Directions in Flat-Panel Displays

Ruth DeJule, Associate Editor -- Semiconductor International, 8/1/1999

Twenty years ago, flat-panel displays were in their infancy, and liquid crystal displays (LCD) were simple monochrome. Today, full-color LCD videos on camcorders are common, and LCD television is entering world markets. Other flat-panel display (FPD) technologies are progressing as well. Plasma TVs with 42-in. rectangular screens, for example, are available from sources including Fujitsu, Hitachi, Mitsubishi, NEC and Pioneer. Plasmaco, owned by Matsushita, leads the pack, demonstrating a 60-in. screen with resolutions of 1,366 X 768 full-color pixels. And by the year 2002, Canon and Toshiba anticipate a field-emission display (FED) TV. Still, LCD controls the industry with >90% of the worldwide flat-panel display market. With more than 40 LCD production lines in operation and signs of saturation in the notebook market, the versatile LCD is moving into new businesses such as desktop monitors. But potential rivals, such as organic LEDs, appear to be gaining ground.

TFT LCD

FPD's leading technology, thin film transistor (TFT) -- often called active matrix (AM) LCD -- is probably the most complex approach imaginable to making a display, according to Stanford Resources' David Mentley. Consisting of a liquid crystal sandwiched between two sheets of glass containing transparent electrodes, polarizers and back light, TFT LCDs position transistors at each pixel. By the mid-1980s, economic circumstances, coupled with sheer critical mass, created an environment in which the necessary LCD components came together. Dozens of companies, mostly in Japan, systematically addressed manufacturing yield, device structure/ transistor architecture, power consumption (reduced from 25 W to 2 W for notebooks), viewing angle and speed.

The LCD40, a 14-in. flat-panel monitor, is just 2.3 in. deep, weighs 10lbs. and uses 90% less desk-top space than a standard CRT display (Source: Mitsubishi)

Amorphous silicon TFTs are typically positioned under each pixel element to switch the pixel on and off. Despite low mobilities, less than 1 cm²/Vsec, a-Si TFTs have high enough off-resistance to reduce crosstalk between pixels and thus enhance on/off contrast. Fabricating a TFT LCD is similar to semiconductor processing, but with a fraction of the masking steps -- approximately six and progressing to four or five. Relying on engineering ingenuity rather than new technologies, TFT LCD technology has progressed markedly in three critical areas: viewing angle, efficiencies and speed.

Most of the panel light is emitted essentially perpendicular to the surface of the display, dropping off on the right and left sides; this is indicated by a drop-off in intensity and color changes. A wider left-to-right angle can be observed when viewed from the bottom of the display than from the top. To attain wider viewing angles, such as those currently exceeding 100°, the pixel is split into two or more sub-pixels, thus spreading the light further to the left and right. There are several approaches to improving the view angle, such as scattering film, in-plane switching and vertical alignment. One approach, multi-domain pixels, illustrates how viewing angle can be increased. This can be done with microgrooves, tens of microns wide, which create a local electric field to line up the liquid crystal, thus influencing the direction of the emitted light. To form the microgrooves, a polymer film is deposited on the liquid crystal and 'rubbed'. For each sub-pixel, the rubbing angle differs, sending light in various directions. Typically, a separate rubbing step is required for each sub-pixel orientation.

Similarly, increased efficiencies have followed straightforward techniques. For example, efficiencies have been improved as much as 20% by replacing standard fluorescent lamps with new designs that spread light from one or two bulbs, and through schemes to open up the active area of the display, such as smaller TFTs and thinner electrodes. Movement toward larger displays has brought greater demand for speed. But while traditional paths have, for the most part, been maintained, the need for speed and resolution has spurred new technologies.

Fig. 1 OLED, consisting of a light-emitting layer sandwiched between electron- and hole-transport layers, may pose a formidable challenge to LCD. (Source: Eastman Kodak)

TFT LCD monitors

Five years ago, an affordable flat-panel monitor was not considered possible. Today, though still 2X to 3X the price of a CRT monitor, the space- and energy-saving TFT LCD monitors have grown in sales, from 1,050,000 units in 1998 to just under 900,000 in the first quarter of 1999, according to Barry Young of DisplaySearch (Austin, Texas). Larger substrates, higher yields, vertical integration, reduction in the number of mask levels, and process and materials innovation have played a part in this cost reduction, according to Carl Steudle, marketing director of TFT display products at Samsung (San Jose, Calif.).

Despite many improvements in CRT technology, CRTs are still larger, heavier and consume more power than LCD displays. More importantly, CRTs face a fundamental tradeoff between brightness and resolution, according to Roger Stewart, vice president of engineering at Alien Technology (Hayward, Calif.). A brighter beam requires more electrons, but ensuing space-charge effects enlarge the spot size, decreasing resolution. So CRT monitors are high-resolution but may be dim, while entertainment systems are brighter but have lower resolution. TFT LCD monitors pose formidable competition in the range of 20- to 35-in. because they combine high resolution and high brightness capabilities.

The need for space is also driving the TFT LCD TV market for 15- to 25-in. screens. While flat-panel TV can be seen in Japan, its introduction into the world market is relatively recent with Sharp's 20-in. screen. One issue has been price. However, in five to seven years, TFT LCD TV is excepted to be comparable in price to a CRT TV, at $10/diagonal inch, according to Young.

Extending TFT LCD

Next-generation glass substrates will be a meter on a side. As panels get larger, faster routing times to pixels along the perimeter of the panel become more of an issue. Borrowed from the toy industry, tab attach is used to position drive chips around the periphery of the screen. Issues arise as panel size increases and with it a growing number of pixels that must be charged in 16 msec/frame. Choosing faster materials is the most straightforward approach. Polysilicon, for example, has mobilities 50X to 100X higher than a-Si, providing higher current, increased frequencies and smaller, faster devices. But fabrication requires additional processing steps, and there are indications of dopant piping at grain boundaries that can adversely affect device performance. There also are switching problems, such as controlling threshold voltages and achieving low leakage currents. Still, higher performance makes polysilicon preferable for high-density displays, and cost makes it more appropriate for smaller displays, Stewart said. Larger, lower-density displays favor amorphous silicon. Eliminating perimeter electronics can mean a 15%-50% cost savings. This can be achieved by integrating drive electronics onto either polysilicon or a-Si displays. Higher-density displays, such as those from dpiX and IBM, profit the most from electronics integration.

Fig. 2 SSD employs conventional vacuum-deposition and screen printing for easy low-cost processing. (Source: Westaim)

Unlike polysilicon technology, integration in a-Si requires no additional processing because the same process is used to form both the pixels and the core; this is accomplished entirely through design changes. The potential cost savings in integrating scanners onto that display is directly proportional to the number of leads eliminated. In amorphous silicon, integration may reduce the number of leads from 5000 to 200 in an SXGA display. This means potential savings of about $0.01/lead or ~$50 per panel. Additional electronics should be put on the display only if doing so reduces the number of leads or the bandwidth of the signals crossing the interface, Stewart said. If there is no interface advantage, it is cheaper per function to implement logic functions in silicon. So while the trend is toward system-on-a-panel, simplification of the interface is its main objective.

IBM's prototype display, the Roentgen, is a 16-in. TFT LCD monitor with a density of 200 pixels/in., about twice that of a typical high-end CRT. The Roentgen has 2560 x 2048 pixels. With only 6 to 8 msec to charge each pixel, aluminum wiring and double-driven gate lines effectively reduce RC time constants by more than 4X, compared to single-sided drive. Important issues are in processing, packaging and making 2000 meters of 5- to 10-mm wire onto glass without breakage, not the fabrication of a-Si TFTs. Reduction of TFT channel lengths (typically 10 to 20 mm) to meet charging requirements is an attractive alternative to changing TFT materials.

New approaches

Fluidic self assembly (FSA) is a new FPD technology that combines traditionally fabricated single-crystal TFTs into a flexible FPD process. Developed by Alien Technology (Hayward, Calif.), this patented process deposits known good devices over rigid or flexible large surface areas with 1 mm accuracies. A wafer of transistors is etched, separating the TFTs into individual nanoblocks. The nanoblocks are suspended in a fluid and flowed into precisely shaped holes that are punched, etched or laser drilled into the FPD substrate. The flow process takes ~1 min. Devices and substrate holes are matched mechanically in all three axes, and held in place with additives in the fluid and through molecular forces. Excess nanoblocks remaining on the surface are removed from the fluid, cleaned, re-suspended and re-flowed. A dielectric is then deposited, followed by standard metallization techniques. Because high-temperature device processing is separate from the display substrate, FSA opens the possibility of forming active matrix electronic back planes on a variety of substrates such as plastic, polyester, polyimides and polycarbonates. Consistent fills of 70 to 200 mm square blocks into 10,000 holes have been demonstrated, according to Glenn Gengel, vice president of manufacturing. Operation of a 1- x 2-in. display containing several hundred pixels has also been achieved. The next step is applying this technique to specific FPD technologies such as AMLCD and OEL.

A growing number of companies are developing liquid-crystal-on-silicon (LCOS) displays, which are manufactured in semiconductor fabs, according to Young. Liquid crystal is positioned between the backplane, which is fabricated on a silicon substrate and a glass frontplate. Light, from an LED for example, is reflected off the frontplate. Nematic liquid crystals are commonly used, producing millisecond switching. One manufacturer, Displaytech (Longmont, Colo.), achieves microsecond switching with ferroelectric liquid crystals, fast enough for full-frame video on a single microdisplay panel. The speed advantage eliminates trailing images (ghosting). Companies including IBM and National Semiconductor are developing these small (0.5-in. to 0.9-in. diagonals), reflective displays for large rear-projection TV and front-projection audio/visual systems.

Organic LED

Among the FPD technologies currently under development, organic light emitting diodes (OLED), sometimes referred to as OEL, may be the most promising, according to Dr. Paul E. Burrows of Princeton University's Center for Photonics and Optoelectronic Materials. This relatively new technology, pioneered by Eastman Kodak, uses thin layers of organic molecules or polymers that can behave much like single-crystal semiconductors. A light-emitting layer sandwiched between an electron-transport and a hole-transport layer can act much like an LED with electrons and holes combining in the emitting layer to produce light (Fig. 1).

Unlike TFT LCDs, OLEDs require no backlighting, top glass nor color filter. Color is achieved through selective doping of pixel sites in the luminescence layer. For each color, an appropriate host and dopant are evaporated through shadow masks, forming amorphous thin films. As with all active matrix displays, transistors at each pixel site maintain an 'on' state for the duration of the frame time. But while LCDs are basically capacitive devices, OLEDs are current-driven, making the higher mobilities and currents of polysilicon TFTs necessary. The higher-leakage currents that reduce the on/off contrast in LCD technologies are inconsequential in OLEDs because the currents are sufficiently below the threshold of light emission, according to Dr. David J. Williams, General Manager of Eastman Kodak Companies Display Technology Alliance.

Organic screens can be fabricated in large areas on inexpensive substrates, possibly metal foils and flexible plastic substrates. Today, OLED detachable faceplates for car audio systems are available from Pioneer. In addition to potentially providing higher brightness at all viewing angles than a-Si TFT LCDs, cost studies indicate OLED will be more cost-effective (Table), according to Stanford Resources' Mentley.

Fig. 3  The 5-in. solid state display contains 2 million colors (7-bit grayscale), and full-motion video capabilities may be 20 times faster than LCDs. (Source: Westaim).

Solid state displays

Originally developed at Sharp, thin film electroluminescence (TFEL) screens are solid state emissive displays with an image quality similar to CRTs. Relatively few companies are involved in this area: Planar (Beaverton, Ore.), Luxell (Toronto) and Westaim Advanced Technologies (Toronto). TFELs consist of an insulator/phosphor/insulator layer sandwiched between patterned electrodes. When voltage is applied across the layers, current flowing through the phosphor creates pockets of excitation that produce light. The light travels up through a color filter before it is emitted from the front of the display (Fig. 2). One approach developed by Westaim addresses cost issues that have confronted this technology. Solid state displays (SSD) use conventional vacuum deposition for the thin layers and screen printing for the thick layers, thus reducing manufacturing costs, according to Don Carkner, manager of applications engineering at Westaim. Substrate materials of ceramic or glass-ceramic make high-temperature processing practical. A 5-in. display with full color capabilities and video speeds has been demonstrated (Fig. 3).

What's next?

In the year 2000, DisplaySearch forecasts a notable 56% increase in the FPD equipment market to $2.3B. Driven by the demand for larger displays, the FPD materials market will see similar growth, to $3.1B. This will support a booming $16B FPD market. Expect to see FPD books replacing heavy textbooks that traditionally are lugged around campus, and more hand-held, wireless Internet-access devices that will allow you to email from virtually anywhere. SID's Alan Sobel sees expansion into microdisplays for use in headmounts and as light valves in projectors. Will we be seeing fellow travelers wearing battery-friendly head-mount displays on airplanes? And how about roll-up computer screens?