2.1 Samples and analysis

Samples of corrosion products were obtained from Units 1 and 2 BWRs. The samples comprised corrosion products collected by cellulose particulate filters (47 mm diameter; 0.45 μm pore-size) using an integrated filtration method. Insoluble corrosion products and radioactive transition metal species were collected from approximately 1000 to 2000 L of water drawn from cross-linked resin beds for iron removal situated at the condensate demineralizer inlet (CDI) and condensate demineralizer effluent (CDE), and from the final feed water sampling port as well. The feed water was cooled and depressurized before being passed through the filters.

Elemental analysis was performed using inductively-coupled plasma-atomic emission spectroscopy (ICP-AES) and radioactivity of the samples was measured using gamma-ray spectrometry. Scanning electron microscopy (SEM) was used to characterize morphology of the corrosion products and X-ray powder diffraction (XRD) was used for phase determination.

2.2 SEM/EDS

Topography of crud was determined using an SEM (JEOL, JSM-6510), which has a fully-integrated energy dispersive X-ray analysis system (EDS). Crud samples were collected on 8 mm² pieces of double-sided carbon conductive adhesive tape mounted on sample holders and then dried under an infrared lamp (250 W) for 1 h.

2.3 XRD

Crud analysis was performed using a Bruker’s D8 Advance diffractometer operated at 40 kV and 40 mA with a ceramic X-ray tube with Cu target (wavelength 1.5418 Å). A vertical theta-theta goniometer of -110°< 2θ < 168° was controlled using stepper motors in conjunction with optical encoders providing the smallest selectable step size 0.001. The scanning range was 2θ=30°-80° at a speed of 1 °/min. Two exchangeable devices were used to detect scattered X-rays: 1) a NaI(Tl) scintillation counter with background of 0.3 cps and dynamic range up to 2d6; and 2) a LynxEYE compound silicon strip detector with lower background (< 0.1 cps) and the same resolution but more than 150 times faster. Moreover, the LynxEYE detector has fluorescence suppression capability, which eliminates the need for secondary monochromators. A reference standard, Corundum (Al₂O₃, NIST SRM 1976a), was used for calibration of the equipment by bringing the diffraction angle to zero prior to sample measurement. The program TOPAS Spectra was used for comparing the sample and reference diffraction spectra (TOPAS spectra) to enable the identification of the fractions of various crystalline phases in the samples.

2.4 ICP-AES

The concentrations of Fe, Co, Cu, Ni, Cr, Zn, and Mn (elements commonly encountered in corrosion products) were detected using ICP-AES (Horiba JY Ultima 2 ICP-AES) and normalized to a total of 100%.

2.5 Analytical flowchart

A flowchart outlining the proposed sampling and analysis method is presented in Fig. 1. The corrosion products were analyzed using a dissolution method. Insoluble iron oxides were obtained at each sampling point in the condensate and feedwater system. The crystalline phase of the corrosion products varied with water chemistry conditions in the BWR. The samples thus included mostly crystalline hematite, magnetite, goethite, lepidocrocite, and non-crystallite under NWC and HWC environments. At first we were to do quantitative and qualitative analysis with XRD. However, the iron-bearing phases were of a poor crystallite structure and could not usually be identified using XRD. Therefore, the crystalline and amorphous forms of iron oxides were separated using a dissolving method with an appropriate chemical solution. Following filtration, the insoluble oxide was re-dissolved in aqua regia and then analyzed again using ICP-AES [6, 7].

Fig. 1.Full-screen View

Analytical flowchart of corrosion products.

A series of experiments was conducted to determine reliability of the techniques used for identification of iron oxides in the condensate systems of the BWR. The experiments are outlined as follows:

(1) Iron oxide standards were tested using the proposed dissolving method(the specifics will be discussed later).

(2) The Fe concentration was analyzed using ICP-AES.

(3) The solubility of the iron oxide standard was calculated from the ICP-AES results.

Experiment data related to the proposed dissolution method are listed in Table 1. The experiment results clearly fit the expected values, indicating that the standard solubility of crystal iron oxides in the forms of hematite, magnetite, and goethite were lower than 2%–7%. However, the solubility of amorphous iron oxide and lepidocrocite exceeded 90%. This indicates that the above pre-tests of crystallite and amorphous compositions in all samples can be calculated from the newly developed dissolution methods by the specific acid dissolution method.

3.1 XRD analysis

Most of the piping used for the heater drain lines and condensate systems of BWRs are made of ferrous materials; therefore, the main component in the corrosion products is iron oxides. The corrosion products removed from different units were comparable in morphology and particle size distribution. All the iron oxides from both BWRs present the following crystallization shapes: γ-FeOOH, α-FeOOH, α-Fe₂O₃ and Fe₃O₄. Generally, quantitative analysis related to the shape of crystallites in iron oxides can be performed using FTIR [8], Mössbauer spectroscopy [9], or XRD analysis [10]. In our preliminary experiment, we performed qualitative analysis using XRD.

Figure 2 illustrates XRD patterns of three samples at different sampling points in Unit 1 operated under HWC. The XRD peak positions were calibrated against the standard using Bruker TOPAS software. This method reduced the error associated with quantitative analysis [11, 12]. Figures 3 and 4 present the fraction and relative concentrations of iron species in condensate influent, effluent and feedwater in Units 1 and 2 operated under NWC or HWC conditions. Under both conditions, the concentrations of α-Fe₂O₃ (space group R3c, a=5.04 Å, c=13.77 Å of unit cell) was of the highest proportions. The condensate demineralizer produced the highest proportion of α-Fe₂O₃ under NWC conditions but it was comparable in proportion under HWC condition. Significant differences in morphological composition of CDI were observed between Units 1 and 2 under NWC and HWC conditions.

Fig. 2.Full-screen View

(Color online) XRD analysis of samples obtained at various sampling points in Unit 1 operated under HWC.

Fig. 3.Full-screen View

(Color online) Fraction of iron species from condensate influent, effluent and feed water in Unit 1 operated in NWC (a) and HWC (b).

Fig. 4.Full-screen View

(Color online) Concentrations of iron species from influent, effluent, and final feed water of Unit 1 (a) and Unit 2 (b).

The corrosion products formed in the condensate demineralizer and feedwater comprised mainly α-Fe₂O₃ (hematite). Hematite is a trigonal crystal structure, which is weakly anti-ferromagnetic at room temperature. Thermodynamic theory and experiment data indicate that the Fe³⁺ oxide phases of hematite and goethite structures are stable. The formation of hematite and goethite depends on the water temperature and pressure. Hematite forms under high temperatures, whereas goethite forms under high pressure. CDI temperatures are maintained at 30 ℃–40 ℃, and corrosion products, such as α-FeOOH and α-Fe₂O₃, commonly form on the pipe walls. However, increasing the temperatures of FW within a pipeline to 200 ℃–240 ℃ tends to cause the formation of α-Fe₂O₃. At higher temperatures, nearly all Fe in reactor water is an insoluble and stable form of α-Fe₂O₃. Kim indicated that extending the immersion time of oxide film on type 316 stainless steel in water at 288 ℃ was required to stabilize the potentials upon changing the water condition from NWC to HWC. This might have a slow transition rate of the α-Fe₂O₃ phase to Fe₃O₄ phase. Medium-level hydrogen injection has been in use for more than three years in the plants in this study. Thus, the reduction of α-Fe₂O₃ to Fe₃O₄ in pure water at 288 ℃ appears to be kinetically inhibited. This can be attributed to the fact that nearly all of the iron crud deposited on the outer cladding surface was in a fully oxidized state, α-Fe₂O₃ [13, 14].

Figure 3 indicates that the relative percentage of hematite increased rapidly in the FW under NWC and HWC conditions. Figure 4 shows that the relative percentages of goethite, lepidocrocite and magnetite in the FW were far lower than those obtained in the condensate. When coolant-borne iron particles pass through a resin bed, most of the large size crystalline is removed. Small residual particles transform into hematite under the high temperatures encountered in the FW.

3.2 Dissolution method and ICP-AES analysis

The samples for ICP-AES analysis were treated with a specific chemical solution to dissolve the iron oxides in the condensate system, and reliability of the treatment was studies with a series of experiments. Following treatment of the crud samples, the amorphous and crystalline structures were separated using a millipore filter. The procedures of dissolving the iron oxide and treating the of samples are outlined as follows:

(1) Add acidic reagents to the samples, disperse them ultrasonically, and heat them to a high temperature;

(2) Samples are separated using a millipore filter at room temperature;

(3) The filtrate is analyzed using ICP-AES;

(4) The insoluble oxides are re-dissolved using aqua regia, and are analyzed using ICP-AES; and

(5) Calculate the ICP-AES data.

Table 2 lists the composition of corrosion products from Units 1 and 2. The amount of amorphous iron oxide in the CDI of Unit 2 is approximately 5% higher than that observed in Unit 1, under NWC and HWC conditions. Table 3 lists the typical composition of iron oxide corrosion products in KNPS condensate since 1993.

3.3 SEM and EDS analysis

The iron oxide corrosion products, including surface morphology, particle size, and composition, were characterized by SEM (Fig. 5). The SEM images of the CDI and CDE samples reveal aggregates of very small dispersible particles (< 2 μm) and elongated particles (>0.45 μm), while the corrosion products in the FW include small dense particles with a partial acicular structure. The large particles in the CDI could be removed from the cooling water by the condensate demineralizer. Between the CDE and FW, the cooling water temperature increases from 42 ℃ to 215 ℃. Under these conditions, corrosion products in the form of small particles would transform into hematite structure. The SEM results indicate that the corrosion products from Units 1 and 2 have similar crystalline structure except for the proportion of particle size. The samples obtained from the CDI of the HWC contain rhomboidal (α-Fe₂O₃), octahedral (Fe₃O₄) and acicular (α-FeOOH) particles, and a higher proportion of rhomboidal (α-Fe₂O₃) particles were generated in the CDE. Both the SEM and XRD results indicate that no significant change is occurred following a switch from NWC to HWC.

Fig. 5.Full-screen View

(Color online) SEM images of corrosion product samples from the CDI, CDE and FW in Units 1 and 2.

Figure 6 shows size distribution of the CDI corrosion products at Units 1 and 2. The small particles in Unit 2 are larger in number than in Unit 1. The size distribution of dispersible particles were as follows: CDI (0.45 μm–1.5 μm), CDE (0.45 μm–2.0 μm), and FW (0.45 μm–1.0 μm).

Fig. 6.Full-screen View

Particle size distribution of corrosion products from the CDI in Units 1 and 2.

The results of EDS analysis are listed in Table 4. The elemental compositions of samples obtained at the CDI, CDE, and FW of Unit 1 include mainly Fe, with minor amounts of Cu, Ni, and Mn. The source of the Cu, Ni, and Mn may be the materials used in the piping.

3.4 Removal efficiency of condensate demineralizer

Table 5 illustrates the removal efficiency of condensate demineralizer in Units 1 and 2. We used the direct data from KNPS operation and calculated results with the experimental data to obtain the removal efficiency. The removal efficiency of the condensate demineralizer in Unit 2 was 2%-15 % lower than that of Unit 1 under NWC and HWC conditions. From Figs. 5 and 6, particle size of the corrosion products in the CDI of Unit 2 is smaller than that of Unit 1. Such small particles are difficult to filter out using a condensate demineralizer. In addition, the quantity of amorphous iron oxide in the CDI of Unit 2 is higher than that in Unit 1, as shown in Table 2.

The removal efficiency of corrosion products depends primarily on the chemical form and particle size of the corrosion product intended for adsorption on the resin surfaces. Smaller particles pass easily through the condensate demineralizer, whereupon crystalline and amorphous structures adsorb on the cation exchange resins due to differences in zeta potential. At approximately pH 7, the corrosion products are positively charged, and the cation exchange resins are negatively charged. Thus, the corrosion products are adsorbed on the surface of the cation exchange resins by an attractive force. However, an amorphous structure has lower potential than does a crystalline structure. As a result, the proportion of crystalline to amorphous material can affect the removal efficiency of corrosion products. It is probable that Unit 2 has a higher proportion of amorphous structures of smaller size, which resulted in lower removal efficiency than that observed in Unit 1. This could be explained by the fact that the removal efficiency of condensate demineralizer appears not to be influenced by whether the removal process occurs under NWC or HWC conditions.