Search

A theoretical analysis is made of the field distribution near the surface of p-n-p structures with double positive edge geometry. This geometry offers in principle the possibility of avoiding the limitations and disadvantages inherent in the use of negative bevel angles. The results show that the reduction of the maximum field at the surface is not as easy as for the case of a simple positive bevel angle and that consequently the passivation of the surface may present more problems. Nevertheless, it is demonstrated that this geometry offers a number of great advantages and presents a real alternative for use in future high-voltage devices. A number of devices, dimensioned for a breakdown voltage of 6 kV, were made with this edge geometry. The measurements show that the reverse current corresponds to the thermally generated current in the space-charge layer and that no significant current flows at the surface. The breakdown voltage corresponds to the theoretical breakdown voltage in the bulk, unlike in the case of double negative beveling. The advantages of this geometry with respect to the conventional negative bevel angle are no significant reduction of active area, no dependence on the doping profile, and no limitation in voltage. A disadvantage, however, is the presence of higher fields at the surface.

Under electrical load the crystal temperature inside a semiconductor power device changes substantially both in space and in time. Its measurement with good resolution in temperature, in time as well as in space, was attempted by detecting the thermal radiation emitted from an operational semiconductor device that was thick enough to be opaque to thermal infrared. A cooled InSb detector is followed by an amplifier allowing for a tradeoff between time resolution (the overall time constant varied from 0.5 to 86 µs) and temperature resolution, such that their product remained constant. For temperature fields periodic in time, sampling methods and signal averaging are applied which further improve temperature resolution with time and space resolution remaining unchanged. A large improvement in speed of response with respect to conventional methods results. For a fixed surface point the time dependence of temperature is obtained. Fixing the observation window in time leads to profiles of instantaneous temperature across a selected line of the exposed semiconductor which are in good agreement with those predicted by a recent theoretical model. In Part I the case of thick slices is treated. The extension to thin slices which are transparent to thermal infrared are treated in Part II.

A model is developed for the dynamic operation of magnetic reed relays. From experimental data, it is found that the processes of dynamic magnetization and mechanical motion can be treated as separate and consecutive in time. Remarkable agreement was found between the experimentally determined static magnetic force and a force calculation based on an assumed domain structure inside the reed tips. The model basically assumes that the flux passes through the air gap in two areas near the tips of the reeds, each area equals half the cross-sectional area of the reed. The dynamic magnetization process is modeled with an improved version of Agarwal's [7] model for eddy-current losses in laminations, and a differential equation for dynamic reed motion is derived. The very good agreement between theoretical and experimental results shows that this model gives a good representation of the dynamics of the system. Furthermore, the findings of the static force model based on domain concepts have basic application to the knowledge of magnetics in addition to the application here to reed relays.

A new method for investigating the field distribution near the surface of p-n junctions is presented. The results show that for negatively beveled junctions, an absolute field maximum exists underneath the surface. This field maximum rather than the field on the surface determines the breakdown voltage of the device. The dependency of the maximum field on various device parameters as base resistivity, bevel angle, and steepness of the doping profile is analyzed. In addition, the influence of surface charges and dielectrics is investigated. Finally, the advantages of positive bevel angles with respect to negative ones are demonstrated. The theoretical results were verified by potential probing measurements on the surface of high-voltage devices.

n/a

Analytical solutions for the forward characteristic of thyristors have been limited to abrupt and constant doping profiles. Average values for the mobilities and the carrier lifetime had to be assumed for each region of the device. Results will be presented here which are based on an exact numerical solution of the transport, continuity, and Poisson equations for the one-dimensional thyristor. Doping, mobility, and lifetime can be varied from point to point. Thyristor structures are much wider and higher doped than the devices treated numerically in the literature. The dependency of the forward characteristic on various device parameters was examined, and the important results were verified experimentally. For high current and a base width that is large compared to the diffusion length, the dependency of mobility on the carrier concentration leads to a base voltage drop significantly lower than that expected from analytical theory. Furthermore, it is shown that for this case the parameters of the highly doped side regions (doping, width, and lifetime) have much less influence than predicted from Herlet's [3] theory.

Experience with beveled p-n junctions does not fully agree with the predictions of Davies and Gentry. Specifically no optimum negative angle as claimed by Huth and Davies was found. A new method for determining the form of the space charge region near the surface shall be presented which avoids the insufficiencies of the model cited. Apart from more accurate results, this paper gives insight into an entirely new phenomenon. In contrast to conventional theory p-n junctions with small negative bevel angles exhibit an absolute electric field maximum underneath the surface. This effect causes a reduction of the breakdown voltage even when the electric field on the surface is sufficiently small. Results will be presented which show the dependency of the absolute field maximum on bevel angle, doping profile and material resistivity. It will be shown that a limit for the use of negative bevel angles exists at about 4 kV breakdown voltage. These theoretical predictions are in excellent agreement with potential probing experiments. Results will also be presented on the influence of dielectrics or charges on the surface.

Infrared microscopy as a means for observing dynamic thermal processes inside a slice out from a silicon power device has been reported by Burtsev et. al. and refined by Jaecklin and Marek. Since the local emissivity depends on doping, instantaneous free carrier concentration and the optics used, interpretation of these radiation measurements is extremely difficult. In the existing literature it has been based on homogeneous emissivity throughout the sample.

A new investigation of the field distribution at and near the surface of p-n junctions with negative bevel angles has shown that an absolute field maximum exists 20?40 ?m below the surface. This field maximum is reduced with smaller bevel angles, but is stronger for structures with a higher breakdown voltage in the bulk.

An analytical approximation for space-charge layers at abrupt p--n junctions under high-injection conditions is presented. In this approximation, an exponential decay of the charge on both sides of the junction is assumed. Predictions from this model are found to be in good agreement with the results of exact numerical solutions.