Flicker (1/F) Noise in Copper Films due to Radiation - Defects.

Changes were measured in the 1/f noise and resistivity in polycrystalline Cu films due to defects induced by 500 keV-1.1 MeV electron irradiation or 1 MeV Kr^+ ion irradiation. The Cu films were irradiated while maintained at 90 K on a cold stage of an electron microscope, and all noise and resistivity measurements were made in-situ. Irradiation with 500 keV electrons increased the voltage noise level in the films by more than an order of magnitude, while the resistivity increased by at most 10%. When the films were annealed at progressively higher temperatures, both the 1/f noise and the resistivity were reduced; however, at lower annealing temperature, the fractional reduction in the induced noise was substantially more than in the added resistivity. These results suggest that a large fraction of the induced noise may be generated by certain "mobile" added defects that are more readily annealed than the majority of the added defects. The temperature dependence of the noise after irradiation and partial annealing indicated that the induced noise was thermally activated in a manner consistent with the Dutta-Dimon-Horn model. Isolated defects created by 1.1 MeV electron irradiation were found to produce substantially higher noise levels than clustered defects resulting from 1 MeV Kr+ irradiation. Isolated In and Be impurities in Cu are known to trap radiation -induced interstitial defects. Measurements on a number of films indicate that the induced 1/f noise is not sensitive to the type and quantity of interstitial traps. A simple "local interference" model is presented, which used calculations by Martin to estimate the 1/f noise magnitude generated by moving defects. This model can account for the induced noise, provided one assumes a sufficient fraction of the added defects to be mobile. Several simple models are examined concerning the identity of the defects responsible for the induced noise. A model which attributes the noise to the motion of vacancy-type defects close to surfaces, grain boundaries, or dislocations is most consistent with the experimental results and theoretical considerations, though none of the models considered provides an entirely satisfactory explanation of the induced 1/f noise.