A brief history of OLEDs

The effects of the electronics and photonics revolutions, enabled by the silicon-based transistor (and its incorporation into integrated circuits), fibre optics, and solid state lasers, are evident in almost every aspect of modern commerce. Yet, far from saturating the market, these devices are predicted to proliferate far beyond anything we have imagined so far.

Whereas roughly 50% of households in the developed countries own one personal computer, industry leaders today predict that in the next decade or so we will all own dozens or even hundreds of computers (most of them embedded in information appliances), all of which will communicate with each other on a network similar to the Internet. We will not know where these computers are, nor will we care, as long as they carry out their functions.

In this new world of 'pervasive computing' (a term coined by Joel Birnbaum of Hewlett-Packard Labs), in which most computing is carried out by distributed resources connected by a utility-like network, the user's awareness of a 'computer' lies only in what he or she sees at the interface: the display and input devices.

Displays, now considered a 'peripheral,' will be the central object from the user's perspective, while the processor becomes peripheral. This vision, however, requires displays that are far different from the current cathode ray tubes and expensive (and slow) liquid crystals, since they must be numerous, compact, and portable. Today display technology is primitive compared to computing technology. Indeed, paper is the preferred medium, resulting in the opposite of the early vision of the 'paperless office.'

While the current display market is many billions of dollars by any estimate, it is this emerging new paradigm that really drives the effort to produce displays that are vastly smaller, lighter, cheaper and more environmentally benign than today's. Organic electroluminescence (EL), a phenomenon first observed and extensively studied in the 1960's, forms the basis for the most likely candidate to serve this role. Because crystalline order is not required, organic materials, both molecular and polymeric, can be deposited far more cheaply than the inorganic semiconductors of conventional LED's.

Patterning is also easier, and may even be accomplished by techniques borrowed from the printing industry. Displays can be prepared on flexible, transparent substrates such as plastic. These characteristics form the basis for a display technology that can eventually replace even paper, providing the same resolution and reading comfort in a long-lived, fully reusable (and eventually recyclable) digital medium.

Until recently, the only significant commercial application of organic materials as active elements in electronics or photonics was xerography. Essentially all photocopiers today use organic photoconductors. Pioneer Electronics was the first company to alter this landscape, by introducing in 1998 a 64 x 256 pixel organic EL monochrome display for automobiles. Several other companies, notably Philips, Seiko-Epson, and an alliance of Kodak and Sanyo, are close behind.

All of these provide displays on glass substrates, using fabrication techniques very similar to those common in the liquid crystal (LCD) industry. However, there are close to a hundred research groups around the world engaged in OLED research and development. Universities, large corporations, government labs, and start-ups are all represented; in several cases new companies have been formed on the basis of university research, and one or more faculty members retain close ties (as for example chief technology officer) to the start-up.

Organic EL is obtained simply by placing a charge-transporting and light-emitting organic material between two electrodes (one of which is transparent) and applying a suitable bias. The organic material may be either a polymer, deposited by various solution processing techniques, or low molecular weight molecules (commonly called 'small molecules'), deposited by evaporation or sublimation in vacuum. Total device thickness (excluding the substrate) is less than 1 micron. When biased, charge is injected into the highest occupied molecular orbital (HOMO) at the anode (positive), and the lowest unoccupied molecular orbital (LUMO) at the cathode (negative), and these injected charges (referred to as 'holes' and 'electrons,' respectively) migrate in the applied field until two charges of opposite polarity encounter each other, at which point they annihilate and produce a radiative state.

The transformation of this academic endeavour to a practical display technology traces its origins to C.W. Tang and S.VanSlyke at Kodak, who introduced the double layer concept in 1987; this was the first publication that combined modern thin film deposition techniques with suitable materials and structure to give a moderately low bias voltage, attractive efficiency, and encouraging lifetimes. Their devices were made by subliming molecules of a triarylamine as hole transporter, followed by aluminato-tris-8-hydroxyquinolate (Alq3) for electron injection and emission, and a magnesium-silver alloy cathode.

In 1990 J. Burroughes, et al., in the research group of Richard Friend at Cambridge University reported a similar single-layer device made with a polymeric medium: poly(phenylene vinylene) (PPV). The physics of charge transport and light emission in these devices is very similar; however significant practical differences arise when one begins to contemplate manufacturing issues.

As solid state devices, organic EL devices are often discussed using terminology drawn from traditional semiconductor physics, in which charge carriers (electrons or holes) are viewed as moving in conduction and valence bands, even though it is generally recognised that even highly conjugated polymers do not behave as fully delocalized one-dimensional solids due to the inevitable conformational defects described by the persistence length. An alternative point of view regards the relevant structural entities as radical cations and anions, and charge motion is through electron exchange reactions. Several authors have provided convincing evidence that the extent of delocalization in a PPV molecule is no more than a few repeat units, in agreement with inferences based on absorption spectra and electroabsorption. We believe the molecular viewpoint to be most appropriate, while recognizing that delocalization plays an important role.

Several reviews have been published to which the reader should refer for more details. The best general discussion of physical aspects is still the monograph of Greenham and Friend. The review in Science by Sheats and co-workers, though slightly dated, is still quite useful as an introduction for a general scientific audience; two other more detailed (but more recent) reviews, concentrating on chemical aspects, are listed at the end of the bibliography. The relevant synthetic chemistry has been extensively described by Holmes et al. In the following discussion we will very briefly list the most salient technical issues and comment on the current state of the art in performance characteristics.

Charge injection. Theoretical modelling of charge injection has been attempted by several approaches. Excellent quantitative results have been obtained by P. Davids and D. Smith, who assume injection to be a combination of thermionic emission and quantum mechanical tunnelling, but with the important feature of a thermionic backflow: if the mobility of carriers in the organic layer is small, the local concentration quickly increases and the probability of reverse hops increases. In this model, which has given good agreement with experiment for single layer devices both with unipolar and bipolar structure, the barrier height derived from modelling is not, in general, what would be calculated from tabulated work functions for the metal electrodes, suggesting some chemical reaction.

Recently J. Bharathan and Y. Yang showed that just 2 Ã…ngstroms of calcium for the cathode followed by aluminium gives identical electrical characteristics to 2000 Ã…ngstroms of calcium. Their data are difficult to reconcile with any mechanism other than 'doping' of the interface by reaction of calcium with the polymer, creating surface states that pin the surface energy levels in a manner somewhat analogous to the pinning of the Fermi level by surface states of an inorganic semiconductor. Additional experiments by J. Sheats and co-workers using zirconium carbide cathodes were best explained by postulating that electrons, injected from a pristine metal into an undoped polymer, quickly become trapped, and these traps are effective luminescence quenching centres. When some doping is present, the charge carriers remain mobile and device performance is correspondingly enhanced. These reactive surface states must be protected from oxygen and water, and complicate the electrode processing compared to liquid crystal display processes.

Charge transport. The carriers move by electron transfer reactions, or hopping, between molecules or polymer segments; even along a polymer chain, conformational disorder interrupts the idealised one-dimensional transport. Mobilities for electrons, which are believed to be more easily trapped (especially by sites containing oxygen), are generally lower than for holes; both are low compared to mobilities in inorganic semiconductors.

Carrier injection balance. Single-layer devices tend to produce unbalanced charge injection and consequently lower efficiency. To prevent this, a charge transport or injection layer (for either holes or electrons, or both) is used. A proper choice of orbital offsets between the layers can block the transmission of the majority carriers, creating a space charge that reduces its current and balances the charge transport. Transport layers can also lower the injection barriers at the electrodes. Creating multiple layer structures is not without difficulties, however, since interlayer mixing is in general undesirable.

Luminescence efficiency. The efficiency of an OLED is a function of several factors: the degree of balance of charge carrier injection, the efficiency of recombination, and the efficiency with which the excited state emits light (the photoluminescence, or PL, efficiency). Recombination efficiency is generally at or near unity; i.e., if two charges come into van der Waals proximity of each other, they are guaranteed to annihilate. However, if the annihilation process has no energetic barrier, then quantum spin statistics dictate that only 25% of the resulting excited states will be useful emissive ones if the state has singlet multiplicity. Although emission from the triplet state is spin forbidden in ordinary organic molecules, progress has been made recently in developing triplet state emitters containing at least one atom of higher atomic weight (which facilitates the triplet-singlet interchange called intersystem crossing).

The excited state may still decay nonradiatively even if it is spin allowed. Much work has been done on understanding and controlling these processes in conjugated polymers, and today most EL materials have quantum yields for PL of greater than 80%. In evaporated molecule systems, the emphasis has been on finding additives that can accept energy from the primary excitation sites (which must provide the charge transport) and radiate with high efficiency; quantum yields similar to the polymers are obtained.

Light extraction efficiency. In accordance with Snell's law, when light passes from a medium of high refractive index to one of lower index, there exists a critical angle of incidence (the angle between the light propagation direction and the perpendicular to the surface) beyond which the light is totally internally reflected; this is the basis for the nearly lossless transport of light over long distances in optical fibres. For the glass-air interface the angle is approximately 43 degrees, and polymers are similar. Thus, a great deal of the light generated by an OLED is trapped inside the glass or plastic substrate and waveguided out to the sides; this is normally about 80% of the total. Current research activity in inorganic LEDs (where the problem is even worse because the index of the inorganic materials is higher still) is directed toward various surface treatments that can greatly increase extraction efficiency. It is reasonable to expect that these losses may be cut at least in half by innovation in this area.

Power efficiency. The fundamental quantity important to the user is power efficiency, since this determines how long the battery will last in a portable device. Power efficiency is measured in lumens per watt. (The lumen is a measure of optical power, scaled to the sensitivity of the human eye; thus there are more lumens in an optical watt at the sensitivity peak of 555 nm than at a wavelength in either the red or blue region.) Incandescent light bulbs used in the home range from about 10 - 15 lumens/watt; fluorescent bulbs can be as much as 80. Currently the best OLEDs are achieving in excess of 20 lumens/watt (for green), although many devices are well below this (often only a few lumens/watt). Progress in improving this number has been rapid during the last few years and will doubtlessly continue; in fact, while it is a long range goal, the use of OLEDs for general illumination may prove feasible.

Power efficiency is affected not only by the quantum efficiency of the device (number of photons out per electron in), but also by the voltage at which the device operates, since according to Ohm's law power is the product of current and voltage. Thus it is important to obtain low operating voltages, which is related to the charge injection barrier issue mentioned earlier. Low voltage is also desirable to avoid expensive voltage upconverters to provide compatibility with batteries. Voltages for the systems likely to see early application are around 3-6 volts.

Colour issues. Different colours are obtained in OLEDs from different materials; hence for a full colour (RGB) display, some patterning of the organic material must occur (in order to get all colours one must have red, green, and blue subpixels at each pixel of the display). This has important ramifications for manufacturing, and while several solutions have been proposed and demonstrated at the laboratory level, it is not yet clear which will be preferred for production.

Organic molecules typically emit light in a band of wavelengths that is rather broad (of the order of 50-70 nm at an intensity level equal to half of the maximum). When three such bands (which may individually appear red, green, and blue) are mixed to obtain an arbitrary colour, the results may not be as satisfactory as when three narrow-band ('saturated') colours are mixed. Various ways to address this include using well-known thin film interference effects to sharpen the emission bands.

Lifetime and environmental sensitivity. The operation of OLED devices generally leads to a more or less steady loss of efficiency. Two types of loss are often distinguished: the quantum efficiency decreases, and the voltage required to maintain constant current increases. Lifetimes depend on the brightness at which the display is operated, and are typically quoted as the time to reach half of an initial brightness of 100 candelas/m2 (which is that of a typical desktop CRT). In the best materials, offered by several companies at the present (in both polymeric and small molecule forms), the lifetime now exceeds 10,000 hours and in a few cases is greater than 50,000 hours. This parameter has also seen substantial and steady improvement in recent years, and can be expected to improve still further.

Every device has temperature and other environmental limits. OLEDs must be hermetically sealed from exposure to water and oxygen; while it has been a challenge to find a suitable flexible, transparent material with the required properties, this has been accomplished. Low temperature does not adversely affect OLEDs as it does LCDs (indeed they become more efficient), but devices must typically be kept below 60 degrees C for operation (their lifetime is substantially shortened when operated at such higher temperatures) and 80 degrees for storage.

Electronics (drivers). A display consists both of a medium for displaying information and a means of impressing this information on the medium. Such means are invariably electronic, and these circuits, called 'drivers,' are an important part of the cost and performance of the product (typically more than half the cost).

Flat panel displays are addressed by rows and columns (hence the term 'matrix'), and are classified as passive matrix (PM) or active matrix (AM). A PM device has driving transistors only at the ends of the rows and columns. A 480-row display (traditional VGA resolution) is driven by applying bias voltage to the first row and holding it while then applying bias to each of the 640 columns simultaneously; then moving on to the next row, and so on.

With no bias applied to the earlier rows, they are no longer emitting light, so one has to quickly go back and repeat the whole process before the human eye notices any flicker (typically about 70 times per second). When the bias is on a pixel, it must be lit up at an intensity high enough that the desired average is obtained; thus for 480 rows, the actual emitting intensity for each pixel is 480 times higher than the continuous intensity that is desired. This high peak current requirement makes the drivers more sophisticated and expensive, among other problems, but there are only a few of them.

In AM displays, there is at least one driver transistor at each pixel (two for OLEDs). The same strobing as just described is carried out, but what it does is to cause a transistor to turn on and conduct, storing a charge on a capacitor. This capacitor in turn is connected to the gate of a second transistor, and causes the latter to conduct charge to the OLED pixel, a process that continues until another signal is applied to discharge the capacitor. Thus the OLED emits continuously. The disadvantage is that while 1120 driving transistors were needed for a PM VGA display, 614,400 additional ones are needed for AM, and they must be placed at each pixel, rather than in groups at the edge. (For colour, separate drivers are needed for each colour subpixel).

Nevertheless, the manufacturing technology to do this has been worked out for the LCD industry, and all high-quality displays are made this way. Since normal integrated circuits are built starting with single crystal silicon wafers, a new technology called 'thin film transistors' or TFTs, has been worked out in which the required materials can be deposited on glass substrates. If the displays are to be flexible, the films must be deposited on plastic or a metal foil.

Manufacturing issues. As mentioned previously, OLED displays are made in much the same way as one makes LCDs. Starting with a glass plate, the transparent conductor (typically indium tin oxide, or ITO) is deposited and any patterning carried out with industry-standard photolithography techniques, followed by the organic layers and the cathode (which may also need to be patterned). The equipment required for these steps can be obtained directly from LCD industry sources. For AM displays, the above structure is built on top of the TFTs, which are essentially the same devices as used for LCDs except that, as mentioned, there must be two at each pixel instead of one. This is the basis of the Seiko-Epson display mentioned in the introduction (made in collaboration with Cambridge Display Technology (CDT)).

One major difficulty that must be addressed is the patterning of the organic materials for full colour. The state of the art as of 1997 was described by J. Sheats in a short Perspective in Science. Other approaches are under development, including various techniques borrowed from the printing industry, and more than one company has now demonstrated quite attractive full colour OLED displays.

The equipment required to fabricate modern LCDs is quite expensive because of the size of the substrate. Vacuum processing equipment capable of handling pieces of glass 1 mm thick and 1 m on a side at high throughput rates) is a technological tour de force, and the capital for a typical plant may come to half a billion dollars. If flexible substrates could be used, processed in a 'roll to roll' fashion (unwinding the web of material from one roll, passing it through the process stations, and winding it up onto another roll), the handling equipment becomes cheaper and the process much faster.

Progress in this area has been held back by several factors. Most important is that web, or roll to roll, processing is inherently high volume; it is not easy to do in a laboratory with a small amount of material. Another issue is the substrate: flexible materials that have sufficiently low permeability to water and oxygen, while satisfying other requirements for OLED application, were not known until recently.

At the present time, however, the great commercial potential of web processing is often mentioned at display-related technical conferences, and there are no fundamental barriers to its development.

Source: http://www.rolltronics.com/intro_oled.htm