II-VI elektrolüminesans cihazları

Abstract

ÖZET Üzerinden elektrik akımı geçirilen bir maddeden ışık yayınımı şeklinde kabaca tarif edebileceğimiz elektrolüminesans işlemi, ilk olarak 1936' da Destriau ile birlikte başlamıştır [ 1 ]. Destriau elektrolüminesans olayını çinko sülfür - çinko oksit toz karışımında, elektrodlar arasında gözlemiştir. 1974 yılına kadar, üzerinde yoğun çalışmalar yapılan elektrolüminesans konusunda, özellikle çinkoselenür üzerine inceleme ve deneyler yapılmıştır. Bu incelemelerde tek kristal çinko - selenürdeki Schottky engelleri ileri ve geri meyilde işleme alındığında elektrolüminesansa ait bazı özellikler gözlenmiştir. En verimli emisyon geri meyilde manganez ile aşılanmış çinko selenürden yapılmış diodların kullanımıyla elde edilmiştir. Bu çalışma sırasında manganez ve alüminyumla aynı anda aşılamanın etkileri de ortaya konmuştur. Manganez kullanıldığında, karakteristik emisyon ile elektrolüminesans daha baskın hale gelmiştir. Burada lüminesans 1713 Cd/m2 çıkmış olup, maksimum güç verimi de 3.1 0`3 gibi en üst değerindedir. incelemeler sonucu lüminesans ömrünün kısa ve parlaklığının az olduğu görülmüştür. Işık yayan diodlarda kimyasal buhar çökme ( CVD ), molecular beam epitaxy ( MBE ) gibi tekniklerle bu problemin kısa bir süre sonra çözülebileceği teorik olarak mümkün gözükmektedir. Bu da bize ışık yayan diodlar ve ışık yayan küçük kristaller yapabilme şansı verecektir. Günümüzde; mavi ve yeşil dalga boylarında ışık yayan diodlar ( LEDs ), kalite kontrol için 10`6 hassasiyetle ölçme yapmak amacıyla kullanılan Laser mikrometrelerinde, gösterge cihazlarında, tıbbi cihazlar ve denizin altından geçen uzun iletişim ağlarında kullanılabilmektedirler. IV'

II-VI ELECTROLUMINESCENT DEVICES SUMMARY Electroluminescence ( EL ) is a process in which light is emitted from a substance as a direct result of the passage of an electrical current through it. The effect was first discovered in II - VI compounds in 1936 by Destriau working with a mixture of powdered zinc sulphide and zinc oxide suspended in castor oil between two electrodes [ 1 ]. On application of on alternating electric field a weak green luminescence was observed. After some initial disbelief, an extensive research effort in the 1950s and 1960s led to the manufacture of a variety of plastic and ceramic electroluminescence lamps. These were of low intensity, but found a range of applications as dark room safelights, exit signs in areas of reduced lighting, clock and dial faces etc. The first matrix display was proposed by Piper in 1953 and this stimulated research into a matrix - addressed TV display, the so - called TV on the wail [ 2 ]. At the same time it was hoped that an improved performance of powder electroluminescence lamps would lead to luminescent walls and ceilings. This first major phase of electroluminescence research was severely reduced in the mid - 1960s when the prospects of achieving these ambitious objectives seemed remote. In addition to problems with matrix addressing, the basic difficulty was that the luminance x life product was insufficient for practical applications. The luminance (brightness ) of a panel could be increased by increasing the drive voltage or frequency, but this was more than offset by a reduced life to half - luminance. The early work has been described in considerable detail in a number of books and review articles. In parallel with the work on powders, a smaller effort was devoted to a study of thin filims of ZnS : Mn, but although Thornton was able to obtain a luminance of 3426 Cd/Wt in 1 952, work proceeded slowly because his films degraded rapidly in a matter o hours [ 3 ]. The first serious attempts to develop a thin film alternative current electroluminescence (ACEL) matrix display device were made by Soxman and his associates at Sigmatron from 1964 to 1970, and although a life of 1000 h was claimed in 1972, the programme was abandoned in the early 1970 [ 4 ]. Meanwhile work had been in progress at the Sharp Corporation in Japan under Mito from about 1950, and in 1974 Inoguchi et al. reported a thin film alternative current electroluminescense sandwich device with much improved stability [ 5 ]. This event, coupled with the discovery of a hysteresis ( memory ) effect in the following year, provided the trigger for a world - wide resurgence of interest in electroluminescence display devices based on ZnS : Mn. Progress has been slow but steady. A 240 x 320 line panel was demonstrated in 1978, and by 1988 three manufacturers were offering monochrome, alternative current electroluminescence display devices with up to 480 x 640 picture elements ( pixels ). Large - area, high - definition monochrome and colour displays arecurrently in development. One manufacturer employs films of ZnS : Mn put down by atomic layer epitaxy ( ALE ), a process pioneered. In contrast, a totally different approach was adopted by Vecht and his group, who developed direct current electroluminescence (DCEL) powder cells, where the grains of ZnS : Mn were coated with copper sulphide [ 6 ]. The panel was subjected to a forming treatment to produce a thin insulating layer of ZnS, in which a high electric field would develop and generate the electroluminescence emission. Panels based on this process have been commercially available for some time. Not surprisingly perhaps, work has also continued on direct current electroluminescence in thin films of ZnS. Some of the recent work is described by Blackmore et al. and Catteli et al. [ 7, 8 ]. No practical device has yet emerged. To summarize, large - area electroluminescence display devices have been produced in both thin film and powder form. They are either a.c. ord.c. driven. Each of the four basic types will be discussed in more detail in section 2.2. These large - area devices are all based on polycrystalline, insulating layers of ZnS. A priori considerations would suggest that higher efficiencies should be achieved in single - crystal devices, where grain boundaries are absent. Much effort has gone into the growth of bulk single crystals, and the deposition of single - crystal epitaxial layers of the II - Vis, as described in Chapters 2 and 3, but no light - emitting diode ( LED ) capable of displacing the III - V compound market leaders has been produced. The main reason for this is that it has proved exceedingly difficult to dope the widegap II Vis to display amphoteric semiconduction. ZnS, ZnSe, CdS and CdSe can all be readily prepared as low - resistivity, n - type semiconductors, but the production of low - resistivity, p - type behaviour is more difficult. Success would appear to be closer at hand with ZnSe where ion implantation and the doping of epitaxial layers with such impurities as Li or N has led to various reports of success. However, even when satisfactory p - type semiconduction is achieved in ZnSe the preparation of a good ohmic contact remains a serious problem. As a result of these difficulties the production of a p n home-junction remains a formidable task. A p - n homojunction of a II - VI compound in forward bias passing 1 00 m A cm`2 and emitting in the green ( hv = 2.5 eV ) would have the surface brightness of a tungsten lamp operating at 3000 K. Blue light emitting diods and lasers would be possible, but no commercially viable device has been produced and research continues. Since homojunctions proved so elusive, attention turned naturally to heterojunctions and to Schottky metal-semiconductors (MS) and metal - insulator - semiconductor ( MIS ) diodes as potential light emitting diods. Reasonable success has been achieved with red, yellow and green emitting devices [ 19 ] but none of these has been able to replace GaAsP and GaP light emitting diods. The electroluminescent properties of Schottky diodes formed from single crystals of zinc selenide containing manganese, manganese and aluminium, aluminium or chlorine are discussed. All such diodes emitted a yellow band of variable width in reverse bias, the maximum of which lay between wavelengths of 5785 ( 2-14 eV ) and 6050 A ( 2 - 05 eV ). Optimum emission was obtained from ZnSe : Mn diodes where luminances of 1713 Cd/m2 were achieved at a power efficiency of % 1 - 9 x 10`3 was obtained at a slightly lower luminance of 1028 Cd/m2. The ZnSe : Mn diodes exhibited the characteristic manganese viemission in reverse bias in a narrow band centred at 5785 A0 ( 2 - 14 eV ). The presence of foreign donors such as aluminium reduced the luminance and the efficiency. This, coupled with a slight shift of the emission to longer wavelengths and a broadening of the emission band, is associated with the onset of the excitation of the self - activated emission which increased with increasing donor content. Theself - activated emission in electroluminescense was found at 6300 A ( 1 - 97 eV ) in ZnSe : Al and at 5900 A, ( 2 - 1 eV ) in ZnSe : CI. Forward bias electroluminescence was only observed in diodes which contained a relatively thick ( ~ 200 A ) insulating layer under the Schottky contact. Forward electroluminescence was at least an order of magnitude less bright than reverse electroluminescence. The work reported here helps to resolve some difficulties which can be described as short life and low brightness. On the other hand.different types of impurities were used to solve the poblems and to have more imformation about electroluminescence. The results show that in reverse electroluminescence, manganese emits a narrow band located at 5785 A (2.14eV), the position of whichdoes not change with temperature. In contrast, crystals doped with aluminium or clorine only, display the self -activated band is much broader than the manganese emission, and with aluminium lies at longer wavelength, 6300 A (1.97eV), than the self-activated emission with chlorine, 5900 A (2.1 OeV). The difference between the wavelengths of the maxima of the two self activated emissions is attributable to the complex responsible for self activated emission. This complex is thought to be an associated foreign donor-zinc vacancy pair, since chlorine and aluminium are both donor type impurities, but occupy different sub-lattices, the energy levels of two complexes will be slightly different. Jones and Woods also report that the self-activated emission with aluminium lies at longer wavelengths than that associated with chlorine [ 32 ]. Despite the appearance of a yellow electroluminescence in chlorine or aluminium- doped zinc selenide, the results show that a higher brightness and higer efficiency are obtained if manganese and foreign donor, say aluminium, only one band can be resolved, althought both the manganese and self-activated emission are present. The with of the observed emission is an indication of relative contributions of the two emissions. The diods, which were used in the experiment, with the highest brightness (1370 Cd/m2) and power efficiency ( % 3.10 `3 at 1028 Cd/m2 ) contained manganese, but no aluminium. Increasing aluminium content reduces the efficiency, because the higher donor content leads to a narrower barrier and therefore a larger current flow at lower voltages in reverse bias, so that the impact excitation process is less efficient. Resuits.which there is no space to describe here, also suggest that the luminescence recombination process in the manganese centre is more efficient than in the self-activated emission centre. One essential feature of the succesful preparation of electroluminescent diodes of zinc selenide is that the material must first be rendered semiconducting by some such treatment as heating in molten zinc. Now both Jones and Woods and Allen et al. were unable to detect the characteristic manganese emission in the photoluminescence of zinc selenide crystals containing manganese which had been heated in molten zinc [31,34 ]. Instead the self-activated emission was observed. Nonetheless, both Allen et al. and we reported here, have shown that the chracteristic emission is excited in electroluminescence when diode is operated in reverse bias [ 19, 34 ]. in forward bias electroluminescence, the results suggest that the manganese emission is partially quenched and in consequence both the manganese and self-activated viiluminescence may be observed. In fact more recent work in this laboratory on samples containing high concenrations (~1 per cent) of manganese has shown that the manganese photoluminescence, although quenhed, is not totally quenched following heat treatment in molten zinc. One illustration of this is provided by the inset to Fig.4.4 which shows characteristic manganese excitation band from the 6A-ground state to the ^2 second excited state at 2.45 eV. The overall situation therefore is clear, the emission observed from manganese doped zinc selenide may consist of the manganese or self-activated emissionor both. The relative proportions observed, manifested in the position and width of the resultant band, depend on the conditionsof excitation. In revers bias the manganese emission is excited preferentially; in forward bias and photoluminescence, where free carriers are plentiful, the manganese emission is reduced to a varying extent depending on the composition of crystal. The fact that some contrubition from manganese emission can be observed under forward bias in semiconducting crystals suggests that the first excited state of Mn2+ lies beneath or close to bottom of the conduction band, in order that the probablity of capturing a thermal electron be finite. Allen et al. have argued, relying largely on the work of Braun et al.photocapacitance of zinc selenide, that the ground state of Mn2+ may lie 0.6 eV above the valence band [ 34, 35 ]. As stated above, our evidence would suggest that the first excited state lies just belove the conduction band. Thus the spectral response of the short-cicuit photocurrent of a ZnSe:Mn diode, (90°K curve, Fig. 4.4), contains no peak appropriate to the firs excitation energy of Mn2+ in ZnSe (i.e. 2.31 eV). It does, however, contain a peak at 2.45 eV corresponding to excitation to the second excited state. We conclude that electrons raised to the first excited state are not ionised to the conduction band, but that electronsin the second excited state are. This implies that the graund state of Mn2+ lies about 0.3 eV above the valence band. The voltage intercept in a plot of C`2 against V was larger than the photoelectric threshold in all diodes examined. With diodes cleaved in air prior to the formation of the gold contact, the difference was slight. With etched diodes the disparity could be as large 2.6 eV. This has been interpreted as indicating the presence of interfacial layers under the gold, which can become as thick as 200 °A or more, depending on the chemical procedure used. Diodes with films about 100 °A thick make satisfactory electroluminescent devices, but their maintenance is distinctly inferior to that of the cleaved diodes. This is largely associated with changes which occur in the interfacial layer during prolonged operation. Long distance fibre-optic communications, laser printers and compact disc players all depend on the same material and device technology-the semiconductor laser. These highly efficient sources are based on gallium arsenide (GaAs) and related lll-V compounds. They operate at infrared, and sometimes red, wavelengths but since the 1960s physicists, materials scientists, and electrical engineers have tried to extend these devices to shorter visible wavelengths. Recent proposals include high density optical memories and display devices, medical diagnostics, and communications through sea water and ice. The media thought most likely to emit such short wavelegths were `direct` bandgap ll-VI semiconductors with bandgap energies exceeding about 2.5 eV. But for three decades researchers trying to produce such devices were frustrated -until 1991 when teams at 3M Corporate Laboratories in St. Paul, Minnesota, and a collaboration from Brown and Purdue universities in the US, led by the autors, independently presented proof of concept demostrations of blue-green semiconductor lasers and light emitting diods. VlllEarly efforts to develop 1 1- VI technology primarily involved growing-bulk crystals or films with equilibrium methods. These bulk -grown crystals generally had large numbers of defects and significant concentrations of backround impurities. About 10 years ago non- equilibrium crystal growth techniques were investigated and the recent breakthroughs in wide-bandgap ll-VI light emitters stem from molecular beam epitaxy (MBE), a non- equilibrium technique. The substantial Japanese effort to develop blue-green semiconductor lasers and light emitting diods has provided the motivation for the commercial production of group II and VI molecular beam epitaxy source materials with high purity. In the 3M and Brown- Purdue laser structures, the elemental sources of Se, Zn and Cd were supplied by the Osaka Asahi company and Sumitomo Electric supplied the ZnS source used for blue devices. A second, vitally important, advantage of the molecular beam epitaxy growth technique is its innate ability to produce layered structures from several constituent materials to build up a `superlattice` composite. Molecular beam epitaxy is an ultra-high- vacuum deposition method capable of controlling thickness on the scale of individual atomic monolayers. This has been exploited in lll-V and group IV semiconductors (silicon and germanium), and has had a profound impact on our ability to tailor the light-emission properties of ll-VI heterostructures. In particular, narrow electron and hole confining layers, quantum wells (QW), enhance the radiative recombination probability of electrons and holes due to the resultant 2-D nature of particles. Epitaxial methods have made possible versatile multilayered structures in ll-VI materials and the physics of the 2-D electronic excitations in these structures can be exploited in laser and light emitting diod applications. The prefered device heterostructure contains a single or multiple quantum well sandwiched by a p-n junction. Applying a positive voltage to the p-side, which contains an excess of holes, and a negative voltage to the n-side, which has an excess of electrons, forces the electrons and holes across the junction. However they become trapped in quantum well(s) and radiatively recombine to produce photons. IX