Search for a command to run...
In those charge-coupled devices (CCD's) which have regions under each electrode which are substantially free of externally applied tangential electric fields, charge motion takes place by space charge assisted diffusion, and this relatively slow process represents a limitation to the operating frequency of CCD's with large plates. In this paper, subject to certain approximations, the equations of motion for CCD charge transfer have been derived, yielding a nonlinear diffusion equation. The solution of this equation by a stable finite difference scheme is described, and the solutions are applied to predicting the operating characteristics of CCD's. The results in synopsis are: (i) The n-channel devices will have lower losses at a given frequency than the p-channel devices, and a higher upper frequency limit. (ii) Higher amplitude signals (more charge) yield lower losses. (iii) Losses can be reduced an order of magnitude by using zero's which carry a substantial amount of charge. For example, with 2-MHz clocking of a 25-μm plate (pad), p-channel 2-phase device, a 2-volt one is diminished by 4 percent per transfer with empty zero's, but with 1-volt zero's, the diminution is 0.26 percent per transfer. (iv) Reasonably efficient CCD operation should be possible up to the 50-MHz range using contemporary design tolerances. (v) Diffusion is important for reaching high transfer efficiencies. The frequency limitations described in this paper can be overcome by using a structure in which the distance between the electrodes and transferring charge is comparable to the electrode width and spacing.
Published in: Bell System Technical Journal
Volume 50, Issue 6, pp. 1721-1740