We present electromigration experiments on single-crystalline silver nanowires. The wires were grown on 4° vicinal silicon (100) substrates by self-organization and were contacted by electron beam lithography. The electromigration experiments were performed in situ in a scanning electron microscope at room temperature with constant dc conditions. In contrast to other experiments we observe void formation at the anode side of the wires. If the current is reversed, the electromigration behavior is also reversed.

It is well known that the Ag/Ag(111) epitaxial system grows three dimensionally because of the existence of a relatively high excess diffusion barrier, DeltaE(s) = 0.13 eV, at the step edges. Several experimental methods have been developed to measure the step edge barrier in this system over a wide coverage range. The probability for an atom to move from a higher to a lower layer depends on both the barrier and the prefactor, so it is important to test whether the prefactors for hopping over a step, nu(s), and for hopping on a terrace, nu(t), are different. We present the results from several experiments on Ag/Ag(111) to conclude that nu(s)/nu(t) = 10(2.0+/-0. 3).

No abstract

Roos and Tringides Reply:There are three questions raised in [1] about the interlayer probability p ͑n s ͞ n t e 2͑DEs͞kT͒ ͒ for Ag͞Ag(111): (i) Is the prefactor ratio n s ͞n t ¿ 1 as suggested in [2 -4]? (ii) What is the correct theory to describe second layer nucleation? (iii) Why does the revised theory [5,6], when applied to Ref.[3], give unphysically high values for n s ͞n t and DE s ? Since Ref.[2] deals with question (i), our reply focuses on n s ͞n t ¿ 1. We have expressed the main conclusion based on second layer nucleation experiments of Ref.[2] as n s ͞n t ¿ 1 since the result n s ͞n t ¿ 1 does not depend on the theory but a specific value does. A specific value n s ͞n t 100 is deduced by considering two additional experiments [2]. Although the revised theory [5] gives a different value, it reinforces the conclusion n s ͞n t ¿ 1. Question (ii) was discussed in the criticism [5,6] of the earlier theory [7]. The third question is still an open challenge.The conclusion n s ͞n t ¿ 1 for Ag͞Ag (111) is supported by three independent experiments: the analysis of second layer nucleation experiments [2 -4], the decay of the reflection high-energy electron diffraction (RHEED) intensity, and the size of the denuded zone [2]. Furthermore, there are three methods of analyzing the second layer nucleation experiments. Reference [1] discusses only one of the three methods, a small part of [2]. This method is the least robust because it uses only part of the experimental information; however, we will show that, even with this method, n s ͞n t ¿ 1 holds.In the original second layer nucleation experiment [3], fits to the fractional occupation curves have shown that n s ͞n t 50, but with large uncertainty. Later, a method was proposed [4] that allows the extraction of n s ͞n t and DE s uniquely. Since this method is far more exact than Eq. (1) of Ref.[2] (criticized in [1]), we review the method. From the expression of the nucleation rate V ͑p 2 ͞2͒F 2 R 5 ͞pD t [5] and the limiting values V max (when all the islands in the ensemble have second layer islands) and V min (when none of the islands in the ensemble have second layer occupation) in the experiment of [3], the dependence of the corresponding island sizes R max , R min vs p is plotted in Fig. 1. By using the measured values of (R min , R max ) [(2 nm, 4 nm) at T 120 K and (3 nm, 7.5 nm) at T 130 K [3] ], we search for the values of p with the best agreement between measured and calculated values n s ͞n t 10 9 , DE s 0.32 eV for the theory of [5] and n s ͞n t 10 3 , DE s 0.13 eV for the theory of [7].A third approximate method in Ref.[2] is based only on R min 3 nm at T 130 K by estimating the number of successful hops f over the step edge barrier, within the time between atom depositions t 1͞FA. The number of step interrogations as pointed out in [1] is 2tD t ͞R min which implies f 2tpD t ͞R min . In Ref.[2] (and in [1]), f . 1 was taken as the condition for a deposited atom to descend the R 3 nm island. However, the experiment was carried out on an ensemble of i...

Photoemission electron microscopy is used to study the thermal decay of Ag islands grown epitaxially on Si(001) surfaces. (2 x 3) Ag reconstructed zones, due to migrating Ag atoms supplied to the surface by the decaying islands, surround each of the islands. The shape of these reconstructed zones depends on the degree of diffusion isotropy in the system. We demonstrate that the imaging of these reconstructed "isocoverage zones" constitutes a unique experimental method for directly observing diffusion fields in epitaxial systems. We describe the dynamics of the thermal decay of the islands and the isozones in the context of a continuum diffusion model.