LOW-FIELD TRAP GENERATION DEPENDENCE ON THE INJECTION CURRENT-DENSITYIN GATE INSULATORS - HOW VALID ARE ACCELERATED HOT-ELECTRON MEASUREMENTS

''Continuous'' low gate insulator field (1 MV/cm) electron injection i n insulated gate field effect transistors using a pulsed injection tec hnique (PIT) was conducted in the dose range 10(13) to 10(17) e/cm(2) over the range of current densities A/cm(2). PIT enables independent c ontrol of insulator fields and injection current densities, while not causing optically induced shallow trap depopulation. As is generally t he case, the threshold voltage shift, Delta V-D Varies monotonically w ith dose, and can be modeled using a defect generation power law requi ring fewer adjustable parameters than is necessary using a first order trapping moael. It was also found that for a given dose the injection current density has a profound effect on the observed magnitude of tr ap generation. Previously, the total dose and insulator field were tho ught to be the only determining factors in trap generation. Based on t hese results, it appears that when the intrinsic defect concentration( s) is to be determined, a very low current density (injection rate) sh ould be used to minimize trap generation effects which would confuse t he issue. It is reported also for the first time that the expected in- use lifetime df the devices, calculated from the injection data, also exhibits a power law dependency on the injection current density. Thes e results raise serious questions about the validity of aggressive inj ection techniques (such as avalanche injection arid Fowler-Nordheim ap proaches), of the existence of ultrasmall cross section electron traps based on such methods, and of aggressive accelerated aging conclusion s, based on the extrapolation of high injection current density data t o end-of-life threshold voltage shifts.