It is now clear that amorphous materials will be the basis of the next great advance in microelectronics. In this paper, I intend to show why the field of microelectronics is presently in a state of crisis, and therefore historically ripe for a basic new approach. I will discuss how our approach using amorphous semiconductors will not only solve the crisis but also spur a new revolution, with the potential of the one of almost 40 years ago. The solid-state revolution, which began in 1947 with the invention of the transistor, was made possible by the ability to make crystalline materials (at that time, germanium) sufficiently free of defects that substitutional dopants could overcome the background noise of other defects and control the electronic transport properties of the semiconductor. Since the early 1930's Bloch, Wilson, and others had laid a sufficient theoretical groundwork in semiconductors so that transistor action could be predicted, demonstrated, and understood even though the point contact transistor had many mysteries associated with it. Figure 1 shows that historical lever that moved the world, the first transistor.

The matrix-addressed liquid crystal display (LCD) has attracted considerable attention in the past few years as an alternative to the cathode ray tube (CRT). This type of display has been demonstrated to have stable performance over a long life as well as the attractive features of small volume, light weight, low power, good brightness and full color. Amor-phous silicon (a-Si) is currently the preferred material for switch devices and plasma enhanced chemical vapor deposition (PECVD)is the preferred deposition technology since the process can be carried out at low temperature, yields material with a low density of defect states and affords good step coverage. The electrical requirements for an a-Si field effect transistor (FET) used as a pixel switch for a LCD include switching time, on current and off current. These parameters depend on the instrinsic characteristics of the amorphous materi-als used, the overall display structure and the device geometry. Present and potential material and geometry limitations will be discussed for different display systems of current interest.

The use of a-Si alloy thin film devices to serve as pixel switching elements in active matrix addressed liquid crystal displays is very attractive. The technology exists to produce uniform high quality a-Si alloy materials over large areas very efficiently and economically. Even though the highest mobility achieved in a-Si TFTs is 1 cm²/V-sec, the currents and speed available with 10-20 μm channel lengths is more than adequate for switching LCD pixels, and the low off-current levels are ideal for holding pixel charge. To date, must of the research and development effort on active matrix LCDs is concentrated on a-Si TFTs. The success of TFT based displays for large area flat panel displays has been limited so far, mainly due to the difficulty of obtaining a high quality gate dielectric by plasma deposition and due to the presence of crossing conductors on the same substrate. Both of these factors increase the probability of defects in the display. When a two terminal sandwich device is used, on the other hand, no gate dielectric is required, and crossing lines are unnecessary, hence a higher yield can be expected.

Properties of "metallic" amorphous silicon alloys and their application to the gate electrode of an FET are described. Amorphous Si-Ge-B prepared by low-pressure CVD and a-Si-Be prepared by RF-sputtering are both metallic in the sense that the density of electron states at and near the Fermi level is finite, though the electron states are localized in contrast to those of metals. The former forms a rectifying junction similar to a metal-semiconductor Schottky-barrier junction with the n-type crystalline semiconductor, while the latter with the p-type one. Barrier heights of 1 V and 0.8 V are obtained for a-Si-Ge-B/n-type GaAs and a-Si-Be/p-type Si junctions, respectively. Such high barriers have not previously been obtained for metal-semiconductor junctions. The performance of FET's is improved by applying these high-barrier amorphous-crystalline junctions to the gate.

A new theory of a-Si Thin Film Transistor (TFT) operation is presented. In addition to the below and above threshold regimes described previously, it predicts two new regimes of operation which occur at very high densities of the induced charge in the a-Si TFT channel. Ina crystalline-like regime the free electron concentration exceeds the localized charge concentration at the a-Si-insulator interface. In a transitional regime (at lower densities of the induced charge) almost all localized states in the energy gap of amorphous silicon near the interface are filled. In the crystalline-like regime, the field-effect mobility is close to the band mobility and the operation of an a-Si TFT is truly similar to the operation of a crystalline field-effect transistor. Our estimates show that the gate voltage necessary to achieve the crystalline-like regime is about 50 V for an a-Si TFT with an insulator 1000 Å thick and a relative permittivity of approximately 3.9. Much lower threshold voltages (close to 8 volts or so) may be achieved by using a gate insulator with a higher dielectric constant.

A silicon bipolar transistor which uses phosphorus doped hydrogenated amorphous silicon deposited by means of the glow discharge technique as a material for the emitter is presented. The advantage of using such material is that its energy band-gap is wider than that of single crystal silicon. Therefore, a barrier for hole injection into the emitter is created at the emitter-base heterojunction, resulting in a higher emitter efficiency. A maximum current gain of 14 at a base Gummel number of 1.35 x 10¹³ cm^-4s is obtained, this value represents a 5-6 fold improvement over a conventional homojunction transistor with an indentical base region.

Because of the growing use of amorphous silicon alloys in microelectronic applications utilizing both the PIN and NIN structures, we have undertaken a study of the limitation imposed by the bulk intrinsic layers. In the former at voltages above approximately 0.8V, the bulk controlled current is recombination limited while in the latter it is space charge limited. As a consequence, the magnitude of the currents as well as the time and frequency responses are very different. Data documenting these differences are presented along with a model to explain these results.

A technique for measuring the electrical characteristics of contacts to doped hydrogenated amorphous silicon (a-Si:H) or other high-resistivity thin film semiconductors is developed. Experimental results for metal and conductive transparent oxide contacts to both n- and p-type a-Si:H are presented and the significance of these values to solar cell applications discussed.

In this paper, recent results of a study of interface formation and microstructural evolution during the growth of hydrogenated amorphous silicon (a-Si:H) will be reviewed. In situ ellipsometry experiments using a photon energy of 3.4 eV have been applied to characterize the thin film nucleation process under different conditions of film preparation. In addition, monolayer sensitivity to changes in the modulation depth of surface roughness layers is achieved during growth. Spectroscopic ellipsometry measurements have been used to establish the absolute thickness of roughness layers at the surface of a-Si:H and at the a-Si0₂/a-Si:H interface. An in situ study of a-Si:H/a-SiNx:H/a-Si:H deposition reveals an asymmetry in the interface structure depending on the order of deposition. These are a sampling of the important results that establish ellipsometry as an ideal in situ probe of the growth morphology of thin film amorphous semiconductors.

We measure picosecond photoinduced absorption in a low defect sample of hydrogenated amorphous silicon. Our results indicate that both the imaginary and real parts of the complex index of refraction contribute to the observed decay of the induced transmittance. We present two methods for obtaining the undistorted decay of the induced absorption. We show that the decay of the induced absorption is significantly different at carrier densities of 5.9 x 10¹⁷ cm" and 2.4 x 10¹⁸ cm'. The lower carrier density decay suggests that the data are consistent with the model for dispersive transport in amorphous materials.

In xerography, a corona charged photoconductor film transforms an optical image into an electrostatic image that is subsequently developed. A variety of photoconductor materials has been employed in commercial reproduction machines. These include alloys of amorphous selenium with arsenic and tellurium, zinc oxide, and cadmium sulfide particles in organic binders; one and two layer organic systems; and, most recently, hydrogenated amorphous silicon. The physics of the charge photogeneration and transport processes are examined and illustrated with the above materials. These processes are shown to determine the reproduction characteristics of the xerographic system.

Amorphous silicon (a-Si) alloy semiconductors are becoming increasingly important in electrophotographic applications. Photoreceptor devices based on these materials exhibit improved performance for many applications and photocopiers and laser printers using these devices are now in commercial production. Fundamental differences in the chemistry and physics of a-Si alloys, as compared to the more conventionally used amorphous chalcogenide alloys, result in differences in the way electrophotographic devices are fabricated and used. This paper will review the construction and operation of a-Si alloy photoreceptors contrasting them with amorphous chalcogenide alloy devices with particular emphasis on aspects which result from fundamental differences in the physics and chemistry of these two classes of amorphous semiconductors.

A-Si has become an attractive alternative for conventional electrophotographic photoreceptors. A-Si photoreceptors have been prepared by other laboratories by plasma deposition with blocking and protection layers. These photoreceptors are highly photosensitive and show low fatigue. Using sputtering we have shown that this technique is capable of produc-ing films with high charge acceptance. The increase of the deposition rate is presently un-der intensive investigation. High rates can be achieved by a higher degree of silane decomposition or by magnetron sputtering together with a higher power level. Deposition rates of more than 20 pm/h have been obtained by both techniques.

The preparation and properties of multilayered glow-discharge a-Si:H photoreceptors are described. It is shown that a high-rate deposition of the intrinsic layer of a Al/p/i/a-Sic structure with no stress is possible by an ordinary glow-discharge method . The composition and thickness of the blocking layers proved to affect particularly the dark decay time, residual voltage and photosensitivity. Inserting a a-Sii-xGex layer into the intrinsic layer was shown very effective in sensitizing the infrared region. A high-speed laser-beam printing system with a printing speed of 20000 lines/min for 8LPI was realized by using this sensitized photoreceptor.

Reactively sputtered a-Si:H is used in a photoconductive target of an image pickup tube. When a-Si:H is used for a color imager, high resistivity and wide optical band gap are required. A blocking type target structure is effective in reducing dark current and producing rapid photo response. Low doping of boron into a-Si:H serves to increase both hole and electron mobilities. The imaging tube using this a-Si:H has high photosensitivity, high resolution, and no sticking or burning. By using high velocity electron beam for scanning, lag property is drastically improved.

A review is given of some aspects of a large-format image sensor which combines an a-Si:H thin-film sensor array with microelectronic chips for readout, interconnected on one substrate. Comparisons are made between the different thin-film photoconductor materials, the various noninjecting contacts for the sensor photodiodes and the circuits for readout as they have been proposed so far. A-Si:H is found to be a superior material for the sensors; however, the readout circuit must be carefully designed to obtain a satisfactory signal at the sensor output.

A contact linear image sensor using a-Si:H heterojunction photodiode array has been developed. The sensor is composed of a pair of LED array, a rod lens array and a document width linear image sensor in which a-Si:H photodiode array and driving ICs are mounted on the same glass The a-Si:H diode has ITO/p-a-SiC/a-Si:H/metal structure. It exhibits as high as 10⁴ photo-to-dark current ratio and quick photoresponse because of excellent blocking characteristics and large built in potential of heterojunction. The contact sensor has been operated with 1 msec/line scanning speed. Performance tests show excellent results with 8 lines/mm resolution. Images have been satisfactorily recorded using a thermal printer. Moreover, thin film image sensor using a-Si:H TFT will be discussed as a lower cost linear image sensor.

Contact type linear image sensor arouse a great interest as a compact image reading component for a facsimile, an image scanner and many other applications. Among those, amorphous silicon linear image sensors which consist of Cr/a-Si:H/ITO structure have excellent features such as capability of fabrication of large area devices and material stability. B4 size, 200spi image sensors are now on mass production. B4 size, 400spi image sensors have been developed. And very large image sensors with 36 inch length and color sensors also have been developed. a-Si:H TFT driven image sensors which can reduce the cost are under development.

Heavy doped a-Si:H:F thin film containing a microcrystalline phase (pc-S:H:F) has properties of a very high dark-conductivity and a large thermoelectric power. Wip this pc-Si:H:F film, high performance thermopile has been developed for power sensors. ' The film was deposited on glass substrate, using SiH₄ + SiF₄ gases, by radio frequeycy glow discharge decomposition. The film showed a largg dark-aonductivity of 33 S cm^-1 (the highest value), and a large thermoelectric power of 250 pV/K for the p-type and -200 pV/K for the n-type, respectively. The thermopile power sensor was produced having power sensitivity of 1.5 mV/mW and the linearity being better than 1%. A new type of high performance wideband level meter, using the amorphous silicon thin-film type thermopile rms-value-detector was developed and has been put into practical use.