|
Results:
A time sequence series of TEM images during electromigration are shown in Figs.1. The swelled part of Cu nanowire (~6nm width), which is indicated as the line labeled P in Fig 1 a, could be a cluster of Cu atoms, The swelled part moved toward the anode electrode as shown in Figs. 1 a to d. This motion is one of the typical behaviour of the Cu nanowire electromigration, where the Cu cluster seems to flow for anode. On the anode electrode, the nanowire seems to be melt and diffuse into anode electrode in Figs. 1 a to d. At Figs.1 d, the swelled part is accumulated on the anode electrode as a ~3 nm extrusion. At Figs.2 c and d the hourglass part appears around the left side of Cu cluster and it becomes thinner, which confirms that the swelled part is a cluster of Cu atoms. This tinning of the hourglass where the Cu atoms removed is also a typical behaviour of electromigration, while this hourglass could be considered as a void at nano-scale. From Fig.1 a to Fig.1 e, time is ~1.4 seconds.
The stresses measured simultaneously with TEM observation are shown in Fig.2, at which a ~ d are corresponding to the timings of images shown in Figs. 1 a ~ d, respectively. From the start (time of 0) to the time point b in Fig.2 the stress shows some constant negative values and the stress sharply decreases with ~1 nN step. Since it is reported that based on the molecular dynamics simulation a few nano-Newton stress causes the slips of the single crystal Cu (111) interface, 1 nN may cause a slip of Cu crystal. The sharp decrease of the stress could be associated with the movement of a Cu atom cluster at one time due to the process of slipping of a Cu crystal plane in the wire rather than a single Cu atom moves one by one, although the crystal plane is not visible at this moment by TEM pictures.
|
|