Values for the activity coefficient of the chloride ion in pore water solution when added as sodium chloride has been produced from electromotive force data using a silver/silver chloride ion selective electrode.
Under normal circumstances steel reinforcement in concrete is in a passive condition due to the high pH environment provided by the hydration of cement. One of the causes of the break down of this passivity is attack by chloride ions . Currently common methods of finding the level of chlorides in a structure involve destructive techniques. The potential of silver\silver chloride electrodes depends mainly on the chloride ion activity in the surrounding electrolyte, and therefore it should be possible to use these electrodes as a non-destructive method of obtaining the chloride ion activity in cement and concrete structures. The term activity is used to explain the deviation in electrolytic properties of real solutions from ideal solutions and is related to the general concentration and interaction of ions in solution, the more concentrated the solution the further the activity will deviate from the concentration, the difference usually being described in the form of an activity coefficient, i.e. activity = coefficient x concentration. In very dilute solutions the activity coefficient is assumed to be unity, but since concrete pore water is approximately pH 13.6 this assumption may not hold and so the activity coefficient must be determined experimentally from a knowledge of the activity. Also before using a silver/silver chloride electrodes in cementitious media it is necessary to ensure that they function in a predictable manner. The purpose of this paper is to first prove this and then to determine the activity coefficient in a simulated pore solution.
Silver/silver chloride electrodes were prepared in a manner similar to that described in Ives and Janz. 25mm long sections of 99.99% pure, 1mm diameter silver wire supplied by Goodfellow Metals of Cambridge were taken and soldered onto screened cables. Heat shrink tubing was cut to an appropriate length and applied to cover the connection between the silver and the cable. These electrodes were then wiped with aceto ne to degrease them and anodised at a current density of 0.4mA cm-2 for approximately 30 to 40 minutes then moved to an aqueous solution of potassium chloride for storage. A molar solution of sodium chloride was added to 100ml of deionised water using a one ml graduated pipette with a least count of 0.01 ml. The exact amount of solution added was recorded and the solution was stirred for approximately one minute. Four silver/silver chloride electrodes were then rinsed with deionised water and immersed into this solution and their potential against a saturated calomel electrode was recorded. The saturated calomel used was in a Luggin probe arrangement, with the probe containing agar agar with 0.1M ammonium nitrate as a salt bridge. Another known quantity of 1M NaCl solution was added and the experiment was repeated until a reasonable range of chloride concentrations were obtained. A simulated pore solution, as used by Yonezawa et al  of a similar composition to that found by Diamond, containing 0.4M KOH and 0.2M NaOH was then used as the base electrolyte and the experiment repeated. The complete procedure was then repeated to avoid any cumulative or weighing errors that may have occurred.
Figure 1 Potential versus concentration (larger image)
Figure 2 Activity versus concentration (larger image)
Table 1 Calculated Activity Coefficients for Sodium Chloride in Simulated Pore Water Solution.
Concentration Activity Coefficients 0.01 0.73 0.02 0.57 0.03 0.50 0.04 0.48 0.05 0.47 0.06 0.46 0.07 0.46 0.09 0.45 0.12 0.44
Figure 1 shows the potentials of the silver/silver chloride electrodes used, measured against a saturated calomel electrode (SCE). From this data the activity of the chloride ion can be calculated by straightforward mathematical manipulation of the Nernst equation:
The potential of the saturated calomel electrode has been calculated at values ranging from 0.2412 V to 0.2467 V. The effect of this variation would be to change the gradient of the line produced in a plot of activity against concentration. It was therefore decided to take already published data on the activity coefficients , and calculate the potential of Eref by fitting the curve produced from the deionised water data to the published activity coefficients. The potential for Eref produced in this manner was then used in equation 2 to calculate the activity of the chloride ion in simulated pore solution and hence find the activity coefficients of the chloride ion in this solution. The data produ ced in this manner is illustrated in figure 2, along with the data the curve was fitted to. The resulting standard potential for the saturated calomel electrode used was found to be 0.2412 V. The maximum deviation between silver/silver chloride electrodes was 3mV at the lowest concentrations used, at higher concentrations the maximum deviation between electrodes was less than 1mV. This suggests a larger error in very dilute solutions, a reasonable statement. This is also supported by the fact that the silver\silver chloride electrode took longer to reach a stable potential in the most dilute solution, yet were almost instantly stable in the higher concentrations.
Table 1 shows the values of activity coefficients in the simulated pore solution. These were found by dividing the calculated activity of the ions in solution by the known concentration of ions in solution.
The most common method of finding the concentration of chloride ions in use is that of pore water expression, as described by Longuet whereby a cylinder of concrete is subjected to triaxial loading and the solution obtained in this manner is then analysed. Since the quantity of solution obtained in this manner is small, usually not more than 5cc, it is diluted between 25 and 100 times before analysis. The so lution is then analysed for OH- and Cl- concentration in various ways. Most commonly used is titration with nitric acid for the OH- concentration and with silver nitrate for the chloride ions, as used by Page in several papers . Some authors  however have used chloride sensitive electrodes in order to find the chloride concentration, or pH electrodes  for the hydroxide ion concentration. Both these methods give activity rather than concentration, as mentioned by Tritthart who found that a glass electrode gave a value of activity of 10-0.45, whereas the measured concentration of hydroxide ions was 10-0.2 which would suggest an activity coefficient of approximately 0.56. Hausmann  stated that the probability of steel corrosion in concrete was dependant mainly on the ratio of chloride to hydroxide ions in solution. Since the above authors all quote this figure an error with a magnitude of approximately 0.5 must occur in those that have determined the activity of one s pecies and compared it with the concentration of the other. This does not occur when comparing concentration with concentration because the activity coefficient is in fact the mean activity coefficient of all ions in the solution.
The activity coefficient of the chloride ion when added as sodium chloride to a simulated pore solution ranges from 0.75 at a concentration of 0.01M to 0.45 at 0.12M. Silver silver chloride electrodes provide the possibility of measuring chloride concentrations in cement and concrete non destructively.