**2. Experiments**

#### *2.1. Material and Composition*

The samples used in this investigation were central wires of prestress strand. Three kinds of samples were chosen: Sample A from Jiangyin, China, Walsin steel cable Co., Ltd., conforming to BS 5896 [21] and using C86D2 grade steel [24] and Samples B and C, according to GB/T 5224 [22] and using YL87B grade steel [25] but from di fferent manufacturers (Xinhua Metal Products Co., Ltd., Xinyu, China and, Walsin steel cable Co., Ltd., Jiangyin, China respectively). The chemical compositions of steel samples were determined by GB/T 4336 [26], and the results are given in Table 1. Besides, the measurement uncertainty for chemical composition is given in Table 2.


**Table 1.** Chemical composition of steel wires (mass fraction)/%.


**Table 2.** Measurement uncertainty for chemical composition (%).

#### *2.2. Specimen Preparation*

The rupture strength *f* t of wires, A, B and C, were measured through axial tension loading tests on three specimens for each wire in lab-air environment, and the average values were retained as its fracture strength: 1971 MPa, 2106 MPa and 2088 MPa for Wires A, B and C, respectively. The relative composite measurement uncertainty for fracture strength values 1971 MPa, 2106 MPa and 2088 MPa is 0.699%, 0.722% and 0.718% respectively.

For the immersion tests, the sample length is 1280 mm, twice as long as the immersion part in the solution, and the diameter of sample is 5 mm. The NH4SCN solution was prepared by dissolving 200 g of analytically pure NH4SCN in 800 mL of distilled water. The stress corrosion testing device is composed of a loading frame and an immersion pool; see Figure 1. During the immersion tests, electrochemical measurements were performed for the steel wires in immersion.

**Figure 1.** Experimental setup for accelerated corrosion tests for prestressed wires.

#### *2.3. Test Procedure*

The samples were wiped, degreased with acetone (CH3COCH3), and air-dried. At least 50 mm in length parts at both ends of the samples were coated with sealant to prevent corrosion. The samples were loaded to 70% of the rupture strength and the loading was sustained, with variation controlled within +2%, during the entire test duration. After loading, the container was sealed to prevent leakage, and the solution was replaced after each test. The NH4SCN solution (Tengtai Chemical Technology Co., Ltd., Suzhou, China) was first deoxidized by introducing nitrogen gas during 2 h, then pre-heated to 50–55 ◦C, and injected into the container and kept at a constant temperature. The solution was filled within 1 min, covering the surface of sample and kept stagnant during the stress corrosion tests. The following cases were tested: (1) Samples A, B and C under stress level 70% *f* t with loading-immersion duration 10 min; (2) Samples A under free stress (0% *f* t) with loading-immersion duration 10 min, and stress level (70% *f* t) with loading-immersion duration 30 min.

During all these tests, the potentiodynamic polarization and EIS (Chenhua Instrument Co., Ltd., Chi600E, Shanghai, China) were used to characterize the corrosion behavior of steel wires. The electrochemical measurements were performed via a three-electrode system containing a stainless-steel auxiliary electrode and a saturated calomel electrode (SCE) as reference, and the potentials hereafter refer to the SCE. The prestressed steel wire, acting as working electrode, was immersed in NH4SCN solution with an exposed working area of 100 cm2. The electrochemical tests were conducted when the open circuit potential (OCP) of the working electrode became stable. The scanning potential of the anodic polarization curve was −1.5–1.0 V vs. OCP, and the scanning rate was 5 mV/s. The sinusoidal voltage excitation signal with disturbance amplitude of 5 mV vs. OCP was used for EIS testing, and the frequency range was 105–10−<sup>2</sup> Hz. All the electrochemical measurements were repeated three times for good reproducibility, and a typical group of data were selected to study for clarity.

#### **3. Electrochemical Analysis**

#### *3.1. Potentiodynamic Polarization Analysis*

The potentiodynamic polarization results are given in Figure 2 and Table 3. For the cases of loading-immersion duration of 10 min, the potentiodynamic polarization curves of wires B and C were on the left side of Wire A, and the curves of B and C were rather close. For the case of Wire A of loading-immersion duration 30 min, the polarization curve of wire A moves towards the bottom right, and the stable range becomes narrower, indicating that longer exposure time promotes the anodic dissolution of steel and the corrosion resistance is weakened [27,28]. Furthermore, compared to the A—10 min case, the case of A-without stress presents a left-upward shift in the polarization curves, showing wider range of the stable interval and the degree of corrosion was minimized [29].

In Table 3, the corrosion potential (*E*corr), corrosion current density (*i*corr), anodic Tafel slope (*b*a) and cathodic Tafel slope (*b*c) values were taken from the polarization curves in Figure 2. Compared to the cases B, C—10 min, the case A—10 min has lower *E*corr, *b*a and *b*c values but larger *i*corr values. By extending the loading-immersion time from 10 min to 30 min (Case A—10 min to Case A—30 min), the value of *i*corr increased from 1.28 × 10−<sup>2</sup> <sup>A</sup>/cm<sup>2</sup> to 1.90 × 10−<sup>2</sup> <sup>A</sup>/cm2, due to the *E*corr descended from −315.5 to −349.0 mV vs. SCE. Compared to Case A—10 min, the stress-free case, A-without stress, has the *E*corr value increased from −315.5 mV to −290.3 mV and *i*corr decreased from 1.28 × 10−<sup>2</sup> <sup>A</sup>/cm<sup>2</sup> to 0.34 × 10−<sup>2</sup> <sup>A</sup>/cm2. From these values, the current density of A was higher than those of B and C under the same corrosion conditions, which indicates that wires B and C had lower corrosion rate via the immersion tests. Moreover, for wires A, the current density increased with the application of tensile stress loading and longer loading-immersion duration, indicating that the corrosion process was accelerated. Tang [30] studied the e ffect of stress on corrosion of *X*70 pipeline steel in neutral solution with microzone electrochemical method. They found that the corrosion rate increased significantly with the stress loading, which confirmed that stress promoted the occurrence of corrosion.

**Table 3.** Electrochemical parameters of polarization curves of prestressed steel wires.


**Figure 2.** Potentiodynamic polarization curves of steel wires in NH4SCN solution.

#### *3.2. EIS Analysis*

Figure 3 provides the experimentally obtained Nyquist curves and the EIS of response evolution of the corrosion products in di fferent cases. Nyquist curves for the prestressed wire samples showed the similar characteristics. All of the Nyquist curves showed incomplete depressed semi-circular arcs that were a ffected by the frequency dispersion, and inductive shrinking occurred in the low-frequency zone, showing that such adsorbents as ferro-thiocyanate film existed on the steel surface and the corrosion was activated during this frequency range [26]. Under the same stress-loading duration (A, B, C—10 min cases), the Nyquist curves showed that the capacitance arc magnitude of B were larger than those of A and C, which suggests a more compact film of corrosion products was formed on the surface of B and the corrosion resistance of B wire is better than A and C wires. The polarization curves are extrapolated from the dynamic electrode behavior of strong polarization, while the EIS is measured under the stable equilibrium state. Therefore, the behavior of the electrode in static equilibrium can be better characterized by EIS results.

Further, as the loading-immersion time increases, from 10 min (A—10 min) to 30 min (A—30 min), the capacitance arc magnitudes are basically the same. By comparing the Nyquist curves of stressed wires (A, B, C under 70% *f* t) and stress free wire (A, 0% *f* t), it was found that the capacitance arc magnitude of stress free wire A is much larger than the wires (A, B, C) under 70% *f* t tensile stress, indicating that the tensile stress promotes substantially the occurrence of corrosion. This observation is consistent with the anodic polarization curves in Figure 2. In Figure 3, the *Z*re values, real part of the impedance, were not the same for di fferent wires. This is due to the fact that the NH4SCN solution was replaced and refilled into the accelerated corrosion container after each individual wire test, resulting in a slight change in the corrosive environment. However, this change in *Z*re value does not change the judgement on the role of tensile stress.

The polarization resistance can be obtained from low frequency impedance, which determines the change in transfer resistance [31]. As shown in Figure 4a, when the influence caused by the change of solution impedance was ignored, the Bode curves of di fferent wires revealed that the impedance modulus of Wire C in the low frequency region was slightly higher than that of Wire A under the same loading-immersion duration. Moreover, the impedance modulus of B was much higher than those of A and C, which indicates that the surface film resistance of B was higher and a significant charge transfer resistance of Wire B was created; therefore, the corrosion resistance of Wire B was higher than Wires A and C. Figure 4b shows the phase angle of three stressed wires (A, B, C—10 min) increased with the frequency under the same stress-corrosion time. The peak value of C was the highest, showing that the surface was smoother, and the pitting corrosion was inhibited. With corrosion time prolonged from 10 min to 30 min, the impedance modulus and phase angle of Wire A decreased, and the corrosion resistance decreased accordingly. When Wire A is under stress-free condition, the resistance of the surface film in the low frequency region was larger than the stressed case of A—10 min, and the resistance decreased also with frequency. Meanwhile, the peak value of stressed A—10 min was lower than the stress-free case, indicating that the formation of compact products was inhibited under tensile stress, and the product film was rougher.

**Figure 3.** Nyquist curves of steel wires of di fferent cases in NH4SCN solution.

The models in literature [32–34] are used to describe the electrochemical processes of the surface films exposed to di fferent conditions, and interpret the obtained EIS curves, see Figure 5. In the figure, *R*sol represents the resistance of the solution, the constant phase angle element CPE1 corresponds to the double-layer capacitance of the interface between the sample surface and solution, *R*1 represented the charge transfer resistance, *R*2 was the surface film resistance, the constant phase angle element CPE2 was used as a substitute for the surface film capacitance, and *n* represented the dispersion exponent. Through this model, the electrochemical parameters can be regressed to represent corrosion kinetics for di fferent cases. In general, the electron transfer resistance determines the impedance at low frequencies, the solution resistance determines the impedance at high frequencies, and the electrochemical corrosion kinetics can be extrapolated by the low frequency impedance.

Due to the heterogeneity of the surface film [35], a constant phase element (CPE) is used to represent the non-ideal capacitance responses of the interface. The CPE was defined as,

$$Z\_{\rm CPE} = \frac{1}{\chi\_0 (jw)^n} \tag{1}$$

where *Y*0 is the admittance magnitude of CPE; ω is the angular frequency; *j* is the imaginary number (*j* 2 = −1) and *n* is the exponent (−1 < *n* < 1). *Y*0 and *n* can be converted into CPE. The *n* value is interrelated to the heterogeneity and smoothness of the surfaces. When *n* values are close to 1, the CPE will approach an ideal capacitance.

The parameters are regressed and given in Table 4 from the EIS data for steel wires in di fferent cases. Under the same stress-corrosion time (70% *f* t, 10 min), the *R*1 value of Wires B and C are higher than Wire A, indicating the transfer of electrons is more difficult in Wire B and C. Therefore, Wires B and C show better resistance to the redox reaction of corrosion relative to A. The *R*2 value of Wire B is the highest among A, B, C—10 min cases, which means that Wire B has the largest surface film impedance and highest compactness. Compared to Wires B and C, the CPE1 and CPE2 values of Wire A are slightly higher, showing Wire A is more prone to corrode. The values from A—10 min and A—30 min cases show a general decrease of resistance and capacitance values, indicating the charge transfer processes were promoted with the immersion time from 10 min to 30 min.

**Figure 4.** Bode curves of steel wires immersed in NH4SCN solution: impedance modulus (**a**) and impedance angle (**b**).

The *R*1 values of Sample A cases showed the longer loading-immersion duration, in immersion solution, tends to decrease R1 value, in other terms, promote the electron transfer and weaken the corrosion resistance. The *R*2 values decreased accordingly by longer loading-immersion duration and tensile stress, possibly attributed to the deterioration of surface film. The evolution of CPE1 and CPE2 increased with prolonged loading-immersion time, indicating that the surface film formed under prolonged loading duration leads to capacitive behaviors, and the corrosion resistance is deteriorated. In addition, with the increase in loading-immersion duration, *n* (*<sup>n</sup>*1, *n*2) values of the double layer capacitance and the surface film capacitance decreased, respectively. These observations confirm that the corrosion of steel wires was gradually intensified. Finally, when comparing the stress-free case of Wire A with other stressed cases, one gets a similar *R*sol value but a much higher *R*1 value in Table 4, meaning the electrons transfer resistance in stress-free wire A is much higher than those stressed wires (A, B or C). When pitting corrosion occurred on the surface of the stressed wire, the potential of pitting corrosion area was lower than that of other parts, which resulted in the area that became active and provided crack core for stress corrosion. The concentrated stress made the crack tip and the surrounding area yield deformation, and then the micro-slip destroyed the surface film of the crack tip again, which accelerated the dissolution of the tip. At the same time, because of the existence of micro cracks, the corrosion resistance of the surface film deteriorated, and the electron transfer resistance decreased. This observation confirms further that the tensile stress, 70% *ft*, substantially decreased the surface electron transfer resistance, thus promoting the SCC of the wires.

**Figure 5.** Equivalent electrical circuit used to represent the measured EIS data.

.



#### **4. Microstructure Analysis**
