P·O 42.5 cement from Harbin Cement Co. LTD was used for all experiments in this study. River sand with a fineness modulus of 2.4 was used as a fine aggregate. The chemical composition of the steel anodes used for the sensor, corresponding to Q235 steel, is given in Table 1. Q235 steel is widely used in civil engineering and especially in coastal constructions in China.

Figure 1 presents a photo and a detailed schematic of the corrosion sensor embedded in cement mortar. Each of the four differently-sized steel anode rings was fixed on a fabricated nylon tower [see Figure 1 (b)], spaced 10 mm from one another. The detailed geometrical design of each steel anode is given in Table 2. Cables connected to each single steel anode, S₁∼S_4, were run out of the cover-zone mortar.

This geometrical sensor design ensures that:

Each steel anode has an equal exposure area of about 668 mm². The surface area of the cathode (more than 3,000 mm²) is sufficiently larger than the anodes to reduce errors due to the lateral current distribution on the surface of the steel anodes [21].

A cathode made of Ti/MMO (mixed metal oxide-coated titanium), which is widely used in the cathodic protection system, was placed 10 mm above anode S₄. A fabricated long-term Ti/MnO₂ reference electrode (RE) was embedded adjacent to the corrosion sensor for potential measurements, and the stable potential of the RE is −40 ± 5 mV referenced to the saturated calomel electrode (SCE) [18]. Each sensor was embedded in the middle of a cubic cement mortar specimen (10 cm × 10 cm × 10 cm). The distance between the upper end of the sensor and the upper surface of the cement mortar was 2 mm.

Figure 3 presents the impedances of each steel anode (S₁–S₄) in the 3% sodium chloride mortar sample C. The diagrams obtained from the different steel anodes are similar in shape to each other. This type of spectrum is usually interpreted by using the model described above, consisting of a CPE (Q_a and Q_c) in parallel with the polarization resistance (R_a and R_c), in addition to the cement mortar resistance (R_p). The adopted frequencies (1 kHz to 50 kHz) are sufficiently high that Y 0 w n cos ( n π 2 ) is much greater than 1 R a or 1 R c in Equations (2) and (3). The impedance curve of the sensor system in the cement mortar exhibits a rectilinear regulation trend (see Figure 3). Consequently, the relationship between Z′ and Z″ can be simplified and is given as follows: (5) Z ″ = a Z ′ + b .where a represents the slope, and b represents the intercept of the impedance curve. The values a and b can be calculated from the data shown in Figure 3. Then, R_s (the red point in the shaded area, corresponding to Z″ = 0) can be obtained.

Most of the existing works on alternating current (AC) approaches to resistance measurements only measure the impedance behavior at a fixed frequency. Generally, the traditional measured results are not exactly the true resistance, due to interference with the interface capacitance between the electrodes and the cement mortar [17]. The adverse effects of the interface capacitance are generally reduced by using the present method.

Figure 4 shows the depth-related resistance between the different anodes (S₁–S₄) and a cathode embedded in different cement mortars (A to D). Mix A, with no added chloride, was investigated as a reference mortar. Each sample was measured six times to give an average resistance value for each electrode pair with very low scatter. The resistance values for this method, presented in Figure 4, represent the measurements from two separate samples. Error range of ± 5% was observed when comparing the resistance of a solution with known conductivity (0.1 mol/L NaCl, 0.011644 S·cm⁻¹) with the calculated resistance value [18].

It is shown in Figure 4 that the resistance is proportional to the chloride content and increases with increasing cover depth. As expected, it was found that the cement mortar without added chloride exhibited a very high resistance, above 1250 ohms. In contrast, a lower mortar resistance (less than 1,000 ohms) is measured in the mortar samples with 1–5% chloride in the mixture. Moreover, the addition of 1% chloride to the cement mortar produces a sharp decrease in the mortar resistance. A difference factor of more than 10 was observed when the resistance values of specimens C and D were compared to the values observed in specimen A. Because the cement mortar has a certain microstructure and chemical composition, the resistance of concrete structures depends entirely on the environmental humidity, temperature and concentration of penetrating ions, etc. [25]. The decrease in mortar resistance with increasing chloride content can be explained by the chloride ion exchange interaction, but this seems to be insignificant when the chloride content is greater than 3% (see the resistance values of mixtures C and D shown in Figure 4).

The relationship between mortar resistance and corrosion rates is still a subject of study. However, measurements of corrosion rates as a function of chloride penetration into the mortar have shown that the mortar resistance has a significant effect on the corrosion rate, especially in the case of mortars with high moisture content [26]. This strongly suggests that, for the investigated cement mortar and environmental factors, measured resistance values could be used to differentiate between the passive and active states of the surrounding steel reinforcements. However, such a measurement would be cumbersome, and in this work the polarization resistance (R_p) was measured and analyzed as discussed in the next section.

The LPR test results for each steel anode (S₁ to S₄) in mortars A and C are shown in Figure 5. Duplicate experiments provided essentially the same results. Figure 5 shows that the difference between the corrosion potential of S₁ and S₄ is approximately 100 mV in mortar A. The recorded corrosion potential tends to increase with depth; this result may be explained by the availability of oxygen at various depths. However, such deviation seems not to be significant in mortar C due to its poor cover-zone quality.

In addition, corrosion potential is a clear indicator of cover mortar quality. A more positive corrosion potential is usually related to a lower corrosion rate in steel [19]. The steel anodes embedded in mix A, depicted in Figure 5 (a), exhibit a shift in the corrosion potential towards a more positive value, indicating the ability to passivate. In contrast, the steel anodes embedded in mix C shift to more negative potential values, less than −300 mV, due to the high chloride content near the surface of the steel anodes. It is found that S₄, which is near the cement mortar surface, showed the lowest corrosion potential, below −340 mV in mix C (versus the RE), which may indicate a higher corrosion risk according to ASTM C 876–91.

The direction of the shift in the polarization curve is another immediate descriptor of the quality of the steel anode surface. As shown in Figure 5 (a), the steel anodes exhibited a negligible current in sample A, further indicating a tendency towards the passive state. Generally, the polarization curve for the steel anodes in mortar A moved towards more positive potentials (higher) as the cover depth increased. In contrast, a nearly opposite trend was observed in the polarization curves of each steel anode in mix C. An increase in current density with increasing electrode potential was observed in both the cathodic and anodic polarization curves. This indicates that the anodic current densities were high enough to guarantee that the steel anodes underwent significant corrosion.

Figures corresponding to mixes B and D (not shown here) show that the steel anodes exhibited similar curves in all three of the chloride-contaminated mortar samples. Previous results, such as polarization curves and corrosion potentials, dealt only with a simplistic prediction of the corrosion risk and a qualitative analysis of the corrosion tendency. For quantitative purposes, a detailed and precise prediction is necessary. The results shown above can be improved by using the R_p values obtained from Equation (4). Figure 6 depicts the R_p values corresponding to the LPR curves obtained from the embedded sensors in all of the cement mortar samples (A to D).

Figure 6 shows that the resistance values measured by the corrosion sensors in each mortar sample follow a descending trend of A>B>C>D, similar to the trend of polarization resistance depicted in Figure 4. In addition, the R_p values recorded by the sensors embedded in the mortar without chloride are two to 10 times higher than the values of the chloride-contaminated samples. This agrees with results presented in prior investigations, obtained from marine concrete structures with similar w/c ratios [27]. Thus, the cement mortar corrosion risk can be estimated to some extent by measuring R_p.

It should be pointed out that the R_p values are higher in S₁ and lower in S₄ than those corresponding to the other anodes embedded in the same mortar. This finding is probably due to the depth-related oxygen availability in cement mortar because linear correlations have been established between the distance from the surface and O₂ concentrations [28]. It is known that the cathodic reaction responsible for the corrosion of steel in concrete requires a certain amount of oxygen. It can only be assumed that, according to what has been previously described, the protective mortar layer surrounding the steel surface plays a decisive role in the corrosion of the steel anode. This assumption can be explained by the lower oxygen availability near S₁, which increases R_p, even when 3% or 5% chloride was added (see Figure 6).

Figure 7 presents a plot of ln R_p versus R_s in the mortar without added chloride, in order to demonstrate the relationship between these two parameters. A nearly linear relationship between ln R_p and R_s was observed. The straight line in Figure 7 is described by the following equation: (6) ln   R p = 0.000716 × R s + 8.7372                 r 2 = 92.5 %

The linear relationship shows that, in the absence of chloride contamination, R_p increases as the mortar resistance increases. However, in this case, the corrosion rate is mainly controlled by the oxygen supply and qualities of the cement mortar. By assuming that all of the electrodes and built-in units of the corrosion sensors were operating at the same temperature and relative humidity, the mortar can be treated as a quantifiable aggressive environment. Thus, this study further confirms that both parameters (R_p and R_s) reflect the corrosion state of the cement mortar, regardless of the current amount of corrosion [29]. This relationship provides the convenience for evaluating the corrosion state of cover-zone concrete which is based on the cover mortar resistance, instead of being determined by R_p which is difficult to obtain on-site. It offers a simple solution, in a limited domain (mortar without chloride contamination), for a complex problem. However, it is apparent that in order to apply this relationship to deduce the R_p values of structural steel in service, we must pay attention to not only the complicated exposure circumstances, but also to the heterogeneity of the concrete due to the addition of the coarse aggregate.

In addition, to determinate the chloride contents using a new developed chloride ions sensor, and to clarify the relationship among the chloride contents, the cover-zone concrete resistance and the polarization resistance, further studies are now in process. Some adjustment is needed to use this relationship to estimate the corrosion state of field concrete structures in which chloride ions diffuse into the concrete. Another point worth emphasizing is that this model was developed at a given constant temperature and relative humidity. More research is required, for example, studying this relationship with temperature sensors, humidity sensors [30,31], chloride content sensors, etc., before the final goal of R_p prediction can be achieved.