Computer modelling and experimental characterisation of amorphous semiconductor thin films and devices

  • Wei Gao

    Student thesis: Doctoral Thesis


    Computer source codes for modelling amorphous semiconductors have been developed during the course of the study. The major program SPIN.F deals with steady state electronic processes for amorphous semiconductor materials and devices under different optical excitation and applied voltage conditions. It uses both standard one-electron state statistics and statistics developed for divalent states to deal with trapping and recombination processes via band tails and defect states. The latest defect pool model has been included into the model as well as other standard defect models.

    Chapter 2 describes the mathematical basis of the SPIN.F modelling algorithms. A Gummel approach is used to solve Poisson's equation and the carrier continuity equations, modified to deal with special conditions in disordered materials, such as trapped space charge. Evaluation of the SPIN.F program accuracy and speed of computation for p-i-n diode structures is described in chapter 3. This testing gives an overview of the capability of the computer program and will be helpful for further applications on amorphous semiconductors.

    A detailed description of the defect pool model and its incorporation into the numerical model is given in chapter 4. The original model formulation and parameters have been proved unsuitable when applied to model p-i-n diode performance. The reason for this is that the sensitivity of the dangling bond density on the Fermi-level position is overestimated giving an excessively high defect density near the doped layers, with the result that the modelled performance of the p-i-n diode is much poorer than for actual devices. A modification to the defect pool expression, changing the effective number of hydrogen atoms involved in defect formation kinetics is shown to give a substantial improvement. When comparing various defect models' fit to p-i-n diode I-V characteristics the best fit is now found for the modified defect pool model.

    Computer simulation and error testing of the Dynamic Inner Collection Efficiency (DICE) method and the associated SVD solution algorithm for probing internal processes in solar cells is described in chapter 5. The DICE method is proved still to be a useful approach to investigate the inner characteristics or processes in semiconductor devices if some data processing treatments can reduce effects of measurement error. Otherwise it is shown that the random error introduced from spectral response measurement seriously affects the resolution of the final results. By minimising the error from the measurement, the accuracy of the final DICE profile may be kept within several percent. However, the work shows that only broad low resolution features may be revealed in practice by the DICE method.

    A detailed study of a "photogating" effect has been carried out and described in chapter 6. A quantum efficiency (QE) much larger than unity has been found both theoretically and experimentally, under reverse bias voltage conditions for the first time. Experimental results show the QE increasing with reverse bias voltage, achieving values over 100 at high reverse bias voltage. From computer simulations, the underlying reason for the photogating effect has been clearly demonstrated, and a good fit has been achieved with experiment. The combination of strongly absorbed blue bias light from the p-side and green probe light from the n-side under high reverse bias gives optimum QE gain, and raises the possibility of an optically controlled amplifier which may be useful in photodetection. It is also shown by modelling that the QE is a sharply peaked function of defect density.

    By using numerical modelling, a recently debated question regarding the carrier type controlling collection in a solar cell has been solved. During charge collection measurement, under reverse bias voltage, or short circuit condition, it is the carrier which has the shorter drift length which controls the behaviour of the device. A detailed analysis on this issue is described in chapter 6, showing a clear picture of the transport processes occurring during the measurement.
    It is demonstrated in chapter 6 that the one dimensional computer model developed in this work can be applied to two or three dimensional situations under certain symmetry conditions. The method is applied to investigate the effects of surface states in coplanar photoconductivity in n-type a-Si:H. Although the model does not fully explain the experimental results, it demonstrates the consistency of this model with similar models proposed by other authors, and proves the validity of the approach.
    Date of AwardSep 1995
    Original languageEnglish

    Cite this