No sooner than the TFT LCD became technologically competitive with the incumbent Cathode Ray Tube display, and gained broader acceptance in the marketplace, challengers for its position have begun to appear. The second wave in flat panel display technology is rapidly approaching production, and promises to exceed the impressive capabilities of the TFT LCD.

The two most mature candidates for the crown of leading flat panel display technology are the Field Emission Display (FED) and the Thin Film Transistor Light Emitting Polymer (TFT LEP). Both of these technologies are emissive, rather transmissive as is the TFT LCD, therefore both offer very wide viewing angles and high contrast ratios in brightly lit environments.

With the impending proliferation of High Definition Television (HDTV) into the marketplace, the first live broadcasts having taken place in the US during Sen John Glenn's well publicised Shuttle flight, flat panel display technology will be will be of key importance. Large and wide screens are not the forte of the CRT display, which suffers geometrical distortions with increasing screen size, and therefore will never be a serious candidate for 0.75 x 1.5 metre domestic cinema environment.

What is no less important is that consumer commodity volume production of large and wide flat panel displays will make them rapidly affordable for computer applications, and since they are designed for a digital drive mechanism, interfacing will not be an issue. Unlike analogue TV technology, HDTV is wholly digital and employs a frame buffering scheme not unlike that in every computer graphics adaptor. Therefore adaptation of large flat panel HDTV displays to computing will happen rapidly, as the technology enters the marketplace.

Both the TFT LEP and the FED promise major improvements over the CRT and therefore there is much merit in exploring these technologies more closely.

Light Emitting Polymer Displays

The Light Emitting Polymer is a very recent scientific discovery. Most conventional organic polymer materials are excellent insulators, indeed most modern insulation materials are polymer based. The reason for this behaviour is simple, since such materials typically posses few if any free electrical charge carriers (electrons or holes), in comparison with semiconductors such as Silicon, or metals. A typical polymer comprises interlocking chains of large organic molecules. Charge carriers in conventional polymers can be liberated only by the application of extremely high electric fields which typically have the effect of damaging the material, literally ripping it apart at the polymer chain level.

Within the last three decades researchers have determined that some types of polymer materials can exhibit properties more traditionally associated with semiconductors or conductors. An example of such are Conjugated Polymers, in which the structure of the polymer chains creates conditions, under which carriers can be freed readily, and achieve useful mobilities along the polymer chain. The number of carriers and their mobility in a material determine its electrical conductivity.

The maturity of modern solid state physics at this time is such, that techniques have been developed to model the behaviour of such materials well enough to produce repeatable and predictable results. Classical semiconductor properties such as bandgap energies, have been observed in conjugated polymers. Many well established component structures in Silicon or other conventional semiconductors can be duplicated in polymers, although with significantly larger geometries and much higher defect densities. We will not be seeing a plastic microprocessor in the near future.

The result of this research has been a concerted effort to produce commercially useful semiconducting and conducting polymers. The first materials to mature to useful level have been conducting polymers, based on polyaniline (known for at least a century now), and polypyrrole. Suitably doped, such materials can exhibit conductivities similar to that of copper, a mainstay of our industry. One of the critical issues for such materials is stability, since they must last at least five to ten years even for consumer products. It would appear that this goal is close to being met now, and some manufacturers, such as Matsushita, are now producing polymer based capacitor electrodes, battery electrodes (incidently a key technology for electrically powered cars since weight is a major issue) and connections on speaker membranes. A long term goal is to replace the traditional etched copper printed circuit board with a polymer based technology.

By the early nineties researchers found another application for semiconducting polymers - the display. By using device engineering techniques developed for Light Emitting Diode (LED) technology, they were able to replicate the recombination mechanism which allows the LED to work - recombining carriers releasing energy as visible band photons of light. The heterojunction structure of dissimilar materials used in GaAs/GaAsP LEDs has its analog in polymer technology, with a similar ability to covert electrical energy into light.

Initially conversion efficiencies were abysmal, but refinement and exploitation of LED engineering methods have now allowed the design of polymer materials and structures which achieve genuinely useful efficiencies, comparable to LEDs, in excess of 5%. Wavelengths have now been demonstrated from the infrared band up to blue. In one half a decade from initial demonstration, the LEP has achieved both the conversion efficiency performance and colour range which LED designers took more than two decades to achieve. Importantly LEPs are electrically fast, at less than a microsecond their switching speeds are competitive with the LED, which is no mean feat for an essentially amorphous thick film material. Drive voltages for LEP emitters are low, of the order of 3V or less, similar to those of semiconductors, which makes them suitable for battery powered applications. Since the LEP is emissive, wide viewing angles can be achieved. In the simplest of terms, an LEP is a light emitting ink, which can be readily printed on to a solid or flexible substrate using inkjet printer technology.

The rapid rate of evolution in this technology owes much to the fact that there is no need for an expensive Silicon Valley fab to produce prototypes, therefore the time to fabricate test a new structure or material variant can be as short as a week.

Some key engineering issues remain before the technology hits the market en masse. The most important at this time is material lifetime, since the polymers will degrade if exposed to the atmospheric oxygen or moisture. Much of the current research effort lies in exploring the longevity of the materials, suitable sealants, and production techniques to avoid contamination. The target is an operating life of more than 20,000 hrs (about 2.3 years nonstop operation).

Another area of major effort is the design of materials which emit suitable red, green and blue wavelengths for practical RGB displays. If the wavelengths are displaced too far from conventional RGB CRT phosphors, a conventional video signal will need to have its "colour vector" rotated and scaled so that images have the correct colour. Otherwise, people viewed on such a display may look either seasick, or hypoxic !

The architecture of displays based upon LEP technology is another major issue. Test displays have been fabricated using passive drive techniques analogous to the passive LCD display (Systems March 98, pp 28). While passive LEP displays do not exhibit the nasty crosstalk properties of passive LEDs, they suffer similar drive capacitance speed limitations.

UK based Cambridge Display Technology (http://www.cdtltd.co.uk), pioneers in the technology and the key developers at this time, have licenced their patents to Seiko Epson, Philips, Hoechst, and Uniax. In a joint effort with Seiko Epson, CDT are now concentrating on the adaptation of the LEP technology to the Thin Film Transistor (TFT) based display substrate technology used with the established LCD. Using essentially similar drive and addressing techniques to the LCD, each TFT acts as a drive transistor switch for an LEP pixel or subpixel.

The glass substrate is produced as for an LCD, the TFT and addressing structures are fabricated on the substrate surface, and the LEP then applied. Using inkjet techniques, two layers are printed on to the substrate, geometrically registered so that the TFT connects to the anode of the LEP heterojunction structure. Between four and six printing passes will be required for an RGB display, depending on the LEP materials required. Since there is no need to laminate with LCD materials and polarising filters, the TFT LEP process is potentially much cheaper than the TFT LCD process, producing a much thinner (ie millimetres) and lighter display with no backlight. While the LEP is a power glutton compared to an LCD, this will be offset by the absence of the backlight tube and its power supply.

The TFT LEP flat panel display is an attractive approach for established manufacturers of TFT LCDs since they can reuse their existing investment in fabrication plant. The down side is that TFT LEP panels will be limited in size and penalised in cost by the TFT/substrate fabrication defect rates, the limitation of TFT LCD panels at this time.

It is reasonable to predict that if no major obstacles are encountered, the TFT LEP display will supplant the TFT LCD display in most applications by the end of the coming decade. My advice is not to invest in chemical manufacturers who specialise in LCD materials!

Field Emission Displays

The Field Emission Display is a cousin to the incumbent Cathode Ray Tube, and is based upon the same idea of exciting an emissive phosphor material with accelerated electrons. That is where the similarity ends alas.

Whereas a CRT employs three electron guns for the whole screen, and uses magnetic or electric fields to scan the beam across the face of the tube, the FED uses one or more miniature "electron guns" or cathodes for each pixel, or subpixel. Since the electrons need only travel a short distance in an FED, the display can be as thin as a centimetre.

The key to the FED is the little known technology of vacuum microelectronics, pioneered by Stanford Research Institute in the early sixties. The central idea in vacuum microelectronics is the fabrication of microscopic vacuum tube structures using the photolithographic and thin film techniques which are today common in the semiconductor industry.

Most current FED designs employ one or more microscopic, conically shaped cathode per each pixel. The cathode sits below a conductive gate structure, and the electron beam from each cathode passes through a hole in the gate, to the face of the FED which forms the anode. A typical design uses the sixties developed Spindt process, using amorphous silicon, and metal, to form the structures. The cathode itself may be silicon, or molybdenum, palladium oxide, carbon or another suitable material.

The physics of the anode-gate-cathode operation are in principle no different from those of the classical triode tube. A high voltage is applied between the cathode and the anode, which causes an electron beam to be emitted from the cathode and accelerated to the anode. Application of a negative voltage to the cathode controls the current in the beam, or shuts it off altogether. Thus the pixel or subpixel phosphor can be turned on or off, or modulated in brightness.

Unlike a CRT, the FED is a "cold cathode" device, in which the electrons are literally ripped out of the cathode by a very stong electric field. Whereas a typical thermionic tube uses a flat cathode, coated with a material which easily lets go of its electrons when heated (those ancients out there may recall the quaint practice of rejuvenating exhausted cathodes by burning off the surface layer with excess current), an FED uses most frequently an unheated conical cathode - the electric field at the tip of the cone is so strong that electrons leap out of the cathode material at room temperature. The best comparison in field strengths can be made by considering that a large CRT uses 35-45 kV across a two foot anode-cathode gap, whereas an FED uses hundreds of Volts up to 10 kV across a gap of millimetres.

The FED is not without its warts. A classical problem in all such devices is contamination of the vacuum by outgassing of materials with the tube. Molecules of species such as a oxygen will be trapped in the crystalline structures of most materials used in the production of the device, and these eventually work their way out and into the vacuum. Once loose in the vacuum, the high electric field ionises them, and the positively charged ions do what physics dictates and collide with the cathode, lodging in its surface and binding with the cathode material. The result is that the cathode can become coated with insulating compounds, such as oxides, and its ability to emit electrons is degraded. The phosphors may also be degraded by loose ions in the vacuum. While classical vacuum tube fabrication methods, such as the use of getters (basically a reactive metal is released after the tube is pumped out and sealed, to trap loose ions), have brought the problem down to a manageable level, it is still considered to be a major limitation to the usable life of the FED. Devices have been fabricated with 10,000 hr operating lives.

Other issues with FED revolve about the phosphors to be used, which must be different from those used in CRTs. Whereas the electron beam in a CRT is highly energetic (ie electrons accelerated over a decent distance), in an FED the distance is short and thus the electrons have only modest energy. To achieve similar brightness to a CRT, either a more sensitive phosphor must be used, or many more electrons pumped into the phosphor. The problem with hitting the phosphor with more electrons is that this tends to be at the expense of operating life. These are engineering issues which are still being worked on at the time of writing.

Fabrication of FEDs is also non-trivial, since these devices combine classical vacuum tube fabrication techniques with semiconductor fabrication techniques.

The addressing architecture for the FED is, as we would expect, based upon a row and column addressing scheme. The two most popular schemes are switched or unswitched anode addressing.

In the former, the phosphors for red, green and blue are laid down as stripes, and overlayed with three sets of anode metallisation, one for each colour. The three subpixels forming each pixel share a single cathode emitter, and gate. Each addressing frame (in time) is divided into a red, green and blue "field". For each colour field/row, the anodes are all connected to a common high voltage, upon which the gates across the entire column of pixels are driven to their respective control voltages. Like this all of the pixels in a row are excited concurrently (essentially the same scheme commonly used in TFT LCDs), and the drive logic cycles its way through the frame, row by row. The drawback of this approach is that high voltages must be switched fast, and dielectric breakdown between anodes can be a problem, limiting the technique to anode voltages of hundreds of Volts.

The alternative scheme is common anode addressing, where each subpixel requires its own cathode emitter and controlling gate shared across a column. In this arrangement, all anodes in a row are switched on at once, and the time is then divided into three slots during which the red, green and blue subpixel gates of each column are driven with their unique control voltages. This approach is frequently used with high voltage FEDs, but triples the amount of gate drive circuitry.

As with passive LCD addressing schemes, the FED is speed limited by the electrical capacitance of the gate electrodes, typically metal over silicon dioxide.

The current state of the art in FEDs are 14 cm diagonal panels, with prototypes as large as 27 cm demonstrated. Resolutions are still modest, with VGA level performance, or lesser, the norm at this stage. Anode drive voltages vary between 300V and 15 kV. Colour palletes are typically in the 16-bit range, although Futaba claim a 24-bit display. Nearly all current designs employ the Spindt process conical cathode technique. Luminescence performance ranges between 50 and 1000 cd/m^2 for RGB displays. Typical contrast ratios are 100:1.

Unlike the TFT LEP, which is very new technology, the FED has been in the development phase for some time now. Motorola, Futaba, Micron and Pixtech are offering production displays, and Canon, Candescent, FED Corp, Fujitsu, Raytheon and Samsung all claim to have working prototypes (readers interested in more detail are referred to the April 98 issue of IEEE Spectrum).

While the FED is still well behind the CRT in the key areas of resolution performance, cost and durability, it would appear that the major technical issues have been overcome and the device is ready to enter volume production.

Conclusions

Clearly the days of the CRT display are numbered. With the three competing technologies of the TFT LCD, TFT LEP and FED, the venerable three gun tube is facing formidable opposition with technological growth potential which the CRT has no more.

Because considerable market pressure for consumer HDTV electronics will develop over the next decade, the computer industry is likely to significant long term benefits in available technology, especially for large desktop screens.

In the shorter term, we will see these maturing display technologies first in the palmtop and laptop marketplace, where modest volume high cost pilot production can be amortised.

What is certain at this time is that flat panel displays have a very bright future.