Corrosion Behavior of Pipeline Carbon Steel under Different Iron Oxide Deposits in the District Heating System

: The corrosion behavior of pipeline steel covered by iron oxides ( α -FeOOH; Fe 3 O 4 and Fe 2 O 3 ) was investigated in simulated district heating water. In potentiodynamic polarization tests; the corrosion rate of pipeline steel is increased under the iron oxide but the increaseing rate is different due to the differnet chemical reactions of the covered iron oxides. Pitting corrosion was only observed on the α -FeOOH-covered specimen; which is caused by the crevice corrosion under the α -FeOOH. From Mott-Schottky and X-ray diffraction results; the surface reaction and oxide layer were dependent on the kind of iron oxides. The iron oxides deposit increases the failure risk of the pipeline and localized corrosion can be occurred under the α -FeOOH-covered region of the pipeline. Thus, prevention methods for the iron oxide deposit in the district pipeline system such as ﬁltering or periodic chemical cleaning are needed. Electrochemical tests were conducted to evaluate the corrosion behavior and properties of the iron oxide deposits. A three-electrode electrochemical cell was constructed with the SPPS 38 carbon steel electrode, uncovered and covered with iron oxide deposits as the working electrode (WE), two pure graphite rods as the counter electrodes (CE), and a saturated calomel electrode (SCE) as the reference electrode (RE). In potentiodynamic polarization tests, the specimens were fully covered with iron oxide deposits (1 cm 2 ), and a stainless steel rod covered with glass for preventing contact with the solution was used for the electric connection with the WE. The dynamic voltage (IR) drop was automatically compensated in the potentiostat because the simulated district heating water has low conductivity (347 µ S/cm). The schematic of the three-electrode electrochemical cell used in this experiment is shown in Figure 2. Before potentiodynamic polarization tests, the working electrode was immersed in the test solution for 6 h until a steady state was reached, and the open-circuit potential (OCP) was obtained. Potentiodynamic polarization curves were measured by scanning the potential from − 250 mV versus OCP to 400 mV SCE at a sweep rate of 0.166 mV/s. The electrochemical impedance spectroscopy (EIS) was performed at the OCP with amplitude of 10 mV at frequencies ranging from 10,000 to 0.01 Hz.


Introduction
A district heating (DH) system provides heat for inhabitants of large cities in an economical manner. Heat from heat and power generating plants and large heat sources is cheaper than heat produced in individual, low-power boiler rooms [1]. There are many merits to DH systems, including increased energy efficiencies, reduced life cycle costs, and elevated control over environmental impacts [2,3]. Additionally, the high efficiencies of DH systems decrease the emission of combustion products into the atmosphere. Because of these advantages, many countries such as Denmark, Russia, Finland, and Canada are developing DH systems, and the market share is reported to be approximately 50% [4,5]. Additionally, a DH system was introduced to Korea in 1987 and applied to various fields.
A DH system becomes more useful with more satellite and planned cities near each megalopolis. A DH system has a long stagnation period due to the construction of numerous urban housing facilities and infrastructure. Additionally, maintenance and improvement work are ongoing. During this period, the pipeline in the distribution part of the DH system is exposed to low temperature (~40 • C) water with little or no flow. The long stagnation condition has problems associated with corrosion such as deposits from solid particles (sand, debris, and iron oxides), ion conductivity, ferrous ions, and dissolved oxygen [6][7][8][9]. The deposits on the internal surface of a pipeline especially cause serious localized corrosion (such as pitting corrosion and crevice corrosion), which is identified as under deposit corrosion (UDC) [10][11][12].
UDC is unlike other corrosion forms, as the conditions, such as pH and concentration of the aggressive species, under the deposit are different from the conditions in regions without deposits. As a result, there will be galvanic corrosion between the areas under the deposit and the areas without  Table 2 lists the chemical composition of the simulated district heating water, which is based on actual district heating water over 10 years as measured in its stagnant condition. To adjust the chemical composition of the test solution, the pH was adjusted with a 0.1 M NaOH solution. Although the heating water was usually soft and pure, the detection of ferric ions (Fe 3+ ) and chloride ions (Cl − ) is attributed to the long stagnation condition [6]. Prior to testing, the solution was de-aerated by purging with N 2 (99.999%) for 2 h because the DH system is blocked from the outside, i.e., it is a closed system so that the amount of dissolved oxygen is very low. The temperature is maintained at 40 • C using a constant-temperature water bath.

Deposition Process
The three iron oxides (α-FeOOH, Fe 3 O 4 and Fe 2 O 3 ) used in this study were purchased from SAMCHUN chemical corporation (Pyeongtaek-si, Korea). The powder X-ray diffraction analysis is shown in Figure 1 and the iron oxides were confirmed with the diffraction database. Because the powder-state iron oxides are difficult to deposit on the specimen, ethanol was mixed with the iron oxides. The specimen was covered with a rubber 'O' shaped ring (diameter: 3 cm, height: 3 mm) and then filled with iron oxide mixed with ethanol, which has the same thickness as the rubber ring, approximately 3 mm, to simulate the UDC environment. Before immersion in the test solution for the electrochemical tests, the deposit-covered specimen was dried in air for 30 minutes to vaporize the ethanol.

Deposition Process
The three iron oxides (α-FeOOH, Fe3O4 and Fe2O3) used in this study were purchased from SAMCHUN chemical corporation (Pyeongtaek-si, Korea). The powder X-ray diffraction analysis is shown in Figure 1 and the iron oxides were confirmed with the diffraction database. Because the powder-state iron oxides are difficult to deposit on the specimen, ethanol was mixed with the iron oxides. The specimen was covered with a rubber 'O' shaped ring (diameter: 3 cm, height: 3 mm) and then filled with iron oxide mixed with ethanol, which has the same thickness as the rubber ring, approximately 3 mm, to simulate the UDC environment. Before immersion in the test solution for the electrochemical tests, the deposit-covered specimen was dried in air for 30 minutes to vaporize the ethanol.

Electrochemical Measurements
Electrochemical tests were conducted to evaluate the corrosion behavior and properties of the iron oxide deposits. A three-electrode electrochemical cell was constructed with the SPPS 38 carbon steel electrode, uncovered and covered with iron oxide deposits as the working electrode (WE), two pure graphite rods as the counter electrodes (CE), and a saturated calomel electrode (SCE) as the reference electrode (RE). In potentiodynamic polarization tests, the specimens were fully covered with iron oxide deposits (1 cm 2 ), and a stainless steel rod covered with glass for preventing contact with the solution was used for the electric connection with the WE. The dynamic voltage (IR) drop was automatically compensated in the potentiostat because the simulated district heating water has low conductivity (347 µS/cm). The schematic of the three-electrode electrochemical cell used in this experiment is shown in Figure 2. Before potentiodynamic polarization tests, the working electrode was immersed in the test solution for 6 h until a steady state was reached, and the open-circuit potential (OCP) was obtained. Potentiodynamic polarization curves were measured by scanning the potential from −250 mV versus OCP to 400 mV SCE at a sweep rate of 0.166 mV/s. The electrochemical impedance spectroscopy (EIS) was performed at the OCP with amplitude of 10 mV at frequencies ranging from 10,000 to 0.01 Hz.

Electrochemical Measurements
Electrochemical tests were conducted to evaluate the corrosion behavior and properties of the iron oxide deposits. A three-electrode electrochemical cell was constructed with the SPPS 38 carbon steel electrode, uncovered and covered with iron oxide deposits as the working electrode (WE), two pure graphite rods as the counter electrodes (CE), and a saturated calomel electrode (SCE) as the reference electrode (RE). In potentiodynamic polarization tests, the specimens were fully covered with iron oxide deposits (1 cm 2 ), and a stainless steel rod covered with glass for preventing contact with the solution was used for the electric connection with the WE. The dynamic voltage (IR) drop was automatically compensated in the potentiostat because the simulated district heating water has low conductivity (347 μS/cm). The schematic of the three-electrode electrochemical cell used in this experiment is shown in Figure 2. Before potentiodynamic polarization tests, the working electrode was immersed in the test solution for 6 h until a steady state was reached, and the open-circuit potential (OCP) was obtained. Potentiodynamic polarization curves were measured by scanning the potential from −250 mV versus OCP to 400 mVSCE at a sweep rate of 0.166 mV/s. The electrochemical impedance spectroscopy (EIS) was performed at the OCP with amplitude of 10 mV at frequencies ranging from 10,000 to 0.01 Hz. To accelerate the corrosion process, potentiostatic tests were performed at an applied potential of −0.65 mVSCE, slightly greater than the corrosion potential (Ecorr) of each specimen for 20 h. The fully deposit-covered specimens were evaluated to observe the corrosion process under the deposits. Additionally, to investigate the difference in corrosion behavior for covered and uncovered deposits, the deposit was designed to cover only half of the specimen. After potentiostatic tests, the iron oxide deposits and corrosion products on the carbon steel surface were removed and cleaned for 10 min in the cleaning solution containing 500 mL HCl (37% concentration), 3.5 g hexamethylenetetramine (C6H12N4), and distilled water to produce 1000 mL. The specimens were then rinsed in distilled water and dried with nitrogen gas.
To analyze the surface properties of uncovered and covered deposit specimens, Mott-Schottky measurements were performed with a PARSTAT 2263 (Princeton Applied Research, Oak Ridge, TN, To accelerate the corrosion process, potentiostatic tests were performed at an applied potential of −0.65 mV SCE , slightly greater than the corrosion potential (E corr ) of each specimen for 20 h. The fully deposit-covered specimens were evaluated to observe the corrosion process under the deposits. Additionally, to investigate the difference in corrosion behavior for covered and uncovered deposits, the deposit was designed to cover only half of the specimen. After potentiostatic tests, the iron oxide deposits and corrosion products on the carbon steel surface were removed and cleaned for 10 min in the cleaning solution containing 500 mL HCl (37% concentration), 3.5 g hexamethylenetetramine (C 6 H 12 N 4 ), and distilled water to produce 1000 mL. The specimens were then rinsed in distilled water and dried with nitrogen gas.
To analyze the surface properties of uncovered and covered deposit specimens, Mott-Schottky measurements were performed with a PARSTAT 2263 (Princeton Applied Research, Oak Ridge, TN, USA). Before the Mott-Schottky measurements, each specimen was immersed in the test solution for 340 h. The Mott-Schottky plots were obtained by sweeping in the positive direction at a frequency of 1 kHz with an amplitude signal of 10 mV, a potential range of −0.7 V SCE to 0.1 V SCE , and a potential step of 25 mV. All electrochemical tests were performed three times to ensure reproducibility.

Surface Characterization
After the potentiostatic tests, the morphology (surface and cross-section) of the corroded specimens was observed using optical microscopy (OM). The specimen after potentiostic test was cut by micro cutting machine, and then produced epoxy mounting. To clear observation of the damaged surface, the surface was cleaned by cleaning solution containing 500 mL HCl, 3.5 g hexamethylenetetramine (C 6 H 12 N 4 ) and distilled water to make 1000 mL for 10 min. Additionally, to investigate the surface condition under the oxide deposits, the specimens were analyzed by X-ray diffraction (XRD, D8 Advanced, Bruker, Billerica, MA, USA) using Cu-Kα radiation operated at 18 kW with a scanning speed of 2 • /min after 340 h of immersion. Before the XRD analysis, the deposit on the specimens was removed by distilled water and dried with N 2 gas to prevent additional oxidation. The XRD patterns were analyzed by the diffraction database. 1 kHz with an amplitude signal of 10 mV, a potential range of −0.7 VSCE to 0.1 VSCE, and a potential step of 25 mV. All electrochemical tests were performed three times to ensure reproducibility.

Surface Characterization
After the potentiostatic tests, the morphology (surface and cross-section) of the corroded specimens was observed using optical microscopy (OM). The specimen after potentiostic test was cut by micro cutting machine, and then produced epoxy mounting. To clear observation of the damaged surface, the surface was cleaned by cleaning solution containing 500 mL HCl, 3.5 g hexamethylenetetramine (C6H12N4) and distilled water to make 1000 mL for 10 min. Additionally, to investigate the surface condition under the oxide deposits, the specimens were analyzed by X-ray diffraction (XRD, D8 Advanced, Bruker, Billerica, MA, USA) using Cu-Kα radiation operated at 18 kW with a scanning speed of 2°/min after 340 h of immersion. Before the XRD analysis, the deposit on the specimens was removed by distilled water and dried with N2 gas to prevent additional oxidation. The XRD patterns were analyzed by the diffraction database. Figure 3 shows the polarization curves in the form of I = f(E) and with the extrapolated Tafel slopes of a bare specimen and different iron oxides (α-FeOOH, Fe3O4, Fe2O3)-covered specimens after a 6 h immersion in the test solution at 40 °C. All of the polarization curves showed active corrosion behavior. In the Fe3O4 and α-FeOOH cases, the current densities of cathodic and anodic polarization curves were increased compared to the bare specimen, and the corrosion potentials were negatively shifted. This indicates that the Fe3O4 and α-FeOOH deposits increased the cathodic and anodic reactions on the surface. Alternatively, in the case of Fe2O3, only the cathodic current density increased and the corrosion potential was not shifted. The corresponding kinetic parameters in the polarization curves, such as the corrosion potential (Ecorr), corrosion current density (icorr), polarization resistance (Rp), and anodic and cathodic Tafel slopes (βa, βc), are listed in Table 3. The icorr and Ecorr values were different for the iron oxide-covered specimens, and icorr was related to the corrosion rate according to Faraday's law:

Potentiodynamic Polarization Tests
where M is the molar mass of the metal (g/mol), z is the number of electrons transferred per metal atom, F is the Faraday's constant, and ρ is the density of the metal (g/cm 3 ). Thus, icorr was increased in all oxide-covered specimens, indicating the deposit state of the iron oxides increased the corrosion of the steel substrate. Also, the polarization resistance was calculated by the following equation [20]: The icorr and Rp are related to the corrosion resistance of the materials, and these were increased or decreased by the covered iron oxide, respectively. Consequently, it means that the corrosion reactions were affected by Fe3O4 and α-FeOOH deposits and the three iron oxide deposits (α-FeOOH, Fe3O4 and Fe2O3) affected the cathodic reactions [21][22][23]. The βa and βc were changed according to the kind of covered iron oxides. In the case of α-FeOOH, the βa was increased while the βc was decreased in comparison with the bare specimen. It indicates that the α-FeOOH influenced the anodic and cathodic reactions on the steel surface. On the other hand, the βa was not significantly changed and βc was changed in the cases of Fe3O4 and Fe2O3, meaning that the Fe3O4 and Fe2O3, mainly had an effect on the cathodic reaction than anodic reaction. Table 3. Electrochemical parameters of the polarization curves of the bare specimen and iron oxides (α-FeOOH, Fe3O4, Fe2O3)-covered specimens in the simulated district heating water at 40 °C.  The corresponding kinetic parameters in the polarization curves, such as the corrosion potential (E corr ), corrosion current density (i corr ), polarization resistance (R p ), and anodic and cathodic Tafel slopes (β a , β c ), are listed in Table 3. The i corr and E corr values were different for the iron oxide-covered specimens, and i corr was related to the corrosion rate according to Faraday's law:

Specimen
where M is the molar mass of the metal (g/mol), z is the number of electrons transferred per metal atom, F is the Faraday's constant, and ρ is the density of the metal (g/cm 3 ). Thus, i corr was increased in all oxide-covered specimens, indicating the deposit state of the iron oxides increased the corrosion of the steel substrate. Also, the polarization resistance was calculated by the following equation [20]: The i corr and R p are related to the corrosion resistance of the materials, and these were increased or decreased by the covered iron oxide, respectively. Consequently, it means that the corrosion reactions were affected by Fe 3 O 4 and α-FeOOH deposits and the three iron oxide deposits (α-FeOOH, Fe 3 O 4 and Fe 2 O 3 ) affected the cathodic reactions [21][22][23]. The β a and β c were changed according to the kind of covered iron oxides. In the case of α-FeOOH, the β a was increased while the β c was decreased in comparison with the bare specimen. It indicates that the α-FeOOH influenced the anodic and cathodic reactions on the steel surface. On the other hand, the β a was not significantly changed and β c was changed in the cases of Fe 3 O 4 and Fe 2 O 3 , meaning that the Fe 3 O 4 and Fe 2 O 3 , mainly had an effect on the cathodic reaction than anodic reaction. Figure 4 presents the EIS results in the form of the Bode phase and impedance plots for specimens covered with different iron oxides (α-FeOOH, Fe 3 O 4 , and Fe 2 O 3 ) and uncovered after 6 h immersion in the simulated district heating water at 40 • C. The Bode plot produces a more detailed explanation of the electrochemical frequency-dependent property than the Nyquist plot [24]. The high frequency spectra indicated local surface defect, whereas the low frequency spectra indicated the process within the film and at the metal/film interface [25]. The absolute impedance value of the specimen covered by iron oxides is less than that of the bare specimen. The absolute impedance value decreases in the following order: α-FeOOH > Fe 3 O 4 > Fe 2 O 3 > bare, which is the same trend as the corrosion current density in the potentiodynamic polarization tests. Additionally, the phase angle maximum decreased in the same order: α-FeOOH > Fe 3 O 4 > Fe 2 O 3 > bare, and the shoulder on the phase angle curve of the bare pipe shifted to a lower frequency. The results were due to inhomogeneity or porosity in the surface film [26], indicating that the surface film deteriorates after being covered with iron oxides, especially in the case of the α-FeOOH-covered specimen.  Figure 4 presents the EIS results in the form of the Bode phase and impedance plots for specimens covered with different iron oxides (α-FeOOH, Fe3O4, and Fe2O3) and uncovered after 6 h immersion in the simulated district heating water at 40 °C. The Bode plot produces a more detailed explanation of the electrochemical frequency-dependent property than the Nyquist plot [24]. The high frequency spectra indicated local surface defect, whereas the low frequency spectra indicated the process within the film and at the metal/film interface [25]. The absolute impedance value of the specimen covered by iron oxides is less than that of the bare specimen. The absolute impedance value decreases in the following order: α-FeOOH > Fe3O4 > Fe2O3 > bare, which is the same trend as the corrosion current density in the potentiodynamic polarization tests. Additionally, the phase angle maximum decreased in the same order: α-FeOOH > Fe3O4 > Fe2O3> bare, and the shoulder on the phase angle curve of the bare pipe shifted to a lower frequency. The results were due to inhomogeneity or porosity in the surface film [26], indicating that the surface film deteriorates after being covered with iron oxides, especially in the case of the α-FeOOH-covered specimen.    Figure 5 shows the equivalent circuit to fit the results of the EIS tests. In this equivalent circuit, the R s is the solution resistance, CPE1 (CPE: constant phase element) is the dielectric property of the film and water absorbed on the film, R film is the electrical resistance resulting from the ionic path of the pores in the film, CPE2 is the capacitance generated by metal dissolution and the electric double layer, and R ct is the resistance associated with metal dissolution. The CPE is applied in this equivalent circuit instead of a perfect capacitor to produce a more accurate fit and interpretation [27,28]. The impedance of CPE is expressed as:

Electrochemical Impedance Spectroscopy (EIS)
where Y 0 is the magnitude of CPE, j is the imaginary unit (j 2 = −1), α is the phase angle of CPE, and n = α/(π/2). The parameter n typically lies between 0.50 and 1.0, and CPE expresses an ideal capacitor at n = 1. The parameter n is often referred to as the frequency dispersion due to surface states like inhomogeneity [29,30] and roughness [31].
Metals 2017, 7, 182 8 of 20 Figure 5 shows the equivalent circuit to fit the results of the EIS tests. In this equivalent circuit, the Rs is the solution resistance, CPE1 (CPE: constant phase element) is the dielectric property of the film and water absorbed on the film, Rfilm is the electrical resistance resulting from the ionic path of the pores in the film, CPE2 is the capacitance generated by metal dissolution and the electric double layer, and Rct is the resistance associated with metal dissolution. The CPE is applied in this equivalent circuit instead of a perfect capacitor to produce a more accurate fit and interpretation [27,28]. The impedance of CPE is expressed as: where Y0 is the magnitude of CPE, j is the imaginary unit ( . The parameter n typically lies between 0.50 and 1.0, and CPE expresses an ideal capacitor at n = 1. The parameter n is often referred to as the frequency dispersion due to surface states like inhomogeneity [29,30] and roughness [31]. Some parameters extracted from the EIS data are listed in Table 4. To calculate the double layer capacitance (Cdl), the following equation was used: In Equation (4), Cdl is the double layer capacitance and fmax is the frequency at which the imaginary component of the impedance reaches the maximum value. The ZSimpwin program (v3.2, Princeton Applied Research, Oak Ridge, TN, USA) was used to fit the EIS data, which is presented in Table 4. Figure 6 shows the Rfilm and Rct of the specimens. Rfilm and Rct decreased for the iron oxide-covered specimens in the following order: α-FeOOH > Fe3O4 > Fe2O3 > bare. The covered iron oxides had the lower corrosion resistance than bare specimen. Additionally, the n value in the case of α-FeOOH was the lowest, indicating an unstable surface film due to pores and defects on the surface. Some parameters extracted from the EIS data are listed in Table 4. To calculate the double layer capacitance (C dl ), the following equation was used: In Equation (4), C dl is the double layer capacitance and f max is the frequency at which the imaginary component of the impedance reaches the maximum value. The ZSimpwin program (v3.2, Princeton Applied Research, Oak Ridge, TN, USA) was used to fit the EIS data, which is presented in Table 4. Figure 6 shows the R film and R ct of the specimens. R film and R ct decreased for the iron oxide-covered specimens in the following order: α-FeOOH > Fe 3 O 4 > Fe 2 O 3 > bare. The covered iron oxides had the lower corrosion resistance than bare specimen. Additionally, the n value in the case of α-FeOOH was the lowest, indicating an unstable surface film due to pores and defects on the surface.

Potentiostatic Tests
Potentiostatic tests were conducted to investigate the corrosion behavior of the iron oxide-covered specimens. Additionally, to identify galvanic corrosion between covered and uncovered specimens, half of the specimen was covered with deposit. Figure 7 shows the specimen covered iron oxides after potentiostatic tests under −650 mVSCE for 20 h in the test solution at 40 °C. As shown in Figure 7a, significant pitting was observed on the α-FeOOH-covered specimen and the size of the pits ranged from 150 to 300 μm. However, as shown in Figure 7b,c, pitting was not observed on the Fe3O4 and Fe2O3-covered specimens. Figure 8 shows the cross section of the α-FeOOH-covered specimen after the potentiostatic test. As shown in Figure 8, pitting was observed on the α-FeOOH-covered specimen and the depth of the pitting was approximately 170 μm. This indicates that the corrosion behavior was different in the α-FeOOH, Fe3O4, and Fe2O3 cases. Figure 9 shows the boundary of the specimens with covered (left side) and uncovered (right side) deposits after potentiostatic tests under −650 mVSCE for 20 h at 40 °C. In Figure 9a,b, the corrosion was accelerated at the boundary of the α-FeOOH and Fe3O4-covered specimens, which was caused by more negative corrosion potential than the bare specimen, i.e., a galvanic effect. Alternatively, the boundary between the Fe2O3-covered and uncovered did not show accelerating corrosion behavior as shown in Figure 9c, which could be due to the absence of the corrosion potential difference [17,[32][33][34]. In the cross-sectional view of the α-FeOOH and Fe3O4-covered specimens in Figure 10, obvious corrosion acceleration was observed. These results indicate that the α-FeOOH and Fe3O4 deposits can cause galvanic corrosion under stagnant conditions, while the Fe2O3 deposit did not cause a galvanic effect.

Potentiostatic Tests
Potentiostatic tests were conducted to investigate the corrosion behavior of the iron oxide-covered specimens. Additionally, to identify galvanic corrosion between covered and uncovered specimens, half of the specimen was covered with deposit. Figure 7 shows the specimen covered iron oxides after potentiostatic tests under −650 mV SCE for 20 h in the test solution at 40 • C. As shown in Figure 7a, significant pitting was observed on the α-FeOOH-covered specimen and the size of the pits ranged from 150 to 300 µm. However, as shown in Figure 7b,c, pitting was not observed on the Fe 3 O 4 and Fe 2 O 3 -covered specimens. Figure 8 shows the cross section of the α-FeOOH-covered specimen after the potentiostatic test. As shown in Figure 8, pitting was observed on the α-FeOOH-covered specimen and the depth of the pitting was approximately 170 µm. This indicates that the corrosion behavior was different in the α-FeOOH, Fe 3 O 4 , and Fe 2 O 3 cases. Figure 9 shows the boundary of the specimens with covered (left side) and uncovered (right side) deposits after potentiostatic tests under −650 mV SCE for 20 h at 40 • C. In Figure 9a,b, the corrosion was accelerated at the boundary of the α-FeOOH and Fe 3 O 4 -covered specimens, which was caused by more negative corrosion potential than the bare specimen, i.e., a galvanic effect. Alternatively, the boundary between the Fe 2 O 3 -covered and uncovered did not show accelerating corrosion behavior as shown in Figure 9c, which could be due to the absence of the corrosion potential difference [17,[32][33][34]. In the cross-sectional view of the α-FeOOH and Fe 3 O 4 -covered specimens in Figure 10, obvious corrosion acceleration was observed. These results indicate that the α-FeOOH and Fe 3 O 4 deposits can cause galvanic corrosion under stagnant conditions, while the Fe 2 O 3 deposit did not cause a galvanic effect.         Figure 11 shows the XRD results for the iron oxide-covered and uncovered specimens in the test solution for 340 h at 40 °C. In the Fe3O4-covered specimen, Fe and a few Fe3O4 peaks were observed. Similarly, in the case of the Fe2O3-covered specimen, Fe, Fe2O3, and a few Fe3O4 peaks were observed. The Fe3O4 and Fe2O3 deposits did not influence the formation of a thick surface film on the steel substrate. Even though the Fe3O4 peaks were observed in the Fe2O3 case, the Fe3O4 peaks overlapped with the Fe2O3-peaks and the intensity was insignificant. Thus, Fe3O4 was formed under the Fe2O3-covered state but the formation and reaction activities were too small. However, in the cases of α-FeOOH-covered and bare specimens, Fe3O4 diffraction peaks were observed. These results indicate that Fe3O4 is the main surface film on the bare state in the test environment, and α-FeOOH influences the formation of the Fe3O4 surface film on the steel substrate compared to the Fe2O3 and Fe3O4 cases.

Mott-Schottky Tests
It is well known that there is a relationship between the surface film properties related to the semiconducting behavior of a surface film and the corrosion behavior of materials [35,36]. These properties and behavior can be investigated by analyzing the curves of a Mott-Schottky plot. Thus, the surface film properties of a steel substrate under iron oxide deposits (α-FeOOH, Fe3O4, Fe2O3) were investigated using a Mott-Schottky plot based on measurements of the electrode capacitance as a function of the electrode potential (E). It was assumed that the capacitance of the space-charge was much less than that of the Helmholtz layer, and thus the electrode capacitance was equal to C.
According to Mott-Schottky theory, the space-charge capacitance and n-type and p-type semi-conductors are given by Equations (5) and (6), respectively: where ε denotes the relative dielectric constant of the layer, ε0 is the permittivity of a vacuum (8.8542

Mott-Schottky Tests
It is well known that there is a relationship between the surface film properties related to the semiconducting behavior of a surface film and the corrosion behavior of materials [35,36]. These properties and behavior can be investigated by analyzing the curves of a Mott-Schottky plot. Thus, the surface film properties of a steel substrate under iron oxide deposits (α-FeOOH, Fe 3 O 4 , Fe 2 O 3 ) were investigated using a Mott-Schottky plot based on measurements of the electrode capacitance as a function of the electrode potential (E). It was assumed that the capacitance of the space-charge was much less than that of the Helmholtz layer, and thus the electrode capacitance was equal to C.
According to Mott-Schottky theory, the space-charge capacitance and n-type and p-type semi-conductors are given by Equations (5) and (6), respectively: where ε denotes the relative dielectric constant of the layer, ε 0 is the permittivity of a vacuum (8.8542 × 10 -14 F/cm), e is the absolute value of the electron charge (1.6029 × 10 −19 C), k is the Boltzmann constant (1.389 × 10 −23 J/K), T is the absolute temperature, E FB is the flat band potential that can be obtained from the extrapolation of 1/C 2 to 0 V SCE , and N D and N A are the donor and acceptor densities, respectively, which can be determined from the slope of the experimental 1/C 2 versus applied potential [37][38][39]. Figure 12 shows the Mott-Schottky plots for the iron oxide-covered and uncovered specimens immersed in the test solution for 340 h at 40 • C. The slope of the Mott-Schottky diagram is negative, indicating p-type semiconducting behavior of the surface films in all conditions. In the iron oxides, the Fe 2 O 3 and α-FeOOH indicate the n-type property whereas the Fe 3 O 4 has p-type property [35]. It means that the Fe 3 O 4 is mainly formed on the surface in all conditions. According to Equation (3), acceptor density (N A ) can be determined from the slope of the experimental 1/C 2 versus E plots, i.e., more negative slope in Mott-Schottky plot suggests the decrease of acceptor density. Thus, the decrease of N A was observed according to the following order: α-FeOOH < bare < Fe 2 O 3 < Fe 3 [36,40]. Thus, the semi-conductive properties of the oxide layer depend on the kind of oxide and the concentration of cation or anion vacancies. Although the increase of the NA is related to the more disordered oxide layer and susceptible to corrosion in the passivity materials, this interpretation is not adequate in this study because the oxide layer was almost not formed on the Fe2O3 and Fe3O4 as shown in XRD results, and localized corrosion was only observed in the case of α-FeOOH-covered specimen. It is therefore the α-FeOOH on the surface and Fe3O4 formed beneath the α-FeOOH have an effect on the pitting corrosion. The thickness (W) of the space-charge layer, which is the surface film at a film formation potential, can be calculated by [41]: In this equation, Ef is the film formation potential, and the other parameters have the same The thickness (W) of the space-charge layer, which is the surface film at a film formation potential, can be calculated by [41]: In this equation, E f is the film formation potential, and the other parameters have the same meanings as those in Equation (5). The film thickness (W) of space-charge layer is proportional to 1/C 2 according to the following equation, obtained using Equations (6) and (7): Thus, W is increased in the following order: α-FeOOH > bare > Fe 2 O 3 > Fe 3 O 4 . This indicates that the iron oxide deposits are not considered a surface film.

Surface Reaction under the Iron Oxide Deposits
The potentiodynamic polarization curves indicated a negative shift in the corrosion potential and an increase in the cathodic and anodic reactions on the iron oxide-covered specimens. The negative shift in the corrosion potential in the α-FeOOH and Fe 3 O 4 -covered specimens means that the increase in the anodic reaction area was due to the surface reaction from the Fe 3 O 4 and α-FeOOH [42]. Additionally, the cathodic and anodic reactions for the α-FeOOH and Fe 3 O 4 -covered specimens increased. Therefore, the oxide deposit does not act as a surface barrier of the substrate, which is due to the porous properties of iron oxide particles [43].
The following anodic and cathodic reactions occur on the steel substrate under the test solution conditions (de-aeration and pH 10). Equation (9) is the anodic dissolution and Equation (10) is the cathodic reaction in an alkaline solution under the de-aeration condition [44]: Additionally, the Fe 3 O 4 formation reactions occurred as shown in the XRD results according to the following Equations [45,46]: The reduction reaction of α-FeOOH occurred and due to these reduction reactions, the cathodic reaction was increased as shown by the polarization curve. In addition, with the increase in α-FeOOH reduction reactions, more Fe 2+ ions were required to satisfy the equilibrium of Equations (13) and (14) such that the anodic dissolution of Equation (9) was increased simultaneously, as shown by the increase in the anodic reaction in the polarization curve. Based on these reactions, the formation of Fe 3 Equations (15) and (16)  Under the Fe 2 O 3 deposit, the following reaction occurs on the substrate [51,52]: Hematite, Fe 2 O 3 , is the final oxidation product among the iron oxides [53]. Thus, reduction reaction, Equation (17), occurs; this increases only cathodic reaction of the Fe 2 O 3 -covered specimen, as indicated in the polarization curve. The formation of Fe 3 O 4 under the Fe 2 O 3 deposit can be induced by Equation (18), but the reaction is limited due to the multiple reaction steps of Fe(OH) 3 -. Thus, the formation of Fe 3 O 4 under the Fe 2 O 3 deposit is difficult. The above reaction results were confirmed by Mott-Schottky and XRD analysis. In the Mott-Schottky results, the p-type semiconducting property and film thickness increased in the following order: α-FeOOH > bare> Fe 2 O 3 > Fe 3 O 4 . Additionally, Fe 3 O 4 , which has p-type semiconducting properties [54], was primarily formed on the α-FeOOH-covered and bare specimens, as shown in the XRD analysis results. The results indicate that different reactions on the surface caused by iron oxide deposits affect the formation of Fe 3 O 4 on the surface. In addition, the semiconducting property of the iron oxide deposits, which means α-FeOOH and Fe 2 O 3 are n-type and Fe 3 O 4 is p-type, was not observed in the Mott-Schottky results [36,55]. This implies that the iron oxide deposit on the specimen could not be analyzed by the Mott-Schottky measurement because the iron oxide deposits did not act as a surface film or barrier but rather as a reactant similar to a catalyst on the surface.

Mechanism of Different Corrosion Behavior under the Iron Oxide Deposits
In the potentiostatic tests, uniform corrosion behavior was indicated in all conditions, except the α-FeOOH-covered specimen. Pitting corrosion only occurred under the α-FeOOH-covered specimen. Uniform corrosion under the iron oxide deposits occurs due to the diffusion of external factors such as water and anions in the test solution, which were associated with porous properties of iron oxides that were revealed by the EIS and Mott-Schottky tests [56]. Under the α-FeOOH, pitting corrosion was observed, indicating that the cathode and anode areas were separated on the substrate.
Localized corrosion of steel can occur due to rust on the steel surface [18]. Two types of localized corrosion such as the UDC and the crevice corrosion were observed in this study. The rust on the steel and iron at the initial stage of corrosion is generally composed of Fe(OH) 3 , α-FeOOH, and Fe 3 O 4 from each transformation. The rust is visualized schematically in Figure 13. Fe(OH) 3 was transformed into α-FeOOH and then α-FeOOH was transformed into Fe 3 O 4 , as described by Equations (11) and (12). Based on these reactions, porous defects such as cracks and voids were developed in the rust layer, and the outside environment reaches the metal. The corrosion in these defects initiates a reaction, but the Fe 2+ and OHions cannot react directly because the remaining α-FeOOH rust obstructs their direct reaction. Due to these separated reactions, FeCl 2 and NaOH are formed in the anode and cathode regions, respectively. Acidification occurs due to the FeCl 2 in the anode region when the anode dissolution is accelerated. This corrosion under the thick rust layer can be considered crevice corrosion [19]. Figure 14 shows the crevice corrosion process under the α-FeOOH rust layer. Localized corrosion occurred under the α-FeOOH-covered specimen, even though the passive behavior was not observed on the polarization curve. developed in the rust layer, and the outside environment reaches the metal. The corrosion in these defects initiates a reaction, but the Fe 2+ and OHions cannot react directly because the remaining α-FeOOH rust obstructs their direct reaction. Due to these separated reactions, FeCl2 and NaOH are formed in the anode and cathode regions, respectively. Acidification occurs due to the FeCl2 in the anode region when the anode dissolution is accelerated. This corrosion under the thick rust layer can be considered crevice corrosion [19]. Figure 14 shows the crevice corrosion process under the α-FeOOH rust layer. Localized corrosion occurred under the α-FeOOH-covered specimen, even though the passive behavior was not observed on the polarization curve. Figure 13. Schematic of the rust film formed on the steel at the initial stage of corrosion. Reproduced with permission from [18]. Copyright Elsevier, 2008.

Conclusions
(1) In the potentiodynamic polarization results, icorr increased in all deposit-covered specimens, which means that the corrosion reaction increased due to the deposit. The α-FeOOH and Fe3O4-covered specimens demonstrate a negative shift in the corrosion potential and an increase in the number of anodic and cathodic reactions, and the Fe2O3-covered specimen only increased the cathodic reaction.
(2) Pitting corrosion was only observed on the α-FeOOH-covered specimen, while uniform corrosion was observed on the others. This occurs due to a crevice corrosion phenomenon, which occurs via localized corrosion under the α-FeOOH rust layer on the steel substrate. Additionally, the corrosion at the boundary between deposit-covered and uncovered pipe was observed on the α-FeOOH and Fe3O4 due to the galvanic effect.
(3) The Mott-Schottky and XRD analyses revealed that surface reactions differ according to the kind of iron oxide deposited, which influences the formation of surface film and the reaction rate of

Conclusions
(1) In the potentiodynamic polarization results, i corr increased in all deposit-covered specimens, which means that the corrosion reaction increased due to the deposit. The α-FeOOH and Fe 3 O 4 -covered specimens demonstrate a negative shift in the corrosion potential and an increase in the number of anodic and cathodic reactions, and the Fe 2 O 3 -covered specimen only increased the cathodic reaction.
(2) Pitting corrosion was only observed on the α-FeOOH-covered specimen, while uniform corrosion was observed on the others. This occurs due to a crevice corrosion phenomenon, which occurs via localized corrosion under the α-FeOOH rust layer on the steel substrate. Additionally, the corrosion at the boundary between deposit-covered and uncovered pipe was observed on the α-FeOOH and Fe 3 O 4 due to the galvanic effect.
(3) The Mott-Schottky and XRD analyses revealed that surface reactions differ according to the kind of iron oxide deposited, which influences the formation of surface film and the reaction rate of anode and cathode sites.
Based on this study, the iron oxides covered as a deposit on the steel substrate had a negative effect on the corrosion in district heating water. α-FeOOH can lead to localized corrosion, which causes failure in the early stage. Thus, stagnant condition prevention and floating matter filtration are important tasks for maintaining the pipeline. It would be helpful to purge with nitrogen to maintain a minimum flow rate rather than to fill the pipe with water during long downtimes.