The effects of high temperature annealing on the conduction and memory properties of MNOS memory devices was studied. The results showed that the conductivity of the nitride increased with annealing time and temperature and that the memory retention qualities of the device were degraded. These results can be explained by the hypothesis that the heat-treatment has increased the number of traps in the silicon nitride. The correlation of the theoretical and experimental results indicates that the traps are uniformly distributed through the silicon nitride layer and are not localized at the oxide nitride interface.Annealing as one step in the fabrication process for producing MNOS memory devices has been used by many workers in this field (1-4). However, no detailed investigation into the effects of this step on the properties of the devices was undertaken. In this study, all samples used were fabricated in an identical manner and then subjected to annealing treatments for various times from 5 to 120 rain and at various temperatures in the range 800~176After the devices were completed, the conduction and memory properties were analyzed in order to provide information about the effects of the annealing on certain parameters which characterize the silicon nitride film.
Fabrication ProcedureAll devices tested were fabricated on phosphorousdoped (111) polished silicon wafers of 0.5-0.88 ohmcm resistivity. The silicon wafers were degreased using 3 min soaks in hot trichloroethelene, acetone, and methanol; followed by a rinse in deionized water down to 10 megohm-cm; a 1 min etch in buffered HF; and a second water rinse. The silicon surfaces were further cleaned using a 5 rain soak in a hot solution of 5 parts deionized water, 1 part ammonium hydroxide, and 3 parts of hydrogen peroxide; a rinse in deionized water; a 5 min soak in a hot solution of 5 parts deionized water, 1 part hydrochloric acid, and 2 parts of hydrogen peroxide; and a final rinse in deionized water.Ellipsometric measurements showed that after the cleaning procedure approximately 10A of silicon dioxide was present on the silicon surface, as has been reported by others (5, 6). Since the primary purpose of this work was to investigate the silicon nitride layer, no additional oxide was grown.Immediately after being cleaned, 750-1000A of silicon nitride was pyrolytically deposited at 700~ on the wafers using silane and ammonia with hydrogen as a carrier gas. The silane and ammonia were in the ratio of 1 to 100 and were purchased from Matherson under the trade name N-gas.Following the silicon nitride deposition, the samples were annealed in dry nitrogen at either 800 ~ , 900 ~ , 950 ~ or 1000~ for various lengths of time. For each batch of samples fabricated, one wafer was left as a control to which the experimental data could be compared. Table I lists the various samples along with their annealing temperatures and times.Next, aluminum was evaporated on top the Si3N4 and photolithography was used to outline circular electrodes 35 mils in diameter giving an are...