The work reported here is an experimental and theoretical investigation of high-frequency electrical noise generated in extrinsic single-crystal n-type germanium at high electric fields. The electric field was pulsed so that the lattice temperature remained near 77°K. During the pulse, the electrons quickly reach a non-equilibrium steady-state due to their gaining energy from the electric field and brought to a steady-state by means of collisions with the lattice vibrations. Previous work has been concerned with noise measurements made at right angles to the electric field direction, where anisotropic behaviour was observed. The present measurements, made in the direction of the electric field, also show a high degree of anisotropy. The electrical noise generated is described by the noise temperature, Tn, obtained by adapting the Nyquist formula to the non-equilibrium case. Measurements of Tn, performed at frequencies of 70Mc/s and 30Mc/s, indicated a uniform noise spectrum in this frequency range for all the samples used. The anisotropy of Tn suggested that Tn was explainable on the basis of the many-valley model of the conduction band of germanium, established by previous experimental investigations of the high-field mobility anisotropy. A feature of the many-valley model is that electrons in different valleys of the conduction band, will in general, exhibit different transport behaviour and as a result, transitions between these valleys will result in a noise phenomenon described as "intervalley noise". However, for measurements of Tn in the <100> direction, the "intervalley noise" will vanish, allowing a direct measure of the electron "heating" due to the electric field, the “hot electron noise”. In the <111> an <110> directions, both intervalley and hot electron noise are expected. Both contributions to Tn are evaluated by means of Barrie's extension to the case of many-valley germanium of Stratton's high-field transport theory. / Science, Faculty of / Physics and Astronomy, Department of / Graduate

Tyler used Hall coefficient measurements to evaluate the gold acceptor concentration in germanium crystallized from a germanium-gold melt at a temperature between the melting point of pure germanium and 17°K below this temperature. The present study reports the results of similar measurements in the temperature range of growth from 12° to 80°K below the melting point of germanium. A crystal grower was designed and constructed for growing the monocrystals of gold-doped germanium. It was found that the <100> direction was the most favorable direction for eliminating dendritic growth on gold-rich phase crystallized as inclusions. The phenomenon of retrograde solubility was observed. The maximum solubility (~2.8 x 10¹⁶ gold acceptor atoms per c.c.) occurred at about 30°K below the melting point of pure germanium. Infra-red absorption in gold-doped germanium was studied between 4.2° and 298°K over the energy range from 0.08 to 0.6 e.v. The magnitudes of the absorption cross-sections for the specimens containing about 3.5 x l0¹⁵ and 1.2 x 10¹⁵ gold acceptor atoms per c.c. were in close agreement with the data of Johnson and Levinstein. The cross-section of absorption near the band gap of germanium for the heavily gold-doped germanium ( ≻ 10¹⁶ gold acceptor atoms per c.c.) is about the same as in the case of lightly doped specimens. Whereas in the case of lightly doped specimens, however, the cross-section of absorption drops sharply to negligible values as the energy of the incident photon decreases, the cross-section of absorption for the heavily doped specimen remains very high. The excess absorption is found to remain substantially the same when measurements are extended to the energy range between 0.032 and 0.044 e.v. At 298°K, the absorption of infra-red radiation in gold-doped germanium shows spectral structure associated with valence band intra-band transitions. For the lightly doped specimens, with increasing concentration, the structure in the spectrum becomes less pronounced. Newman and Tyler have reported studies of the same effect in gallium-doped germanium. Our studies indicate that the effect is more pronounced in gold-doped germanium than in gallium-doped germanium. When intra-band absorption is peeled off from the absorption spectra at 195° and 298°K, the threshold of absorption is at a higher energy for the higher temperatures. The transmission of infra-red light through both pure and lightly gold-doped germanium is found to be enhanced by about 16 per cent at 4.2°K compared to that at higher temperatures (≻60° K). The determination of surface reflectivity at 4.2°K gives 0.29 and correspondingly a refractive index (μ) of 3.3 This shows an anomalous behaviour of {equation omitted} at low temperatures. It is known that materials with the diamond structure reveal similar extraordinary behaviour in other lattice properties such as the temperature dependence of their coefficient of thermal expansion and Debye temperatures. / Science, Faculty of / Physics and Astronomy, Department of / Graduate

Measurements have been made on a variety of germanium and silicon junction diodes in order to determine departures from the ideal behavior both under illuminated and dark conditions. The diodes included alloyed, grown and diffused junction types and the experimental study was largely confined to forward voltage. Carriers injected by a forward voltage into the bulk regions of a p-n junction in Ge or Si reduce the resistances of these regions, A model of a photodiode consisting of an ideal diode (Shockley, 1949) in series with this carrier modulated resistance produces good agreement with the experimentally observed dark chs.; in one Ge alloy diode the agreement was exact up to at least 0,7V forward voltage. For large forward currents (density of injected carriers comparable to injected carrier density) the Shockley expression relating I and V is no longer valid nor consequently is the conductivity modulation theory. This theory, making use of the experimental dark chs, and the experimental dynamic capacity measurements, is able to deduce many of the parameters of the diode material e.g. Ƭ𝞺, Pո,Ψo. The chs. of the diode dark and illuminated have been found to intersect at large forward currents. Theoretical investigation of the condition for crossover in various models of the diode is made. The physical meaning of this condition is that at some forward voltage the resistance of the diode is sufficiently reduced on illumination to offset the effect of the opposed internal photo-e.m.f.. For an ideal diode the relation between the short circuit photo-current and open-circuit photo-e.m.f. (photochs.) should simply be the same as the dark forward chs., except for a reversal of sign of the current. The fact that the experimental photochs falls below the ideal photochs is attributed to internal resistance in the diode and an expression is developed from this assumption; the theoretical expression is largely consistent with experiment. Several other models are examined for the non-ideal behavior of junction diodes which attempt to take into account the possible effects of the resistance of the bulk material and of the electrode-semiconductor contact; the implications of non-linear recombination are also investgated. It is found that none of these alternative models is as successful as the conductivity-modulation scheme in explaining the dark, the illuminated and the photocharacteristics. / Science, Faculty of / Physics and Astronomy, Department of / Graduate

The d-c. and a-c. characteristics of p-n-p diodes were measured and interpreted on the basis of present theories. The equivalent parallel capacitance of these diodes was found to decrease as (f)⁻³/₂ at high frequencies in accordance with diffusion theory; from these measurements the lifetime of a hole in the base was estimated to be about 10 microseconds. The current fluctuations in the diode were measured by comparing them with a standard noise diode, and the results represented as a shunt noise current generator. From the frequency spectrum of the noise it was deduced that it consisted mainly of excess or 1/f noise except at low currents and high frequencies, and here the results were interpreted as a combination of shot and thermal noise. The deduced hot carrier temperature was compared with the theoretical values of Shockley and poor agreement was obtained, indicating inadequacy of the present theory. / Science, Faculty of / Physics and Astronomy, Department of / Graduate

Germanium space-charge regions have been studied under conditions of double-depletion achieved by applying a reverse bias to both junctions of a transistor structure. The generation and distribution of carriers between the two junctions are described in terms of models for the thermal generation of carrier pairs and for potential distribution in a cylindrical system. Measurements over a wide range of temperature reveal that generation of carriers occurs through the medium of one or more sets of recombination centers and not by direct transition between the valence and conduction bands even at elevated temperatures. The Shockley-Read recombination-generation theory is applied to obtain the activation energies associated with the recombination centers. The impurity density, base width and junction areas are estimated from measurements of punchthrough voltage and junction capacitance. For some specimens the capacitance measurements made with one junction floating confirm the sharp increase in capacitance at punchthrough noted by Barker. The distribution of current between the junctions when both are equally reverse biased is found to be roughly proportional to their areas. It is also shown that control of the reverse current across one junction may be achieved by variation of the reverse-bias on the other junction. The mechanism of this interaction is considered in terms of the diffusion of carriers between the two space-charge regions. The phenomen of slightly non-saturating reverse current is explained in terms of the expansion with reverse bias of the space-charge region within the base. An expression relating the base current and reverse bias in a cylindrical system shown to give good agreement with experiment for one specimen. For the units in which the increase in current with voltage is appreciable, space-charge expansion cannot account for the increase and two other mechanisms are considered: avalanche multiplication in the bulk and along the surface, and a high generation rate on the surface. The surface conduction channels on the base region are investigated in two ways. Maximum floating junction potentials are calculated from measured values of the common emitter amplification factor (using the Shockley 1949 theory) and these are compared with directly measured potentials. A second method involves direct measurement of a.c. surface conductance between collector and emitter when both junctions are reverse-biased to prevent bulk conduction. Both tests reveal very small degrees of surface conductance on all specimens. / Science, Faculty of / Physics and Astronomy, Department of / Graduate

Platinum pentafluoride has been prepared, by a new preparative method, in a more stable, pure and crystalline form than the material already described. It has been shown, by x-ray powder photography, to be isomorphous with other noble metal pentafluorides and almost isodimensional with rhodium pentafluoride. A tetrameric structural unit like that observed in ruthenium pentafluoride is also assumed for platinum pentafluoride. Its magnetic properties have been shown to be representative of a third transition series d⁵ ion in a distorted octahedral environment. The nature of the bonding in germanium difluoride is discussed in the light of the crystal structure, which has been deduced from data obtained from single crystals prepared in this work. The products of interaction of chlorine or bromine with the difluoride are consistent with the structural findings. The ¹⁹F n.m.r. spectra of the mixed chlorofluorides (GeFCl₃, GeF₂Cl₂, GeF₃CI, GeF₄ ) and bromofluorides, indicate that intermolecular exchange between these compounds must be extremely slow, at least in the absence of a catalyst. It has been shown that germanium difluoride is so strong a reducing agent that it reduces iodine pentafluoride to iodine below room temperature. The powerful reducing properties of the difluoride were also illustrated by the reduction of platinum tetrafluoride to the metal at room temperature. Attempts to reduce tungsten hexafluoride at 300° led to the formation of Ge₂WF₈. In glass the germanium difluoride reacted with tungsten hexafluoride and the container to yield WO₂F. / Science, Faculty of / Chemistry, Department of / Graduate

No description available.

No description available.

No description available.

Afin d’adresser la problématique liée aux limitations des interconnections métalliques en termes de débits notamment, la photonique Si s’est imposée comme une technologie de choix. Un des composants de base des circuits photonique Si est le photodétecteur : Il permet de convertir un signal optique en signal électrique. Les photodétecteurs à base de Ge sur Si ont montré leur potentiel et offrent la meilleure alternative aux photodétecteurs III-V, pour une intégration dans les circuits photoniques Si.Dans ce contexte, les photodiodes à base de Ge su Si ont été étudiées. L’optimisation des photodiodes p-i-n a permis l’obtention de résultats à l’état de l’art. Une nouvelle approche utilisant une double hétéro-jonction latérale Si/Ge/Si a été proposée afin d’augmenter la responsivité mais aussi afin de proposer une meilleure solution d’intégration, avec les modulateurs Si notamment. Pour augmenter encore la sensibilité des récepteurs, l’utilisation de photodiodes à avalanche est cependant nécessaire. La structure SACM (Separate Absorption Charge Multiplication), combinant le faible bruit de multiplication du Si et l’absorption du Ge aux longueurs d’onde télécom, a d’abord été étudiée. Des modèles ont été développées afin d’optimiser le fonctionnement, et ces photodiodes ont été fabriquées et caractérisées. Les résultats obtenus sur des photodiodes éclairées par la surface (produit Gain-Bande passante de 560GHz à seulement -11V) sont très encourageant pour une intégration avec un guide d’onde. D’autre part, les photodiodes p-i-n en Ge sur Si, ont été étudiées en avalanche. La faible largeur de la zone intrinsèque a permis de diminuer le bruit de multiplication par effet « dead space », et le fonctionnement à 10Gbits/s pour un gain de 20 et une puissance optique de seulement -26dBm, pour une tension de -7V, sans utilisation d’amplificateur (TIA), a pu être démontré. Ces développements ouvrent ainsi la voie vers des récepteurs rapides, à faible consommation électrique et grande sensibilité. / To address the issue related to the limitations of metallic interconnects especially in terms of bitrate, Si photonics has become the technology of choice. One of the basic components of photonic circuits is the photodetector: It allows to convert an optical signal into an electrical signal. Photodetectors based on Ge on Si have shown their potential and offer the best alternative to III-V photodetectors, for integration into Si photonic circuits. In this context, the Ge on Si photodiodes have been studied. The optimization of pin photodiodes enabled the achievement of state of the art results. A new approach using a double lateral Si/Ge/Si heterojunction was proposed to increase the responsivity but also to provide a better integration solution, especially with Si modulators. To further increase the sensitivity of the receivers, the use of avalanche photodiodes, is however necessary. SACM (Separate Absorption Charge Multiplication) structure, combining Si low multiplication noise and Ge absorption at telecom wavelengths was first studied. Models have been developed to optimize the devices, and the photodiodes have been fabricated and characterized. The results obtained on the surface illuminated photodiodes (Gain-bandwidth product of 560GHz only -11V) are very encouraging for waveguide integration. On the other hand, Ge on Si pin photodiodes have been studied in avalanche. The small width of the intrinsic region contributed to the multiplication noise reduction thanks to "dead space" effect, and operation at 10Gbps for a gain of 20 and an optical power of -26dBm at only-7V, without using amplifier (TIA), have been demonstrated. These developments open the way to fast, low power consumption and high sensitivity receivers.