This article reports an experimental and theoretical investigation of mercu
ry dissolution from dental amalgams immersed in neutral (noncorrosive) and
acidic (corrosive) flows. Atomic absorption spectrophotometric measurements
of Hg loss indicate that in neutral flow, surface oxide films formed in ai
r prior to immersion persist and effectively suppress significant mercury r
elease. In acidic (pH 1) flows, by contrast, oxide films are unstable and d
issolve; depending on the amalgam's material composition, particularly its
copper content, two distinct mercury release mechanisms are initiated. In l
ow copper amalgam, high initial mercury release rates are observed and appe
ar to reflect preferential mercury dissolution from unstable Sn8Hg (gamma (
2)) grains within the amalgam matrix. In high copper amalgam, mercury relea
se rates are initially low, but increase with time. Microscopic examination
suggests that this feature reflects corrosion of copper from grains of Cu6
Sn5 (eta') and consequent exposure of Ag2Hg3 (gamma (1)) grains; the latter
serve as internal mercury release sites and become more numerous as corros
ion proceeds. Three theoretical models are proposed in order to explain obs
erved dissolution characteristics. Model I, applicable to high and low copp
er amalgams in neutral flow, assumes that mercury dissolution is mediated b
y solid diffusion within the amalgam, and that a thin oxide film persists o
n the amalgam's surface and lumps diffusive in-film transport into an effec
tive convective boundary condition. Model II, applicable to low copper amal
gam in acidic flow, assumes that the amalgam's external oxide film dissolve
s on a short time scale relative to the experimental observation period; it
neglects corrosive suppression of mercury transport. Model III, applicable
to high copper amalgam in acidic flow, assumes that internal mercury relea
se sites are created by corrosion of copper in eta' grains and that corrosi
on proceeds via an oxidation-reduction reaction involving bound copper and
diffusing hydrogen ions. The models appear to capture the correct time depe
ndence of each dissolution mechanism and to provide reasonable fits to the
experimental data.