Manufacturing emissive displays
01/01/1999
Manufacturing emissive displays
Ed Korczynski, Senior Technical Editor
A look ahead
Each issue of Solid State Technology in 1999 will feature an article looking at one aspect of semiconductor/thin film processing for the year 2000 and beyond. Rather than simply focusing on obvious challenges, each article will discuss some likely solutions to critical problems in each technology area.
The series starts below with a look at the manufacturing of the many new emissive flat panel displays that will challenge LCDs for both new and established applications. These displays will be used in factories, automobiles, and shopping areas to communicate information, as well as in personal receiving devices and other consumer applications.
In February, remarkable efforts to stretch the use of optical lithography to smaller and smaller circuit dimensions will be explored by Senior Technical Editor Pieter Burggraaf. This will be followed by forward-looking articles on wafer cleaning (March), packaging and assembly (April), deposition (May), ion implantation (June), chemical-mechanical planarization (July), materials (August), etching (September), gas flow and handling (October), metrology/test (November), and automation
obotics (December).
New emissive flat-panel displays are currently reaching volume production, with the potential to take significant market share from LCDs. Though emissive displays employ submicron scale structures composed of exotic new materials, one of the biggest manufacturing problems is the millimeter-scale sealing of the finished display. The emissive display industry can be likened to the IC industry of the 1970s, with much new equipment having to be designed and built by the end-users.
Emissive flat panel displays (FPDs) are based on engineered materials, such as phosphor pixels, which emit light directly after electric or photonic switching. Standard active and passive matrix liquid crystal displays (LCDs), in contrast, are essentially light-valves that modulate the transmitted luminance of a white backlight through color filters.
Though LCDs for notebook computers still represent the vast majority of current sales, nascent production of emissive designs will alter the display marketplace of the next millenium. A proliferation of emissive options could significantly change the way we view information (Table 1), allowing for the development of new types of portable and fixed displays.
Existing emissive designs serve high-profit-margin custom applications in military and industrial markets, and low-profit/low-resolution segmented applications. Volumes in these sectors will always remain low, representing a small percentage of overall global FPD revenues. Several companies are now trying to develop new designs and manufacturing processes for high-resolution commercial-scale emissive displays, however. Most potential markets have more than one competing display technology that meet performance criteria (i.e., brightness, contrast, power-consumption, size, and lifetime), so manufacturing cost will be the primary gate for success.
 |
Different FPD designs
LCDs will continue to show strength for all current applications, but newer emissive displays will challenge LCDs in most new and in some entrenched positions. Microdisplays (most of which are liquid crystal on silicon [LCOS] designs) combine traditional liquid crystal technology with silicon wafer CMOS manufacturing. Processing challenges are generally easier than those found in standard direct-view LCD production. Most FPD production - that for LCD, plasma display panel (PDP), and field emission display (FED) panels - can be broken down into manufacturing the "backplane," the "frontplane," and final assembly. The front- and backplane are commonly manufactured on two lines and are only brought together at final assembly.
FED. Field emission displays (FEDs) are bright, rugged, and have exceptional viewing angles. FEDs rely upon the
stimulated emission of photons from a phosphor-coated front plate. FEDs are conceptually similar to cathode ray tubes (CRTs), with hundreds to thousands of micron-scale electron cold emitters oriented behind each pixel instead of the single rastered electron-beam of a CRT (Fig. 1). Emitters are switched by the signals that come from row and column drivers that define their cathodes and gates; the potential between the gates and the anode frontplate accelerates electrons toward the frontplate phosphors (usually through focusing apparati).
FEDs have finally hit the mainstream, with Motorola unveiling 2.9- and 5.6-in. low-voltage models (intended for palm-top, instrument, and automotive applications), and Candescent Technologies breaking ground on a 32,000-m2 fab to produce high-voltage FEDs in San Jose, CA. Power consumption is the main limitation for low-voltage phosphor designs, with 15% of the pixels in a 5.6-in. low-voltage FED draining 2-3 W. High-voltage designs consume only about half of this power and currently have longer lifetimes, though they require larger internal volumes that are more difficult to assemble.
EL. Electroluminescent (EL) displays are also rugged and have wide viewing angles, and can operate over extremely wide temperatures. They are thus under consideration for automotive and industrial applications. There is a wide variety of EL technologies under development: monochrome and color, thin-film active and passive matrix elements, large workstation monitors, and head-mounted displays (HMDs).
Unlike LCD and FED designs, EL displays inherently consume too much power to meet the battery conservation requirements of portable applications. EL, then, is limited to applications that can plug into a big battery or a wall outlet (such as automotive or instrumentation).
 |
Figure 1. Schematics showing the conceptual similarities between CRTs and FEDs: both accelerate emitted electrons to strike pixel phosphors.
PDP. Plasma display panels (PDPs) can be likened to an array of microminiature integrated fluorescent light tubes. The ignited cold plasma (xenon, usually diluted with helium or neon) inside each tube excites phosphors that emit different colors. Compared to other FPD technologies, PDP designs are easier to scale to large diagonal dimensions, so they are leading the race to produce 35- to 55-in. diagonal wall-mounted monitors.
PDP production costs remain very high, due in part to unoptimized direct manufacturing costs, but mainly to low yield. Current PDP retail prices are in the US$5000-10,000 range, with uncertainty over how to bring about enough manufacturing efficiency to drop prices by the factor of 10 needed to stimulate wide demand.
OLED. Organic light emitting diode (OLED) displays are much simpler to build than LCDs, PDPs, or FEDs. Highly engineered organic molecules that exhibit electroluminescence can be patterned into arrays of simple diodes. Multilayer or phase-separated single-layer OLED materials can be formed that produce 300-30,000 cd/m2 light output (well above minimum luminance levels) and have lifetimes above 10,000 hours.
Monochrome, low-resolution OLED displays have been in production for many years, but high-resolution (often full-color) designs are a relatively new concept. Requiring extremely complex new materials, these newer OLEDs have great potential, but it is too soon to predict eventual performance and cost.
OLED displays can be fabricated on flexible substrates, allowing for the development of roll-up displays, high-tech greeting cards, and wearable displays. The greatest early high-volume use of OLEDs will probably be in high-resolution monochrome applications that are extremely cost sensitive.
Commonalities
In manufacturing, challenges are found in establishing high-volume production that were not evident during prototype manufacturing development. The inevitable yield losses that result tend to keep product prices high, limiting market penetration. PDPs still cost US$5000-10,000 because of yield losses, and FEDs are just ramping into volume after years of manufacturing development.
All backplate processing - whether for field emission cathode tips or for TFTs - is fairly common in terms of the types of steps and therefore the types of tools needed to form structures. In general, manufacturers can talk to established equipment companies and purchase the necessary tools with modifications.
Processing commonality arises because all microelectronics manufacturing involves the creation of a set of very small features of some form or another. It always involves cleaning, lithography, and etching tools to create features. "That`s why the commonality is almost a given when you`re in microelectronics," stated US Display Consortium (USDC) Chief Technical Officer Bob Pinelle. "It`s just a variation on a theme, then, of how you`re going to make those features using thin-film or thick-film technology."
Thus, even though the backplates contain the most complex structures, the processing technology needed to form these structures can be leveraged from existing high-yielding LCD production. The greatest manufacturing challenges for new emissive displays will be in the frontplate and the assembly operation. OLEDs, though emissive displays, fall into a slightly different category because of their dramatically simpler designs and process flows.
Novel enabling processes
FED. Most FED companies use lithography to define the size and location of micron-scale cold cathodes. Candescent Technologies has licensed a remarkable process from Lawrence Livermore National Labs to create random arrays of 0.15-?m-diameter emitters (Fig. 2). Lithography is used only to define the array areas, not to define each emitter. Ion implants through a dielectric layer leave damage trails that can be controllably etched to create thousands of uniform emitter sites beneath each pixel (using less power in operation due to their smaller size).
FEDs are much more difficult to package than LCDs, because instead of sealing the glass and backfilling with liquid crystal material, one must create and maintain a high-quality vacuum. Plates must be placed within microns of each other, that gap must be controlled across comparatively huge dimensions, and a 10-6 to 10-9 torr vacuum must be formed with no moisture present.
For any emissive technology, sealed display components will outgas under the required level of vacuum. Thus getter structures must be located throughout the inside of the display to capture any outgassed contaminants that would otherwise degrade either emitter tips or phosphor dots. Getters must be of sufficient quantity and widely distributed such that any contaminants are readily attracted to a local getter structure.
The phosphor elements themselves can be a source of outgassing contaminants during operation. Either due to impurities in the source material or to molecular breakdown, aging phosphors can outgas and seriously degrade emitter tips. One solution is to encapsulate the phosphors with a transparent film to isolate them physically from the internal vacuum.
Cleaning, aligning, evacuating, and sealing in vacuum are all essential to create a long-lived emissive display. Companies are taking different approaches to sealing in a vacuum. Though it is conceptually simple to perform the entire operation inside larger vacuum chambers, it is practically very difficult to accomplish. Thus, one approach is to perform some of the steps at atmospheric pressure inside a cleanroom, and then pump-down, pinch-off, and seal (similarly to CRT production). It is a tradeoff between the manufacturing cost (incorporating yield losses) and the final display reliability.
PDP. PDP processing relies upon both thin- and thick-film deposition tools, and the ability to work with extremely large substrates. Thin-film tools are modifications of commercial LCD equipment, while thick-film tools are derived from printed circuit board production. Screen printing equipment deposits phosphor paste through a mask. Lithography tools can form 50-?m lines with >1 panel/min (Fig. 3).
Phosphor efficiency for PDPs is very low at ~1 lumen/W, compared to common fluorescent lightbulbs that use similar technology to achieve 80 lumens/W. Fluorescent lights, though, use a mercury plasma with phosphors optimized for the spectral output of mercury, while today`s PDPs use the same phosphors with a xenon plasma that emits an entirely different spectrum (a very inefficient pairing). Improved xenon-based phosphor materials and structures, as well as optimized gas mixtures and pixel geometries, can be expected to increase efficiency. Display brightness can increase and power consumption decrease with increased luminous efficiency.
The benefit of higher resolution will be needed to offset the higher ASPs of PDPs; HDTV resolution will be expected across 42- to 55-in. diagonal displays. Driver circuitry (currently running at ~180 V) will be correspondingly dense, with cost and speed constraints. It is estimated that 75% of the direct cost in manufacturing high-resolution PDPs will be in the flip-chip or chip-on-glass driver chips.
Assembly (involving sealing, evacuation, gas filling, and drying) can take 4-6 hours, due to slow gas conductance at the low temperatures required post-seal. The plasma volume inside a PDP is susceptible to outgassing contamination, though less so than in FEDs. The phosphors are typically protected from physical contact with the plasma by a thin dielectric layer such as MgO.
EL. The main limitation to EL technology`s taking market share from LCDs is, once again, manufacturing cost. Though EL displays are comparatively simple to manufacture, using screen-printing and similar printed circuit board production processes, current processing occurs on relatively inefficient first-generation lines. Also, low production volumes preclude economies of scale in direct material costs.
Dielectric insulators are needed in EL designs. One promising equipment technology development is the award of a US$1.1 million matching funds contract from USDC to Intevac (Santa Clara, CA) to develop reactive sputtering tools for the high-volume deposition of dielectrics. Intevac will supply matching funding to the project, which is also intended to develop rapid thermal annealing for EL phosphors. Total USDC funding of equipment and materials projects has reached $52.5 million, with another $70 million matched by industry (resulting in 20 products that are either commercialized or in beta-testing).
 |
Figure 2. SEM showing Candescent Technologies` 0.15-?m FED tips. The Candescent fabrication process uses ion implant damage trails through a dielectric to form tips without requiring lithography.
 |
Figure 3. ORC Technologies new ProForm 9000 panel printer is designed for Gen.IV processing of PDPs. The system uses scanned 365-nm light and a 5-7 mil. proximity gap to achieve 50-?m resolution.
The two biggest challenges to EL`s winning rugged high-brightness applications from LCDs is cost and power consumption. Both of these critical parameters are gated by the thick-film phosphor processing; higher deposition rate processes are needed to produce more robust phosphor materials. Organic EL materials have recently been shown to offer dramatically higher luminescent efficiency than older inorganic phosphors (particularly in the blue), though deposition and anneal costs must be reduced. Durel Corp. (a joint venture between 3M and Roger Corp.) recently began selling micro-encapsulated phosphors that can be simply screen-printed without the need for an additional encapsulation step.
EL devices require high voltages to drive current through the high capacitive loads of the phosphors. For large screen displays, advances in low-resistance transparent electrode technology is needed. High-resolution EL displays will require high-speed and high-current drivers to satisfy the fast row-scan times with high capacitive loads.
 |
OLED. A staggering variety of processes have been used to create OLED devices: screen-printing, spin-coating, PVD, CVD, and layered molecular self-assembly. The first products will probably be passive, monochrome, and low-resolution (1/4 VGA) displays that try to compete with low-cost passive matrix LCDs. Very large passive matrix designs are reliability-limited by the Joule heating of long connection lines, so it may be necessary to develop active matrix polymer TFTs.
The biggest integration challenge is likely to be the encapsulation layer needed to isolate the main OLED material from the atmosphere. Encapsulation should be particularly difficult for the hot-running passive OLED displays that are first planned; active designs would run cooler and be easier to encapsulate, if they eventually reach production.
Though only recently conceived and requiring extremely complex organic molecules, OLED displays may quickly reach volume production due to inherent simplicity in fabrication. For LCDs, FEDs, and PDPs, complex structures are built that require many interdependent components. Each component may have several manufacturing challenges, so that the complete process becomes very difficult to control to acceptable yield. OLED displays are so simple in comparison that once materials are engineered, the process integration work for manufacturing should go very quickly. Also, it is likely that many different OLED materials will be successfully developed, so a proliferation of displays should not be dampened by intellectual property battles over a single workable solution.
OLED display production can readily occur on thin, flexible substrates (such as metal foils, metal-coated plastic foils, and flexibly thin glass), using continuous rather than batch processing. High-volume manufacturing may look much more like the continuous web-based production of photographic film, for example, instead of the batch production methods employed by most FPD lines.
Manufacturing standards
"You can draw an analogy to where the silicon IC industry was 30-35 years ago, when they couldn`t agree on how big the wafer should be, or how to carry it," declared USDC president and CEO Mike Ciesinski. "Moving beyond this is an early indicator of the business maturing. So we`ve now got FPD groups arguing about the digital standard for interfacing laptop displays, how big the substrate is going to be, how to carry it, and so on." Standards activities are always driven by industry economics, especially when the movement to a next-generation process technology becomes prohibitively expensive for a single company to develop.
Different panel manufacturers have developed custom tooling to create assembly/sealing processes. Each manufacturer`s process is unique, and the lack of commonality prevents the short-term evolution of a supplying equipment industry.
For example, since high-volume FED sealing is only now being attempted, it is too soon to predict which approach offers the best combination of yield, direct cost, and display reliability. If no single approach is proven to be the obvious winner, then the FED industry will probably be forced into rapid consortium development. The USDC`s Pinelle stated, "I don`t personally feel that this has happened yet, but I think that it will be a necessary development for FEDs to become highly competitive and capture a meaningful part of the LCD market."
Manufacturing costs for most FPDs would be reduced if production could move to larger substrates. The drivers are nearly identical to those moving the semiconductor industry to 300-mm wafer production, with the same equipment cost-of-ownership issues determining the speed of the transition. Current performance projections from FPD equipment suppliers have effectively stalled the introduction of Gen. IV processing (Table 2), except for PDPs.
Conclusion
The vast majority of the equipment and processes needed to manufacture LCDs - 60-80%, depending upon the design - can be applied to new emissive technologies (i.e. FED, PDP, EL). Much of the remaining processing technology can be borrowed from semiconductor, CRT, and magnetic storage head production. Established equipment manufacturers in these industries can thus expect to win orders from emissive display companies.
LCD manufacturers want to take desktop monitor market share from CRTs, and FED and PDP manufacturers want to take market share from both LCDs and CRTs. Given that most technologies can produce displays that meet performance specifications, the overwhelming concern for electronic system integrators will be cost.
The greatest barrier to lowering production costs is the lack of off-the-shelf toolsets. Standard tools cannot be developed when manufacturing standards do not exist, however. Fabrication requires novel frontplate and assembly processes, many of which are in transition from development to pilot-line production. With no obvious show-stoppers, however, the future for emissive displays looks bright indeed.
ED KORCZYNSKI is Senior Technical Editor for Solid State Technology. He received his BS degree in Materials Science and Engineering from the Massachusetts Institute of Technology. He has more than 10 years of engineering and management experience in process development and equipment marketing. His current interests are thin films, process integration, and plasma and vacuum technology. He is a member of theMaterials Research Society. Solid State Technology, 1700 S. Winchester Blvd., Suite 210, Campbell, CA ph 408/370-4833, e-mail [email protected].