Pore structure affects almost all the physical properties of materials and, thus, is important in engineering practice. Engineering desired physical properties by microstructural modification of the porosity and/or the solid phases has practical importance and, therefore, there is considerable research activity in developing relationships between the desired physical properties and the pore structure. The ability to vary the level of porosity, the pore size distribution, pore shape, and pore connectivity opens new opportunities for developing materials not only for conventional uses but also for novel applications such as electronics.

The pore size distribution in cement paste over the range of pore sizes interrogated by high pressure mercury intrusion porosimetry may be described by a mixture of two lognormal distributions. The compound distribution of pore sizes may be given as:

where p(x) is the probability density function of pores of size x, f and (1−f) are the weights of sub-distributions, μ1 and μ2 are the location parameters of sub-distributions, and σ1 and σ2 are the shape parameters of sub-distributions. These two sub-distributions may represent the larger and smaller capillary pores respectively. The changes in the sub-distributions and the compound distribution as functions of curing age and water-to-cement ratio are discussed.

We show that the scaling of the dynamic permeability κ(ω) by its static value κ0 and the frequency ω by a characteristic frequency ω0 particular to the medium results in a dimensionless function , with , which is dominated by the geometry of the throat regions in a porous medium. If the pore cross sectional area S varies slowly near the throat, i.e. dS/dz ≃ 0 where z is the distance normal to the cross section, then is an approximate universal function independent of microstructures. When scaling holds, the dynamic permeability κ(ω) is shown to contain only two pieces of geometric information, and the knowledge of either the low-frequency or the high-frequency asymptotic constants of κ(ω) would enable one to deduce the other missing parameters. In particular, since the high-frequency asymptotic parameters of κ(ω) can be related to the electrical formation factor and the volume-to-surface ratio, the static permeability value κ0 may be directly deduced from such non-permeability measurements.

Initial observations of the influence of pore structure on material properties of cement paste are described. The total porosity is held constant and pore structure is altered by use of a superplasticizer. Corresponding changes in strength and fracture energy are observed. Changes in the cement paste pore structure are evaluated by analysis of images obtained by optical microscopy. The technique is a direct method of measurement that avoids assumptions about pore geometry.

Cement paste of water:cement ratios less than about 0.3 usually are not workable, but workability can be maintained at these lower water:cement ratios by using superplasticizers. A typical explanation [1] of the mechanism behind the effectiveness of superplasticizers is that they adsorb on the surface of cement particles and adjust the surface charge so that the particles become deflocculated. The individual particles flow more easily than larger flocs.

Pore structure and the surface area of the pores are the most important characteristics controlling the properties of porous materials. Many techniques, based largely on nitrogen or water vapour isotherms or mercury porosimetry, have been used to determine the pore structure of solids. Values obtained from these methods have been relatively reliable for materials with pore structures that remain stable on removal or addition of water.

A series of techniqu1es were employed to characterize the pore microstructure of a group of colloidal silica-alkali silicate gels. Methods used included mercury penetration porosimetry, nitrogen desorption measurements, transmission electron microscopy, single and multiple small angle neutron scattering, nuclear magnetic resonance relaxation measurements, x-ray tomography and dielectric constant measurements. Comparison of results of pore volume, pore size and size distribution as well as homogeneity of the pore structure are presented.

A brief discussion is presented of some of the characteristics of hardened cement paste (hcp) and hcp in concrete that bear on the measurement of pore size distribution. The postulates underlying several present and potential measurements are discussed individually, and an attempt is made to assess the applicability of each in light of the peculiar characteristics of hcp. Some of the problems inherent in each procedure are briefly described.

The importance of knowing the pore size distribution of a material, and the utility of mercury intrusion, are briefly reviewed. Mercury intrusion results can be used to obtain the specific surface of a material. Under the right circumstances, this method has many advantages over vapor sorption, and a mercury intrusion test may be desirable for this reason alone. The procedure and limitations are described. Modern instruments allow for the measurement of mercury extrusion as well as intrusion. This opens the possibility of measuring two pore size distributions. One is reversibly intruded, and the other is not. The use of these sub-distributions in seeking correlations with material properties and behavior is discussed. The applied-pressure/diameter-intruded relationship of the Washburn equation depends upon the surface tension and contact angle. Inter-metallic solutions with mercury as the solvent offer the possibility of reducing the surface tension and/or the contact angle. This allows the use of lower pressures to intrude the same diameter pore, or the intrusion of smaller pores at the same pressure. The practical application of using such solutions is described.

The pore structure (i.e. total pore volume, surface area and pore-size distribution curves) was measured using mercury porosimetry and nitrogen sorption. Hydrated portland cement (type I) of water-cement (w/c) ratios 0.3, 0.4 and 0.6 by weight was analyzed at three degrees of hydration (i.e., 30%, 50% and 80%; 70% for the 0.3 w/c system) corresponding to low, intermediate and high levels of hydration. The effect of curing temperature (3°, 23°, and 43°C) on pore structure was also studied. The two techniques were evaluated as well on porous Vycor glass, which has a narrow pore size distribution in the size range accessible to both. Results obtained by both techniques on porous Vycor glass agreed well. However neither technique can be used alone to study the entire pore structure in well-hydrated cement due to the wide range in pore sizes and the presence of micropores. Due to the unstable pore structure in cement a specimen treatment procedure such as methanol replacement, combined with volume-thickness (V-t) analysis, is necessary in order to measure the micropores. At low hydration values the pore structure can be estimated by mercury intrusion porosimetry (MIP). At higher hydration values, however, this technique underestimates total pore volume and surface area due to the presence of micropores which MIP cannot determine. In the pore size range of overlap, higher pore volumes were obtained with MIP. Nitrogen V-t analysis shows that micropores are more pronounced with lower w/c ratios. This finding is consistent with pore size distribution curves obtained by MIP. For a given w/c ratio and degree of hydration the total pore volume measured by MIP was found to be independent of curing temperature in the temperature range studied. At any w/c ratio, capillary porosity is controlled by degree of hydration alone.

The pore-gel microstructures of hydrating cement and effects of their interaction with the environment have been studied using the non-destructive technique of small angle neutron scattering (SANS). The results presented for the cement microstructures are statistically representative of such structures over macroscopic sampling volumes.

In backscattered electron images of polished sections of cement paste pores can be identified down to a size of about 0.05 microns. Moreover, the contrast between pores and the solid phases is sufficient to allow the pores to be distinguished and quantified by an image analyser. There is a good correlation between measures of porosity obtained by this technique and those obtained by methanol absorbtion methods despite the lower limit to resolution.

The bse method can also be used to study the distribution of the porosity in space. However, as only two dimensional sections can be examined, there are difficulties in determining the connectivity by this method. The possibility of using serial section reconstruction is examined and discussed.

Despite the difficulty in relating two dimensional characterisation to three dimensional properties, several techniques have been used with some success for sandstones. Results from these techniques for cement paste are presented and discussed.

While there is presently a great deal of interest in measurement of permeability, there are a wide variety of non-standard methods reported. The author's experience combined with a critical review of the literature is that:

1. (a) results vary with operating variables of each test,

2. (b) the coefficient of variation of results is large (typically 30 to 50%),

3. (c) most methods cannot provide quantitative results in the range of interest for high strength, concretes and those containing supplementary cementing materials (10−15 m/s or less),

4. (d) most methods do not represent the type of boundary conditions experienced by most critical concrete field conditions (i.e. fluid on one side and vapour on the other side),

5. (e) Indirect measures of permeability such as rates of absorption, or resistivity measurements are easier to perform and relate directly to permeability.

The activity of research in the area of permeability is similar to that described in the technical literature from the 1920's and 1930's. Much of that early work has abandoned for reason (c) above.

It is attempted to review some of the previous experience, describe the problems and make suggestions for improved measurement of permeability.

With a new method relying on measurements of water vapour flow from a specimen and on measurements of distribution of relative humidity in the specimen at steady state conditions, it is possible to calculate the moisture permeability and to determine its dependence on the relative humidity.

The hydraulic conductivity of three concretes with a high ratio of water to cementitious solids (w/cs) was measured using a test device in which confining and driving pressures are controlled separately (triaxial, or Hassler cell). Test variables included: confining pressure (c), driving pressure (d), ratio of these two pressures (c/d), and sample length. The objective was to determine the effect of these variables on measured permeability (K).

For these concretes, the driving pressure has a strong influence on K values, which generally are significantly higher at lower driving pressures. The c/d ratio controls K to a lesser extent, and not in a consistent direction. K increases as c/d increases for specimen lengths of 3X the maximum aggregate size or longer. For specimens the length of which is closer to the aggregate size, K values decrease with increasing c/d, making specimen length an important variable. Removal of entrapped air from the system and presaturation of the specimens also are important.

A gamma ray spectrometer using an Americium source has been built. The source and the detector are mounted on two separate towers and can move vertically. The studied sample is mounted on a platform that can move horizontally. The movements of the towers and the platform are controlled and synchronized by a computer which is also used to define the parameters of each test, to run the test, and to collect the data. This equipment has been used to analyze the capillary moisture transfer through two cement pastes.

The permeability of cement paste to water was determined by using a triaxial permeability cell. A new technique of specimen conditioning, based on cyclic flow reversal, was used for the early attainment of equilibrium flow conditions. Effects or cement type, age, and silica fume addition, were investigated. With the specimen conditioning used, equilibrium flow conditions were achieved on the first day of the test itself. The coefficient of permeability was found to decrease with an increase in the degree of hydration. The use of silica fume was found to decrease the permeability However, the permeability did not particularly depend upon the aotunt of silica fume added.

Permeabilities to water and diffusion of ionic species in cementitious grouts, pastes and mortars are important keys to concrete durability. Investigations have been made of numerous materials containing portland and blended cements, and those with fine-grained filler, at room temperature and after prolonged curing at several elevated temperatures up to 90°C. These constitute part of studies of fundamental material relationships performed in order to address the question of long-term durability. In general, the permeabilities of the materials have been found to be low [many <10−8 Darcy (10−13 m·s−1)] after curing for 28 days or longer at temperatures up to 60°C. The results obtained at 90°C are somewhat more complex. In some sets of studies of blended cement pastes with w/c varying from 0.30 to 0.60 and cured at temperatures up to 90°C the more open-pore structure (at the elevated temperature and higher w/c) as evident from SEM microstructural studies as well as mercury porosimetry are generally correlated also with a higher permeability to liquid. The degree of bonding and permeability evident in paste or mortar/rock interfacial studies present somewhat more conflicting results. The bond strength (tensile mode) has been shown to be improved in some materials with increased temperature. The results of permeability studies of paste/rock couples show examples with similar low permeabilities, and some with increased permeability with temperature.

Ionic diffusion studies also bring important bearing to understanding the effect of pore structure. The best interrelationships between chloride diffusion and pore structure appear to relate diffusion rate to median pore size. Similar results were found with “chloride permeability” test.

Air permeability was determined for concretes of variable porosity (w/c ratio of 0.33, 0.50, and 0.67). The reproducibility of the test and the ability to characterize the permeability of different concrete were evaluated. Results indicate that air permeability test gives suitable reproducibility with a margin of error of 10%, which tends to improve with increase in w/c ratio. The difference in air permeability of concretes with the w/c ratios investigated are distinguishable by this technique. Furthermore air permeability coefficients used were modified with equations derived in accordance to Darcy's law, and the resulting air permeabilities compared. Data calculated with an air permeability coefficient based on mean radius pore (corrected equation) appears to give a more realistic value with greater differentiation between the ranges. Values were determined on specimens after 28 days, maintained at room temperature (50% RH), then again after an additional 2 days at 60°C in a ventilated oven. Oven-dried specimens exhibit significantly greater air permeability.

This paper presents a new test method for rapidly assessing the resistance of a concrete to chloride ion diffusion.