Ultra-broadband photoconduction (UBPC) is used to measure the sub-gap trap density of amorphous indium gallium zinc oxide (a-IGZO) thin film transistors (TFTs) across the bandgap, from the valence band mobility edge to within 0.15 eV (~6kBT) of the conduction band mobility edge. Figure (a) is a plot of the total integrated trap density (Ntraps, green) and the sub-gap density of states (dNtraps/dE, blue) as measured by UBPC for traps located just below the conduction band mobility edge (E-EC = 0). Three Gaussian sub-gap states are observed with peak energies of 0.9, 0.7, and 0.4 eV below the conduction band mobility edge. Note that the width of the 0.4 eV peak (180 meV) is much larger than that of the other two deeper peaks (45 meV). All three of these peaks are likely associated with oxygen vacancies, but it is unclear why the 0.4 eV trap is so much broader than all the other deeper oxygen vacancy derived donor trap states.The black dotted curves included in Figure (b) are experimental EKV mobility (μEKV) versus overvoltage (VG-VON) plots obtained at three different temperatures (0 °, 20 °, and 100 °C). The red dashed curves are simulated mobility curves obtained assuming that only conduction band tail states act as electron traps to degrade the mobility. Since these red simulated curves grossly overestimate the experimental mobility, other traps must be operative in establishing the black experimental mobility curves. In fact, the inclusion of the 0.4 eV trap measured by UBPC in addition to the conduction band tail states in the simulation (blue curves) leads to excellent agreement between experimental and simulated mobility for all three temperatures (0 °, 20 °, and 100 °C). In summary, it is demonstrated how the mobility of a-IGZO can be simulated by including the experimentally observed DoS of a 0.4 eV centered Gaussian trap state.
Read full abstract