Hydrogen passivation and its effects on carrier trapping by dislocations in InP/GaAs heterostructures

Publisher Website
Google Scholar
Abstract
In previous work we reported on the stable passivation of dislocations in InP/GaAs heterostructures by plasma hydrogenation (Chatterjee et al., Appl. Phys. Lett. vol. 65, p. 58, 1994). In this article we investigate and compare the trapping kinetics and general trapping properties of dislocations in strain relaxed p‐InP grown on GaAs by metalorganic chemical vapor deposition prior to and after hydrogen passivation using deep level transient spectroscopy (DLTS) and current‐voltage‐temperature (I‐V/T) measurements to determine the complete role of hydrogen passivation in these heterostructures. Three hole traps, T1A, T1B, and T2, were detected and attributed to dislocations in heteroepitaxial p‐InP which displayed the logarithmic capture kinetics, extended dependence on fill pulse time, and broadened DLTS features expected for dislocation related traps. Quantitative analysis of the DLTS characteristics revealed progressive asymmetry in DLTS peak shape, an increase in characteristic peak width, and a decrease in activation energy as fill pulse time is increased until saturation values were reached. These observations are explained on the basis of a distribution or band of energy states for each trap resulting from the interaction of electrically active sites either between closely spaced dislocations or along dislocation cores within the strain‐relaxed InP. For fill pulse times increasing from 1 μs to 10 ms, activation energies for T1A decreased monotonically from 0.80 to 0.65 eV, for T1B from 0.56 to 0.45 eV, and for T2 from 0.45 to 0.35 eV, with saturation occurring at the upper and lower limits for each trap, which indicates a qualitative measure of the energy spread for each trap. Plasma hydrogenation was not only found to passivate dislocations by reducing the trap concentration from ∼6×1014 to ∼3×1012 cm−3 for a 2 h exposure, but also strikingly altered their basic trapping properties. The qualitative measure of energy spread for the T1A and T2 traps were narrowed from ∼100 to 150 meV to ∼20 to 30 meV after a 2 h hydrogen exposure, whereas T1B was not detected after passivation. In addition, a simultaneous reduction in fill pulse saturation time, DLTS peak broadening, and peak shift as a function of hydrogen exposure time were observed. These observations suggest that hydrogen passivation modifies the dislocation trapping characteristics toward a more point defectlike behavior due to an increase in the average spacing between electrically active dislocation sites. This in turn reduces the interactions between these sites and narrows the distribution of states within each defect band. Further, reverse bias I‐V/T measurements revealed that the near midgap trap T1A, which was found to dominate the space charge generation current prior to passivation, is no longer dominant after hydrogen passivation. Instead a 2 h hydrogen treatment shifted the dominant center to an activation energy which more closely matches the shallow T2 level.