Environment
Sensing Stress Corrosion Cracking in Phosphorus Copper Tubes
2024 Volume 65 Issue 12 Pages 1566-1574
Details
2024 Volume 65 Issue 12 Pages 1566-1574
The leakage of fluorocarbon refrigerants due to corrosion of phosphorus copper tubes, which are widely used in air conditioners, is a growing environmental threat. In this study, to prevent this leakage, which is caused by stress corrosion cracking, we attempted to detect corrosion using a sensor that comprises a galvanic couple consisting of the copper tube and a carbon electrode. On testing in the presence of ammonia, the sensor detected a rise in background current, revealing the formation of a water film in which ammonia gas dissolves, with current noise identifying localized corrosion and oscillation of the background current indicating gas leakage. These phenomena can be detected by simply monitoring the current.
Air conditioners use heat pump technology that is much more energy-efficient than alternative products such as electric heaters, because they can extract several times more heat energy than the power they consume. The further improvement of this technology and wider uptake of heat pump technology are expected to help realize decarbonization. On the other hand, fluorocarbon refrigerants such as R410a and R32, which have global warming points of 2090 and 675 respectively, are still in wide use. Only about 40% of the refrigerants currently in use are recovered [1] and the remainder leaks into the atmosphere. International regulations have been progressively tightened to stop this leakage. For example, under the Montreal Protocol, most fluorocarbon refrigerants are to be completely phased out.
Propane is a promising alternative refrigerant for air conditioners. Air-to-Water systems using it have already been commercialized in Europe [2] because propane leakage is measured from the outdoor part of the circuit only. But since propane is flammable, it is dangerous if corrosion of refrigerant tubes causes it to leak indoors. Risk assessments [3, 4] and leakage prevention technologies are therefore urgently needed to guarantee the safe use of propane-based air conditioners. Leak detection methods have been proposed, such as remote monitoring of equipment [5, 6] and ultrasonic sensing [7], but the downside of these techniques is that there is a time lag between the occurrence of the leak and its detection. It is therefore important to detect in advance potential refrigerant leakage caused by corrosion.
Phosphorus copper tubes are commonly used in air conditioners; however, when ammonia in the environment acts on areas that experience tensile stress, such as bent parts, stress corrosion cracking (SCC) can occur [8]. With the goal of preventing refrigerant leakage due to SCC, we propose a model for the mechanism of SCC in phosphorus copper tubes [9]. In this study, we describe a sensing method for preventing refrigerant leakage due to SCC in copper tubes. In a previous paper, we detected the current noise associated with the development of SCC in phosphorus copper tubes by using a galvanic couple comprising the copper tube and a platinum wire immersed in an ammonia solution [9]. SCC sensing is enabled by applying this principle, but there are two problems.
The first problem is variations in environments. SCC occurs mainly in the hairpin bend parts of heat exchangers in indoor units, where condensate will be generated on the copper tube surfaces while cooling the air. If ammonia gas dissolves in this condensate, SCC can occur. Our previous study used an ammonia solution with a specified concentration, not condensate that might include ammonia.
The second problem is high cost. The platinum wire used as the counter-electrode (CE) in the previous study is expensive, making it impractical to apply it directly to commercial air conditioners. The requirements of a counter-electrode are that it must be electrically conductive, highly resistant to corrosion, have a higher electrical potential than copper in an ammonia solution. It must also be inexpensive and simply and easily molded. We experimented with sensing SCC by adopting carbon (graphite), which meets these requirements, as the counter-electrode and forming a galvanic couple with the phosphorus copper tube acting as the working electrode.
Spiral grooved phosphorus copper tube (P: 0.28 mass%, quality code: OL, bottom thickness: 0.23 mm, outer diameter: 6.35 mm) was used in this experiment. It was bent using a hairpin bender (Taiyo, TB-HU-7-1105-FA) such that the radius of curvature was 10.8 mm. To prevent residual stress relaxation, no heat treatment was performed. After removing the processing oil, the surface of the copper tube was polished with #1500 polishing paper and degreased with acetone. A sensor with the shape shown in Fig. 1(a) was fabricated using this copper tube. It was masked with tape (Fig. 1(c), outer diameter: 1.0 mm) at the position shown in Fig. 1(a), where SCC was most likely to occur in our previous test [9]. The entire hairpin bend part, including the masked and straight parts, was then coated with epoxy resin (Fig. 1(d)). The masking was next peeled off to create an exposed part (Fig. 1(e)). The epoxy resin around the exposed copper part (about 150 µm thick) acts as an electrical insulating layer. A conductive layer was formed at the same time as bonding a lead wire (coated copper wire) to the epoxy layer around the exposed part of the copper (Fig. 1(f)) using conductive silver paste (TPS, MCN-DJ002). Next, carbon paste (Micro to Nano, EM-Tec C38) was applied to the surface of the conductive layer and dried to form a cathode layer (Fig. 1(g), about 50 µm thick). The surrounding carbon layer surface was again masked (Fig. 1(h)) such that the diameter of the exposed part of the carbon electrode was 3.0 mm. Finally, the carbon layer was coated with epoxy resin (Fig. 1(i)) and the masking tape was peeled off (Fig. 1(j)).
Schematic diagram of sensor (a) and the process for making the electrodes (b)–(j).
One end of the copper tube was sealed with a flare cap, and the other end was sealed with a valve and attached to a digital pressure gauge (Nagano Keiki, GC61). The copper tube was injected with helium gas to an internal pressure of about 4 MPa (the design pressure of a conventional air conditioner). To be able to observe the timing of cracking, acoustic emission (AE) signals were measured by attaching an acoustic emission (AE) sensor (NF, AE-901U) to the copper tube. The amplification factor was 70 dB, and AE waves with a higher intensity than environmental noise and a frequency of 20 kHz or higher were filtered for easy counting. The lead wires were joined to the exposed part of the copper tube by soldering for the measurements of potential and current.
2.2 SCC testing in the typical operating environment of air conditionersWe conducted an SCC test using the fabricated sensor and experimental system shown in Fig. 2, adopting the typical operating environment that air conditioners experience. As shown in the Figure, the part of the working electrode (WE, the exposed part of the copper tube) and the counter-electrode (CE, carbon) were sealed in a polyethylene container (volume: 400 mL) with 100 mL of 1 mass% ammonia solution and ambient air, with the WE and CE exposed to the gas phase. The copper tube was arranged such that the surfaces of the WE and CE were perpendicular to the water surface. In this state, the copper tube was cooled using a Peltier element (AS ONE, ZCP-15150), and condensate was generated on the surface of the WE.
Schematic diagram of SCC test.
This experiment was performed three times. In the first version, we measured working electrode potential using a reference electrode (RE, Ag/AgCl (saturated KCl) electrode, Meidensha, HX-R6). The tip of the RE was provided with an agar salt bridge prepared using a saturated KCl solution, with the tip of the salt bridge placed close to the WE. This held a meniscus of condensate between the WE and RE, which tended to stagnate. In actual air conditioners, the hairpin bend parts of the heat exchanger are normally equipped with a stiffener (resin cover or similar) to protect against impacts, so the hairpin bends stay in contact with the condensate during the cooling process. This experiment showed that Cl− ions from the KCl in the salt bridge are present in the condensed water.
To test for the presence or absence of this effect, a second experiment was conducted using a resin rod (dummy RE) of the same shape as the salt bridge used in the first experiment. The dummy RE was placed close to the WE such that the condensate remained but there was no elution of Cl− ions. The current flowing through the external circuit between the WE and CE was measured using a zero-shunt ammeter (Syrinx, SZRA-204). At that time, the direction of the anode reaction at the WE was positive. These parameters, as well as the tube temperature and internal pressure, were monitored with a data logger (Graphtec, GL240). When the RE was used, the corrosion potential of the WE was also measured using the data logger.
Furthermore, to demonstrate that the SCC observed in the first and second experiments is not simply due to galvanic corrosion caused by the use of carbon electrode (CE), control experiment without CE (Fig. 1(e)) was carried out using the dummy RE. At this time, only the internal pressure, AE, and tube temperature were measured using the data logger.
All of the above parameters and AE signals were measured at a sampling interval of 1 s, and the test was performed until SCC occurred and the internal pressure had decreased sufficiently. After the test, the sensor was immediately taken out of the container, and the water droplets on the sensor were removed and allowed to dry naturally. After drying, the surface and cross-section around the SCC generation sites were observed. Samples for cross-sectional observation were prepared by cutting, embedding in resin and mirror polishing to expose the cut surfaces. An optical microscope (KEYENCE, VHX-6000) and scanning electron microscope (SEM; Hitachi High-Tech, FlexSEM 1000II) were used for the observations. Before using the SEM, the samples were coated with Pt.
2.3 Copper tube immersion test in an ammonia solution containing chloride ionsTo investigate the effect of Cl− ions eluting from the salt bridge, we conducted immersion tests of copper tubes in an ammonia solution. Figure 3 is a schematic representation of the test system.
Schematic diagram of immersion test of copper tube in ammonia solution containing chloride ions.
The test solution (volume: 100 mL) and atmosphere were sealed in a polyethylene container with a volume of 400 mL, and the working electrode (WE: phosphorus copper tube, P: 0.28 mass%), reference electrode (RE: Ag/AgCl (saturated KCl)), and counter-electrode (CE: graphite rod, outer diameter: 4 mm) were immersed in it. The outer diameter of the copper tube was 6.35 mm. It was coated with silicone resin such that the length of the exposed part was 30 mm from the tip. The tip was also sealed with silicone resin so as not to contact the test solution to the inside of the tube.
Before testing, the exposed part of the WE was polished using #1500 polishing paper, then degreased with acetone. Three types of test solutions were used: 1 mass% ammonia solution, 1 mass% ammonia solution in which 5 mass% NaCl (Cl−: 0.9 M) had been dissolved and 1 mass% ammonia solution in which 6.4 mass% KCl (Cl−: 0.9 M) had been dissolved. These salts dissociated into Cl− ions and their corresponding cations in the ammonia solution.
The potential of the WE and current flowing through the external circuit between the WE and CE were measured. A zero-shunt ammeter (Syrinx, SZRA-204) was used to measure the current, and a data logger (Graphtec, GL240) was used to monitor the output from the ammeter and measure the potential. The direction of the current was positive for the direction in which the anode reaction occurred at the working electrode. The test time was 38 h, and the sampling interval was 1 s.
Figure 4 shows the measurement results of the SCC test with the RE close to the WE surface. To explain this data step by step, the test period was divided into four stages (I–IV) as shown in Fig. 4(a).
Time series data measured during the SCC test with the reference electrode brought close to the working electrode. The left-hand figure (a) shows the data for the entire duration. The right-hand figures (b), (c), (e) are enlarged views of characteristic segments, comparing the current and potential. Figure (d) shows an enlarged view of a characteristic segment, comparing the current and AE.
Stage I (0–1.6 h): At this stage, the current gradually increased from 0 to 0.23 µA (Fig. 4(b)), and the potential was about −0.02 V. About 1.6 h after the start of the test, the current rose sharply to 1.6 µA and the potential dropped sharply to −0.26 V. The potential then shifted to the background of −0.15 V over about 0.2 h. At this stage, no oscillations in internal pressure were observed, and almost no AE was detected.
Stage II (1.6–21 h): As shown in Fig. 4(c), a large number of spike-like current peaks on the anode side were observed at this stage, with a background current of about 1.0 µA. All these current peaks were waveforms that rose rapidly and then gradually decayed (RR-type noise according to Inoue’s classification [10]). For example, focusing on the interval from 5 h to 5.5 h enlarged in Fig. 4(c), the current rose sharply from 1.0 µA to 2.7 µA at 5.09 h, rapidly decayed to 1.8 µA over about 300 s, and then slowly decayed to the original background over about 0.3 h. The increase in current noise and the decrease in the potential noise corresponded well. The potential noise was waveforms that rapidly dropped and then gradually recovered (RD-type noise according to Inoue’s classification [10]). The potential noise corresponding to the current noise observed at 5.09 after the start of the test rapidly shifted from −0.13 V to −0.24 V, rapidly recovered to −0.19 V over about 300 s, and then gradually decayed to the original background over about 0.3 h. Although there was no change in internal pressure during Stage II, the first AE was detected at 20.987 h after the start of the test (Fig. 4(d)), and continuous AE was observed thereafter. At the timing of the first AE, a weak RR-type current noise of about 0.2 µA was also detected.
Stage III (21–38.7 h): From 21 h to 25 h after the start of the test, which corresponds to the initial part of Stage III, the background current gradually increased from 0.8 µA to 2.5 µA, while the potential remained at about −0.11 V. Subsequently, from 25 h to 38.7 h, the background current gradually decreased to about 1.5 µA, and an oscillation with an amplitude of about 0.05 µA was observed. Corrosion potential oscillations of about 10–30 mV to the noble side were also observed over the background of about −0.11 V (Fig. 4(e)). Although there was no clear correspondence between these weak current and potential oscillations, relatively strong RR-type current noise was observed at this stage, as was corresponding RD-type potential noise. For example, in Fig. 4(e), RR-type current noise was observed at 28.49 and 28.66 h after the start of the test, and the potential noise was correspondingly detected. In the interval between 25 h and 38.7 h during the test period, the internal pressure gradually fell from 3.91 to 3.63 MPa, and AEs were continuously detected throughout this interval. This decrease in internal pressure did not correspond to a decrease in tube temperature.
Stage IV (38.7–41.1 h): From 38.7 h to 40.0 h after the start of the test, the current was close to 0 and potential was 0–20 mV. After 40.0 h, current increased and potential decreased. At this stage, a notably large AE was observed at 38.7 h after the start of the test, after which the internal pressure dropped sharply from 3.63 MPa to 0 MPa.
Figure 5 shows the measurement results for the SCC test using the dummy RE. In this Figure, no stage corresponding to Stage IV in Fig. 4 was observed. Therefore, in the following explanation, the test period is divided into three stages (I–III) as shown in Fig. 5(a).
Time series data measured during the SCC test using the dummy reference electrode. The left-hand figure (a) shows the data for the entire duration. The right-hand figures (b), (c), (d), (e) are enlarged views of characteristic segments, comparing the current and AE.
Stage I (0–4 h): In the period between the start of the test and 1.7 h into it, the current gradually increased from 0 to 1.8 µA; the current then remained in a more or less steady state until the end of this stage (Fig. 5(b)). At this point, no fluctuations in internal pressure were observed, and almost no AEs were detected.
Stage II (4–43.8 h): As shown in Fig. 5(c), a large number of current noise events on the anode side were recorded. At this stage, all the current noise was of the RR type. For example, from 19.34 h to 19.4 h after the start of the test shown in Fig. 5(c), the current rose sharply from 1.5 µA to 3.0 µA at 19.362 h, then rapidly decreased to 2.3 µA over about 30 s, and then gradually decayed to the background level over about 0.12 h. Although there was almost no change in internal pressure at this stage, the first AE was detected 26.82 h after the start of the test, followed by continuous AEs up to 28.4 h (Fig. 5(d)). During these AE occurrences, there were also continuous current noise events on the anode side.
Stage III (43.8–95 h): At this stage, weak oscillations of approximately 0.5 µA in amplitude were observed against the background current of around 2.0 µA (Fig. 5(e)). The oscillations were not correlated with fluctuations in tube temperature. At this stage, the internal pressure was gradually decreasing, with AE events continuously occurring. There was no clear correspondence between the occurrence of AE events and the current noise or oscillations.
In the control experiment without CE, the test was carried out for 120 h. In this case, of course, we cannot measure the external current and distinguish the stage, but a similar drop in internal pressure and associated continuous AE was observed at 100.5 h after the start of the test. This clearly indicates that the result of the first and second experiments was not specific to galvanic corrosion.
Figure 6 shows the observation results of the surface and a cross-section of the samples after each test. A comparison of the surfaces showed that in the test using RE, the area near the CE was covered with a blue film, whereas in the tests using dummy RE, a blue film was observed only on the WE surface. In addition, in only the system using the RE, a white precipitate was observed in the center part of the copper surface. This precipitate is assumed to be KCl, and the amount is estimated from the image to be approximately 0.7 mg. When we assume that the diameter of the water droplet during the test is the same as the CE, 3 mm, and that all the KCl is dissolved in the hemispherical water droplet, the Cl− ions concentration would be 0.014 M. From the cross-sectional images, pitting, intergranular corrosion and cracking were observed in all the samples, with the penetrations caused by cracking. In the system using the RE, a large smooth-walled pit with a diameter of about 300 µm and a depth of around 100 µm was observed. In contrast, in the system using CE and the dummy RE, an uneven pit with a diameter of approximately 700 µm and a depth of around 45 µm was observed. Intergranular corrosion was evident at the bottom of this pit, with countless cracks. These were observed similarly even when the CE was removed.
Optical and SEM images of the sample surface and cross-section after the SCC test. The entire cross-sectional images were taken by SEM and the other images were taken by optical microscope. The left-hand figure shows the results when using a reference electrode (RE), the middle figure shows the results when using a dummy reference electrode and the right figure shows the results when using a dummy reference electrode without CE.
Figure 7 shows the immersion test results in ammonia solutions. From the beginning of the test to 10 h, the corrosion current and potential in all solutions decreased just after immersion, followed by a gradual increase. However, where the ammonia solution contained a large number of Cl− ions, which is higher concentration than estimated in the previous section, the corrosion potentials were approximately 20–50 mV lower than with no Cl− ions, and the corrosion currents were approximately 2–4 times higher.
Results of immersion testing in ammonia solution. The upper figure shows the current flowing between the working electrode and counter-electrode, and the lower figure shows temporal changes in the potential of the working electrode.
From 14 h to 18 h after the start of the test, a rapid increase in potential of approximately 100 mV was observed in all solutions, accompanied by a discontinuous decrease in current. When Cl− ions were present, the potentials at the beginning and end of the rapid potential increase were approximately 30 mV lower than when Cl− ions were absent. After the rapid potential increase, the corrosion potential stabilized at around −0.14 V in the solution without Cl− ions, and the corrosion current oscillated at approximately 2 µA in amplitude against a background of about 0.2 mA. In the solution containing a high concentration of Cl− ions, corrosion potentials stabilized at around −0.17 V, and the corrosion currents oscillated by approximately 20 µA in amplitude against a background of about 0.7 mA.
We discuss here phenomena corresponding to the corrosion current observed in Section 3.1. The phenomena occurring at each stage are schematically illustrated in Fig. 8. We will cover them from Stage I to IV, in the light of the immersion test results using the ammonia solution presented in Section 3.2.
Schematic diagram of corrosion current and phenomena during SCC test of phosphorus copper tube.
(1) Stage I
At this stage, the current rose from 0 to several microamperes, and when the corrosion potential was measured, it decreased from around 0 to −0.26 V. Since this interval represents the early stage of the test, we believe it consists of the formation of a water film on the WE surface and the dissolution of ammonia gas in this film. In other words, it is presumed that a water film covers the WE and CE, into which oxygen and ammonia gas dissolve. The dissolved oxygen acts as an oxidizing agent, causing the dissolution of copper ions from the WE, thus reducing the solution resistance of the water film. As a result, this process leads to the formation of a localized battery between WE and CE, initiating the flow of galvanic current. The potential of −0.26 V reached in the early stage is a potential observed in the immersion tests (Fig. 7) during the rapid increase in potential, regardless of the presence of Cl− ions. We have reported that when the ammonia solution does not contain Cl− ions, the rapid increase in potential corresponds to the transition of the stable range of monovalent copper [Cu(NH3)2]+ ions to divalent [Cu(NH3)4]2+ ions [11]. As seen in Fig. 7, the rapid potential increase in the presence of Cl− ions is similar to that seen in the absence of Cl− ions, although it showed about 30 mV lower potential. In the process of water film formation, it appears that copper initially dissolves as monovalent [Cu(NH3)2]+ ions, regardless of the presence or absence of Cl− ions. The ions rapidly concentrate and the stable phase of copper shifts to divalent [Cu(NH3)4]2+ ions.
Based on this, when focusing on the potential in the test using RE (Fig. 4(a)), copper was in the stable range of divalent [Cu(NH3)4]2+ ions for the majority of the test period. Therefore, assuming that the entire galvanic current obtained throughout the entire test period corresponds to the dissolution reaction of copper into divalent ions, the time-integrated value was calculated to be approximately 200 mC. By applying Faraday’s law, this translates to a copper dissolution volume of 7.36 × 10−3 mm3. This corresponds to the volume of a cylinder with a diameter of 300 µm and a height of 100 µm, approximately matching the size of the corrosion pit observed in Fig. 6. Similarly, when applying the calculations while using a dummy RE (Fig. 5(a)), the time-integrated value of the galvanic current was approximately 693 mC. This translates to a copper dissolution volume of 2.54 × 10−2 mm3, corresponding to the volume of a cylinder with a diameter of 610 µm and a height of 45 µm, approximately matching the size of the corrosion pit observed in Fig. 6. These findings indicate that the corrosion of copper occurring during the test is predominantly governed by galvanic corrosion with the counter-electrode, rather than self-dissolution. This is likely due to the waterline of the droplet covering the WE and the CE being closer to the CE surface, making it easier for oxygen to be supplied to the surface. As a result, a predominantly cathodic (oxygen reduction) reaction is assumed to occur on the surface of the counter-electrode.
(2) Stage II
At this stage, a large number of anodic RR-type current noise events were observed, accompanied by RD-type potential noise when using the RE. We have reported that RR-type current noise is obtained in response to SCC in phosphorus copper tube [9]. It has also been reported that the current noise corresponding to the phenomenon of slip dissolution associated with SCC in SUS 316 shows a similar pattern [12]. RD-type potential noise has been observed in the pitting corrosion process of pure iron in NaCl aqueous solution containing NaNO2 [13]. This rapid potential decrease corresponds to the progress of pitting corrosion, while the subsequent decay corresponds to re-passivation within the pit. The RR-type current noise and the corresponding RD-type potential noise thus appear to indicate the rapid increase in current (decrease in potential) caused by the formation of an active new surface resulting from the breakdown or cracking of the passive film, followed by the gradual recovery to the background due to the deposition of the film. We have also reported that localized corrosion such as intergranular corrosion occurs in phosphorus copper tubes [11], and film deposition occurs as a result of the concentration of copper ions [9]. It is likely that the RR-type current noise events observed in this study also correspond to localized corrosion such as pitting, intergranular corrosion or cracking.
To be able to schematically explain this, we describe the partial polarization curves in Fig. 9(a), (b), based on the anodic polarization curve of phosphorus copper tube in an ammonia solution [11]. The generation of an active new copper surface due to the breakdown of the film activates the anodic reaction. Consequently, the intersection with the cathodic reaction (oxygen reduction) shifts towards higher current and lower potential (Fig. 9(a)). As the concentration of copper ions near the anode surface increases, the current gradually approaches a steady state due to film deposition. Referring to Fig. 4(a), (b), RD-type potential noise shifting to approximately −0.30 V was observed. These potentials correspond to the potential before the rapid increase during the immersion test in the ammonia solution (Fig. 7). At this potential, copper is in the stable range of monovalent [Cu(NH3)2]+ ions [11]. Therefore, when such potential noise events occur, the rapid dissolution of monovalent [Cu(NH3)2]+ ions occurs, followed by a transition to the dissolution of divalent [Cu(NH3)4]2+ ions and the subsequent deposition of the film [9]. Even in the case of dissolution of only divalent [Cu(NH3)4]2+ ions, the intersection point of the partial polarization curves of the anodic and cathodic reactions shifts to higher currents and lower potentials (Fig. 9(a)). However, due to the decrease in the equilibrium potential of the anodic reaction itself, the extent of this shift becomes greater (Fig. 9(b)). In the test using the RE, the RD-type potential noise observed at 5.09 h after the start of the test (Fig. 4(c)) rapidly shifted from −0.13 V to −0.24 V, and then swiftly recovered to −0.19 V over approximately 300 s, followed by a gradual decline to the background level over the next 0.3 h. The potential of −0.19 V, which corresponds to the inflection point, closely matches the potential range of −0.18 V to −0.22 V reached immediately after the rapid increase observed in the immersion test shown in Fig. 7. The inflection point therefore appears to mark the transition of the anodic reaction, monovalent [Cu(NH3)2]+ ions dissolution to divalent [Cu(NH3)4]2+ ions dissolution. This inflection point in the RD-type potential noise is also observed in the corresponding RR-type current noise.
Schematic diagram of the partial polarization curve corresponding to the noise events and oscillations in current and potential.
RR-type current noise with a similar envelope was observed in tests using the dummy RE, but its decay rate and occurrence frequency were different from those in systems using RE. In the test using dummy RE, the decay rate of the RR-type current noise observed at 19.362 h after the start of the test is noteworthy. It rapidly decayed from the peak to the inflection point over approximately 30 s, followed by a gradual decay to the background level over the next 0.12 h (Fig. 5(c)). This decay rate is clearly faster than that in the system using RE. Focusing on the occurrence frequency of the RR-type current noise events observed in Stage II, in the test using the dummy RE, RR-type current noise with peak intensities of more than 1 µA, for example, were observed four times over a period of 39.8 h, whereas in the test using RE, they were observed 13 times over a period of 19.4 h. The cause of this difference is not clear, but we presume it may be attributed to the influence of Cl− ions. As shown in Fig. 6, in the sample using RE, the entire CE was covered with a blue film, and a white precipitate was observed at the center of the WE. Regarding the blue film, it is likely that copper ions accumulated in the condensation due to the dissolution of copper, the liquid near the graphite surface shifted towards a higher pH due to the cathodic reaction (O2 + 2H2O + 4e → 4OH−), and the thermodynamically stable phase (copper hydroxide) was then deposited in the surrounding area. Regarding the white precipitate, it appears that KCl contained in the agar salt bridge dissolved in the condensed water and precipitated during the evaporation of the water film. Additionally, the presence of a smooth pit with a diameter of approximately 300 µm indicates that the copper surface was in an environment that permitted relatively easy dissolution due to the presence of Cl− ions. It is reported that cuprous oxide (passivate film) is very soluble in chloride rich waters [14]. Therefore, Cl− ions contribute to the destruction of passivated films and increase the occurrence of localized corrosion. In contrast, in the systems using the dummy RE, where Cl− ions were absent, the film was less susceptible to rupture and easy to re-passivate. As a result, the shape of the corrosion pit was uneven. It appears likely that these differences affect the progression of localized corrosion and the deposition behavior of the film, and that they correspond to the shapes and frequencies of current noise events. Current noise that does not correspond to AE observed in Stage II likely indicates pitting or intergranular corrosion at the WE, while the RR-type current noise events corresponding to AE indicate the cracking of copper or film.
(3) Stage III
At this stage, continuous slow leakage of helium gas is occurring, as evidenced by the continuous AE and loss of internal pressure. Figures 9(c) and (d) show, in schematic form, partial polarization curves that explain the changes in current and potential. When gas bubbles are generated due to gas leakage, oxygen supplied to the CE surface dissolves faster due to the agitation of the condensed water and rise and fall of the water surface. As a result, the partial polarization curve of the cathodic reaction (oxygen reduction) shifts towards an increase in diffusion-limiting current. The intersection with the anodic reaction therefore shifts towards a higher current and more noble potential (Fig. 9(c)). Furthermore, the contact area between the working electrode and the condensed water decreases due to the presence of gas bubbles. In this case, while the partial polarization curve of the anodic reaction shifts towards the low current side, the water film around the CE, which is relatively distant from the bubbling area of the WE, is less likely to oscillate. The intersection of the partial polarization curves of the anodic and cathodic reactions therefore shift towards low current and more noble potential (Fig. 9(d)). This suggests that the current oscillations and potential oscillations towards the noble side observed at this stage can be attributed to these phenomena. The RR-type current noise events appear to correspond to localized corrosion, as seen in Stage II. However, because of the continuous occurrence of AE, it is not possible to determine whether they are due to pitting, intergranular corrosion or cracking.
(4) Stage IV
At this stage, continuous and faster helium gas leakage than in Stage III is occurring, as evidenced by the continuous AE and a significant decrease in internal pressure. In this situation, a continuous water film between the WE and CE cannot be maintained, leading to a temporary interruption of the current. As the internal pressure further decreases, bubble generation slows somewhat, and a continuous water film reappears between the WE and CE. Furthermore, strong agitation of the water film leads to significant current oscillations.
In this study, we have proposed a galvanic couple sensor suitable for monitoring the SCC of phosphorus copper tubes. The sensor can be made using inexpensive materials and processes. The working electrode is made of the same material as the actual equipment, and the counter-electrode is made of carbon (graphite). The space between them is insulated with epoxy resin. Using this design, it was found that the corrosion environment and the occurrence of localized corrosion can be adequately monitored in the form of galvanic currents.
During the test, the galvanic current behavior was classified into four stages, and the phenomena occurring at each stage were interpreted as follows. In the initial Stage I, a rise in the background current was observed. Here, the formation of a water film occurs, and the dissolution of ammonia into it creates a localized battery between the working and counter-electrodes. In Stage II, numerous current noise signals were observed toward the anodic side, above the background current. These current noise signals correspond to localized types of corrosion such as pitting, intergranular corrosion and cracking. In Stage III, current oscillations with a different shape from the current noise were observed, corresponding to oscillations of the water film and contact area due to gas leakage. In the final Stage IV, the oscillations of the background current show an even larger amplitude, corresponding to a higher flow rate gas leak.
These phenomena can therefore be detected by simply monitoring the current. By applying our galvanic couple to air conditioners, it is theoretically possible to forecast refrigerant leaks without the need for AE sensors or pressure gauges. For example, place a sensor with a dummy tube of a shape similar to that shown in Fig. 1 where SCC is most likely to occur. Galvanic couple accelerate the corrosion of the sensor itself, but it can be viewed from a safety perspective.
