Next: Description of microstructure
Up: Main
Previous: Main
The microstructure of cement paste is known to be complex. When starting with a disordered suspension of irregularly-shaped cement particles, which have a wide size distribution, and which subsequently undergo random growth due to a hydration reaction that occurs after mixing with water, one could expect no less.
However, in the last ten years or so, from the field of the physics of random materials, have come an ever-increasing number of growth or aggregation models. These models, using very simple growth rules that have an element of randomness in them, have been shown to produce extremely complex aggregated structures, often with fractal morphology. Some of the models include the diffusion-limited aggregation (DLA) model [1], the cluster- cluster model [2], and the Eden model [3]. In light of these models, it is not unreasonable to suggest that the complex microstructure of cement paste could be simulated using a few, relatively simple growth rules, which are iterated many times cyclically throughout the hydration process.
The model presented in this paper represents an attempt to implement the idea of simulating the microstructure of cement paste via the iteration of a few simple growth rules. One should note that any such rules, while of necessity being compatible with known cement hydration chemistry, need not slavishly follow each step of the hydration process, but rather can be thought of as summarizing chemical details to produce simple algorithmic rules of growth. In this manner,the rules are based more on the physical mechanisms of microstructural development than on the chemical mechanisms. The ultimate test of such a scheme is, of course, how well the final simulated material compares with the real material, in geometric and topological features like phase distribution and pore space connectivity, and in physical properties like transport coefficients.
Having a model that gives a simulated cement paste microstructure, one can proceed to use the model to investigate scenarios of experimental interest, at a level of resolution and understanding impossible in ordinary experiments. We have chosen to study, as a first application of the model, the interfacial zone between aggregate and cement paste in concrete or mortar [4]. By aggregate is meant either small sand particles or larger rock particles, both of which act as inert fillers in the cement paste matrix. Careful image analysis of back-scattered scanning electron micrographs by Scrivener and co-workers [5] have shown that a region, on the order of 10-50 micrometers thick, exists in the cement paste around the aggregate particles in ordinary portland cement concrete, in which the porosity increases as one moves through the zone, reaching a maximum at the aggregate surface. Porosity in this case is defined as the water-filled space left over from the space around the original unhydrated cement particles. Some studies have also shown a similar increase in calcium hydroxide through the interfacial zone [6,7]. Our goal is to simulate the relevant processes and geometry, and attempt to duplicate experimental observations and gain new insight into the underlying causes of the interfacial zone.