Search

Amorphous thin film tetrahedrally based alloys have recently achieved an efficiency which permits them in volume production to become competitive to fossil fuels. In this paper, the next stage of development which is to raise stabilized efficiencies to 18% is discussed. It is shown how the scientific and technological problems can be viewed and understood and point to means of solution.

We report the effect that thermal annealing in inert and oxidizing atmospheres, and with and without encapsulating layers, has on the structure of tungsten/carbon [W/C] multilayer thin films. This study focuses on the tungsten component and deals mainly with multilayers where the ratio of thickness of tungsten layers is equal to or greater than for the carbon layers (that is, γ ≤ 0.5). This is in contrast to prior studies where the tungsten layer thickness was generally held constant and the carbon layer was varied. Thermal annealing in inert atmospheres produces reactions and other structural changes in the tungsten and carbide layers which depend on the as-deposited multilayer structure which depends, in turn, on the thickness of the tungsten layer. In samples where both the tungsten and carbide fractions of the multilayer are completely amorphous as deposited, which is the case for thin tungsten layers (thickness of tungsten (tw)4 nm/period), the reactions in the tungsten layer forming crystalline tungsten and tungsten carbide occur at annealing temperatures above 900°C. The layer pair spacing, or period, (d), in this group shows an expansion of up to 10-15% of the original value as has been reported in the past. Changes in both the tungsten and carbide layers, and their interfaces, contribute to changes in d spacing and relative thickness of the high and low Z components. When the tungsten layer thickness exceeds 4 nm per period the tungsten is partially crystallized in as-prepared samples. In such multilayers interfacial reactions, producing an oriented partially crystalline W2C/C superlattice, occur at temperatures of 600°C and below. The fact that W2C crystallites in one period can form a structure which is correlated to W2C crystallites in neighboring layers is somewhat surprising, since layers are presumably still separated by amorphous carbon which is still visible via Raman. The expansion of the layer pair spacing is relatively small (5%) in this group and, more importantly, mostly involves increases in the thickness of the high Z components. Samples annealed in air at temperatures below 300°C are progressively destroyed by the oxidation of both tungsten and carbide layers. Encapsulation of similar multilayers with a thin (30 nm) dielectric layer of any of several types can retard oxidation to 600°C. The silicon-containing encapsulants generally perform better. Failure at this temperature is seen to occur from pinhole formation.

Hydrogen $({\mathrm{H}}_{2})$ dilution of the source gas is known to be a key factor in producing hydrogenated amorphous silicon films that demonstrate a high degree of optoelectronic stability. In this work, we investigate, using Raman spectroscopy and high-resolution transmission electron microscopy (TEM), whether microstructural differences exist between such films and those made with no ${\mathrm{H}}_{2}$ dilution (i.e., that have greater instabilities). The key variable is the ${\mathrm{H}}_{2}$ dilution, which ranges from none to very high levels, producing amorphous and microcrystalline silicon films. The TEM results show that embedded within the amorphous matrix are chainlike objects (CLO's) having \ensuremath{\sim}3 nm widths, \ensuremath{\sim}30 nm lengths, and showing a high degree of order along their length. Such order implies vanishing levels of bond-angle distortion (BAD). These CLO's are present in all samples investigated, but their density increases with the level of ${\mathrm{H}}_{2}$ dilution. The Raman spectra show a TO band centered at $490 {\mathrm{cm}}^{\ensuremath{-}1}$ $(37\ifmmode\pm\else\textpm\fi{}3{\mathrm{cm}}^{\mathrm{\ensuremath{-}}1}$ full width). Quantitative analysis shows this band to exist in all samples investigated, but increases in magnitude with increasing ${\mathrm{H}}_{2}$ dilution. In the highest dilutions when microcrystallites are observed, the band is distinctly evident. Its position and width are also consistent with very low (crystallinelike) levels of BAD \ensuremath{\sim}0\ifmmode^\circ\else\textdegree\fi{}. It is thus likely the $490 {\mathrm{cm}}^{\ensuremath{-}1}$ Raman band is a signature of the intermediate ordered CLO's.

Tungsten/carbon (W/C) multilayer thin films with a nominal d spacing varying from 2.5 to 14 nm were prepared by magnetron sputtering technique. The thicknesses of the W and C layers were varied from 0.5 to 12 nm. The multilayers were subjected to isochronal anneals in a quartz tube furnace in the range of 300 to 1000 C under high purity Ar flow conditions. X-ray diffraction, Raman scattering and Auger depth profiling were used to characterize the structure of the as-prepared and annealed films. It is found that an overcoat layer of silicon nitride (30-50 nm) prevents the multilayers from oxidation during the 1 hr heat treatment at temperatures as high as 1000 C in Ar flow. In all studied W/C multilayers, the carbon layers are amorphous (up to 12 nm). The tungsten layers are also amorphous when their thicknesses are less than 5 nm. Tungsten layers thicker than 5 nm show crystalline W peaks in addition to the amorphous W feature. Annealing of samples with a silicon nitride protective layer results in several structural changes which depend on annealing temperature, d spacing, the as-deposited W layer structure and the layer thickness ratio of W to C. For W layer thicker than C layer and W layer thickness > 4 nm and/or C layer thickness < 1 nm, the multilayers show the initial crystal formation of microcrystalline W2C occurring at C-W interfaces (that interface in which C was deposited on W) after 600 C anneal, followed by a second crystallization of a-W or a-W and WC at W-C interfaces (W was deposited on C) at the annealing temperature of 900 C. They reveal a relatively small (< 5 %) or essentially no layer expansion. For those multilayers having thin W layers (2 nm) and the same or thicker C layer thicknesses, the initial crystallization takes place at both W-C and C-W interfaces at 900 C or higher. The crystal formed is a-W or a-W and WC. The layer pair period of the multilayers in this group increases monotonically with increasing annealing temperature. Expansion is up to 16 % of the original d spacing and occurs in both W and C layers at approximately equal rates. The expansion in all multilayers is interpreted to be associated mainly with the structural ordering processes in the amorphous W and C layers.© (1992) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

We report on evidence that fluorine, properly incorporated into a-Si, replaces weakly bonded hydrogen and improves the material stability under light soaking. Our fluorinated amorphous silicon (a-Si:H:F) is made by if glow discharge at high deposition temperatures up to 430 C from a gas mixture of SiH4 or Si2H6 and F2. These a-Si:H:F films show much lower density of states in the light soaking saturated state than device quality a-Si:H prepared in the same deposition system. It is evident from our results that fluorine incorporated into the network at such high deposition temperature makes for a new configuration which minimizes dangling bonds and other defects.

The thermal activation energy, AE, of the conductivity in amorphous chalcogenide semiconductors is usually about half of the optical gap energy E. This is commonly interpreted as (intrinsic) conduction at the mobility edge by carriers whose concentration is governed by the Fermi energy EF which is pinned near the gap center. Small changes in glass composition and the presence of impurities usually have little effect on the properties except for small changes in EO and AE with composition. The insensitivity to impurities is ascribed to the fact that in a material lacking long-range order each foreign atom can satisfy its valency requirement,(1) and, thus, will not act as an electrically active center. The origin of the various states in the gap (DECs) (2,3) and the pinning of EF is attributed to those DECs which are inherent valence alternation defect centers.

The position of the Department of Energy (DoE) reiterated specifically at this meeting (1) is that photovoltaic solar energy conversion will not begin to make a large-scale contribution to our overall energy production until 1986 at the earliest, and more probably the year 2000. It is the theme of this paper that the basic materials problems which have led to such pessimistic projections have already been solved and that, given an intensive program of development, economically competitive photovoltaic cells can be mass produced in the near future.

The US Department of Energy is probably the best known body to have claimed that photovoltaic conversion devices cannot compete with fossil tuels or uranium until at the earliest 1986, or more likely 2000. Such statements are based on the known current technological problems of three types of silicon—single crystal, polycrystalline, and amorphous. With amorphous silicon the chief difficulty has been the high density of states (as explained by Taylor on p 672) which restricts the necessary current flow.

On Sir Nevill’s 80th birthday, I wish to continue the discussion which 1 outlined in my contributions to his 65th and 75th birthday festschrifts [1,2], and I look forward to further exploring these concepts in the festschrift for his 85th.

We view amorphous materials as being ideal matrices for the encoding of information by optical or other means.1–10 The result can be the storage and reproduction of information serially or in parallel. The laser is ideal for the digital storage of information. One can achieve an exceedingly high density (one micron spots closely spaced) with a high signal to noise ratio, and utilize the same laser for reading, writing, and erasing. For higher densities, scanning electron beams are used.8

There are three key issues that challenge amorphous photovoltaics, namely, efficiency, stability, and production. In this paper we will review the current status of our amorphous photovoltaics at ECD in terms of these three issues and demonstrate how we have broken the efficiency-stability-production barrier.

The electronic nature of Ovonic threshold switching (1) has been definitively established.(2–8) Electronic action is also the initiating factor in our memories and the electronic concept has proven useful in establishing information encoding in an imaging manner as well.(9–22) It is now appropriate to elucidate the physical and chemical basis for such action.

It is fitting to discuss in this festschrift for Hellmut Fritzsche the phenomena which first interested him in the amorphous field.

It is well known in crystalline tetrahedral semiconductors that conventional incorporation of atoms by substitutional doping increases the electrical conductivity by many orders of magnitude. Such doping is the basis of the crystalline semiconducting industry and has recently been extended to amorphous silicon and germanium, with hydrogen appearing to play an important compensating role.(1–3)

The holographic response (Keneman, 1971; Ohmachi and Igo, 1972) of evaporated AS2S3 at 4880A has been ascribed to photo-polymerization of a hard-sphere As4S6 glass (deNeufville et al., 1973). The effect of illumination is to shift the location of the optical absorption edge (ΔE=-0.06eV), and to increase the refractive index (An=0.08). A comparable increase in refractive index occurs during annealing of evaporated As2S3 at 189 C, near its glass transition temperature, accompanied by a smaller edge shift (Δn=-0.03eV). While the optical absorption edge can be reversibly cycled by exposure at 25°C and annealing at 180°C (ΔE=-0.03eV), the accompanying index changes are relatively small (ΔE=**0.01). Furthermore, the X-ray diffraction pattern is unaffected by cycling, suggesting that the optical effects arise from the creation (exposure) and elimination (annealing) of a nonequilibrium concentration of trapped charge carriers rather than from an innate structural change.

An important difference between crystalline and amorphous structures is that while in crystals the local environment is the same everywhere, in amorphous materials one finds a large spectrum of 3-dimensionally varying spatial and bonding relationships. This gives rise to unique energy interactions, orbital overlaps and bonding configurations of the lone pair electrons associated with the group VI chalcogen atoms which do not occur in crystals. We present the ‘ball and spoke’ models we have built to test out the concepts of lone pair relationships discussed earlier (Ovshinsky 1972, Ovshinsky 1973, Ovshinsky and Fritzsche 1973).

There is a need for alternative energy sources. Photovoltaics are an attractive possibility, but their application has been limited by economic considerations in single-crystal materials, and for physical reasons such as grain boundaries in poly-crystalline materials. Amorphous semiconductors are especially attractive in this regard because they are basically much less expensive than their crystalline counterparts and because they possess a direct band gap with a high value for the optical absorption coefficient. We report here the development of a new alloy that eliminates the physical problems associated with the silicon-hydrogen alloys.

Energetic processes have been described for fast and controllable ordering and disordering of amorphous semiconductors.(1, 2, 3, 4, 5) Concomitant with these changes are alterations of various physical parameters, all of which we use for the temporary or permanent storage and display of information (6). The present paper reports on some physical characteristics associated with photostructural changes reflecting imaging processes in amorphous materials.

A subject of great scientific and technological interest, amorphous materials is intimately connected with the topic of this meeting. This is especially so since one of the preferred methods of making amorphous films for photovoltaic devices utilizes plasmas in which dissociation of gases is associated with dc electric fields or radio frequency excitation. Such creation of plasmas by nonthermal means generates excited states in the gas, ultimately producing different chemical species from that of the ground state molecules. The excited states have a high energy content and, in some cases, a unique electronic distribution, and are thus much more reactive.

Amorphous Si:F:H with desirable properties for photovoltaic applications can be fabricated by the glow discharge of SiF4 and H2. The preparation conditions influence the properties of the resultant alloy. For instance, altering the ratio of SiF4 to H2 from 80 to 5 can alter the localized state density from 1019cm−3eV−1 to ≃ 1016cm−3eV-1, respectively. The conduction mechanisms are altered and there are vast changes in the photoconductivity as the density of recombination centers is decreased. The lower density of states achieved in the a-Si:F:H alloy reflects in the ease of doping. In addition, the lower density of states in a-Si:F:H alloy should result in a wider depletion region than reported for the a-Si:H alloy when fabricated within the device configuration. Results of C-V measurements using Au Schottky barrier devices confirm this.

We have grown amorphous silicon-germanium alloys by rf glow discharge decomposition of disilane, germane, silicon tetrafluoride and hydrogen. Films with an optical band gap in the range of 1.72 to 1.2e V have been made. The optimized films show the following properties: 1) incorporation of germanium is uniform across the length of the sample; 2) hydrogen is bonded to silicon and germanium uniformly without any preference; 3) the slope of the valence band tail remains constant as the optical gap is lowered down to 1.25eV and 4) use of silicon tetrafluoride in the gas mixture improves the quality of the material. Photothermal deflection spectroscopy and space charge limited conduction were used to obtain information about the deep states. Even for the lowest band gap material, the density of states at the Fermi level is about 10 17 cm −3 eV −1 which demonstrates the high quality of the material.

Previously we have shown that amorphous silicon-based alloys containing F and H (1–7) possess desirable properties for photovoltaic application. Amorphous tetrahedral semiconductors ordinarily possess a very large density of states which act as traps leading to low values for drift mobility and low recombination lifetimes of free carriers. However, Spear and his group (8–10) reported that a-Si decomposed from SiH4 gas by r.f. glow discharge and deposited on a heated substrate produces a film which has a comparatively low density of states. However, materials prepared in this way possess a large concentration of H.(11) Because of the relatively low density of states, these types of films can be doped n type, albeit using relatively large concentration of the dopant. However, p-type doping using B2H6 is generally accompanied with the decrease in the band gap, indicating alloying action rather than conventional doping.(9) Furthermore, the material has apparently a sufficiently large carrier lifetime that efficiencies of photovoltaic devices exceeding 5% have been reported.(12)

It is fitting that we honor Norbert Kreidl by considering the future of glass for there is a dramatic “phase” transformation presently taking place in the area of glass science and technology. The glass meetings of old were centered on the much debated subject, “What is glass?” In a sense, this reflected not only scientific uneasiness but a search for identity, for the very basis of glassy materials, their inherent disorder, was the cause of much insecurity. The crystalline field had as its bedrock the crystal lattice, the order, if not the boredom, of repetitive atoms which looking in all directions saw sharply definable and identifiable neighbors over relatively long distances. It is no wonder that glassy materials from time immemorial through the 1950’s appeared to be associated only with wide band gap materials whose optical and mechanical properties were of paramount importance.

The way we work, indeed, our entire industrial civilization, depends upon materials. We have long used noncrystalline solids in the form of glasses for their passive properties, such as their inertness, transparency, and in everyday use as containers. They are artifacts of an ancient time, not only on earth but in the heavens—on the moon glassy materials are widespread.

We have made a study of the resistivity, ρ(T), and the structural and superconducting properties of amorphous Mol-xSix Besides the usual interest in studying a metal to nonmetal transition, this system has the additional feature that Si changes from acting metallic for x << 1 to acting nonmetallic for x ~ 1. Murarka et al. (1) measured the resistivity of both amorphous and crystallized cosputtered films of MoSi and a comparison of their data with ours will be made below. We have observed a large increase in dρ(297)/dx with increasing x at x = x0 ≡ 0.63 ± 0.05 which is most likely associated with an electronic transition involving the Si valence electrons. A tentative model for this transition is presented below. Besides observing this increase in d ρ(297)/dx, the specific results pertinent to superconductivity are: 1. The initial linear decrease of the superconducting transition temperature T­c is 0.075 K/at.% Si. This is consistent with the e/a dependence found by Collver and Hammond (2) for near neighbor amorphous 4d metals if one assumes that each Si atom contributes 4 electrons to the conduction band. 2. Supperconductivity persists up to at least 64 at.% i.e., slightly into the concentration region where the resistivity begins to increase. As x→xo ­from below, the superconducting transition width ΔT­c increases rapidly until at x ≃ xo the width ΔTc becomes comparable to Rc. 3. For x > xo, Tc, if it exists, is less than 1.5K.

A NEW TYPE of amorphous semiconductor alloy has recently been reported [1]. This new material has silicon and fluorine as its main structural components. It is multi-elemental and includes hydrogen and can also include other elements such as oxygen without deleterious effects. This Si: F: H alloy, prepared by the glow-discharge decomposition of SiF4 mixed with hydrogen, has been reported to overcome a number of problems of a-Si and a-Si: H. The a-Si: F: H alloy is highly photoconductive, possesses a lower density of states in the upper half of the band gap than the a-Si: H alloy and is devoid of any photostructural changes. In order to gain further information about the electronic and vibrational states of this amorphous semiconductor, we have investigated the electrolyte electroreflectance (EER) and Raman spectra of several samples of this material.

THE AMORPHOUS-CRYSTALLINE transition in a-Si and a-Ge has been studied for many years [1,2]. However, unlike electrical and optical properties, no substantial modifications on the RDF (radial distribution function) for the amorphous phases have been reported for different methods of preparation or annealing conditions short of crystallization. More recently, Barna et al. [3] have found a small shift of the second peak in the RDF towards smaller r, and a more pronounced peak at 5 A for the a-Si: H prepared by glow discharge technique. These authors concluded that a higher degree of local order is present in the glow discharge a-Si: H although they stated that there exist only minor differences in amorphous silicon samples with different preparations. Earlier Raman measurements [4] have not been shown to be sensitive to preparation conditions, however, recent results indicated that Raman measurements can indeed differentiate various types of amorphous silicon samples. Tsu et al.

Amorphous solids, as opposed to crystalline solids, have an atomic structure with no long-range periodicity. Since the quantum theory of solids was developed in the period before the importance of amorphous materials was evident, the symmetries resulting from periodicity were exploited to simplify the calculation of physical properties. This led to a rapid understanding of the general behavior of crystals, but also obscured both the fundamental reasons for this behavior and the greater range of properties which could be achieved in amorphous solids. Even after amorphous semiconductors were discovered and characterized, an inordinately large effort was expended into understanding the properties of those amorphous materials with simple crystalline analogues. It is the purpose of this paper to emphasize the new modes of behavior when the constraints imposed by long-range periodicity are removed.

There are many dogmas which have been uncritically accepted in the amorphous field that are basically unsound, the chief one being that the short-range order surrounding each atom is the same as in the “corresponding” crystal and that all that is different is some fluctuations in bond lengths and angles. We propose to show that short-range order is not conserved in amorphous materials, since space is related to energy in a basically different manner. The consequences of this lead to entirely new local structures than are found in crystalline materials and to the fact that many of these structures have no analogs in crystals and are inherently anticrystalline in nature.

Overall conversion efficiency of 6.6% has been obtained for a photovoltaic device over an active area 0.73 cm2 using amorphous Si : F : H alloy in a MIS configuration.

A roll-to-roll plasma deposition machine for depositing multilayered amorphous alloys has been developed. The plasma deposition machine has multiple deposition areas and processes a stainless steel substrate 16 in wide continuously. Amorphous photovoltaic thin films (less than 1 μm thick) with a six-layer structure (p-i-n-p-i-n) are deposited continuously in a single pass onto a roll of stainless steel substrate 16 in wide and 1000 ft long. Mass production of low cost tandem solar cells utilizing roll-to-roll processes is now possible. A commercial plant utilizing this plasma deposition machine for manufacturing tandem amorphous silicon alloy solar cells is now in operation.

I feel that the best way for me to write a paper that honors Sir Nevill Mott is to continue a discussion on points of our mutual concern for Mott is not one who rests on his laurels but is in ceaseless struggle with nature, using the power of his mind as a searchlight into the unknown; in the case of amorphous materials, into the world of disorder. It is unfortunate that this negative emotive word, disorder, is used to describe amorphous materials since I believe that the operative description of the uniqueness of amorphous materials is really “freedom”-freedom from restrictions of crystalline symmetry, freedom from stoichiometry, and especially freedom of choices of bonding and nearest neighbor relationships. The multiple bonding and orbital relationships permitted by the removal of crystalline restrictions, while numerous, are specific and, in the manner and number in which they co-exist in an amorphous material, are definable as well as being unique.

Previously, we have reported, in heavily As- or P-doped Si:F:H alloy systems, the appearance of a Raman peak lying intermediate between 522 cm-1 for c-Si and 480 cm-1 for amorphous Si.(1,2) Whenever such Raman peak is observed, electrolyte-electro-reflectance (EtR) peaks appear around 2 eV, together with those associated with c-Si at 3.4 eV and 4.5 eV. We have explained these observations in terms of an intermediate range order or a “microcrystalline phase.” Now we have found similar observations in moderately P-doped samples. On glass substrates EER may be observed when the volume fraction of crystallinity has passed 0.16, the critical density, ρ cr, in percolation processes. (3,4) However, on stainless steel substrates, EER has been observed for ρ cr < 0.16, indicating that unlike conductivity, EtR requires only the existence of relatively sharp electronic density of states.

The structural, optical, and electrical properties of amorphous multilayer films containing up to 180 double layers of a-Si:H/a-SiNx and a-Si:H/a-SiOx were studied. For a-Si:H layer thickness d >100 A the transport properties are dominated by space-charge doping with a-SiNx positive and a-SiOx negatively charged. For d A quantum-well effects increase the optical and electrical gaps. A d =12 A multilayer film shows no evidence for the predicted loss of extended states in two-dimensional disordered systems.

In the first paper of this conference, Prof. Stuke presented a detailed review of the optical and electrical properties of the basic materials, Se, Ge and Te, which can be formed in both the amorphous and crystalline form1). In the final day of this meeting we would like to tie the basic physics learned from these discussions to more material ends — that is to the design of materials which can be utilized for the basic differences between crystal and amorphous forms and, in particular, the differences in optical properties.

ORDER-DISORDER transformations such as ordering of magnetic or ferroelectric domains are used in many information storage devices. The structural transformation of amorphous semiconductors from the amorphous to a more ordered state can be used for the same purpose if the structure transformation can be achieved fast and reversibly.1 Amorphous semiconductors have then the advantage over other information storage devices in that they can be cheaply produced and easily shaped in many different configurations. Furthermore, for semiconductors the difference between the physical properties of the amorphous and the crystalline state is particularly large. This enables one to retrieve and read the binary information, stored in the form of the structural state, with good signal to noise ratio despite extensive miniaturization.

Description of a double-injection space-charge process which can account for the principal features of amorphous threshold switching, and discussion of its applicability to chalcogenide glasses under various operating conditions.

This paper describes structural studies of the reversible memory type electrical transition1) in amorphous films of Ge-Te2) based alloys. The films, as evaporated, are high resistance intrinsic-like semiconductors. After subjection to a sufficiently high electrical field to bring about a transition to a low resistance state, and a further delay during which there is current flow, a portion of the material is transformed to a semipermanent low resistance state. That is, the material remains in a low resistance state without further application of an electric field. The material may be switched back to a high resistance state through application of a current pulse with a rapid turn off. This paper reports initial results of studies designed to elucidate structural aspects of the phase transition in these materials.

Radial distribution studies of amorphous Ge χ Te 1−χ alloys have been calculated from X-ray diffraction data for x = 0.11 and 0.54. These show peaks at approximately 2.7 A and 4.2 A. The absence of a peak at the crystalline GeTe first neighbor separation of 3 A is shown to imply that the local coordinations in the amorphous materials are different from those in crystalline GeTe. The areas under the first neighbor peaks indicated that it is very unlikely that the coordination numbers of the Ge and Te atoms are the same. These areas are consistent with models in which the average Ge and Te coordinations are four and two, respectively. This implies that the similarity in the optical absorption edges of crystalline and amorphous GeTe cannot be explained by similarities in short range structures. The similarities in bonding over a wide range of compositions in these amorphous alloys appear to lend support to Mott's picture for the failure of many impurities in many amorphous semiconductors to add significantly to the conductivity. The form of bonding postulated appears to be inconsistent with a microcrystalline picture for these alloys.

A three-fold classification of amorphous semiconductors into (i) elemental, (ii) covalent alloys, and (iii) ionic and tightly bound amorphous materials is proposed. The experimental evidence supporting a simple band model for the amorphous covalent alloys is presented. The present understanding of the reversible switching effects and of the switching with memory is discussed.

This paper will briefly review our past work in amorphous switching and discuss some of the operational concepts underlying our continued effort. In the past, periodicity of atomic structure has been the basis of the understanding of solid state physics. Part of the problem in understanding the amorphous state is the terms themselves, for the words “disordered” and “amorphous” for non-crystalline materials have misleading connotations as they project an image of a material with lack of specific structure. In this paper we will describe some of our working hypotheses which pay particular attention to spatial relationships of the localized structural units as a means of developing unusual electronic and electrical changes in amorphous materials.

We have observed a high-speed crystallization of amorphous semiconductor films and the reversal of this crystallization back to the amorphous state using short pulses of laser light and evidenced by a sharp change in optical transmission and reflection. This optical switching behavior is analogous to the memory-type electrical switching effect in these materials which has received wide attention1 since the observation by S. R. Ovshinsky2 of both threshold and memory switching in amorphous semiconductors. In this letter, we propose a model which closely relates the optical and electrical switching behavior, and shows that the phase change from amorphous to crystalline state is not only a thermal phenomenon but is directly influenced by the creation of excess electron-hole carriers by either the light, or, for the electrical device, by the electric field. The reversibility of the phenomenon in this model is obtained through the large difference in crystallization rates with the light on or off.

We describe here a rapid and reversible transition between a highly resistive and a conductive state effected by an electric field which we have observed in various types of disordered materials, particularly amorphous semiconductors1,2 covering a wide range of compositions. These include oxide- and boron-based glasses and materials which contain the elements tellurium and/or arsenic combined with other elements such as those of groups III, IV, and VI.

The influence of illumination and annealing on thin films of amorphous As2Se3 and As2S3 has been studied through their effects on the structure and optical properties. It is shown that the dominant photostructural change in evaporated films is the polymerization of the as-deposited molecular (As4S6 or As4Se6) glass through either illumination or heating. This irreversible polymerization gives rise to a large absorption edge shift to smaller energy and to an appreciable index of refraction change. In addition there is a small reversible absorption edge shift, due presumably to trapping of photo-generated electrons and holes which is larger in As2S3 than in As2Se3. This reversibility persists in both well-annealed evaporated films and in sputtered films which appear to be structurally indistinguishable both from each other and, aside from defects or voids, from the bulk glass. The reversible effect, called photo-darkening, is accompanied by a negligible refractive index change when compared with the irreversible polymerization change. The application of the present results to photochemical effects is discussed.

Performance and reliability of amorphous semiconductor devices that deal with the handling of information in the form of switching, modulation, storage, and display are discussed. Structural changes between a disordered and a more ordered state and the concomitant large change in many material properties offer the possibility of using amorphous semiconductors for high-density information storage and high-resolution display devices. The structural changes can be initiated by various forms of energy such as an electrical pulse, a short light pulse, or a brief light exposure. Many materials show good structural reversibility. The sensitivity of an amorphous photostructural film is amplified by several orders of magnitude by first forming a latent image by photonucleation and subsequent dry development by heat or radiation. Examples of optical contrast and resolution in image formation are given. The major differences between crystalline and amorphous semiconductors are briefly outlined.

Amorphous covalent alloys particularly of group-IV, -V, and -VI elements are readily formed over broad ranges of composition.1–6 They have been described as low-mobility electronic intrinsic semiconductors with a temperature-activated electrical conductivity σ = σ 0×exp(-ΔE/kT) which sometimes extends well into the molten state.2,3,7 They remain intrinsic with changed ΔE when their composition is changed.1,5,7 These alloys transmit infrared light up to an exponential absorption edge from which an energy gap E g is estimated.1,2 The value of E g usually is smaller than 2ΔE, often by as much as 10–20%.7,8 Photoconductivity9 and recombination-radiation10 measurements have been interpreted as giving evidence for the presence of localized states in the gap.