We are studying the rate of attack of carbonic acid on cement using two types of tests.
In the flow-through experiment, fresh solution passes continuously over the cement, so that the maximum rate of corrosion is observed (i.e., transport of reactants and products is fast enough so that the rate is controlled by chemical kinetics). The photo above shows the series of rings of reaction products formed on a cylinder of cement paste during such a test. We monitor the composition of the liquid entering and leaving the reaction cell. Samples of cement are collected periodically, so that the radial variation in composition and crystal phases can be quantified, and the microstructure can be determine using scanning electron microscopy. The apparatus and some of the results are shown below. The details are described in our publications.
In the batch experiments, the sample consists of a disk of stone (either Berea sandstone or Salem limestone) with a hole drilled through it that is filled with cement. The faces of the disk are covered with sheets of teflon and stainless steel so that the carbonic acid must enter radially. The hole is drilled off-center, so that the acid has to diffuse different distances through the stone to reach the cement. Consequently, the depth of attack is greatest where the cement is closest to the surface of the stone, and decreases gradually around the perimeter. The diffusion coefficient of water in the stone was measured using NMR and the pore structure of the stone was determined using mercury intrusion porosimetry. The permeability of the disks was measured before exposure and after various periods of exposure (from 1 to 12 months). As soon as the boundary between the cement and the stone had visible corrosion, the permeability of the disk leaped up, owing to flow through the small gap created at the boundary. The corrosion was examined using electron microprobe analysis, scanning electron microscopy, and X-ray diffraction. The kinetics of attack are now being simulated using sophisticated models of diffusion and corrosion. The goal is to refine the models so that they accurately reproduce the results of these experiments. Then we will use the models to predict the rate of attack on an abandoned oil well. The following photos show the batch setup and the formation of a reaction rim between the sandstone and cement after a few months of exposure.
Ed Matteo extended this work by making time-lapse videos of the corrosion process and obtaining precise values for the rate of attack. That work is described in several papers.