intended to be used as a server on a local area network, to permit fast and easy access from office workstations (such as P E R Q s or Apollos, say) to heterogeneous information stored on optical discs. A paper by I. Page (UK) entitled 'Disarray: a 16 x 16 raster-op processor' described a novel display processor architecture to manipulate bit-maps faster than a conventional (SISD) yon Neumann machine. Two-dimensional bit-maps are the lowest-level data objects and the machine contains 256 one-bit processors organized as a 16 × 16 array to manipulate them. There is now working prototype hardware and preliminary performance results are very encouraging. M a y we hope that some enterprising manufacturer will realize these ideas in a product? The paper concluded with some ideas to improve the architecture stillfurther and to integrate the array into a powerful single-userworkstation. Proceedings of the conference, edited by S. Turk, will be published by North-Holland. Pre-conference seminars The pre-conference seminars, now a
key feature of Eurographics conferences, were again well supported. Topics included 'Introduction to computer graphics', 'Interactive techniques', 'Graphics systems and standards' and 'Geometric modelling'. Judging by comments heard, all seem to have been very satisfactory. Exhibition Because of difficultiesin moving equipment into Yugoslavia, the exhibition this year was very small. However ICL did have a P E R Q on display and this aroused great interest. Summary Considering the cost of travel to Yugoslavia, the conference was very well attended. The technical content of the programme was good though not outstanding. However that does depend on one's viewpoint. For the 100 or so delegates from Eastern Europe, the impact was described as equivalent to that made by Siggraph on someone from the UK attending for the first time. A larger conference could undoubtedly have been held elsewhere in Europe, but this is not what Euro-
graphics is about. The Association aims to promote computer graphics in Europe, and this is what this conference succeeded in doing. It brought together groups from the host nation w h o had previously had littlecontact, and they saw what is happening elsewhere in the world. At the end of the day, one was reminded of the remark at the end of the firstinvited lecture, 'Science acts as a bridge of understanding among people and nations'. There are signs that Eurographics '83 acted as such a bridge for those attending. Such was the true worth of this conference. Eurographics '84 will be held in Copenhagen, Denmark, from 12-14 September. The familiar format of technical conference, pre-conference tutorials, exhibition and industrial seminars will be followed. Planning is at an advanced stage and this promises to be a large and lively conference and exhibition. Information is available from : DIS Congress Service, Linde
Alle 48, DK 2720 Vanlose, Copenhagen, Denmark. A. Sweeney, H. Watkins Rutherford Appleton Laboratory Didcot
International workshop on electroluminescence B a d Stuer, G D R , 1983
This workshop follows on from that held on the same subject in Lidge three years ago, whose proceedings were published in J Luminesc 23 (1981). This time there were 57 delegates, 38 from the G D R and 19 from other countries. The foreign visitorsincluded two each from the USA, Japan, France and the USSR, five from the U K and one from each of West Berlin, the Federal German Republic, Czechoslovakia, Finland, Hungary and Poland. The meeting was organized around 12 one-hour invited talks plus a tutorial introduction to the subject from the Conference Chairman, Dr G.O. Muller of the Zentralinstitut fiirElektronenphysik (ZIE) in East Berlin. Additionally, 21 contributions were presented
in the form of posters. The majority of these were introduced by their principal authors in five-minute periods plus ten minutes for discussion, grouped together in association with the most relevant invited talks. This formula proved quite successful, and a lot of discussion was stimulated in both the lecture and poster sessions. The technical programme contained scientific and technological aspects, both strongly present as is appropriate to the current status of this subject. Concerning the latter, this reviewer was encouraged to hear of strong continuing interest in mature stages of device development and even moderate scale production. The recent retrenchment of interest in high field, thin film electroluminescence (HFTFEL) amongst several large electronics companies in the USA and Europe is
already well known. However, one of these companies in the USA has fostered the emergence of a small facility dedicated to the topic, Planar Systems. The technology of the Sigmatron company, a leading pioneer in the alternating current-driven (a c) version of HFTFEL, now resides in an enthusiastically managed new American company, Sigmatron Nova, whose activities were described in some detail at the workshop. The currently most significant firm involved in a c HFTFEL devices, the Sharp Company of Japan, was represented by the key figure in their early research and development of the devices, Dr Toshio Inoguchi. Maintenance of the active layer brightness is no longer the severe problem traditionally associated with HFTFEL, with 104 - 1 0 s h achieved for the a c devices. Production yield has been a
DISPLAYS. J A N U A R Y 1984
considerable problem for complex matrix~addressed displays, but yields of about 50 per cent are currently claimed by both Sharp and Sigrnatron Nova. We heard that total device production of 3 000 units per month was established by Sharp in April 1983, in the form of a message display unit and a more complex unit which can also handle graphic information at high resolution and with a pleasing appearante.
A number of other major electronics companies in Japan are reported to have significant research and development on these EL devices, a major contender for large area active displays. In addition, there is a significant amount of university research into the subject, particularly on key topics like alternative colours, sought by the use of both host and activator alternatives to the n o w overwhelmingly predominant ZnS activated by M n z2+ n (3d s ). This work was reviewed by Dr Inoguchi and Professor H. Kobayashi from Torrori University in Japan, both partly in the context of information from the large conference Japan Display '83, held at Kobe, Japan in the previous week*. This conference covered all types of display, both active and passive, but the EL type received appropriately significant attention. Further healthy industrial interest in EL also continues in Europe, at Lohja in Finland, Phosphor Products and Lucas in England (not represented at the workshop), with strong interest at a number of other types of institution, particularly ZIE in G D R , the Royal Signals and Radar Establishment in the UK, also the Kiev Institute of Semiconductors (KIS) in the USSR, Oregon State University in the USA, Osaka University in Japan, the Universities of Hull and St Andrews in the UK, the University of Paris, France and the Technical University in West Berlin. All of these activitieswere presented. The key roles of Dr Inoguchi (Sharp, Japan) and Dr Natalie Vlasenko (KIS, U S S R ) in the development of a c H F T F E L devices were acknowledged by all.They met for the firsttime at Bad Stuer! The workshop fallsalmost at the tenth anniversary of the SID conference paper in 1974 in which Dr Inoguchi and Dr Mito announced remarkable progress in brightness and maintenance achieved at Sharp Laboratories for ZnS :Mn a c H F T F E L panels. The technical discussions were strongly focused on ZnS :Mn, with
some attention to ZnSe:Mn mainly for mechanism-evaluation reasons. The substantial progress in the understanding in this system achieved in recent years was highlighted by Dr Muller in his introductory review and involve concepts some of which were clarified further during the workshop. The a c H F T F E L structure contains at least five T F layers, the active ZnS :Mn being sandwiched between insulating films, coated in turn by conducting electrode structures, often one reflectingand one substantially optically transmitting. T w o transmitting electrodes m a y also be used, particularly with film deposition technologies which favour optically clear active layers. The production devices tend to be more complex for a variety of reasons, for example the desirability of an Am: 03 layer over the usual Y:O3 insulating layer and adjacent to the Al electrode. The current Sigmatron Nova process is said to involve ten deposited layers. Electrons are injected into the active layer through energy barriers, whose detailed properties are believed to depend to a significant degree on insulator-semiconductor interface electrical states about which very little is yet known. The energy of these hot electrons is maintained in the high electric field imposed near the peaks of the applied voltage. The resulting electron motion is described by some as ballistic,although lattice scattering processes remain very important over the active layer thickness of perhaps 0.5/Ira. In any case, the M n :+ activators, whose energy level positions relative to the ZnS band structure remain ill-defined,become excited by impact from these hot electrons, with subsequent 4T I ~ 6 A I internal relaxation. This relaxation ismainly radiative up to about I per cent concentration of Mnzn, and is responsible for the well k n o w n yellow-gold luminescence observable both in ZnS and ZnSe. Good agreement exists between laboratories on the optimum M n concentration, near 0.8 per cent. The strong decrease of EL efficiency above this value is consistent with a concentration quenching effect which also strongly reduces the decay time. The detailed form of the conduction band structure is very important for efficient excitation. The properties of the ZnS and ZnSe hosts seem to be nearly optimally matched to the M n 2+ energy states in this regard. Allen (St Andrews, UK) suggested that conditions in direct
*A report on thisconference willappear in the next issueof Displays.
DISPLAYS. JANUARY 1984
current Schottky barrier E L and a c H F T F E L m a y differ, avalanche multiplication being more likely in the latter case. Recent work at the ZIE has revealed the impact length for this excitation process and a more accurate value for the cross-section r of the M n 2+ ions, about 4 x 10 -16 c m 2 , roughly 2 x larger than the previously accepted value. This rather large crosssection is believed to be close to the m a x i m u m which can occur in this type of process. In fact, its magnitude was considered by Dr Hagston of Hull University to be i0-I00 × larger than is realisticfor isolated Mnzn, although he was more particularly concerned with rare-earth activators, the subject of experimental work at Hull. This consideration led to the proposal that the radiative R E or M n 2+ centres in EL exist in some form of complex, perhaps involving Vzn and/or donor species. Although this concept had some attractions for several people at the workshop there is no independent evidence for such a complex, for example from optical spectroscopy. In addition severe doubts were raised about the basis of the calculations of ~and concerning the processes of internal energy transfer within the complex 'molecule' which are required by known properties of the system. The decay time of EL is observed to decrease with brightness, but no true activator saturation seems to occur. Instead, extra non-radiative pathways are postulated to involve interactions between unexcited Mn. The 'memory effect' observed for M n concentrations [Mn] in the range i-5 per cent was discussed by several speakers. Lack of stability in the operating voltage margin between switch on and switch off has so far prevented exploitation of this effect in the devices currently available commercially, although Inoguchi and Mito reported on the potential of the effect for matrix-addressed displays as early as 1975. The existing consensus opinion is that adequate improvements in control of this interesting effect are unlikely for the forseeable future, perhaps because understanding of it remains rather poor. The 'memory margin' is temperature dependent, becoming completely quenched above 200°C. It now seems likely that deep donor states are responsible for this behaviour, rather than the hole states advocated in some previous work. At least two groups of interface trap states appear to exist in this high range of [Mn], in addition to traps present in
'typical' Y203 insulating layers. Considerable attention was given to extra luminescence bands which appear in the red and infra-red spectral region in this same high range of [Mn]. It seems improbable that any of them are due to simple pairs of Mnzn as has sometimes been suggested. However, the exact nature of the Mn-induced complexes responsible for them remains obscure. It is not even known whether these centres involve M n directly, or whether their appearance at high [Mn] is simply a consequence of transfer of excitation between isolated Mnzn and the unknown centres, through the opening of the percolation pathways between M n z n which was discussed in some detail. Strong changes are observed in the line shape of the broad band luminescence of ZnS:Mn with layer thickness t in the range where t ~ X, the wavelength of the emitted light. This has sometimes been attributed to light gain effects, involving stimulated emission. However, it was shown that these effects are due to interference, mainly a classicalproblem of electromagnetic wave theory, including the effects of the metal electrodes. The limitations in the EL brightnessvoltage relations well-known for these devices were shown to involve a limitation in the charge which can be delivered by the dielectric layers in the a c L H F T F E L structures, rather than optical saturation of the active layer. Considerable attention was paid to the question of optimization of the properties of the insulating layers which control this quantity. Present estimates indicate that the Y203/ZnS: Mn/Y203 film reported from Sharp in 1974 had properties close to the optimum when allowance is made for the efficiency for light extraction from relatively perfect thin films! This inherent limitation sets a m a x i m u m brightness level of only about 100 ed m -2 for an extensively multiplexed x - y addressed display. However, such an output can permit displays of good appearance and contrast, at least for non-alfresco use, even without the integral black .layer currently favoured by Sigmatron Nova, as is clear from the current range of Sharp devices. The latterbenefit from strong reflection at the rear electrodes. This problem of insulator-limited charge delivery does not occur for direct current-driven devices. This potential advantage of d c drive is most valuable at high levelsof multiplexing. However, the workshop consensus was that
significant doubts exist as to the ability of d c devices, which must contain some series element for current control, to deliver brightness-voltage relationships sufficiently abrupt to permit adequate contrast at such extensive levels of multiplexing. It must be said that the relatively small amount of extant work on d c H F T F E L was not well represented in Bad Stuer, however ! Although the current status of silicon technology readily permits construction of the addressing and driving circuitry of present-day a c H F T F E L devices, which require peak voltages of about 200 V, there is significant cost benefit in reduction to i00 V or a littleless.Various avenues for achieving this were discussed, particularly the use of improved insulators with high dielectric constant which absorb a smaller fraction of the applied voltage. These improved insulators are as yet untested with regard to stability and maintenance, however. Dead layers were reported by several workers, nearly 0.05 g m wide in ZnS :Mn. It is not yet clear what governs this property for M n activators, though photoluminescence depending upon neutral donors was shown to be quenched by electric field effects at both the substrate interface and the outer surface in single crystal ZnSe grown on GaAs. The M n activator may not exhibit this phenomenon. However, the ZnS :Mn layers in EL devices are polycrystalline. Additional scattering or phase separations effects may occur in the firststages of layer growth, since electron microscopy reveals much smaller crystallitesizes in the first 0.i/~m. Studies of the effects of annealing confirm that improved crystallization is necessary for high EL efficiency. However, the electric field distribution is not simply related to microstructure and crystalline perfection alone is not a sufficient criterion. Processes which involve low deposition temperatures with no freed for subsequent annealing, such as metal organic chemical vapour deposition ( M O C V D ) and the atomic layer epitaxy (ALE) process of Lohja (Finland) may produce improved material quality, particularly in the initialgrowth. Recent experiments at ZIE have raised doubts about the mechanism proposed by Lohja for A L E growth, but there is no question about the regular structure and high optical quality of these layers. Measurements of M n luminescence fine structure confirm x-ray and electron microscopy conclusions that the A L E
material provide hexagonal sites for Mnzn, while cubic sites predominate for MOCVD. Electron beam evaporation provides higher quality material than sputtered, and the ZnS deposition may be more controllable than with thermal evaporation. Hot wall deposition systems are particularly favoured at several laboratories. One particular problem for the future is the introduction of improved techniques for assessment, both at the research level and particularly for in situ use during commercial device production. Very clean conditions are generally agreed to be essential at least for commercial production of complex panels. A burn-in period is also a general requirement, 5 0 - 1 0 0 h depending on the manufacturer. Threeterminal control of individual lightemitting elements in a matrix display would be of benefit, so there is some interest in future thin-film transistor developments for EL as well as for liquid crystal devices. A variety of other systems have been investigated for a c HFTFEL, particularly in Japan, the UK and in France. Most of the work reported at the workshop involved rare-earths, usually also in ZnS, though CdF2 and LaO2 S are also discussed in the literature. The best performance is reported for Tb in ZnS, with Tb incorporated as TbF3 rather than as the metal (Kobayashi, Japan). The resulting green EL is distinctly less efficient than ZnS:Mn, however. Other activators are Sm (red), Pr (white) and Tm (blue), in descending order of brightness. Schanda (Hungary) in reviewing visibility, contrast and human factors also showed that ZnS:Mn yellow is to be preferred over green. He also pointed out that visual acuity may be significantly compromised in large area displays in which information is displayed against a black background. The workshop was enjoyed by all participants on both scientific and social levels. The venue, a holiday centre for workers in the GDR coal industry, was comfortable and set in attractive lakeside surroundings, isolated and so free of distractions! The excursion half-day involved a visit to the studio and museum of the German sculptor/author Barlach and a delightful evening banquet in a Renaissance Palace in Gustrow.
P.J. Dean Royal Signals and Radar Establishment, Great Malvern
DISPLAYS. J A N U A R Y 1984