AbstractMuch of the work presented within this thesis is concerned with the development of computer programs capable of simulating the transient current pulses that would be expected for amorphous materials possessing known density of states distributions. Two simulation methods are used, the Monte Carlo technique and the finite difference method. While the Monte Carlo method makes few basic assumptions, the transients produced are noisy and the computer execution times are long. The newly developed finite difference method circumvents both of these problems. Comparison of the finite difference results with those obtained from Monte Carlo simulations shows that the transients produced are essentially the same. The finite difference method can therefore be used to produce accurate, noise free simulated transients in relatively short times.
The simulation programmes are used to investigate the accuracy of three methods proposed as techniques for obtaining the functional form of the density of states from 'time of flight' current transients. The three analysis methods used are the Marshall - Allen technique, the 'TROK' method and the integral technique. The last of these was developed during the course of this work. These three methods were each used to analyse the transients that would be produced by an amorphous semiconductor possessing either an exponential or linear distribution of tail states. These two distributions were chosen as they are both currently proposed in the literature as possible candidates for the density of states within amorphous silicon.
Examination of the results produced when each of these analytical methods was applied to the same transients revealed major differences in the calculated results. The TROK method could not distinguish between the two different types of tail state distribution, and interpreted linear tail simulated data in terms of an exponential distribution of characteristic temperature 312K, a figure very close to that frequently quoted in the literature. The Marshall - Allen technique could differentiate between the two distributions but could not reproduce the correct functional forms. The integral technique distinguished between the two distributions and also gave them their correct functional form.
To complement the computer simulation work above, an experimental computerized 'time of flight' measurement system was developed and constructed by the author. This system was used to obtain transients over a wide range of field and temperature, as is required by the integral technique. Unfortunately, due to the nature of the sample used, only a small energy range within the tail states could be studied. When analysed, two functional forms for the distribution proved to be possible; these were an exponential tail (T - 93K) and a linear tail (ΔE - 0.165eV). The exponential tail would, however, not be consistent with the observation of anomalously dispersive transients above 100K so it was rejected as an option.
|Date of Award||May 1989|