Failure analysis of copper pipes used in the heat exchangers in fan coil units

3.1. Optical microscope analysis

When all the samples were examined by naked eye, there was no clear failure on the sample surfaces. Thus, all samples’ cross sections were examined by optical microscope in order to observe and find evidence for the possible corrosion and failure modes. Figure 2 shows the images of both failed and defect free copper pipe cross sections. It can be seen that there are no cracks or micro tunnel formations in the C1 and C2 samples. On the other hand, tunnel-like formations and defects were discovered in the C3 sample. When the working conditions and the failure type of the C3 sample considered, tunnel formations could be evaluated as the main reason of leakage. Since there was no visible corrosion damage on the sample inner surface and the material was operating in a fluid-liquid environment at 70-80 °C, it was determined that the corrosion in the C3 sample was similar to ant-nest corrosion (formicary corrosion). Ant-nest type corrosion occurs on the surface of the copper pipe in the form of very fine pinhole-like leaks that are not visible to the human eye. When examined under a microscope, ant-nest type corrosion pits show a tunnel structure connected to each other along the copper pipe (Moss, 1969Moss, A.K. (1969). The corrosion of copper and copper alloys. Australian Corrosion Engineering 13 (5), 5-11.; Peltola and Lindgren, 2015Peltola, H., Lindgren, M. (2015). Failure analysis of a copper tube in a finned heat exchanger. Eng. Fail. Anal. 51, 83-97. https://doi.org/10.1016/j.engfailanal.2015.02.016.). In addition studies in the literature showed copper pipes that were exposed to formic acid and similar corrosive environments resulted with tunnel structure and failed because of ant nest corrosion (Situmorang and Kawai, 2018Situmorang, R.S., Kawai, H. (2018). Investigating the mechanism behind ‘ant nest’corrosion on copper tube. Materials 11 (4), 533. https://doi.org/10.3390/ma11040533.; Zhou et al., 2018Zhou, J., Yan, L., Tang, J., Sun, Z., Ma, L. (2018). Interactive effect of ant nest corrosion and stress corrosion on the failure of copper tubes. Eng. Fail. Anal. 83, 9-16. https://doi.org/10.1016/j.engfailanal.2017.09.013.; Cozzarini et al., 2020Cozzarini, L., Marsich, L., Schmid, C. (2020). Ant-nest corrosion failure of heat exchangers copper pipes. Eng. Fail. Anal. 109, 104387. https://doi.org/10.1016/j.engfailanal.2020.104387.).

3.2. SEM and EDS analyses

Figure 3 and Fig. 5 shows SEM images of C1, C2 and C3 samples, respectively. When the images of the C1 sample were examined, micro cracks and contamination-like structures were observed in various regions of the sample surface (indicated by arrows). The presence of two structures was considered as a result of SEM analysis in Fig. 3a. When the C2 sample was examined by naked eye, a dark brown structure was seen easily on the surface of the sample. In Fig. 4, it was seen that dust-like structures deposited on the sample surface with different ratios in the different regions on the sample surface. In addition, microcrack formation which thought to be related to bending stresses applied during sample preparation process was observed at various regions in both C1 and C2 sample. SEM images of the C3 sample in Fig. 5 showed that the inner surface of the sample was apparently covered with a black corrosion product. It was observed that the amount of corrosion product accumulated on the surface was much higher than C2 sample. Also, the amount of corrosion product deposition on the surface showed regional differences and there were various microcracks and large cavities on the surface of these. In addition, it was thought that different structures exist due to contrast differences in Fig. 5a. According to SEM results, it was determined that the corrosion products on the surface increased over time and at the same time, these corrosion products were too high amount in the samples exposed to corrosion. In order to observe leakage locations more clearly, corrosion products on the surface were removed by gentle cleaning and it was demonstrated in Fig. 6. It was seen that there were various cracks and pits on the surface that could cause leakage, indicated by the arrow in Fig. 6.

When the literature was examined, similar pits and cracked structures were found in the corrosion studies of copper pipes used in heat exchangers and air conditioners. In a study by Peltola and Lindgren (2015)Peltola, H., Lindgren, M. (2015). Failure analysis of a copper tube in a finned heat exchanger. Eng. Fail. Anal. 51, 83-97. https://doi.org/10.1016/j.engfailanal.2015.02.016., similar fractured structures were found in a corrosion study of the copper pipe used in the heat exchanger. Authors determined that ant-nest corrosion occurred and caused damage in copper pipes. It was stated that ant-nest corrosion products were dark colored on the copper pipe surface. Zhou et al. (2018)Zhou, J., Yan, L., Tang, J., Sun, Z., Ma, L. (2018). Interactive effect of ant nest corrosion and stress corrosion on the failure of copper tubes. Eng. Fail. Anal. 83, 9-16. https://doi.org/10.1016/j.engfailanal.2017.09.013. examined a copper pipe damaged in an air conditioning system. The authors determined that samples were subjected to ant-nest corrosion with the occurrence of various pits. In a study conducted by Bastidas et al. (2006)Bastidas, D.M., Cayuela, I., Bastidas, J.M. (2006). Ant-nest corrosion of copper tubing in air-conditioning units. Rev. Metal. 42 (5), 367-381. https://doi.org/10.3989/revmetalm.2006.v42.i5.34., copper pipes damaged by ant-nest corrosion that were used in air conditioners for 2-3 months were examined. In that study, they determined that various cracks and pits were formed interior of the copper pipe. Similarly, Cozzarini et al. (2020)Cozzarini, L., Marsich, L., Schmid, C. (2020). Ant-nest corrosion failure of heat exchangers copper pipes. Eng. Fail. Anal. 109, 104387. https://doi.org/10.1016/j.engfailanal.2020.104387. observed corrosion formations in heat exchangers and detected black corrosion product accumulation and ant-nest corrosion which was similar to our study. In our study, when the SEM images of the corroded C3 sample found in Fig. 6 were compared with the literature, pits and cracks similar to ant-nest corrosion were observed. As can be seen in Fig. 5, an accumulated layer was formed on top of these pits. The fact that this accumulated layer was different in color (black) from the copper pipe in the SEM-BSD analysis was shown in Fig. 5a. This color difference indicates the presence of a different structure. In addition, crystal structures were seen around the corrosion hole in SEM images. Similar structures have been specified as copper oxide structures, which are specified as corrosion products in various studies (Peltola and Lindgren, 2015Peltola, H., Lindgren, M. (2015). Failure analysis of a copper tube in a finned heat exchanger. Eng. Fail. Anal. 51, 83-97. https://doi.org/10.1016/j.engfailanal.2015.02.016.; Vazdirvanidis et al., 2019Vazdirvanidis, A., Papadopoulou, S., Papaefthymiou, S., Pantazopoulos, G., Skarmoutsos, D. (2019). Failure investigation of Cu-DHP tubes due to ant-nest corrosion. Int. J. Struct. Integr. 10 (3), 325-336. https://doi.org/10.1108/IJSI-09-2018-0056.). It has been stated that these structures were related to the ant-nest corrosion starting point (Peltola and Lindgren, 2015Peltola, H., Lindgren, M. (2015). Failure analysis of a copper tube in a finned heat exchanger. Eng. Fail. Anal. 51, 83-97. https://doi.org/10.1016/j.engfailanal.2015.02.016.).

Figure 7 demonstrated elemental analysis of C1 sample inner surfaces by EDS analysis. The observed contamination-like structures were revealed that these regions contain elements such as Cu, O, Fe, P and C. It was determined that the light-colored areas were metallic copper and contained only Cu, while the gray and black areas contained Cu, O and Fe elements. Figure 8 showed elemental analysis of C2 sample inner surface. As a result of the EDS analysis, Cu, O and Fe elements were determined on the sample surface along with small amounts of Si, Ca and S. While the light-colored regions contained only Cu, O and Fe, elements such as Si, Ca and S were also seen in the regions where dark deposition was higher. It was seen that Fe element concentration increased in the dark areas where the contamination and amount of deposition increases. The heat exchangers from which the copper tube samples were taken are connected to installations consisting of steel tubes. Thus, steel pipes which were connected to the heat exchangers, were determined as one of the main reasons for the existence of Fe element on the inner surface of samples. Furthermore, another reason for the existence of elements such as Fe, Si, Ca and S was thought to be mains water, since it circulates through these installation pipes and copper pipes.

The elemental analysis of the inner surface of failed sample C3 was given Fig. 9. It was seen that the light-colored region contains Cu and O, and the dark colored corrosion product covered region contains Cu and O as well as Fe and Si elements. It was determined that the corrosion product formed on the surface contains high amounts of Fe and O elements. Kuźnicka (2009)Kuźnicka, B. (2009). Erosion-corrosion of heat exchanger tubes. Eng. Fail. Anal. 16 (7), 2382-2387. https://doi.org/10.1016/j.engfailanal.2009.03.026. examined the corroded copper heat exchanger that has been used for 5 years and detected pits and surface microcracks formations. Their EDS analysis showed that the corrosion products formed on the inner surface consisted of elements such as Fe, Si, C, and Ca. Similarly, Lachowicz (2020)Lachowicz, M.M. (2020). A metallographic case study of formicary corrosion in heat exchanger copper tubes. Eng. Fail. Anal. 111, 104502. https://doi.org/10.1016/j.engfailanal.2020.104502., performed EDS analysis on the corrosion products on the inner surface of a damaged copper pipe and their results showed that while the light-colored copper pipe region consisted of Cu and O elements, dark colored corrosion products contained Fe, C, O and Cu. They determined the damage was due to ant-nest corrosion and water-insoluble residual particles accumulated and surface cracks on the surface played an important role on the initiation of corrosion. According to literature, when small amount of elements such as Fe, Mn and Ni are in the copper structure, the color of corrosion products can shift to orange, brown or even black (Jones, 2001Jones, R.H. (2001). Environmental effects on engineered materials. 1st Edition, CRC Press.). In the context of literature knowledge, it was understood that the color of the inner surface of our samples was changed because of contaminating elements such as Fe, Si, P and Ca. It was thought that elements such as Fe and Si deposited on the surface of C3 sample could be carried to the surface by the water flow through the pipe like others.

For better understanding of corrosion formation, elemental analysis of the gently cleaned leakage area of the C3 sample were given in Fig. 10. While the analysis made from the pit’s edge and corrosion product area showed the occurrence of Cu and O along with Fe, analysis carried out inside the pit revealed the existence of a high percentage of C element. When the studies in the literature were examined, it was seen that the amount of C was higher in the EDS analysis performed from the leakage area compared to the other regions. It was stated that the reason for this is the formic acid-like structures that play an active role in the formation of ant-nest corrosion (Zhou et al., 2018Zhou, J., Yan, L., Tang, J., Sun, Z., Ma, L. (2018). Interactive effect of ant nest corrosion and stress corrosion on the failure of copper tubes. Eng. Fail. Anal. 83, 9-16. https://doi.org/10.1016/j.engfailanal.2017.09.013.). Bastidas et al. (2006)Bastidas, D.M., Cayuela, I., Bastidas, J.M. (2006). Ant-nest corrosion of copper tubing in air-conditioning units. Rev. Metal. 42 (5), 367-381. https://doi.org/10.3989/revmetalm.2006.v42.i5.34. pointed that the formic acid which is essential in ant-nest corrosion can be generated by the pyrolysis of ethylene glycol or hydrolysis of various organic solvents that contains chlorine. Therefore, it was concluded that the EDS results obtained in our study were compatible with the literature and that C element in the leakage area was caused by acidic formations.

3.3. XRD and FTIR analysis

X-ray diffraction analysis was performed on the samples which were taken from each pipe. The analysis was applied to the inner parts of the copper pipes. Figure 11 shows the XRD pattern of C1, C2 and C3 samples.

Although O and Fe elements were detected on the surface of C1 sample by EDS, in the XRD analysis of the C1 sample only metallic Cu peaks were seen. Since there were no other phases beside copper, it was understood that corrosion and oxidation occurrences did not occur in sample C1. XRD analysis results of C2 copper pipe, which was used for 2 years, showed that there were two different phases in the structure. One of them was metallic Cu and the other one was Cu₂O (copper (I) oxide). It was understood that contamination product was formed in the copper pipe during the 2-year usage period. Since we detected Fe and O elements by EDS on both C1 and C2 samples, we can conclude that these samples were exposed to contaminating environment from two months to two years usage. However, this formation did not cause serious damage to the pipe. Zhou et al. (2018)Zhou, J., Yan, L., Tang, J., Sun, Z., Ma, L. (2018). Interactive effect of ant nest corrosion and stress corrosion on the failure of copper tubes. Eng. Fail. Anal. 83, 9-16. https://doi.org/10.1016/j.engfailanal.2017.09.013. obtained similar metallic Cu phase by XRD for copper pipes that were not exposed to formic acid environment. They also revealed that the exposure of copper pipes to the formic acid resulted with the formation of Cu₂O phase. In accordance, our findings showed that even without corrosion formation, a considerable amount of Cu₂O can deposit on the inner surface of samples. As a result of the XRD analysis of C3 sample, 4 different phases (compounds and elements) such as Cu, Cu₂O, CuO and Fe₃O₄ were detected on the inner surface of the copper pipe. It was found that this sample contained a secondary copper oxide phase along with iron oxide phases that were different from the C2 sample. Cozzarini et al. (2020)Cozzarini, L., Marsich, L., Schmid, C. (2020). Ant-nest corrosion failure of heat exchangers copper pipes. Eng. Fail. Anal. 109, 104387. https://doi.org/10.1016/j.engfailanal.2020.104387. found Cu, CuO and Cu₂O phases in a copper pipe which was failed due to ant-nest corrosion. They also stated that there was a black corrosion product on the surface of the failed pipes which was similar to our study. In a different study, Bastidas et al. (2006)Bastidas, D.M., Cayuela, I., Bastidas, J.M. (2006). Ant-nest corrosion of copper tubing in air-conditioning units. Rev. Metal. 42 (5), 367-381. https://doi.org/10.3989/revmetalm.2006.v42.i5.34. examined the ant-nest corrosion formation under presence of organic acids such as formic, acetic, propionic and butyric acid. The results showed that various copper oxide formations can be found on the surface of corroded samples along with as copper formate and copper acetate formations. The literature studies showed that oxide formations can be detected inside the pitting holes or around the leakage areas as corrosion products. Thus, formation of different oxides could be attributed to corrosion formation or the amount of corrosion. Lee et al. (2016)Lee, S.K., Hsu, H.C., Tuan, W.H. (2016). Oxidation Behavior of Copper at a Temperature below 300 ºC and the Methodology for Passivation. Mater. Res. 19, 51-56. https://doi.org/10.1590/1980-5373-MR-2015-0139. stated that the Gibbs free energy change to form Cu₂O is smaller than CuO and CuO can be formed either by oxidation of Cu or secondary oxidation of Cu₂O. In this context, CuO formation in the C3 sample could be caused by excessive amount of Cu₂O formation which can be correlated with corrosion. On the other hand, the black corrosion product structure formed on the surface also contained Fe₃O₄ phase apart from other studies. It was not clear whether the Fe₃O₄ formed due to pollution of mains water, steel pipes connected to the heat exchanger or corrosion. However, when the studies in the literature were examined, it was stated that contamination-induced Fe atom causes damage to progress faster. It was thought that Fe compounds may damage passive copper oxide layer (Kuźnicka, 2009Kuźnicka, B. (2009). Erosion-corrosion of heat exchanger tubes. Eng. Fail. Anal. 16 (7), 2382-2387. https://doi.org/10.1016/j.engfailanal.2009.03.026.). In addition, Lachowicz (2020)Lachowicz, M.M. (2020). A metallographic case study of formicary corrosion in heat exchanger copper tubes. Eng. Fail. Anal. 111, 104502. https://doi.org/10.1016/j.engfailanal.2020.104502. stated that impurities such as oxide or phosphides may act as potential corrosion initiation sites for ant-nest corrosion. Thus, it can be concluded that the amount of Fe ions on the surface and the formation of Fe₃O₄ phase may increase the possibility of corrosion formation directly or indirectly, but existence of C in pit holes plays a key role in corrosion. On the other hand, May et al. (1991)May, P.M., Ritchie, I.M., Tan, E.T. (1991). The corrosion of copper in ethylene glycol-water mixtures containing chloride ions. J. Appl. Electrochem. 21 (4), 358-364. https://doi.org/10.1007/BF01020222. states that protective films of copper(I) oxide and copper(I) chloride are more likely to form in the ethylene glycol-water solution than in aqueous, especially at low chloride concentrations, and this could be inhibit corrosion of the copper components of the cooling system. However, it was stated that the conditions in a vehicle cooling system are favorable for the oxidation of ethylene glycol and breakdown products such as glycolic and formic acids were often found in coolants; and this cause the pH of the fluid to decrease. Under these conditions, it was reported that the formation of copper(I) oxide and copper(I) chloride protective films are less likely even in the presence of chloride ions and thus corrosion of copper components is expected. FTIR spectroscopy was carried out for evaluating the functional groups (especially C-Cl stretching) in the corrosion products of C3 sample. Figure 12 shows the FTIR spectra of corrosion products of C3 sample. Bands around between the 850 cm^-1 and 550 cm^-1 (Khan et al., 2016Khan, M.Z.H., Hossain, M.I., Halder, P.K., Hasan, M.R., Al-Mamun, M.R. (2016). Fuel properties of pyrolytic tyre oil and its blends with diesel fuel-towards waste management. IJEWM 18 (4), 335-348. https://doi.org/10.1504/IJEWM.2016.081835.) corresponds to C-Cl stretching caused by presence of Cl in mains water circulating in copper pipe. Thus, the presence of Cl ions in the water in the system is proven.

All in all, we can conclude that main reason of corrosion was depended on the amount of ethylene glycol type additives, chloride ion concentration and pH value of water which resulted with the formation of formic acid in coolant water.