small displays at low cost. (Footprint becomes less of a concern for diagonals greater than 3 in.)

These advances will enable the production of a-Si-based AMOLED displays in a full range of sizes – including, someday, large high-definition televisions (HDTVs) – by enabling existing capitalized fabrication facilities to be exploited once more for this new market opportunity. Yet, one essential breakthrough for successfully converting a-Si into an AMOLED contender does not involve advances in OLED materials, fabrication techniques, or packaging. It has been accomplished within the pixel electronics itself.

Managing Vth Shift

One problem with devices made with a-Si is their tendency to alter their properties when under electrical stress. More specifically, the threshold voltage (Vth) of an a-Si TFT shifts upward over time. If left unmanaged, this Vth shift causes current output – hence OLED luminance – to eventually fall to zero. In response, several research groups have introduced a-Si circuits designed to compensate for this shortcoming by using multiple transistors instead of the conventional two (Fig. 1), thereby applying an engineered solution requiring only a mask-design change to a materials-based problem.

Amorphous-silicon AMOLED circuits fall into one of two categories. They are either voltage-programmed or current-programmed, and they typically have three, four, five, or more transistors (Fig. 1). Some designs use the OLED itself as an integral part of the circuit, while others do not. Not surprisingly, each solution yields unique advantages and disadvantages from either an electrical or display-performance perspective, and no one circuit meets all requirements. But with clever design and layout techniques, a particular circuit can nonetheless provide good all-round performance.

 

fig_1_tif

Fig. 1: The current output of a conventional two-transistor (2T) a-Si OLED pixel circuit (far right) decreases with time because of a continuous and irreversible shift in threshold voltage. On the left are three of the many a-Si circuits that have been designed to compensate for this shortcoming by using a greater number of transistors.

 

fig_2_tif

Fig. 2: Accelerated lifetime data for the IGNIS four-transistor (4T) pixel circuit is shown. The increase in drive current to maintain constant luminance compensates for OLED-material degradation; the drive current has increased by less than a factor of 2 over an effective 30,000 hours of use.

 

The main criterion for determining circuit performance is long-term stability, a measure of constant OLED current supply to maintain constant luminance. While a handful of manufacturers have already demonstrated impressive AMOLED prototypes that clearly prove the potential of a-Si, the Achilles heel has always been how to demonstrate stability and long lifetime. For instance, in a device in which both the circuit and OLED exhibit different degradation rates, how is it possible to de-couple one from the other?

At IGNIS Innovation, Inc., we have developed a four-transistor (4T) current-mirror circuit that, unlike some other designs, does not require the OLED for proper operation. Therefore, its stability performance can be tested separately from the OLED itself. The stability of the circuit can be demonstrated by using a testing platform to program and monitor the performance of single pixels stressed under typical and accelerated video-driving conditions.

One IGNIS circuit has operated for 10,000 hours in real-time operation, but at high current loads, which has the effect of accelerating the life test by a factor of three. Thus, the pixel circuit shown has been operating for an equivalent lifetime of 30,000 hours, and the OLED current has not quite doubled (Fig. 2). The rising OLED current is an optional designed-in feature of the circuit to compensate for OLED degradation. Different circuits have also been stressed at both high current and high temperature, and have demonstrated similar equivalent lifetimes. Therefore, the long-term operation of an a-Si circuit for AMOLEDs has been demonstrated for the first time in a practical setting, proving its suitability for mobile applications (which have a 5000-hour requirement) and approaching television requirements (40,000 hours) as well.

Correcting Previous Assumptions

Contrary to popular belief, an a-Si circuit for AMOLEDs provides several important benefits besides economy and, now, adequate stability.

Aperture ratio. In standard AMOLED production, a bottom-emission architecture is used, so it is necessary to minimize TFT circuit area in order to maximize aperture. A higher aperture ratio is generally desired for bottom emission to reduce pixel current density (mA/cm2) and thus prolong OLED lifetime. Proper layout of the circuit is important, and circuit area depends primarily on the size of the TFTs used, which is dictated by the width/length (W/L) ratio and governed by manufacturing process constraints and line overhead. The minimum W/L in high volume is typically 7 μm/4 μm. The total circuit area is generally described as the area of the switching transistors (Aswitch) plus the area of the drive transistors (Adrive) plus the area of the interconnects, storage capacitor, etc.(Aoverhead).

While Aswitch and Adrive both vary by design, Aoverhead is a constant for both a-Si and LTPS technologies. It is surprising that an a-Si 2.2-in. QCIF+ (176 RGB x 220) cellular-telephone AMOLED that uses four transistors per circuit has an aperture ratio of 35%, which compares to 45% for an LTPS-based display using only two transistors per circuit, with both being limited by the overhead area. For small displays, the final choice of backplane technology is left to the manufacturer and is based on application, final form factor, power requirements, and cost. It is a decision that is neither easy nor clear-cut at this time.

 

fig_3_tif

Fig. 3: Pixel aperture ratios for small (top) and large (bottom) AMOLEDs using the bottom-emission structure are compared.

 

On the other hand, a-Si is the choice for larger displays, especially HDTV applications. The advantage is that a-Si pixel circuits do not have to scale linearly with increasing pixel size to deliver the higher drive currents that the larger OLED subpixels require. For example, although the subpixel area increases by 7.5 times going from a 2.2-in. QCIF AMOLED with a luminance of 300 cd/m2 to a 32-in. WXGA HDTV with a luminance of 800 cd/m2, the subpixel area of the four-transistor circuit increases by only 2.5 times. For HDTV, this translates into an impressive aperture ratio of 78%, which helps to prolong OLED lifetime by maintaining current densities at manageable levels (Fig. 3).

OLED placement. Placing the OLED on the drive TFT of the pixel circuit has important implications for the overall behavior of the AMOLED device. Because a-Si is only an n-MOS material, the conventional non-inverted PIN OLED stack must be located at the source of the drive TFT. Although this involves a simpler manufacturing process, the circuit becomes dependent on the characteristics of the OLED material itself when the OLED is located at the source.

The increasing voltage drop across the OLED stack as degradation occurs causes knock-on effects that the pixel circuitry must progressively be able to manage. While not a fatal flaw, it is a situation that circuit designers must respect. Conversely, OLED drain positioning is possible if the stack is flipped to an NIP structure. Drain positioning is feasible because inverted NIP OLED stacks are already under development – and should be in production within the next 1–2 years – and they would be ideal for the a-Si circuit because they would make the circuit immune to undesirable effects from the OLED.

Driving an a-Si Circuit

Both voltage-programmed and current-programmed a-Si circuits deliver the drive current to the OLED required by the input data signal, but they do so by different mechanisms. The designer must decide how to drive the pixel circuit.

Each mechanism has its advantages. Voltage programming typically produces a fast response, even over large areas; current programming provides excellent current accuracy and can better compensate for environmental and mechanical stresses. Each also has its disadvantages. Voltage programming imparts poor management of process-induced stresses or increasing OLED voltage drop, while current programming typically has a slow programming speed. There seems to be no a-Si-driver solution that can be relied upon to be practical for a broad range of applications. A confusing array of potential possibilities for driving a-Si-based AMOLED displays is thus presented, each of which applies to only a narrow range of products. So why not develop a driver that combines both techniques into one and can use off-the-shelf LCD drivers at the same time to keep costs low?

At IGNIS, we have implemented this idea in a novel hybrid-driver solution that can use both voltage and current programming (Fig. 4). In this system, voltage programming is the primary addressing function because it is fast and can be used in displays of all sizes and resolutions. Current programming is used only occasionally, when the system must be re-calibrated to manage the increasing threshold voltage of the TFTs and the OLED, and occurs in a way that is transparent to the user. The design leverages standard source/gate and controller ICs from the LCD industry, and requires only a simple current source and switch matrix to alternate programming between the voltage and current regimes for the affected columns. The design has been demonstrated in a prototype display using the 4T circuit and will next be applied to larger and higher-resolution demonstrators.

 

fig_4_tif

Fig. 4: A hybrid-driver solution has been developed that combines the advantages of voltage and current programming, and can be implemented with a standard display controller and standard LCD drivers.

 

Conclusion

With all the advances occurring in the OLED industry, including the demonstration of capable drive circuits based on a-Si, many are becoming convinced that a-Si is a viable lower-cost alternative to LTPS technology for implementation in active-matrix products. Considering the ongoing problems with LTPS, the flood of new a-Si processing capacity coming on line, and access to what are virtually drop-in solutions for a stable a-Si backplane, it is no surprise that many are looking to a-Si to power the AMOLEDs of the future. •


Corbin Church is VP of Operations at IGNIS Innovation, Inc., 1010 Sherbrooke St. W., Suite 818, Montreal, Quebec H3A 2RT, Canada; telephone 514/396-0212, fax 514/396-3511, e-mail: cchurch@ignis.ca. Arokia Nathan is CTO at IGNIS Innovation; telephone 519/888-4803, e-mail: anathan@ ignis.ca.