Introduction

Cement paste, the binding phase of concrete, is primarily a calcium silicate hydrate (C-S-H) that develops its interconnected solid structure from a chemical reaction that consumes approximately 0.25 g water for every 1 g cement, with an additional 0.15 g water per 1 g cement incorporated into the hydrated gel structure [1]. Although the majority of the hydration reaction takes place during the first month, the hydration during the first few days has the greatest effect on the overall performance of the concrete. It is during this initial hydration that the availability of water is critical to proper C-S-H formation. Most concrete elements are made with sufficient water for complete hydration, and can, therefore, hydrate sufficiently under sealed conditions. When evaporation from the surface occurs, however, moisture transport through the concrete may be insufficient to replace lost water from either the environment or the interior of the concrete. Failure to maintain saturation at the surface can have detrimental effects on concrete surface properties and, possibly, on the overall concrete structural element performance. Industrial curing practices incorporate a number of methods for either maintaining a humid environment at the concrete surface or using chemical or physical barriers to minimize moisture loss [2].

Because proper curing practices constitute a financial burden that is proportional to their duration, and because curing beyond a certain period will have no effect upon concrete performance, an optimum curing practice is sought. The ideal curing regimen maintains saturation within the concrete surface for the shortest period of time such that additional effort would have negligible effects on the desired concrete properties. The ACI Manual of Practice [2] recommends 7 d curing for concretes made with Type 1 cement, regardless of the water-cement ratio. Moreover, "Natural curing from rain, mist, high humidity,...regarded as sufficient to provide ample curing..." (ACI 308-92, Section 2.10) may be considered, even though sufficiently 'high humidity' is neither defined, nor a lower limit established.

The more important question, however, is the combined effects of curing duration and the level of concrete saturation required for normal hydration with respect to saturated conditions (100 % RH). The minimum sufficient level of saturation is not known because the effects of exposure to environments below 100 % relative humidity (RH) have not been sufficiently quantified so that a rational basis for curing procedures can be developed.

A number of previous studies have attempted to quantify the effect of environmental RH on portland cement paste hydration. In the experiment reported by Powers [3] that quoted previous results of Gause and Tucker [4], fresh cement paste was placed into sealed glass jars. Periodically, the jars were opened, and the relative humidity was measured in the air space over the paste. The measured relative humidity decreased monotonically, terminating at a value in the range of 80 % to 90 %, depending upon the initial water-cement (w/c) mass ratio. As cement paste hydrated, the chemical reaction consumed available free water. When all the surface water had disappeared, continued hydration reduced the RH inside the jar. At the same time, the reaction products were filling the initially water-filled pore space, decreasing the effective pore diameters, and reducing the RH at which the pores could still remain saturated via the Kelvin-Laplace effect [5]. Within a sealed system, there is a competition between the decreasing RH due to water consumption and the decrease in the equilibrium RH due to pore size reduction. Based on the results reported by Powers, the rate of RH reduction due to reaction was greater than the reduction in equilibrium RH environment due to pore size reduction, and the reaction terminated because the pores were unable to remain saturated.

In a related experiment, Parrott, Killoh, and Patel (PKP) [6] reported continued hydration for 0.59 w/c pastes at relative humidities as low as 55 %. Their specimens were 3 mm thick slices cut from the middle of a prism after two days of hydration. Coming from the middle of the prism, the samples were initially able to draw water from the surrounding paste. As such, the samples represented (virtually) saturated samples at the time of exposure. The degree of hydration (DOH) was measured after 14 d and 90 d exposure to various relative humidities, and the values were compared to the DOH of continuously saturated specimens. The ratio R of the DOH for a sample hydrating at a relative humidity $\phi$ to the DOH of a sample hydrating under saturated conditions was approximated by a power law relationship [6]:

$$R = \left(\frac{\phi - 0.55}{0.45}\right)^4$$ (1)

The ratio R for values of $\phi$ less than 0.55 is, by definition, zero. Using Eqn. 1, the rate of hydration for exposure to a 90 % RH environment is 0.37 times that for a continuously saturated specimen.

Unfortunately, neither experiment demonstrates the full complexity of the effects that an unsaturated environment (<100% RH) during initial hydration can have on microstructural development. The rate of hydration in an unsaturated environment will depend upon the age of the cement paste at the time of exposure, and the time required for the specimen to reach equilibrium with the unsaturated environment. A young specimen having large capillary pores and exposed to an unsaturated environment could lose sufficient free water through evaporation that the rate of hydration at the surface could virtually cease. Conversely, a specimen initially hydrating under saturated conditions for a sufficiently long time would contain pores small enough that water would not evaporate into an unsaturated environment, and the specimen could continue to hydrate in an unsaturated environment at a rate nearly equal to saturated conditions. Therefore, the rate of hydration for a specimen exposed to an unsaturated environment should depend, in part, upon the duration of the initial period of hydration under saturated conditions.

To test this theory, an experimental plan was developed to study the hydration and microstructural development in specimens exposed to an unsaturated environment. In the experiment, the unsaturated environment consisted of 90 % RH air at 25 ºC. A 90 % RH environment was chosen because the common assumption, based largely on Powers [3], is that hydration continues indefinitely in a 90 % RH environment. Moreover, 90 % RH could be considered sufficiently 'high humidity' for use as natural curing.

Thin cement paste specimens were initially exposed to a 100 % RH environment. After either 6 h, 12 h, 3 d, or 7 d exposure to a 100 % RH, each specimen was moved to a 90 % RH environment. Degree of hydration measurements from these exposed specimens, as a function of time, were compared to corresponding values from samples continuously exposed to 100 % RH. Differential scanning calorimetry (DSC) measurements were used to detect freezable water within a specimen as a function of time.

Arguably, a 100 % RH exposure is not identical to moist curing. A 100 % RH environment can, however, supply water to surface pores as needed, provided the rate of vapor transport through the atmosphere is faster than the rate of water consumption due to hydration. Further, due to the ionic strength of the pore fluid, the pore fluid is likely in equilibrium with a 95 % to 97 % RH environment. Therefore, it is likely that the 100 % RH environment will form a film of liquid over the surface of the paste. Moreover, results will show that the rate of hydration for the specimens exposed to 100 % RH was identical, to within experimental error, to independent measurements made on cement pastes in contact with water.

A similar experiment to this one has been performed by Bager [7]. In that experiment, specimens were exposed to 58 % RH after water curing for 2 d, 3 d, 7 d, and 10 d. Bager used mercury intrusion porosimetry to characterize the microstructure, DSC to determine total freezable water, and electrical conductivity and permeability measurements to characterize the effect of exposure on transport. The experiment reported herein extends the work of Bager by using the temperature dependent DSC data to characterize the developing microstructure and the DOH to characterize macroscopic effects, both as a function of time. In this way, the time-dependent effects of early exposure can be studied in more detail.