Cement-based materials are unique in that the same media used for the hydration reactions and microstructural development, water, is also largely responsible for the degradation of the materials in service . While many degradation processes, such as the diffusion of chloride ions to induce corrosion of the steel reinforcement in reinforced concrete, occur over many years, others are induced by thermal and hydric loads very early in the life cycle of the structure. For example, one common problem, especially for newer high performance lower water-to-cement ratio (w/c) concretes, is early age cracking caused by self desiccation and autogenous shrinkage [2,3], due to the chemical shrinkage occuring during the cement hydration process [4,5].
As a cement paste hydrates under sealed conditions, chemical shrinkage occurs because the hydration products occupy less space than the original reactants [4,5]. For a typical portland cement, this chemical shrinkage has a magnitude of about 0.06 g water/g cement reacted [4,5,6]. In a sealed unrestrained system, initially, the entire 3-D cement paste volume will physically shrink to match the chemical shrinkage, but once the cement sets (establishing a solid framework), this shrinkage will be resisted and instead, empty porosity will be created within the capillary pore system of the cement paste microstructure. The water/air menisci created in these empty pores will in turn induce an autogenous shrinkage (similar to a drying shrinkage ) of the cement paste that may result in cracking under restrained conditions if the tensile strain capacity of the material is exceeded. Cracking may occur globally if the system is externally restrained or proceed locally from the surfaces of the non-shrinking aggregate particles [8,9]. The magnitude of these autogenous stresses will depend on the radius of the pores being emptied as shown in Fig. 1 and Table 1; hence, the initial pore size distribution of the cement paste should have a strong influence on this process. Since the initial pore size distribution is controlled to a large extent by the initial arrangement of the cement particles, cement fineness should be one material parameter that can be engineered to control/avoid autogenous shrinkage. While computer modeling and some existing data in the literature have indicated that this should be the case , this paper presents experimental results for a cement clinker ground to four widely different finenesses which quantitatively demonstrate the validity of this hypothesis. It should be noted that while Hua et al.  and Dela  have both demonstrated that stress relaxation and creep play a major role in the autogenous response of the hardening cement paste, here, we are focusing only on the effects of cement PSD on internal relative humidity, autogenous deformation, and eigenstresses without attempting to quantify the significant influence of creep and stress relaxation on the measured experimental quantities.