Many wet sprinkler systems installed with copper tub-ing in Korea have been suffering from water leakage for the last 10 years due to pitting corrosion failureof copper tubes [1,2]. Pressurized air remaining over water in the copper tubing of the wet sprinkler systems have made the copper tubes vulnerable to pitting corrosion , just like stagnant water with an air pocket causes water line corrosion in the water supply system . Sediment deposi-tion at the bottom of copper tubes was identified as anoth-er cause of pitting corrosion failure of the sprinkler copper tubes [2,4].
There are inorganic and organic corrosion inhibitors used for copper. Chromates, molybdates and tetraborates belong to the inorganic corrosion inhibitors for copper. Inorganic inhibitors have an advantage over organic in-hibitors in that they are stable over a relatively wide tem-perature range , while they have some shortcomings such as toxicity and weak corrosion inhibition efficiency . Organic corrosion inhibitors are more popular because of their much higher corrosion inhibition efficiency than inorganic counterparts. Azole compounds including ben-zotriazole (BTAH), amine, and amino acid are the most important organic corrosion inhibitors for copper . BTAH has been known for more than 70 years since 1947  as the most effective corrosion inhibitor for copper and its alloys. When copper or its alloy is immersed in a BTAH solution, a protective layer of polymerized Cu-BTA complex is formed rapidly either on copper oxide or copper surface which prevents corrosion reaction. It has been used not only for industrial applications where copper is almost free of corrosion, but also for con-servation of archaeological artifactssuffering from corro-sion since Madsen first used BTAH in this field in 1967 . BTAH is classified as a mixed type inhibitor since it suppresses both oxygen reduction and copper oxidation [9,10]. Synergistic effects of a combination of BTAH with other organic compounds have been studied by Golfomitsouand Merkel.
Initiation of corrosion pits in freshly installed copper can be prevented by using a corrosion inhibitor generating a high pitting potential such as BTAH .However, it is believed that once corrosion pits are initiated on the copper tube, it is difficult to stop their propagation up to the stage of pin hole formation and water leakage. It is not clear whether BTAH could be effective even for preventing propagation of already existing corrosion pits. The fact that BTAH has been used successfully for the conservation of archaeological objects suffering not just from general corrosion but also from pitting corrosion gives us a hint that BTAH can be used even for preventing propagation of corrosion pits formed in the sprinkler cop-per tubes. Corrosion pits of copper are covered with a thick cap of different corrosion products of cuprite(CuO), cuprous oxide (Cu2O), and malachite (Cu2(OH)2CO3)par-ticles, which may make penetration of BTAH molecules into the corrosion pits difficult. It is also not sure whether BTAH can form corrosion inhibitive Cu-BTA composite layer on copper or copper oxide inside the corrosion pit which is under a highly acidic as well as chloride ion-con-taining environment. Musianiand Mengoli showed by an electrochemical and SERS (surface enhanced Raman spectroscopy) that at low pH, BTAH molecules are adsorbed weakly on a copper surface and displaced from surface sites by chloride ions. Brusic et al. re-ported that in acidic BTAH solutions (pH<2) a thick film is formed, which is not protective enough because of poor polymerization of the film. All the research works on the pH dependence of corrosion inhibition of BTAH has been done on clean copper or copper alloys in acidic solutions. However, there has been no such study yet on the for-mation of corrosion inhibitive Cu-BTA composites inside the corrosion pits, particularly those in the sprinkler cop-per tubes.
In this study, the authors tried to reduce the occurrence of water leakage in the sprinkler copper tubes by stopping the propagation of corrosion pits, using BTAH. The au-thors also investigated whether Cu-BTA film was formed well on the inner structure of the corrosion pits of one sprinkler copper tube by using optical microscopy, scan-ning electron microscopy (SEM), energy dispersive spec-troscopy (EDS), X-ray photoelectron spectroscopy (XPS), and X-ray diffractometry (XRD).
2. Experimental procedures
2.1 BTAH treatment of copper tubes in the wet sprinkler system
Copper tubes of 277 wet sprinkler systems for the total of 829 households at an 8-year old apartment complex were filled with 400 ppm BTAH-water solution after vac-uum evacuation of the tubes from the pressure gauge valve on the sprinkler alarm valve, using a patent technique  and a system shown in Fig.1. The copper tubes could be filled with BTAH-water solution without air pockets forming in the copper tubes. Some sediments at the bottom of the tubes could be removed as well by vacuum evacuation. The purity of BTAH (CAS No. 95-14-7) was better than 99.0%, as specified by the manufacturer, Samchun Chemicals of Korea. Water leakage from the sprinkler systems of the households was monitored for 161 days.
Fig. 1 Schematic diagram of the system for filling BTAH-water solution in the sprinkler copper tubing after vacuum evacuation.
A piece of the copper tubing was cut from the apartment fire sprinkler system which leaked water after being filled with 400 ppm BTAH-water solution for 36 days. Small specimens with corrosion pits were cut from the copper tube by a saw and used for microscopical, compositional and structural characterization of the corrosion pits.
2.2 Microscopical, compositional and structural characterization
The general morphology of the corrosion pits was ex-amined by optical microscopy. SEM and EDS of a FEI Inspect F50 system and a Hitachi Regulus 8230 were used to characterize the surface morphology and measure the composition of the inside of the corrosion pits in the cop-per tube. The acceleration voltage of 10 kV was used for the SEM and EDS analyses. XPS of a ThermoFisher Scientific K-Alpha+ was used to measure the surface com-position and depth profile of the inside of corrosion pits after the cap of corrosion products was removed. The size of the X-ray source for the XPS analysis was 200 μmin diameter. Ar ion sputtering with an Ar ion gun energy of 500 eV and 10 mA was used for depth profile. The sputtering rate for the depth profile that was based on the SiO2sputtering rate was 1.25 nm s-1. A micro-XRD of D8-discover was used to identify the materials in the corrosion pits using Cu Kα radiation at 40 kV and 40 mA. A rough estimate by Beer-Lambert law for attenu-ation of X-ray gives the XRD analysis depth of 60 μm. The X-ray beam size on the specimen was 800 μm in diameter. The exact position of the micro-XRD measure-ment was selected by moving the X-Y stage with the help of a laser beam as shown in Fig. 2.
Fig. 2 Alignment of the specimen for the micro-XRD measurement with the help of a laser beam; the bright spot at the center shows the laser beam on the corrosion pit found in the region 3 (see Fig. 5).
3.1 Decrease of water leakage rate of the wet sprinkler systems by the BTAH treatment
Each of the sprinkler systems for the 829 households has been filled with 400 ppm BTAH-water solution for 142 days on the average. During the monitoring of 161 days, water leakage occurred from only two households, 36 and 133 days, respectively, after filled with BTAH-wa-ter solution in their sprinkler systems. This is about 90% reduction from a year earlier in the water leakage rate at this apartment complex.
3.2 Characterization of corrosion pits on a sprinkler copper tube treated with BTAH
The sprinkler copper tube in the ceiling of a living room which leaked water 36 days after it was filled with BTAH-water solution was cut and studied. Fig. 3 shows the outer surface of the sprinkler copper tube of 25 mm in inner diameter from which water leaked. The tube was divided into four regions for identification.
Fig. 3 The outer surface of the sprinkler copper tube which leaked after being filled with BTAH-water solution for 36 days.
The two pin holes through which water leaked are shown on the outer surface in region 4 of Fig. 3. Fig. 4 shows the inside morphology of the corrosion pit con-taining two pin holes. All the corrosion products were removed by a tweezer for visual inspection. As small cor-rosion pits grew, they were probably combined to become two giant corrosion pits. The inside surface of the corro-sion pits is covered with dark red cuprous (Cu2O) particles. It seems that as the giant corrosion pit at the bottom became so wide and deep (about 4 mm in width and more than 1 mm in depth), the thickness of the tube became very thin, forming pin holes. The bigger pin hole is shown as a white dot in Fig. 4.
Fig. 4 The inner structure of the corrosion pits from which water leaked.
The corrosion pits found in regions 2 and 3 were studied thoroughly by optical microscopy, SEM, EDS, XPS, and XRD. Fig. 5a is the optical microscopic image of the cor-rosion pit in region 3 before removing its cap. Fig. 5b and c are the optical microscopic and SEM images, re-spectively, taken after the cap of the corrosion pit was removed. Table 1 shows the composition measured by EDS of areas A~D indicated in Fig. 5b and c. At area A which is the rim part of the cap consisting of bluish green malachite (Cu2(OH)2CO3) particles and black cu-prite (CuO) particles, a high nitrogen concentration near 10 at.% is measured. At area B under the malachite layer, the nitrogen concentration was 1.3 at.%. Area B has a very high concentration of chlorine. It is generally be-lieved that chloride ions are attracted into the corrosion pit by the positive ions of hydrogen and copper generated in the pit during the propagation stage . No nitrogen was measured by EDS at area C which looks like the cop-per base material just under the Cu2O layer, while a small concentration of nitrogen was measured at the small pit D. Chlorine was not detected by EDS at areas C and D.
Table 1 EDS analysis results of A~D areas marked in Fig. 5 (at.%)
Fig. 5 Corrosion pit found in region 3: (a) optical microscopic picture before the cap was removed; (b) optical microscope and (c) SEM pictures after the cap was removed.
Because the EDS analysis depth with the acceleration voltage of 10 KeV is around 1 μm, the EDS analysis re-sults in Table 1 are only the average concentrations over the volume as big as 1 μm in depth. In order to study the surface structure of areas B, C and D, XPS survey spectra of those areas were obtained as presented in Fig. 6. XPS peaks observed are Cu 2p, Cu LMM, O 1s, N 1s, C 1s, and Cl 2p. The appearance of the N 1s peak as well as the high C 1s peak in the XPS spectra indicates that the Cu-BTA film was formed on the surface. The C 1s spectrum for area C (Fig. 7) shows two peaks, one at 284.5 eV with high intensity and the other around 288.5 eV. The intense C 1s peak at 284.5 eV is due to carbon in the aromatic ring of BTAH molecule [17,18] while the broad peak around 288.5 eV may be from contaminant carbon . The surface concentrations at these areas ob-tained by XPS are given in Table 2. The chlorine composi-tion at area B is higher than those at areas C and D. At areas C and D, about 20 at.% copper, about 9 at.%nitro-gen and more than 40 at.% carbon was measured by XPS while about 80 at.%copper, almost no nitrogen, and less than 8 at.%carbon was measured by EDS. These differ-ences between the XPS and EDS results indicate that a very thin Cu-BTA film consisting of carbon, nitrogen, and copper was formed at the surface. The XPS analysis depth is as large as 10 nm . Therefore,detection of oxygen, copper, and chlorine at the surface by XPS may not mean that copper oxide and chloride compounds are embedded in the thin Cu-BTA film whose thickness is less than 5 nm, as shown later in the XPS depth profiles of area B in region 3 and area I in region 2 (Fig.8 and Fig. 12). The position of these compounds in the whole structure will be discussed in the next paragraph.
Fig. 6 XPS survey spectra at areas (a) B, (b) C, and (c) D indicated in Fig. 5.
Fig. 7 C 1s XPS spectrum at area C.
Table 2 Surface concentrations at areas B, C, and D measured by XPS (at.%)
Fig. 8 shows the XPS depth profile of area B. Since the etch depth in the XPS depth profile was converted from the sputtering time based on the sputtering rate of SiO2, there could be some discrepancy between the etch depth and the converted one. The nitrogen concentration is about 10 at.%at the surface. It decreases continuously to 0 at.%up to 37 nm, and it is not detected at the larger depth. The Cu-BTA film thickness may be much smaller than 37 nm, however, because the unevenness of area B could have made nitrogen detected even at the etch depth larger than the Cu-BTA film thickness. Instead, the etch depth at the inflection point of the composition profile of nitrogen which is 2 nm could be considered as the thickness of the Cu-BTA film. The high oxygen concen-tration measured at the outermost layer of the surface may mean that oxygen atoms of the underlying Cu2O layer are located almost at the same depth as the constituent elements of the Cu-BTA film as a model of the Cu-BTA film on Cu2O shows . The chlorine concentration in-creases rapidly in the surface region of 5 nm in depth. Considering that the XPS analysis depth is as large as 10 nm, this rapid increase of the chlorine concentration at the surface may mean that the chloride compound is not included in the Cu-BTA film. In other words, the chlorine concentration measured at the zero etch depth might have come from the layer underneath the Cu-BTA film, not from the outermost Cu-BTA film. A high chlor-ine concentration more than 25 at.%was measured at the etch depth larger than 80 nm, which is consistent with the EDS result of Table 1. At the etch depth larger than 130 nm (not shown in Fig. 8), the ratio of chlorine to oxygen concentrations is close to 2.
Fig. 8 XPS depth profile of area B of the corrosion pit found in region 3.
Since a high chlorine concentration more than 25 at.%was measured by both EDS and the XPS depth profile, it would be interesting to study the structure of the chlor-ide compound under the Cu-BTA film. An X-ray dif-fraction pattern of Cu2O and CuCl peaks was obtained on the surface of area B (Fig. 9). This result is consistent with the XPS depth profile in which only copper, chlorine, and oxygen exist at the etch depth larger than 80 nm. The strong peaks of Cu2O and weaker peaks of CuCl in-dicate that the Cu2O layer is beneath a composite of CuCl and Cu2O. The peak marked with ‘?’ could not be identified.
Fig. 9 XRD spectrum of area B of the corrosion pit found in region 3.
It was confirmed by the XPS measurements on areas B, C, and D in region 3 that all parts of the corrosion pit in region 3 are coated well with Cu-BTA films. This was only possible because BTAH molecules could pene-trate the corrosion pit through all corrosion products in-cluding CuO, Cu2(OH)2CO3, and Cu2O layers.
To confirm whether the deep position of the corrosion pit found in region 2 was coated with the Cu-BTA film, the corrosion products in the left half of the corrosion pit shown in Fig. 10 were scraped offby a tweezer. The areas indicated by the symbols of E, F, G, H, and I in Fig.10 were analyzed by EDS (Table 3). The depths of these positions are in the alphabetical order. In other words, area E is at the shallowest location in the corrosion pit while area I is at the deepest location in the corrosion pit. Nitrogen was detected on all positions analyzed. The deeper the position in the corrosion pit is, the smaller the nitrogen composition becomes. Areas E and F consisting of bluish green malachite particles and dark red cuprous oxide particles have relatively high nitrogen concentration. Noticeable chlorine concentrations of 4.1 and 23.3 at.%were detected at areas H and I, respectively. The H is the area where Cu2O particles are gathered together while the I is the bottom of the corrosion pit from which almost all Cu2O particles were scraped off.
Fig. 10 Optical microscope image of the corrosion pit found in region 2 after the left half of the cap was removed.
Table 3 EDS analysis results of the areas E~I marked in Fig. 7 of the corrosion pit found in the region 2 (at.%)
To confirm whether the Cu-BTA film was formed at the two deepest positions, H and I, XPS survey spectra and surface concentrations of these areas were obtained as presented in Fig. 11 and Table 4, respectively. The small N 1s peaks (about 5 at.%), as well as the high C 1s peaks (more than 40 at.%), show that the Cu-BTA film was formed well on the surface of these two deep regions. Fig. 12 shows the XPS depth profile of area I. The nitro-gen concentration is 5 at.%at the surface, and it is not detected at the etch depth larger than 2 nm, indicating that a Cu-BTA film thinner than 2 nm was formed on the surface. The carbon and oxygen concentrations also decrease rapidly while the copper and chlorine concen-trations increase rapidly on the outermost layer of the surface. Detection of oxygen and chlorine at the surface does not necessarily mean that oxygen and chloride com-pounds are embedded in the Cu-BTA film for the same reason given when interpreting the XPS depth profile of area B of the corrosion pit in region 3. At the etch depth larger than 5 nm, the ratio of chlorine to oxygen concen-trations is close to 2, just like at area B of the corrosion pit in region 3.
Fig. 11 XPS survey spectra at areas (a) H and (b) I.
Table 4 Surface atomic concentrations at areas H and I measured by XPS (at.%)
Fig. 12 XPS depth profile of area I of the corrosion pit found in region 2.
The area I was analyzed by micro-XRD, and Fig. 13 shows its XRD spectrum. Peaks of comparable heights were observed for Cu and Cu2O, indicating that a thin Cu2O layer is present on the Cu base material. It should be noticed that CuCl peaks were also observed, which is consistent with the detection of more than 23 at.%Cl by both EDS analysis and the XPD depth profile. The peak marked as ‘?’ in Fig. 13 could not be identified.
Fig. 13 XRD spectrum of area I of the corrosion pit found in region 2.
Based on the information gathered by EDS, XPS, and XRD analyses, we propose the schematic structure of Fig. 14 for the Cu-BTA film and the underlying materials formed inside the corrosion pit of the sprinkler copper tube. The outermost layer is the Cu-BTA film with the underlying layer of Cu2O and CuCl phases sitting on the Cu2O layer. If the chloride concentration in the water in-side the corrosion pit is diluted, the layer of Cu2O + CuCl in the structure will shrink in thickness, which is the case for areas C and D of the corrosion pit in region 3. The way how this kind of structure could be formed in the chloride containing water by BTAH is discussed in the next paragraph.
Fig. 14 Schematic structure of the Cu-BTA film and the underlying materials formed inside the corrosion pit of the sprinkler copper tube.
In metallic materials such as Fe and Cu, chloride ions are attracted electrically into the corrosion pits by positive metallic ions and hydrogen ions, accelerating the growth of corrosion pits [16,22]. Several research groups dis-cussed the mechanism of how chloride ions suppress rath-er than accelerate corrosion of copper in the presence of BTAH. BTAH has been known for a long time to behave as an efficient corrosion inhibitor of copper in chloride solutions [23,24]. Evans et al. explained the corrosion inhibition effect of BTAH in chloride solutions by for-mation of a CuCl layer on the surface of Cu which inter-acts with BTAH molecules. This blocking substance would render high efficiency of corrosion inhibition. Hashemi et al. examined the surface layer formed on copper immersed in a 3 wt% NaCl solution containing BTAH by using XPS and X-AES (X-ray induced Auger spectroscopy). They concluded that a CuCl layer is formed on Cu and this CuCl layer acts as a base layer for for-mation of the Cu-BTA composite layer whose thickness is determined during the stage of CuCl formation.
On the other hand, Tromans et al. showed that a Cu-BTA composite layer could be grown on Cu in the presence of CuCl2-. According to their mechanism, a mon-olayer of Cu-BTA is formed on the copper surface. Then, (Cu-BTA)ncomposite is formed by the help of CuCl2-within the diffusion distance of CuCl2-. Finally, this com-posite is joined to the Cu-BTA monolayer forming the final Cu-BTA composite layer.
According to the above three mechanisms chloride should be present in the solution for formation of the Cu-BTA protective layer on pure Cu without the Cu2O on the surface. Because Cu2O particles or layers are al-ways found inside the corrosion pits in the sprinkler cop-per tubes, there would be no problem in forming the Cu-BTA protective layer inside the corrosion pits. Chloride ions in the solution would make the formation of the Cu-BTA protective layer on the Cu2O phase inside the corrosion pit a lot easier. Even further, chloride ions became a constituent of the whole structure by forming the CuCl phase. What is implied in the schematic structure of Fig. 14 is that Cu needed to form the CuCl phase is supplied from the Cu2O layer on the Cu material rather than from Cu+in the water. In other words, some parts of the Cu2O layer are replaced with CuCl by a reaction between the Cu2O layer and chloride ions in water. The composition profile in Fig. 12 follows this schematic structure given in Fig. 14. Milosev and Kosec  and Rao et al. also reported that the Cu-BTA protective layer is formed on Cu and Cu-Ni alloy, respectively in the chloride solution. However, they could not observe any CuCl phase in their structure. It seems that the struc-ture of the Cu-BTA complex layer depends delicately on the environment such as chlorine concentration and pH in the solution.
The composite structure of Cu2O/(CuCl+Cu2O)/Cu-BTAformed well inside the corrosion pit of the sprinkler copper tube seems to be useful in preventing corrosion of the copper material in the corrosion pit under the highly acidic and chloride-containing environment.
BTAH was successfully used for reducing the water leakage rate of the sprinkler copper tubes at an apartment complex by about 90%. All the inner parts of the corrosion pits of the sprinkler copper tube were coated well with Cu-BTA films, which is believed to have helped to reduce the water leakage rate. The Cu-BTA film is about 2 nm thick at the bottom of the corrosion pit. The Cu-BTA film was formed on a layer consisting of CuCl and Cu2O phases. This complex structure seems to be useful in stop-ping the growth of corrosion pits in the sprinkler copper tubes and reducing the water leakage rate resultantly.
The authors would like to acknowledge the financial support from Korea Land & Housing Corporation for this research.