Worst-case Results Research Articles

A measurement technique of time-dependent dielectric breakdown in MOS capacitors

The statistical nature of time-dependent dielectric breakdown characteristics in MOS capacitors was evidenced by testing large numbers of capacitors fabricated on single wafers. A multi-point probe and automatic electronic visual display technique are introduced that will yield statistical results which are necessary for the investigation of temperature, electric field, thermal annealing, and radiation effects in the breakdown characteristics, and an interpretation of the physical mechanisms involved. It is shown that capacitors of area greater than 2 × 10 −3 cm 2 may yield worst-case results, and that a multi-point probe of capacitors of smaller sizes can be used to obtain a profile of non-uniformities in the SiO 2 films.

Transistor Blocking Oscillator Analysis

Two commonly used basic transistor blocking oscillator circuits are considered: 1) a common-emitter with collector-to-base transformer coupling, and 2) a common-base circuit with collector-to-emitter transformer ccupling. The pulse widths of the blocking oscillators are assumed to be dependent on the transistor and circuit parameters rather than external timing devices, i.e., delay lines. Also the transformers used in the circuits are assumed to operate linearly and exhibit no saturation effects. A review of the basic principles of operation is given and simplified equivalent circuits are included. A rise time equation is given in terms of the circuit parameters for the common-base blocking oscillator. The deterioration of the rise time with loading is shown in the equations. Pulse width equations, suitable for design purposes, are given for each type of circuit. A linear worst-case analysis shows the response of the common-base circuit to a step voltage trigger. This linear trigger analysis is an approximation due to the inherent nonlinearities associated with the transistor during the turn-on interval. However these worst-case results compare favorably with experimental rise time results. A trigger pulse of sufficient amplitude and a pulse duration of 0.05 <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">\mu</tex> sec simulates the voltage step function adequately since the circuit attains a regenerative mode during that period of time. The pulse width stability of the circuits is emphasized, and the common-base circuit is shown to have a better pulse width stability than the common-emitter circuit. A pulse width stability equation is obtained by considering partial derivatives of the pulse width equation.