Sinking CO2 composite particles consisting of seawater, liquid CO2, and CO2 hydrate were produced by a coaxial flow injector fed with liquid CO2 and artificial seawater. The particles were injected into a high-pressure water tunnel facility to permit determination of their settling velocities and dissolution rates. Injections were performed at fixed pressures approximately equivalent to 1200-m, 1500-m, and 1800-m depths and at temperatures varying from approximately 2 to 5 degrees C. Immediately after injection, the cylindrical particles were observed to break away from the injector tip and often aggregated into sinking clusters. The seawater flow in the tunnel was then adjusted in a countercurrent flow mode to suspend the particles in an observation window so that images of the particles could be recorded for later analysis. The flow would often break or cause rearrangement of some of the clusters. Selected individual particles and some clusters were studied until they became too hydrodynamically unstable to follow. in general, the flow required to suspend clusters or individual particles decreased with time as the particles dissolved. For example, one particle was produced and observed for over 6 min at an average pressure of 15.022 MPa and an average temperature of 5.1 degrees C. Its sinking rate, determined from the flow required for stabilization, changed from 37.2 to 3.3 mm/s over this time. Particle sinking rates were compared to correlations from the literature for uniform cylindrical objects. Reasonable agreement was, observed for short times; however, the observed decrease in sinking velocity with time was greater than that predicted by the correlations for longer times. Particle dissolution rates, based on changes in diameter, were also determined and varied from 5 to 11 mu m/s. A pseudo-homogeneous mass transfer model was used to predict single-particle dissolution rates. Good agreement was achieved between experimental dissolution data and the modeling results. (C) 2008 Elsevier Ltd. All rights reserved.