Hydration and percolation at the setting point

a r t i c l e i n f o The setting of cement paste is widely understood to be caused by percolation of the links that are created by overlap of hydration products on the surfaces of reacting grains of clinker. Percolation theory predicts that the elastic modulus will increase with a certain functional form, but few attempts have been made to demonstrate this behavior quantitatively. We discuss the appropriate variables to use for this test of the theory, and show that the percolation probability is proportional to time only over a narrow time interval. We compare the measured and predicted degree of hydration at the percolation threshold, and show that the hard-core/soft-shell model strongly overestimates the amount of hydration at the setting point. The discrepancy is attributed to agglomeration of particles in the paste, which reduces the amount of hydration needed to link the particles into an elastic network. The setting of cement paste is generally understood to be a perco-lation process in which the hydration products that form on the surface of clinker particles intersect, leading to the formation of clusters that eventually join into a continuous elastic network. The percola-tion of the solid and pore phases has been investigated using numerical simulations, the earliest of which was by Bentz and Garboczi, using CEMHYD [1]. That paper predicted a high degree of hydration at the percolation threshold, but later work produced lower, more realistic , values [2]; the difference apparently resulted from increasing the spatial resolution used in the simulation [3]. Similar studies have been done using HYMOSTRUC to predict the fraction of connected solids [4]. The connection between the percolation of solids and the rise in elastic modulus has also been examined by comparing various measures of rigidity, such as the Vicat test or sound velocity, with the degree of connectivity simulated with CEMHYD [5–7] or HYMOSTRUC [4,8]. Only a few studies have tried to quantify the change in properties near the setting point in terms of percolation theory, but each of those analyses has defects that will be discussed. The purpose of this paper is to reexamine the use of percolation theory for interpreting the setting behavior of cement paste. The analysis will be demonstrated by using data for the ultrasonic pulse velocity and chemical shrinkage obtained in an earlier study [9]. We will then compare the observed threshold to the prediction of the …