In the process of studying silver particle growth on graphite surfaces, we observed something rather unusual. We found that when silver was first deposited on a graphite surface, and then redissolved, that not all of the silver was removed. Moreover, AFM images of the graphite surface revealed that the silver which ws left behind (at positive potentials) was dispersed as very small, subnanometer silver particles. These particled presisted on the graphite surface for up to an hour - even when the voltage was held 0.5 V positive of the stripping potential for silver.

1. The anodic dissolution of silver microparticles from a graphite electrode surface potentiostatted at +500 mV vs. Ag/Ag⁺ occurs rapidly until the mean diameter of these particles is reduced to 0.4 - 1 nm. Because we have deposited an arbitrary amount of silver (the equivalent of several monolayers), the time scale for this initial oxidation is meaningless. The main point is that a sharp decrease in the oxidation rate is observed when silver particles are reduced in diameter to 0.4 - 1 nm. Here's what they look like to an AFM:
   

2. Under these conditions, the remaining 0.4 - 1 nm clusters are metastable and possess a half-life on the graphite surface of approximately an hour. Characterization of these clusters by electron diffraction (see above) shows their composition to be silver metal having the FCC crystal structure.

3. The presence of 0.4 - 1.0 nm silver clusters on the graphite surface is associated with a diminution of the nucleation overpotential for the redeposition of silver onto the surface in cyclic voltammetry experiments. This effect is seen in the data shown below:
   
   
   Electrode surfaces on which silver has been electroplated, and from which all silver has been removed (following anodic stripping for more than an hour), exhibit a nucleation overpotential equal to, or greater than, that of a freshly cleaved graphite electrode surface. Consequently, it is reasonable to attribute the "facilitated renucleation" of silver in these experiments to the presence of metastable silver clusters on the electrode surfaces. It is unnecessary to postulate that the intercalation of silver into the graphite electrode surface is responsible for this behavior.

The underlying reason for the slow oxidation kinetics exhibited by subnanometer-sized silver clusters will be the subject of further study. One aspect deserving of closer scrutiny is the possibility that silver particles undergo a metal to nonmetal transition during dissolution, and that interfacial electron transfer to the resulting nonmetallic silver clusters is slow for a variety of reasons. A second possibility involves the existence on these clusters of a strongly coordinated anion layer that impedes their dissolution.