- Email: [email protected]
Simulating sulfuric acid dew point corrosion of enamels with different contents of silica
Corrosion Science 127 (2017) 201–212
Contents lists available at ScienceDirect
Corrosion Science journal homepage: www.elsevier.com/locate/corsci
Simulating sulfuric acid dew point corrosion of enamels with different contents of silica
MARK
⁎
Ken Chena,c, Minghui Chenb, , Zhongdi Yub, Qunchang Wangb, Shenglong Zhuc, Fuhui Wangb a
College of Material Science and Engineering, University of Science and Technology of China, Hefei 230000, China Key Laboratory for Anisotropy and Texture of Materials (Ministry of Education), School of Material Science and Engineering, Northeastern University, Shenyang 110819, China c Laboratory for Corrosion and Protection, Institute of Metal Research, Chinese Academy of Sciences, Shenyang 110016, China b
A R T I C L E I N F O
A B S T R A C T
Keywords: A: Glass B: Weight loss B: SEM B: Raman spectroscopy C: Acid corrosion
Corrosion of three bulk enamels with different silica contents in sulfuric acid was studied. Catastrophic corrosion happened for enamel with 54.6 wt% silica. Its weight loss reached 36.90 mg/cm2 after seven-day corrosion. A porous leached layer with tremendous cracks formed at surface during corrosion, which favored acid invading and further corrosion. On the contrary, the durability of enamels with silica content over 60 wt% was guaranteed by their highly connected silicate network and a developed dense gel layer that clogged corrosion pits. Their weight losses were less than 0.2 mg/cm2 after seven-day corrosion.
1. Introduction Sulfuric acid dew point corrosion frequently occurs in chemical, power generation industries and marine engine cylinder liners because of condensing gases [1–6]. One of the most serious corrosion components in thermal power plants is the gas–gas heat (GGH) exchanger in desulfurization system (DSS). Burning of the S-containing fuel can produce SO2 and minor SO3, which will react with water to form sulfuric acid vapor in GGH [7,8]. At the cold end, its temperature is only 80–100 °C, and such acid vapor begins to condense either as acid droplets or as a thin liquid layer on GGH surface, leading to the sulfuric acid dew point corrosion of the GGH [9,10]. Because GGH is mostly applied in the consideration of environmental protection, taking into account the cost, it is too expensive to prepare it with high alloyed steel [11]. On the other hand, ordinarily, cost effective low alloyed steels cannot resist such harsh corrosion. Thus, applying corrosion resistant coatings on the cheap low-alloyed steel pieces is the ordinary practice in industry, among which borosilicate enamel coating attracts much attention [12–15]. In that case, the life-span of GGH depends mainly on the corrosion resistance and stability of its enamel coating. However, few researches are available on the corrosion behavior of enamel coating in acid solution, and the only few literatures are mainly focused on the simple comparison of corrosion rate of enamels with different formulations and preparing processes. Rodtsevich et al.[16] compared the weight losses of seven kinds of titanium-boron-silicate
⁎
Corresponding author. E-mail addresses: [email protected], [email protected] (M. Chen).
http://dx.doi.org/10.1016/j.corsci.2017.08.012 Received 26 May 2017; Received in revised form 15 August 2017; Accepted 16 August 2017 Available online 25 August 2017 0010-938X/ © 2017 Elsevier Ltd. All rights reserved.
enamels in 4% acetic acid solution. It was found that enamels with the ratio B2O3: R2O (R2O: Li2O, Na2O, K2O) close to 1 had the highest resistance to corrosion in acetic acid solution. Then Tavgen et al. [17] demonstrated that a satisfactory acid resistance of such titanium-boronsilicate enamels can be achieved by adjusting the contents of B2O3 and TiO2 to 8–10% and 20–22%, respectively. Afterwards, the effect of ratio of three alkali oxides on acid resistance was studied. Results indicated that the titanium-containing enamels with 2–4% K2O, 2–4% Li2O and 8–12% Na2O showed the lowest corrosion rate in acetic acid solution. More recently, Goleus et al. [18] found that corrosion rate of enamels could be reduced significantly by reducing the contents of fluorine, anhydrous boron and alkali oxides, as well as introducing zirconium dioxide to its composition simultaneously. In addition to the composition, the preparing process of enamel coatings affects corrosion rate as well. Svetlov [19] enhanced the corrosion resistance of titanium-containing enamels by adding silica and highly siliceous glasses during the powder milling process. Jiang et al. [20,21] conducted a systematic study of milling addition on corrosion of enamel coating, detecting that adding 10–12 wt% ZrSiO4 [20], 5–7 wt% mullite [21] or 2–5 wt% nano-sized SiO2 [22] can significantly decrease the corrosion rate. However, neither corrosion kinetics nor microstructure evolution of enamel during acid corrosion was carefully studied. Thus a quantitative relationship between the composition and corrosion resistance is still lacking. Considering the decisive role of silica on the microstructure and acid corrosion resistance of an enamel coating [23–27],we developed our
Corrosion Science 127 (2017) 201–212
K. Chen et al.
instruments Co., Oxford, U.K.). Test condition is set as: Gun pressure: 1.90 e−7 Pa; Emission current: 162 μA; HV: 25 KV; WD: 10 mm. Electron probe microanalysis (EPMA-1610, Shimadzu, Kyoto, Japan) technique was employed to study the details of the leached layer. Test condition is set as: Acc.V: 15.0KV; Beam Size: 1 μm; B.C.: 20 nA. Phase constituent was characterized by using XRD (X’PertPRO, PANalytical Co., Almelo, Holland, Cu Ka radiation at 40 kV). To in situ observe the corrosion process of enamel bulk, a mark was made at the middle of sample surface by Vickers impresser. This mark is very small, and should not have a significant impact on corrosion behavior of the bulk. Due to the low conductivity of enamel, a thin carbon layer was always deposited on it before SEM observation, which will affect more or less its further corrosion. In order to avoid such an influence during the cyclic corrosion test, surface morphologies of enamel bulks were observed by a laser scanning confocal microscopy (LEXT-OLS4, Olympus Co., Tokyo, Japan) after each cycle of corrosion, instead of SEM. Raman spectra of pristine samples were obtained by LabRAM HR Evolution Raman spectrometer. Test condition is set as: YAG laser: 532 nm; laser power: 10% of 100 mW; exposure time: 10 s; scan rounds: 6 and scanning range: 300–1300 cm−1.
Table 1 Nominal composition of enamels (wt%).
Si54 Si60 Si64
SiO2
B2O3
Al2O3
Na2O
K2O
CoO
CaF2
54.62 60 64.62
12.32 10.86 9.61
5.96 5.25 4.65
12.32 10.86 9.61
5.96 5.25 4.65
2.46 2.18 1.90
6.36 5.60 4.96
study that follows: firstly, analyze corrosion kinetics of three enamels with different silica content in sulfuric acid to simulate their sulfuric acid dew point corrosion; secondly, record the microstructure evolutions on the surface of these enamels with prolonging corrosion time, to explore the underlying protection mechanism of enamel coating against acid corrosion and the effect of silica content on corrosion resistance. 2. Experiment 2.1. Preparation of samples The nominal composition of three enamels is shown in Table 1. Depending on their silica content, the three kinds of enamel were named as Si54, Si60 and Si64, respectively. Corrosion of steel matrix underlying the enamel coating may occur due to some defects introduced by the preparing process of enamel coating, which might affect the evaluation on acid resistance of the enamel itself. For this reason, enamel bulks, instead of enameled steel plates, were used to compare their corrosion resistance in this experiment. The preparation method of enamel powder has been previously described in detail in [28]. Enamel biscuits with diameter of 13.5 mm and height of 6 mm were prepared by tableting enamel powders through a tablet machine (XL140409, Hebi Lixin instrument Co., Hebi, China). Thereafter, the enamel biscuits were sintered for 5 min at proper temperature (Si54: 700 °C; Si60: 720 °C and Si64: 740 °C) in muffle furnace to form regular shape bulks. In order to simulate the firing process of enameled steel plate, the sintered enamel bulks were further heat-treated following the same procedure of enamel coating, i.e. firing at 880 °C for three minutes and then cooling in air.
3. Result 3.1. Macro morphology Fig. 1 shows macro morphologies of the three kinds of enamel bulks after acid corrosion for different times. Before corrosion (0 d), all the enamel bulks were blue. Their surface was smooth and glossy. However, for Si54, it lost luster and the color faded from blue into white after only one day of corrosion. Differing from Si54, Si60 lost its luster gradually within the first three days of corrosion, and the color transformed to brown. However, with the corrosion continuing to seven days, its surface, incredibly, turned blue and luminous back again. Si64 maintained its gloss and color for the entire corrosion duration. 3.2. Corrosion kinetics Fig. 2 shows corrosion kinetics of the three kinds of enamel bulks at 80 °C in 30% H2SO4 solution. The weight loss of Si54 is catastrophic (36.90 mg/cm2 at seventh day), and far higher than that of Si60 and Si64 (0.1950 mg/cm2 and 0.093 mg/cm2 at seventh day, respectively). For Si54, its weight loss shows nearly a linear relationship with the slope kSi54 = −3.72 mg/(cm2 d). For Si60, its weight loss is small, but is still twice as that of Si64 after seven days of corrosion. At the first three days, its rate of weight loss is relatively high, which slows down thereafter and follows a nearly linear relationship with the slope kSi60 = −0.0102 mg/(cm2 d). For Si64, it is found that the stage of high corrosion rate occurs only within the initial one day, follows by the linear section with the slope kSi64 = − 0.0097 mg/(cm2 d).
2.2. Corrosion test In actual working condition, the temperature of cold end of heat exchanger is about 80 °C. In order to simulate the hot-cold and wet-dry cyclic working environment, the following steps of cyclic corrosion test were carried out: (1) Enamel bulk was rinsed in ethanol for 2 min, dried at 120 °C and weighed; (2) Enamel bulks were immersed in 10 ml 30% (vol.) H2SO4 solution in a Teflon container with condensing device; (3) The container was then fixed in water bath that has already been heated to 80 °C; (4) Bulks were taken out from the container after 24 h corrosion, flushed by deionized water for 2 min (83.3 ml/s of the flow rate) at room temperature to remove the residual solution on their surface; (5) After rinsing in ethanol for 2 min, all the corroded enamel bulks were dried for 1 h in the hot-air oven at 120 ± 5 °C. After a further two-hour holding in the desiccator, they were cooled down to room temperature and their masses were recorded. The above five steps constitute a cycle of acid corrosion. Before the next cycle proceeds, the acid solution used in the former cycle was replaced by a fresh one. According to their corrosion kinetics, enamels with higher silica content enter into a steady stage of corrosion after 4 ∼ 7 days soak, so a total of seven cycles was carried out. Each group of enamel contained three parallel bulks.
3.3. Changes of the chemical composition at surfaces Elemental analysis at surface of the corroded enamel bulks after different corrosion times was carried out by EDS, shown in Fig. 3. For Si54, the elements of Na, K and Co were hardly detected by EDS on surface after one day of corrosion. However, their contents slightly increase thereafter. The contents of Al and Ca also slump rapidly after one day of corrosion. After three days of corrosion, the content of Al improves, following the footsteps of Na, K and Co, but that of Ca decreases continuously with prolonging corrosion time. Anyway, the content variations of Na, K, Co, Al and Ca during the corrosion stage of 2 ∼ 7 days were very small and totally in the scope of measuring error. After the initial one day of corrosion, the percentage of Si goes up from 19.7 to 23.1, and is basically unchanged thereafter. In conclusion, the element content changed substantially during the initial one day of corrosion, and was almost unchanged thereafter.
2.3. Characterization Surface and cross section of the samples were analyzed by SEM (Inspect F 50, FEI Co., Hillsboro, OR) and EDS (X-Max, Oxford 202
Corrosion Science 127 (2017) 201–212
K. Chen et al.
Fig 1. Macro morphologies of the three model enamel bulks (Si54, Si60 and Si64) after corrosion for different times in 30% H2SO4 solution at 80 °C.
3.4. Surface and cross-sectional microstructures 3.4.1. Microstructure evolution of Si54 The surface morphology of Si54 after one day corrosion is shown in Fig. 4a. There were many corrosion cracks on the surface, with crystals of different shapes and sizes developed on them. The chemical composition at points of “A” (precipitated crystal) and “B” (surface area) was analyzed by EDS. EDS results are summarized in Table 2. As shown in Table 2, the precipitated crystal (A) is rich in calcium and sulfur, while the surface area (B) region is rich in silicon. In addition, XRD was used to determine the phase of the crystals. As is shown in Fig. 4c, strong diffraction peaks of anhydrite (CaSO4) are detected, except a broad peak at around of 250 (an indication of the existence of amorphous phase). Fig. 4b shows the cross-sectional microstructures of Si54 after acid corrosion for one day. A 370 μm thick corrosion layer is observed, which is reported as well in many references [29–31] and has been named as the leached layer (LL). This LL was full of netlike cracks. Furthermore, some cracks even penetrated directly from the outermost surface to the enamel/LL interface. Bubbles are found in pristine enamel bulks which are formed during the process of enamel sintering. After corrosion, these bubbles with the similar shape and size are detected as well in the leached layer, indicating that the main part of the LL should be the remnants of enamel bulks after corrosion (or dissolution) rather than the precipitates from the acid solution. Furthermore, some cracks and bubbles in the LL are filled with white solids, as shown in Fig. 4d. Their chemical composition is detected by EDS and shown in Table 2. Combined with previous XRD results, those white solids should also be anhydrite. With corrosion continuing, the crosssectional microstructure is basically unchanged, except that the thickness of LL grows from 370 μm at the first day to 670 μm at the third day, and then to 940 μm at the seventh day. In order to investigate the corrosion behavior of Si54 in details, acid corrosion of Si54 was conducted for a short time. Fig. 5a shows the
Fig. 2. Corrosion kinetics of the three model enamel bulks in 30% H2SO4 solution at 80 °C (inset is the enlarged corrosion kinetics of Si60 and Si64).
For Si60, the contents of Na, Al, K, Ca and Co (they are all modifiers in the network of enamel) drop down continuously within the first three days of corrosion. In contrast to the network modifiers, upward trend of the silica content can be observed for the first three days. Thereafter, there is no obvious change for the composition of Si60 at surface with prolonging corrosion time. For Si64, the contents of Na and Ca drop significantly (from 8.83 to 4.80 and from 1.29 to 0.85, respectively), and the Co content reduces slightly after the first day of corrosion. To be noted, even the largest content change in Si64, i.e. from 8.83% to 4.80% for Na, when compared with the changes recorded with Si54 or Si60. In the subsequent corrosion process, the decreasing rates of Na and Ca slowed down, and the composition of Si64 changed very little with prolonging corrosion time. 203
Corrosion Science 127 (2017) 201–212
K. Chen et al.
Fig. 3. Function of element content at surface of the three model enamel bulks vs. corrosion time in 30% H2SO4 solution at 80 °C.
the surface. This agrees with the results reported in literature [30]. In the location of original corrosion pits, these particles stacked layer by layer, gradually blocked the corrosion pits. When corrosion is executed for seven days, almost all the corrosion pits are blocked and micro cracks appear on the surface. The cross-sectional morphologies of Si60 after three, five and seven days of corrosion are shown in Fig. 8. Corrosion pits with diameters of 10 μm and depth of 7 μm are seen after three days of corrosion, which is consistent with the surface characteristics (Fig. 7a). With regard to the fifth day, BSE-SEM image shows that a gray layer of 1.3–3.7 μm thick grows at the outermost of the corroded enamel bulks. This gray layer consists with the gel layer (GL) that typically observed at surface of other corroded glass systems [29]. In addition, the outer surface of the gel layer is flat, while the inner surface (i.e. the gel/enamel interface) is very rough. Combining with the surface morphologies (Fig. 7), it seems that the corrosion pits formed in the first three days are blocked exactly by such a gel layer when corrosion time prolonging to five days. Immediately under the gel layer, a bright white layer of 1 μm thick can be detected. This may be the result of leaching of Na of enamel during the acid corrosion. After corrosion for seven days, the inner surface of the GL becomes flat. Besides, the bright white layer grows to ∼1.75 μm thick, and it is brighter than that of five days corrosion, indicating that more Na have been leached out.
surface morphology of Si54 after corrosion of only one hour. It can be observed that a great deal of corrosion cracks have already been developed on enamel surface with small CaSO4 crystals precipitated there. Some CaSO4 crystals went deeply into cracks, which was probably the reason why the crystals were hard to be washed away. Fig. 5b presents the cross-sections BSE image of Si54 after corrosion of 1 h. It is shown that the thickness of the LL reaches to 40 μm. There are numerous cracks in this layer. For one thing, a high-magnification SEM image of a crack (Fig. 5d) reveals that countless nano-sized pores are evenly distributed in front of the crack. For another, the interface between the LL and the enamel bulk is very rough and most cracks dwell near the interface. To better understand the distribution of elements in the leached layer of Si54, electron probe microanalysis (EPMA) was employed. Fig. 6 provides details of distribution of the elements Si, S, Ca, Na and K in the corroded enamel. In the pristine enamel bulk, its element is uniformly distributed, whereas in the leached layer, Si and S are enriched while K and Na are almost leached out. As the original distribution of Ca in enamel is uniform, the distribution of CaSO4 crystals should be the same if its formation is due to the direct reaction between enamel and acid solution. But, the fact is that the distribution of Ca is random and has a high degree of consistency with cracks where the high surface energy favors precipitation of new solid phases. Thus, it could be speculated that Ca2+ in the enamel bulk has been dissolved into the acid solution and subsequently precipitation in the cracks.
3.4.3. Microstructure evolution of Si64 With regard to Si64, its corrosion kinetics suggests that the rate of weight loss alleviates after only one day of corrosion. Consequently, microstructures of Si64 after one day and seven days of corrosion were examined by SEM. Fig. 9 presents the surface morphologies of Si64 after corrosion. Different sizes of corrosion pits ranging from 4.6 to 18.8 μm were seen at the enamel surface after acid corrosion for one day (as shown in Fig. 9a). It is like the case of Si60 after three days of corrosion, but the corrosion pits here are relatively narrower and shallower. With respect to the size of corrosion pits, the surface feature of Si64 enamel after one day of corrosion is similar to that of Si60 after five or seven days
3.4.2. Microstructure evolution of Si60 Fig. 7 shows surface morphologies of Si60 after three, five and seven days of corrosion. As shown in Fig. 7a, many corrosion pits with diameter of ∼12 μm form after three days of corrosion. Meanwhile, cracks initiated and spread from their edges. With increasing corrosion time to five days, such corrosion pits disappear and are replaced by many nano pores, as indicated by arrows in Fig. 7b and e. From the macroscopic point of view, the surface becomes flat after five days corrosion. However, a higher magnification of SEM image reveals that the original mirror surface becomes coarse. Nano-sized amorphous particles cover 204
Corrosion Science 127 (2017) 201–212
K. Chen et al.
Fig. 4. Microstructures: (a) surface, (b) cross section (c) XRD pattern and (d) high magnification cross section of Si54 enamel bulk after one day corrosion in 30% H2SO4 solution at 80 °C.
layer (LL) with the thickness of about 2.7 μm is found at the outermost surface. This LL is quite flat. Besides, neither distinct corrosion pits nor gel layer could be detected at surface. When corrosion is executed for seven days, the solo LL has transformed to be a double-layer structure (Fig. 10b). A continuous and dense GL develops at the outmost surface. Its thickness is only 0.9 μm, and the inside and outside interface of this layer was very flat.
Table 2 Chemical composition derived from EDS analysis at points A, B (in Fig. 4a), C (in Fig. 4b), and D, E (in Fig. 9a) (at%).
O Na Al Si K Ca Co S
A
B
C
D
E
77.88 0.00 0.00 0.74 0.00 9.88 0.00 11.50
73.78 0.00 0.56 23.10 0.00 0.36 0.00 2.20
69.08 0.00 0.00 0.00 0.00 14.56 0.00 16.36
67.43 2.27 1.08 25.61 1.72 0.86 0.39 0.65
67.14 4.80 1.81 23.09 1.91 0.85 0.41 0.16
3.5. In situ characterization by LSCM Laser scanning confocal microscopy (LSCM) is employed to observe the clogging progress of corrosion pits formed on the surface of Si60 and Si64. An impress was marked by the Vickers impresser at the surface of enamel bulks before acid corrosion. This mark favors orientation and we can analyze the corrosion behavior at a certain location after different corrosion times. It is thus considered, though not exactly, as an in situ characterization. In situ observation of Si54 is given up as its corrosion occurs catastrophically, which makes it hard to track the original mark. Fig. 11 shows the surface morphologies and 3D model of Si60 after corrosion for different times. In order to exhibit the pore morphology, a reversal treatment of the surface topography is carried out and shown as the 3D images, which means that the bulges in the 3D images corresponds to the sink of the pits on the actual surface. As shown in Fig. 11b, plenty of corrosion pits appear after one day of corrosion.
corrosion, but here its surface is still mirror-like rather than consisting of many nano-sized amorphous particles. EDS analysis at corrosion pits (D) and near area (E) is conducted and reveals that Na, K and Al are specially lean at the corrosion pits, while the content of Si is higher than that of the surrounding area. As shown in Fig. 9d, after seven days of corrosion, the surface is covered by nano-sized amorphous spherules, liking the case of Si60 after seven days corrosion. However, as observed from high-magnification SEM images, the extent of interconnection of these amorphous particles here is so high that the surface becomes almost free of pores. The cross-sectional morphologies of Si64 after corrosion are shown in Fig. 10. After one day of corrosion, as is shown in Fig. 10a, a white 205
Corrosion Science 127 (2017) 201–212
K. Chen et al.
Fig. 5. Microstructures of Si54 enamel bulk after one hour corrosion in 30% H2SO4 solution at 80 °C: (a) surface, (b) cross section, (c) high magnification images of crack and anhydrite, and (d) porous structure of leached layer.
connectivity of the three kinds of enamel, Raman microspectra have been obtained and shown in Fig. 13. According to the reports of Raman characteristics of many glasses or enamels [32–34], the band at ∼1080 cm−1 belongs to the vibration of antisymmetric stretching of SiOb-S (bridge oxygen band with Si), the region at ∼950 cm−1 belongs to the vibration of antsymmetric stretching of Si-Onb (non-bridge oxygen band with Si), and the band at ∼330 cm−1 corresponding to CaF2. As is shown in Fig. 13, with increasing the content of SiO2 from 54% to 64%, the intensity of the peak of 950 cm−1 falls down obviously while the intensity of the peak of 1080 cm−1 extends sharply, implying that the degree of network connectivity of Si64 is the highest. In addition, with the drop of CaF2 content, the intensity of 330 cm−1 peak slumps significantly.
Amongst them, a big pit is marked by an arrow near the impress. Afterwards, the number and depth of pits ascend and reach the extreme at the third day. Cracks begin to cover the whole surface. However, after 4 days, some corrosion pits are clogged, as shown by the arrow mark in Fig. 11d. With the corrosion continuing, most of the corrosion pits disappeared and the surface became smooth again. The 3D images inserted in Fig. 11 reveal as well that the depth and number of the corrosion pits increase from 0 day to three days corrosion, and decrease thereafter. When the Si60 sample is corroded for seventh days, there were only few shallow pits on the surface. The size of impression at the upper left corner keeps almost unchanged before and after corrosion. Fig. 12 shows the surface morphologies and 3D model of Si64 after corrosion for different times. The corrosion morphology of Si64 is generally similar to Si60. The only difference is the time that the corrosion pits begin to be clogged. As to Si60, only one day of corrosion, plenty of corrosion pits form at surface. However, for Si64, the number and size of the corrosion pits begin to reduce thereafter. As shown in Fig. 12c, the enamel surface has become smooth again and only few pits are observed after only three days of corrosion. This surface topography is similar to the Si60 that has been corroded for five days. After corrosion of three days, its surface morphology keeps almost unchanged. To be noted, the degree and the speed of healing of the corrosion pits of Si64 are not as good as Si60.
4. Discussion In terms of corrosion resistance of the three enamel bulks with different silica content, there are two issues that need to be addressed. The first one involves the tremendous difference in corrosion kinetics between Si54 and Si60/64. The second issue is to explore the self-repairing mechanism of corrosion pits in Si60/64. 4.1. Effect of silica content on thickening speed of the leached layer
3.6. Raman microspectroscopy
In this study, it is found that corrosion rate of these enamel bulks in sulfur acid strongly corresponds to their silica contents. As shown in Fig. 2, the content of SiO2 increases by only 5.5 wt% from the Si54
In order to qualitatively determine the degree of network 206
Corrosion Science 127 (2017) 201–212
K. Chen et al.
Fig. 6. EPMA maps of Si, S, Ca, K and Na of Si54 enamel bulk after one hour corrosion in 30% H2SO4 solution at 80 °C.
Although few investigations have been conducted on the corrosion mechanism of enamel in sulfuric acid, the massive researches on aqueous corrosion of borosilicate glass shed light on the understanding of this special corrosion. Referring to the view of Gin et al. [25] and the results of our experiment, we subdivide the corrosive layer into the following three layers: the leached layer (LL), the gel layer (GL) and the
sample to the Si60 one, leading to the sharp reduction of corrosion rate by two orders of magnitude. Combining with their microstructure evolutions, it can be speculated that the reason for the distinct of corrosion resistance between Si54 and Si60/Si64 is probably the difference in thickening speed of the leached layer on acid corrosion, which is directly determined by the compactness of silica network.
Fig. 7. Surface morphology of Si60 enamel bulks after corrosion for: (a) three day, (b) five day, (c) seven day and corresponding high magnified surface morphology: (d) three day, (e) five day and (f) seven day in 30% H2SO4 solution at 80 °C.
207
Corrosion Science 127 (2017) 201–212
K. Chen et al.
Fig. 8. Cross-sectional microstructure of Si60 enamel bulks after corrosion for: (a) three day (SE), (b) five day (BSE), (c) seven day (BSE) in 30% H2SO4 solution at 80 °C.
Schematic diagram of the corrosion process of Si54 and Si60/Si64 are shown in Fig. 14. For Si54, its Raman spectra shows that, because of its low content of silica, the content of the non-bridging oxygen is tremendous and there are ample interconnected areas with loose networks. As shown in Fig. 14a, ion exchange is progressive in the depth direction of the Si54 sample, but in the absence of hydrolyzing many B/ Al-O-Si and/or Si-O-Si bonds. At the same time, processes of ion exchange, hydration and hydrolysis are strongly coupled. On the one hand, the holes formed in the ion exchange process and the channels generated by hydrolysis of the network, lead to the invasion of more acid solution into the enamel bulk and the occurrence of hydration in the deeper interior. On the other hand, the interior hydration will trigger more ion exchange and network hydrolysis reactions [36]. Such mutual promoting processes result in more and more interconnected regions with loose networks, forming exactly the loose LL, as shown in Fig. 14b. In addition to Na+ and K+, Ca2+ in the enamel is also leached out into the acid solution. Thereafter, it combines with the sulfate ion in the solution to form precipitate of calcium sulfate. As the surface of cracks and original bubbles in enamel provides nucleation sites, calcium sulfate crystals are always detected there (as shown in Figs. 4 and
crystalline phase layer (CPL), and the corrosion process into the following five steps: [23,25,26,35,36] (1) Hydration, that is, molecular water as a solvent going into the enamel; (2) Network hydrolysis, i.e. hydrolysis of B/AleOeSi bond or SieOeSi bond:
M − O − M + H2 O → M − OH + HO − M (M = Si, B, Al)
(1)
(3) Ion exchange. Modifier cations are replaced by H+:
Si − O − R + H+ → Si − OH + R+
(2)
(4) Condensation. Silanols produced by hydrolysis or ion exchange condensate:
Si − OH + OH − Si → Si − O − Si + H2 O
(3)
(5) Precipitation of certain phase at surface, such as the anhydrite on Si54 and the gel layer on Si60/Si64.
Fig. 9. Surface morphology of Si64 enamel bulks after corrosion for: (a) one day, (b) seven day and corresponding high magnified surface morphology: (c) one day, (d) seven day in 30% H2SO4 solution at 80 °C.
208
Corrosion Science 127 (2017) 201–212
K. Chen et al.
Fig. 10. Cross-sectional BSE image of the Si64 enamel bulks after corrosion for: (a) one day, (b) seven day in 30% H2SO4 solution at 80 °C.
well as the rate of corrosion rate. As shown in Fig. 14e, a GL precipitates thereafter at surface through the above-mentioned corrosion steps of (3) and (4), i.e. dissolution of Si-O-R into the acid solution under the action of H+ (step (3): ion exchange) and precipitation of Si-O-Si at enamel surface due to the reaction between SieOH and SieOH in solution (step (4): condensation). Specifically, with regard to Si60, though its compactness of silica network is higher than Si54, there are still some areas with comparatively loose network. As a result, ion exchange, hydration and hydrolysis reactions proceed at a relatively fast rate at such areas. Consequently, these areas are preferentially corroded to form corrosion pits, as shown in Fig. 8a. This stage corresponds very well to the initial three days corrosion. However, once such loose networks are depleted, the number of corrosion pit reaches the maximum and the leaching out of mobile ions and hydration reaction are alleviated [37]. For this reason, the thickening rate of the LL is reduced. With the formation of
5a). Once the calcium sulfate crystals grow out of the cracks and reach at the outer surface of LL, they will act as crystal seeds, favoring the formation of a discontinuous CPL. Indeed, these calcium sulfate crystals block part of the crack, as shown in Fig. 14c. However, considering the large quantities of cracks and the high porosity of the LL, the acid solution can still penetrate into the LL easily and contact the surface of the pristine enamel. For this reason, the corrosion rate of Si54 only declines slightly after three days. For Si60/Si64, its Raman spectra reveals that, because of its high content of silica, the content of the non-bridging oxygen is less and the compactness of silica network is higher than Si54. As shown in Fig. 14d, ion exchange and hydration can only be in progress at the surface. Further corrosion into the interior of Si60/Si64 relies on the hydrolysis of more and more B/AleOeSi and/or SieOeSi bonds (Eq. (2)). Hence, the reaction rate of ion exchange and hydration is reduced. At such a case, as shown in Fig. 14e, the thickening speed of the LL decreases, as
Fig. 11. LSCM image and corresponding 3D model at the same position of a Si60 sample after corrosion in 30% H2SO4 solution at 80 °C for different times: (a) pristine, (b) one day, (c) three day, (d) four day, (e) five day and (f) seven day.
209
Corrosion Science 127 (2017) 201–212
K. Chen et al.
Fig. 12. LSCM image and corresponding 3D model at the same position of a Si64 sample after corrosion in 30% H2SO4 solution at 80 °C for different times: (a) pristine, (b) one day, (c) three day, (d) four day, (e) five day and (f) seven day.
less than that of Si60. Accordingly, only narrow and shallow corrosion pits can be formed. In addition, ion exchange and hydration processes into the interior of Si64 are strongly limited by its compact network. For this reason, as shown in Fig. 8c and Fig. 10b, the thickness of the LL of Si64 is thinner than that of Si60 after corrosion for the same time. 4.2. Self-repairing of corrosion pits In this study, it is also found that, for enamels with high silica content, the corrosion pits formed at the initial stage of corrosion can be self-repaired. As shown in Figs. 11 and 12, in situ observation showed that a large number of corrosion pits formed at surface of the Si60/Si64 enamel bulks within the high-speed corrosion stage (three days and one day, respectively). Thus, the surface became very rough (as the 3D images revealed). However, as the corrosion time increases, the corrosion pits are gradually repaired and the surface becomes flat again. As mentioned earlier, the [SiO4] network of some areas at the enamel surface are relatively loose. So these areas will be preferentially corroded. Consequently, corrosion pits are formed there. According to the corrosion model of glass, proposed by Spierings [38], acid solution at the glass surface can be supersaturated with amorphous silica after the initial corrosion stage. Then, the dissolved silica begins to nucleate at hydroxylated enamel surface, especially at the corrosion pits where the concentration of amorphous silica in acid solution is high, and grow to larger silica spherules [39]. Finally, a GL forms upon the leached layer, repairing the etched enamel surface. Regarding to the repairing speed, since the [SiO4] network of Si64 is more compact than that of Si60, the dissolution degree of Si64 is lower than Si60. There is no doubt that the rate of GL formation is strongly dependent on the degree of enamel dissolution. To be exact, the former is in direct proportion to the later. So, the growth rate of GL on Si64 is lower than that on Si60. For this reason, the degree and the speed of healing of the corrosion pits of Si64 are not as good as Si60.
Fig. 13. Raman spectra of the three model enamel bulks.
GL, finally, the corrosion process of enamel enters a low-speed corrosion stage. The corrosion process of Si64 is similar to that of Si60, but the highspeed corrosion stage is shortened from three days to one day, according to the corrosion kinetics (Fig. 2) and the variation trend of the content of flux cations at surface (Fig. 3). By comparing the Raman spectra of Si60 with that of Si64, it can be found that the intensity of peak at 1070 cm−1 of Si64 is higher than that of Si60, while its intensity of peak at 950 cm−1 is lower than that of Si60. This result indicates that the overall connectivity of tetrahedron silica network of Si64 is higher than that of Si60. Moreover, the results of in situ observation of corrosion show that the number and depth of corrosion pits of Si64 are lower than Si60. Therefore, the loose network area of Si64 is 210
Corrosion Science 127 (2017) 201–212
K. Chen et al.
Fig. 14. Schematic diagram showing the corrosion processes of: (a-c) Si54, and (d-f) Si60/Si64. [4] R. Cordtz, J. Schramm, A. Andreasen, S.S. Eskildsen, S. Mayer, Modeling the distribution of sulfur compounds in a large two stroke diesel engine, Energy Fuels 27 (2013) 1652–1660. [5] R. Cordtz, J. Schramm, R. Rabe, Investigating SO3 formation from the combustion of heavy fuel oil in a four-stroke medium-speed test engine, Energy Fuels 27 (2013) 6279–6286. [6] R. Cordtz, S. Mayer, S.S. Eskildsen, J. Schramm, Modeling the condensation of sulfuric acid and water on the cylinder liner of a large two-stroke marine diesel engine, J. Mar. Sci. Technol. (2017) 1–10. [7] M.T. Kim, S.Y. Chang, O.Y. Oh, J.B. Won, H.W. Park, Failure analysis of enamelcoated carbon steel heating elements of gas–gas heater for flue gas desulfurization system, Eng. Fail. Anal. 14 (2007) 686–693. [8] W. Pongsaksawad, S. Kaewkumsai, S. Sorachot, E. Viyanit, Failure of porcelain coated heating elements used in regenerative air preheaters, Mater. Test. 57 (2015) 185–191. [9] B. Zarenezhad, A. Aminian, A multi-layer feed forward neural network model for accurate prediction of flue gas sulfuric acid dew points in process industries, Appl. Therm. Eng. 30 (2010) 692–696. [10] J.M. Blanco, F. Peña, Increase in the boiler’s performance in terms of the acid dew point temperature: environmental advantages of replacing fuels, Appl. Therm. Eng. 28 (2008) 777–784. [11] X. Cheng, F. Sun, S. Lv, X. Li, A new steel with good low-temperature sulfuric acid dew point corrosion resistance, Mater. Corros. 63 (2011) 598–606. [12] M. Chen, W. Li, M. Shen, S. Zhu, F. Wang, Glass coatings on stainless steels for hightemperature oxidation protection: mechanisms, Corros. Sci. 82 (2014) 316–327. [13] Y. Xiong, S. Zhu, F. Wang, Synergistic corrosion behavior of coated Ti60 alloys with NaCl deposit in moist air at elevated temperature, Corros. Sci. 50 (2008) 15–22. [14] W. Li, S. Zhu, C. Wang, M. Chen, M. Shen, F. Wang, SiO2-Al2O3-glass composite coating on Ti-6Al-4 V alloy: oxidation and interfacial reaction behavior, Corros. Sci. 74 (2013) 367–378. [15] M. Chen, W. Li, M. Shen, S. Zhu, F. Wang, Glass–ceramic coatings on titanium alloys for high temperature oxidation protection: oxidation kinetics and microstructure, Corros. Sci. 74 (2013) 178–186. [16] S.P. Rodtsevich, S.Y. Eliseev, V.V. Tavgen, Low-melting chemically resistant enamel for steel kitchenware, Glass Ceram. 60 (2003) 23–25. [17] S.P. Rodtsevich, V.V. Tavgen’, T.S. Minkevich, Effect of alkali metal oxides on the properties of titanium containing glass enamels, Glass Ceram. 64 (2007) 244–246. [18] V.I. Goleus, T.I. Nagornaya, O.N. Rubanova, T.I. Kozyreva, O.B. Gurzhii, Chemical stability of titanium enamel coatings, Glass Ceram. 69 (2012) 274–275. [19] V.A. Svetlov, Increasing corrosion resistance of titanium enamel coatings by modifying the slip, Glass Ceram. 46 (1989) 293–295. [20] H. Qian, J. Li, W. Jiang, Influence of mill addition of ZrSiO4 on properties of the enamels, Mater. Rev. 27 (2013) 372–374. [21] J. Li, Y. Wang, F. Li, R. Li, W. Jiang, Influence of mill addition of mullite on properties of cast iron enamels, Glass Enamel 41 (2013) 9–12. [22] H. Qian, A study on increasing acid resistance of cast iron enamel by milling addition of SiO2 nano-powder, J. Sh. Ins. Tec. 7 (2007) 45–48. [23] S. Gin, P. Jollivet, M. Fournier, F. Angeli, P. Frugier, T. Charpentier, Origin and consequences of silicate glass passivation by surface layers, Nat. Commun. 6 (2015) 8. [24] C. Cailleteau, F. Angeli, F. Devreux, S. Gin, J. Jestin, P. Jollivet, O. Spalla, Insight into silicate-glass corrosion mechanisms, Nat. Mater. 7 (2008) 978. [25] S. Gin, C. Guittonneau, N. Godon, D. Neff, D. Rebiscoul, M. Cabié, S. Mostefaoui, Nuclear glass durability: new insight into alteration layer properties, J. Phys. Chem.
It should be noted that, in this study, we concentrate on only the corrosion behavior of model enamels in acid solution. In fact, the practical environment that the enamel coatings serve in is much more complex than that of our experiment. Notwithstanding its limitation, this study does suggest that the content of silica has a great influence on the acid resistance of enamel. When using the similar composition to design enamel coating for GGH, to achieve a satisfactory protection, the content of silica in the enamel should be at least 60 wt%. 5. Conclusion From the above study, the following conclusions can be drawn: (1) For enamel with lower content of silica (54.6 wt%), disastrous corrosion happened. The weight loss reached 36.90 mg/cm2 after seven days of corrosion. Ion exchange, network hydrolysis, hydration reacted quickly in it for its low connectivity of silicate network. Consequently, a loose and thick leached layer with tremendous cracks is formed during corrosion, which cannot prevent the further intrusion of acid solution. (2) For enamel bearing higher content of silica, its highly connected silicate tetrahedron network and the fact that corrosion pits are all clogged by a dense gel layer, ensure the durability of the enamel in the acid solution. After seven days of corrosion, the weight loss of Si60 and Si64 were only 0.195 mg/cm2 and 0.093 mg/cm2, respectively. Acknowledgements This project is financially supported by the National Natural Science Foundation of China (No. 51471177), the Youth Innovation Promotion Association CAS (No. 2016178), and by the Fundamental Research Funds for the Central Universities (No. N160205001). References [1] Y. Yang, T. Zhang, Y. Shao, G. Meng, F. Wang, In situ study of dew point corrosion by electrochemical measurement, Corros. Sci. 71 (2013) 62–71. [2] X. Wang, Y. Gao, K. Li, J. Yan, Y. Li, J. Feng, Effect of yttrium on the corrosion behaviour of 09CrCuSb alloy in concentrated sulphuric acid, Corros. Sci. 69 (2013) 369–375. [3] R. Ebara, F. Tanaka, M. Kawasaki, Sulfuric acid dew point corrosion in waste heat boiler tube for copper smelting furnace, Eng. Fail. Anal. 33 (2013) 29–36.
211
Corrosion Science 127 (2017) 201–212
K. Chen et al.
[32] S.A. Brawer, W.B. White, Raman spectroscopic investigation of the structure of silicate glasses. I. The binary alkali silicates, J. Chem. Phys. 63 (1975) 2421–2432. [33] S.A. Brawer, W.B. White, Raman spectroscopic investigation of the structure of silicate glasses (II). Soda-alkaline earth-alumina ternary and quaternary glasses, J. Non-Cryst. Solids 23 (1977) 261–278. [34] H. Tsuda, W.L. Jongebloed, I. Stokroos, J. Arends, Combined raman and SEM study on CaF2 formed on/in enamel by APF treatments, Caries Res. 27 (1993) 445–454. [35] S. Gin, Open scientific questions about nuclear glass corrosion, Procedia Mater. Sci. 7 (2014) 163–171. [36] B. Bunker, Molecular mechanisms for corrosion of silica and silicate glasses, J. NonCryst. Solids 179 (1994) 300–308. [37] W. Vogel, Phase separation in glass, J. Non-Cryst. Solids 25 (1977) 170–214. [38] G.A.C.M. Spierings, Wet chemical etching of silicate glasses in hydrofluoric acid based solutions, J. Mater. Sci. 28 (1993) 6261–6273. [39] T. Geisler, T. Nagel, M.R. Kilburn, A. Janssen, J.P. Icenhower, R.O.C. Fonseca, M. Grange, A.A. Nemchin, The mechanism of borosilicate glass corrosion revisited, Geochim. Cosmochim. Acta 158 (2015) 112–129.
C 115 (2011) 18696–18706. [26] S. Gin, I. Ribet, M. Couillard, Role and properties of the gel formed during nuclear glass alteration: importance of gel formation conditions, J. Nucl. Mater. 298 (2001) 1–10. [27] R. Hellmann, S. Cotte, E. Cadel, S. Malladi, L.S. Karlsson, S. Lozanoperez, A. Seyeux, Nanometre-scale evidence for interfacial dissolution-reprecipitation control of silicate glass corrosion, Nat. Mater. 14 (2015) 307. [28] C. Ken, C. Minghui, W. Qunchang, Z. Shenglong, W. Fuhui, Micro-alloys precipitation in NiO- and CoO-bearing enamel coatings and their effect on adherence of enamel/steel, Int. J. Appl. Glass Sci. (2017), http://dx.doi.org/10.1111/ijag.12284. [29] A. Tournie, P. Ricciardi, Glass corrosion mechanisms: a multiscale analysis, Solid State Ionics 179 (2008) 2142–2154. [30] T. Geisler, A. Janssen, D. Scheiter, T. Stephan, J. Berndt, A. Putnis, Aqueous corrosion of borosilicate glass under acidic conditions: a new corrosion mechanism, J. Non-Cryst. Solids 356 (2010) 1458–1465. [31] L. Dohmen, C. Lenting, R.O.C. Fonseca, T. Nagel, A. Heuser, T. Geisler, R. Denkler, Pattern formation in silicate glass corrosion zones, Int. J. Appl. Glass Sci. 4 (2013) 357–370.
212
Contents lists available at ScienceDirect
Corrosion Science journal homepage: www.elsevier.com/locate/corsci
Simulating sulfuric acid dew point corrosion of enamels with different contents of silica
MARK
⁎
Ken Chena,c, Minghui Chenb, , Zhongdi Yub, Qunchang Wangb, Shenglong Zhuc, Fuhui Wangb a
College of Material Science and Engineering, University of Science and Technology of China, Hefei 230000, China Key Laboratory for Anisotropy and Texture of Materials (Ministry of Education), School of Material Science and Engineering, Northeastern University, Shenyang 110819, China c Laboratory for Corrosion and Protection, Institute of Metal Research, Chinese Academy of Sciences, Shenyang 110016, China b
A R T I C L E I N F O
A B S T R A C T
Keywords: A: Glass B: Weight loss B: SEM B: Raman spectroscopy C: Acid corrosion
Corrosion of three bulk enamels with different silica contents in sulfuric acid was studied. Catastrophic corrosion happened for enamel with 54.6 wt% silica. Its weight loss reached 36.90 mg/cm2 after seven-day corrosion. A porous leached layer with tremendous cracks formed at surface during corrosion, which favored acid invading and further corrosion. On the contrary, the durability of enamels with silica content over 60 wt% was guaranteed by their highly connected silicate network and a developed dense gel layer that clogged corrosion pits. Their weight losses were less than 0.2 mg/cm2 after seven-day corrosion.
1. Introduction Sulfuric acid dew point corrosion frequently occurs in chemical, power generation industries and marine engine cylinder liners because of condensing gases [1–6]. One of the most serious corrosion components in thermal power plants is the gas–gas heat (GGH) exchanger in desulfurization system (DSS). Burning of the S-containing fuel can produce SO2 and minor SO3, which will react with water to form sulfuric acid vapor in GGH [7,8]. At the cold end, its temperature is only 80–100 °C, and such acid vapor begins to condense either as acid droplets or as a thin liquid layer on GGH surface, leading to the sulfuric acid dew point corrosion of the GGH [9,10]. Because GGH is mostly applied in the consideration of environmental protection, taking into account the cost, it is too expensive to prepare it with high alloyed steel [11]. On the other hand, ordinarily, cost effective low alloyed steels cannot resist such harsh corrosion. Thus, applying corrosion resistant coatings on the cheap low-alloyed steel pieces is the ordinary practice in industry, among which borosilicate enamel coating attracts much attention [12–15]. In that case, the life-span of GGH depends mainly on the corrosion resistance and stability of its enamel coating. However, few researches are available on the corrosion behavior of enamel coating in acid solution, and the only few literatures are mainly focused on the simple comparison of corrosion rate of enamels with different formulations and preparing processes. Rodtsevich et al.[16] compared the weight losses of seven kinds of titanium-boron-silicate
⁎
Corresponding author. E-mail addresses: [email protected], [email protected] (M. Chen).
http://dx.doi.org/10.1016/j.corsci.2017.08.012 Received 26 May 2017; Received in revised form 15 August 2017; Accepted 16 August 2017 Available online 25 August 2017 0010-938X/ © 2017 Elsevier Ltd. All rights reserved.
enamels in 4% acetic acid solution. It was found that enamels with the ratio B2O3: R2O (R2O: Li2O, Na2O, K2O) close to 1 had the highest resistance to corrosion in acetic acid solution. Then Tavgen et al. [17] demonstrated that a satisfactory acid resistance of such titanium-boronsilicate enamels can be achieved by adjusting the contents of B2O3 and TiO2 to 8–10% and 20–22%, respectively. Afterwards, the effect of ratio of three alkali oxides on acid resistance was studied. Results indicated that the titanium-containing enamels with 2–4% K2O, 2–4% Li2O and 8–12% Na2O showed the lowest corrosion rate in acetic acid solution. More recently, Goleus et al. [18] found that corrosion rate of enamels could be reduced significantly by reducing the contents of fluorine, anhydrous boron and alkali oxides, as well as introducing zirconium dioxide to its composition simultaneously. In addition to the composition, the preparing process of enamel coatings affects corrosion rate as well. Svetlov [19] enhanced the corrosion resistance of titanium-containing enamels by adding silica and highly siliceous glasses during the powder milling process. Jiang et al. [20,21] conducted a systematic study of milling addition on corrosion of enamel coating, detecting that adding 10–12 wt% ZrSiO4 [20], 5–7 wt% mullite [21] or 2–5 wt% nano-sized SiO2 [22] can significantly decrease the corrosion rate. However, neither corrosion kinetics nor microstructure evolution of enamel during acid corrosion was carefully studied. Thus a quantitative relationship between the composition and corrosion resistance is still lacking. Considering the decisive role of silica on the microstructure and acid corrosion resistance of an enamel coating [23–27],we developed our
Corrosion Science 127 (2017) 201–212
K. Chen et al.
instruments Co., Oxford, U.K.). Test condition is set as: Gun pressure: 1.90 e−7 Pa; Emission current: 162 μA; HV: 25 KV; WD: 10 mm. Electron probe microanalysis (EPMA-1610, Shimadzu, Kyoto, Japan) technique was employed to study the details of the leached layer. Test condition is set as: Acc.V: 15.0KV; Beam Size: 1 μm; B.C.: 20 nA. Phase constituent was characterized by using XRD (X’PertPRO, PANalytical Co., Almelo, Holland, Cu Ka radiation at 40 kV). To in situ observe the corrosion process of enamel bulk, a mark was made at the middle of sample surface by Vickers impresser. This mark is very small, and should not have a significant impact on corrosion behavior of the bulk. Due to the low conductivity of enamel, a thin carbon layer was always deposited on it before SEM observation, which will affect more or less its further corrosion. In order to avoid such an influence during the cyclic corrosion test, surface morphologies of enamel bulks were observed by a laser scanning confocal microscopy (LEXT-OLS4, Olympus Co., Tokyo, Japan) after each cycle of corrosion, instead of SEM. Raman spectra of pristine samples were obtained by LabRAM HR Evolution Raman spectrometer. Test condition is set as: YAG laser: 532 nm; laser power: 10% of 100 mW; exposure time: 10 s; scan rounds: 6 and scanning range: 300–1300 cm−1.
Table 1 Nominal composition of enamels (wt%).
Si54 Si60 Si64
SiO2
B2O3
Al2O3
Na2O
K2O
CoO
CaF2
54.62 60 64.62
12.32 10.86 9.61
5.96 5.25 4.65
12.32 10.86 9.61
5.96 5.25 4.65
2.46 2.18 1.90
6.36 5.60 4.96
study that follows: firstly, analyze corrosion kinetics of three enamels with different silica content in sulfuric acid to simulate their sulfuric acid dew point corrosion; secondly, record the microstructure evolutions on the surface of these enamels with prolonging corrosion time, to explore the underlying protection mechanism of enamel coating against acid corrosion and the effect of silica content on corrosion resistance. 2. Experiment 2.1. Preparation of samples The nominal composition of three enamels is shown in Table 1. Depending on their silica content, the three kinds of enamel were named as Si54, Si60 and Si64, respectively. Corrosion of steel matrix underlying the enamel coating may occur due to some defects introduced by the preparing process of enamel coating, which might affect the evaluation on acid resistance of the enamel itself. For this reason, enamel bulks, instead of enameled steel plates, were used to compare their corrosion resistance in this experiment. The preparation method of enamel powder has been previously described in detail in [28]. Enamel biscuits with diameter of 13.5 mm and height of 6 mm were prepared by tableting enamel powders through a tablet machine (XL140409, Hebi Lixin instrument Co., Hebi, China). Thereafter, the enamel biscuits were sintered for 5 min at proper temperature (Si54: 700 °C; Si60: 720 °C and Si64: 740 °C) in muffle furnace to form regular shape bulks. In order to simulate the firing process of enameled steel plate, the sintered enamel bulks were further heat-treated following the same procedure of enamel coating, i.e. firing at 880 °C for three minutes and then cooling in air.
3. Result 3.1. Macro morphology Fig. 1 shows macro morphologies of the three kinds of enamel bulks after acid corrosion for different times. Before corrosion (0 d), all the enamel bulks were blue. Their surface was smooth and glossy. However, for Si54, it lost luster and the color faded from blue into white after only one day of corrosion. Differing from Si54, Si60 lost its luster gradually within the first three days of corrosion, and the color transformed to brown. However, with the corrosion continuing to seven days, its surface, incredibly, turned blue and luminous back again. Si64 maintained its gloss and color for the entire corrosion duration. 3.2. Corrosion kinetics Fig. 2 shows corrosion kinetics of the three kinds of enamel bulks at 80 °C in 30% H2SO4 solution. The weight loss of Si54 is catastrophic (36.90 mg/cm2 at seventh day), and far higher than that of Si60 and Si64 (0.1950 mg/cm2 and 0.093 mg/cm2 at seventh day, respectively). For Si54, its weight loss shows nearly a linear relationship with the slope kSi54 = −3.72 mg/(cm2 d). For Si60, its weight loss is small, but is still twice as that of Si64 after seven days of corrosion. At the first three days, its rate of weight loss is relatively high, which slows down thereafter and follows a nearly linear relationship with the slope kSi60 = −0.0102 mg/(cm2 d). For Si64, it is found that the stage of high corrosion rate occurs only within the initial one day, follows by the linear section with the slope kSi64 = − 0.0097 mg/(cm2 d).
2.2. Corrosion test In actual working condition, the temperature of cold end of heat exchanger is about 80 °C. In order to simulate the hot-cold and wet-dry cyclic working environment, the following steps of cyclic corrosion test were carried out: (1) Enamel bulk was rinsed in ethanol for 2 min, dried at 120 °C and weighed; (2) Enamel bulks were immersed in 10 ml 30% (vol.) H2SO4 solution in a Teflon container with condensing device; (3) The container was then fixed in water bath that has already been heated to 80 °C; (4) Bulks were taken out from the container after 24 h corrosion, flushed by deionized water for 2 min (83.3 ml/s of the flow rate) at room temperature to remove the residual solution on their surface; (5) After rinsing in ethanol for 2 min, all the corroded enamel bulks were dried for 1 h in the hot-air oven at 120 ± 5 °C. After a further two-hour holding in the desiccator, they were cooled down to room temperature and their masses were recorded. The above five steps constitute a cycle of acid corrosion. Before the next cycle proceeds, the acid solution used in the former cycle was replaced by a fresh one. According to their corrosion kinetics, enamels with higher silica content enter into a steady stage of corrosion after 4 ∼ 7 days soak, so a total of seven cycles was carried out. Each group of enamel contained three parallel bulks.
3.3. Changes of the chemical composition at surfaces Elemental analysis at surface of the corroded enamel bulks after different corrosion times was carried out by EDS, shown in Fig. 3. For Si54, the elements of Na, K and Co were hardly detected by EDS on surface after one day of corrosion. However, their contents slightly increase thereafter. The contents of Al and Ca also slump rapidly after one day of corrosion. After three days of corrosion, the content of Al improves, following the footsteps of Na, K and Co, but that of Ca decreases continuously with prolonging corrosion time. Anyway, the content variations of Na, K, Co, Al and Ca during the corrosion stage of 2 ∼ 7 days were very small and totally in the scope of measuring error. After the initial one day of corrosion, the percentage of Si goes up from 19.7 to 23.1, and is basically unchanged thereafter. In conclusion, the element content changed substantially during the initial one day of corrosion, and was almost unchanged thereafter.
2.3. Characterization Surface and cross section of the samples were analyzed by SEM (Inspect F 50, FEI Co., Hillsboro, OR) and EDS (X-Max, Oxford 202
Corrosion Science 127 (2017) 201–212
K. Chen et al.
Fig 1. Macro morphologies of the three model enamel bulks (Si54, Si60 and Si64) after corrosion for different times in 30% H2SO4 solution at 80 °C.
3.4. Surface and cross-sectional microstructures 3.4.1. Microstructure evolution of Si54 The surface morphology of Si54 after one day corrosion is shown in Fig. 4a. There were many corrosion cracks on the surface, with crystals of different shapes and sizes developed on them. The chemical composition at points of “A” (precipitated crystal) and “B” (surface area) was analyzed by EDS. EDS results are summarized in Table 2. As shown in Table 2, the precipitated crystal (A) is rich in calcium and sulfur, while the surface area (B) region is rich in silicon. In addition, XRD was used to determine the phase of the crystals. As is shown in Fig. 4c, strong diffraction peaks of anhydrite (CaSO4) are detected, except a broad peak at around of 250 (an indication of the existence of amorphous phase). Fig. 4b shows the cross-sectional microstructures of Si54 after acid corrosion for one day. A 370 μm thick corrosion layer is observed, which is reported as well in many references [29–31] and has been named as the leached layer (LL). This LL was full of netlike cracks. Furthermore, some cracks even penetrated directly from the outermost surface to the enamel/LL interface. Bubbles are found in pristine enamel bulks which are formed during the process of enamel sintering. After corrosion, these bubbles with the similar shape and size are detected as well in the leached layer, indicating that the main part of the LL should be the remnants of enamel bulks after corrosion (or dissolution) rather than the precipitates from the acid solution. Furthermore, some cracks and bubbles in the LL are filled with white solids, as shown in Fig. 4d. Their chemical composition is detected by EDS and shown in Table 2. Combined with previous XRD results, those white solids should also be anhydrite. With corrosion continuing, the crosssectional microstructure is basically unchanged, except that the thickness of LL grows from 370 μm at the first day to 670 μm at the third day, and then to 940 μm at the seventh day. In order to investigate the corrosion behavior of Si54 in details, acid corrosion of Si54 was conducted for a short time. Fig. 5a shows the
Fig. 2. Corrosion kinetics of the three model enamel bulks in 30% H2SO4 solution at 80 °C (inset is the enlarged corrosion kinetics of Si60 and Si64).
For Si60, the contents of Na, Al, K, Ca and Co (they are all modifiers in the network of enamel) drop down continuously within the first three days of corrosion. In contrast to the network modifiers, upward trend of the silica content can be observed for the first three days. Thereafter, there is no obvious change for the composition of Si60 at surface with prolonging corrosion time. For Si64, the contents of Na and Ca drop significantly (from 8.83 to 4.80 and from 1.29 to 0.85, respectively), and the Co content reduces slightly after the first day of corrosion. To be noted, even the largest content change in Si64, i.e. from 8.83% to 4.80% for Na, when compared with the changes recorded with Si54 or Si60. In the subsequent corrosion process, the decreasing rates of Na and Ca slowed down, and the composition of Si64 changed very little with prolonging corrosion time. 203
Corrosion Science 127 (2017) 201–212
K. Chen et al.
Fig. 3. Function of element content at surface of the three model enamel bulks vs. corrosion time in 30% H2SO4 solution at 80 °C.
the surface. This agrees with the results reported in literature [30]. In the location of original corrosion pits, these particles stacked layer by layer, gradually blocked the corrosion pits. When corrosion is executed for seven days, almost all the corrosion pits are blocked and micro cracks appear on the surface. The cross-sectional morphologies of Si60 after three, five and seven days of corrosion are shown in Fig. 8. Corrosion pits with diameters of 10 μm and depth of 7 μm are seen after three days of corrosion, which is consistent with the surface characteristics (Fig. 7a). With regard to the fifth day, BSE-SEM image shows that a gray layer of 1.3–3.7 μm thick grows at the outermost of the corroded enamel bulks. This gray layer consists with the gel layer (GL) that typically observed at surface of other corroded glass systems [29]. In addition, the outer surface of the gel layer is flat, while the inner surface (i.e. the gel/enamel interface) is very rough. Combining with the surface morphologies (Fig. 7), it seems that the corrosion pits formed in the first three days are blocked exactly by such a gel layer when corrosion time prolonging to five days. Immediately under the gel layer, a bright white layer of 1 μm thick can be detected. This may be the result of leaching of Na of enamel during the acid corrosion. After corrosion for seven days, the inner surface of the GL becomes flat. Besides, the bright white layer grows to ∼1.75 μm thick, and it is brighter than that of five days corrosion, indicating that more Na have been leached out.
surface morphology of Si54 after corrosion of only one hour. It can be observed that a great deal of corrosion cracks have already been developed on enamel surface with small CaSO4 crystals precipitated there. Some CaSO4 crystals went deeply into cracks, which was probably the reason why the crystals were hard to be washed away. Fig. 5b presents the cross-sections BSE image of Si54 after corrosion of 1 h. It is shown that the thickness of the LL reaches to 40 μm. There are numerous cracks in this layer. For one thing, a high-magnification SEM image of a crack (Fig. 5d) reveals that countless nano-sized pores are evenly distributed in front of the crack. For another, the interface between the LL and the enamel bulk is very rough and most cracks dwell near the interface. To better understand the distribution of elements in the leached layer of Si54, electron probe microanalysis (EPMA) was employed. Fig. 6 provides details of distribution of the elements Si, S, Ca, Na and K in the corroded enamel. In the pristine enamel bulk, its element is uniformly distributed, whereas in the leached layer, Si and S are enriched while K and Na are almost leached out. As the original distribution of Ca in enamel is uniform, the distribution of CaSO4 crystals should be the same if its formation is due to the direct reaction between enamel and acid solution. But, the fact is that the distribution of Ca is random and has a high degree of consistency with cracks where the high surface energy favors precipitation of new solid phases. Thus, it could be speculated that Ca2+ in the enamel bulk has been dissolved into the acid solution and subsequently precipitation in the cracks.
3.4.3. Microstructure evolution of Si64 With regard to Si64, its corrosion kinetics suggests that the rate of weight loss alleviates after only one day of corrosion. Consequently, microstructures of Si64 after one day and seven days of corrosion were examined by SEM. Fig. 9 presents the surface morphologies of Si64 after corrosion. Different sizes of corrosion pits ranging from 4.6 to 18.8 μm were seen at the enamel surface after acid corrosion for one day (as shown in Fig. 9a). It is like the case of Si60 after three days of corrosion, but the corrosion pits here are relatively narrower and shallower. With respect to the size of corrosion pits, the surface feature of Si64 enamel after one day of corrosion is similar to that of Si60 after five or seven days
3.4.2. Microstructure evolution of Si60 Fig. 7 shows surface morphologies of Si60 after three, five and seven days of corrosion. As shown in Fig. 7a, many corrosion pits with diameter of ∼12 μm form after three days of corrosion. Meanwhile, cracks initiated and spread from their edges. With increasing corrosion time to five days, such corrosion pits disappear and are replaced by many nano pores, as indicated by arrows in Fig. 7b and e. From the macroscopic point of view, the surface becomes flat after five days corrosion. However, a higher magnification of SEM image reveals that the original mirror surface becomes coarse. Nano-sized amorphous particles cover 204
Corrosion Science 127 (2017) 201–212
K. Chen et al.
Fig. 4. Microstructures: (a) surface, (b) cross section (c) XRD pattern and (d) high magnification cross section of Si54 enamel bulk after one day corrosion in 30% H2SO4 solution at 80 °C.
layer (LL) with the thickness of about 2.7 μm is found at the outermost surface. This LL is quite flat. Besides, neither distinct corrosion pits nor gel layer could be detected at surface. When corrosion is executed for seven days, the solo LL has transformed to be a double-layer structure (Fig. 10b). A continuous and dense GL develops at the outmost surface. Its thickness is only 0.9 μm, and the inside and outside interface of this layer was very flat.
Table 2 Chemical composition derived from EDS analysis at points A, B (in Fig. 4a), C (in Fig. 4b), and D, E (in Fig. 9a) (at%).
O Na Al Si K Ca Co S
A
B
C
D
E
77.88 0.00 0.00 0.74 0.00 9.88 0.00 11.50
73.78 0.00 0.56 23.10 0.00 0.36 0.00 2.20
69.08 0.00 0.00 0.00 0.00 14.56 0.00 16.36
67.43 2.27 1.08 25.61 1.72 0.86 0.39 0.65
67.14 4.80 1.81 23.09 1.91 0.85 0.41 0.16
3.5. In situ characterization by LSCM Laser scanning confocal microscopy (LSCM) is employed to observe the clogging progress of corrosion pits formed on the surface of Si60 and Si64. An impress was marked by the Vickers impresser at the surface of enamel bulks before acid corrosion. This mark favors orientation and we can analyze the corrosion behavior at a certain location after different corrosion times. It is thus considered, though not exactly, as an in situ characterization. In situ observation of Si54 is given up as its corrosion occurs catastrophically, which makes it hard to track the original mark. Fig. 11 shows the surface morphologies and 3D model of Si60 after corrosion for different times. In order to exhibit the pore morphology, a reversal treatment of the surface topography is carried out and shown as the 3D images, which means that the bulges in the 3D images corresponds to the sink of the pits on the actual surface. As shown in Fig. 11b, plenty of corrosion pits appear after one day of corrosion.
corrosion, but here its surface is still mirror-like rather than consisting of many nano-sized amorphous particles. EDS analysis at corrosion pits (D) and near area (E) is conducted and reveals that Na, K and Al are specially lean at the corrosion pits, while the content of Si is higher than that of the surrounding area. As shown in Fig. 9d, after seven days of corrosion, the surface is covered by nano-sized amorphous spherules, liking the case of Si60 after seven days corrosion. However, as observed from high-magnification SEM images, the extent of interconnection of these amorphous particles here is so high that the surface becomes almost free of pores. The cross-sectional morphologies of Si64 after corrosion are shown in Fig. 10. After one day of corrosion, as is shown in Fig. 10a, a white 205
Corrosion Science 127 (2017) 201–212
K. Chen et al.
Fig. 5. Microstructures of Si54 enamel bulk after one hour corrosion in 30% H2SO4 solution at 80 °C: (a) surface, (b) cross section, (c) high magnification images of crack and anhydrite, and (d) porous structure of leached layer.
connectivity of the three kinds of enamel, Raman microspectra have been obtained and shown in Fig. 13. According to the reports of Raman characteristics of many glasses or enamels [32–34], the band at ∼1080 cm−1 belongs to the vibration of antisymmetric stretching of SiOb-S (bridge oxygen band with Si), the region at ∼950 cm−1 belongs to the vibration of antsymmetric stretching of Si-Onb (non-bridge oxygen band with Si), and the band at ∼330 cm−1 corresponding to CaF2. As is shown in Fig. 13, with increasing the content of SiO2 from 54% to 64%, the intensity of the peak of 950 cm−1 falls down obviously while the intensity of the peak of 1080 cm−1 extends sharply, implying that the degree of network connectivity of Si64 is the highest. In addition, with the drop of CaF2 content, the intensity of 330 cm−1 peak slumps significantly.
Amongst them, a big pit is marked by an arrow near the impress. Afterwards, the number and depth of pits ascend and reach the extreme at the third day. Cracks begin to cover the whole surface. However, after 4 days, some corrosion pits are clogged, as shown by the arrow mark in Fig. 11d. With the corrosion continuing, most of the corrosion pits disappeared and the surface became smooth again. The 3D images inserted in Fig. 11 reveal as well that the depth and number of the corrosion pits increase from 0 day to three days corrosion, and decrease thereafter. When the Si60 sample is corroded for seventh days, there were only few shallow pits on the surface. The size of impression at the upper left corner keeps almost unchanged before and after corrosion. Fig. 12 shows the surface morphologies and 3D model of Si64 after corrosion for different times. The corrosion morphology of Si64 is generally similar to Si60. The only difference is the time that the corrosion pits begin to be clogged. As to Si60, only one day of corrosion, plenty of corrosion pits form at surface. However, for Si64, the number and size of the corrosion pits begin to reduce thereafter. As shown in Fig. 12c, the enamel surface has become smooth again and only few pits are observed after only three days of corrosion. This surface topography is similar to the Si60 that has been corroded for five days. After corrosion of three days, its surface morphology keeps almost unchanged. To be noted, the degree and the speed of healing of the corrosion pits of Si64 are not as good as Si60.
4. Discussion In terms of corrosion resistance of the three enamel bulks with different silica content, there are two issues that need to be addressed. The first one involves the tremendous difference in corrosion kinetics between Si54 and Si60/64. The second issue is to explore the self-repairing mechanism of corrosion pits in Si60/64. 4.1. Effect of silica content on thickening speed of the leached layer
3.6. Raman microspectroscopy
In this study, it is found that corrosion rate of these enamel bulks in sulfur acid strongly corresponds to their silica contents. As shown in Fig. 2, the content of SiO2 increases by only 5.5 wt% from the Si54
In order to qualitatively determine the degree of network 206
Corrosion Science 127 (2017) 201–212
K. Chen et al.
Fig. 6. EPMA maps of Si, S, Ca, K and Na of Si54 enamel bulk after one hour corrosion in 30% H2SO4 solution at 80 °C.
Although few investigations have been conducted on the corrosion mechanism of enamel in sulfuric acid, the massive researches on aqueous corrosion of borosilicate glass shed light on the understanding of this special corrosion. Referring to the view of Gin et al. [25] and the results of our experiment, we subdivide the corrosive layer into the following three layers: the leached layer (LL), the gel layer (GL) and the
sample to the Si60 one, leading to the sharp reduction of corrosion rate by two orders of magnitude. Combining with their microstructure evolutions, it can be speculated that the reason for the distinct of corrosion resistance between Si54 and Si60/Si64 is probably the difference in thickening speed of the leached layer on acid corrosion, which is directly determined by the compactness of silica network.
Fig. 7. Surface morphology of Si60 enamel bulks after corrosion for: (a) three day, (b) five day, (c) seven day and corresponding high magnified surface morphology: (d) three day, (e) five day and (f) seven day in 30% H2SO4 solution at 80 °C.
207
Corrosion Science 127 (2017) 201–212
K. Chen et al.
Fig. 8. Cross-sectional microstructure of Si60 enamel bulks after corrosion for: (a) three day (SE), (b) five day (BSE), (c) seven day (BSE) in 30% H2SO4 solution at 80 °C.
Schematic diagram of the corrosion process of Si54 and Si60/Si64 are shown in Fig. 14. For Si54, its Raman spectra shows that, because of its low content of silica, the content of the non-bridging oxygen is tremendous and there are ample interconnected areas with loose networks. As shown in Fig. 14a, ion exchange is progressive in the depth direction of the Si54 sample, but in the absence of hydrolyzing many B/ Al-O-Si and/or Si-O-Si bonds. At the same time, processes of ion exchange, hydration and hydrolysis are strongly coupled. On the one hand, the holes formed in the ion exchange process and the channels generated by hydrolysis of the network, lead to the invasion of more acid solution into the enamel bulk and the occurrence of hydration in the deeper interior. On the other hand, the interior hydration will trigger more ion exchange and network hydrolysis reactions [36]. Such mutual promoting processes result in more and more interconnected regions with loose networks, forming exactly the loose LL, as shown in Fig. 14b. In addition to Na+ and K+, Ca2+ in the enamel is also leached out into the acid solution. Thereafter, it combines with the sulfate ion in the solution to form precipitate of calcium sulfate. As the surface of cracks and original bubbles in enamel provides nucleation sites, calcium sulfate crystals are always detected there (as shown in Figs. 4 and
crystalline phase layer (CPL), and the corrosion process into the following five steps: [23,25,26,35,36] (1) Hydration, that is, molecular water as a solvent going into the enamel; (2) Network hydrolysis, i.e. hydrolysis of B/AleOeSi bond or SieOeSi bond:
M − O − M + H2 O → M − OH + HO − M (M = Si, B, Al)
(1)
(3) Ion exchange. Modifier cations are replaced by H+:
Si − O − R + H+ → Si − OH + R+
(2)
(4) Condensation. Silanols produced by hydrolysis or ion exchange condensate:
Si − OH + OH − Si → Si − O − Si + H2 O
(3)
(5) Precipitation of certain phase at surface, such as the anhydrite on Si54 and the gel layer on Si60/Si64.
Fig. 9. Surface morphology of Si64 enamel bulks after corrosion for: (a) one day, (b) seven day and corresponding high magnified surface morphology: (c) one day, (d) seven day in 30% H2SO4 solution at 80 °C.
208
Corrosion Science 127 (2017) 201–212
K. Chen et al.
Fig. 10. Cross-sectional BSE image of the Si64 enamel bulks after corrosion for: (a) one day, (b) seven day in 30% H2SO4 solution at 80 °C.
well as the rate of corrosion rate. As shown in Fig. 14e, a GL precipitates thereafter at surface through the above-mentioned corrosion steps of (3) and (4), i.e. dissolution of Si-O-R into the acid solution under the action of H+ (step (3): ion exchange) and precipitation of Si-O-Si at enamel surface due to the reaction between SieOH and SieOH in solution (step (4): condensation). Specifically, with regard to Si60, though its compactness of silica network is higher than Si54, there are still some areas with comparatively loose network. As a result, ion exchange, hydration and hydrolysis reactions proceed at a relatively fast rate at such areas. Consequently, these areas are preferentially corroded to form corrosion pits, as shown in Fig. 8a. This stage corresponds very well to the initial three days corrosion. However, once such loose networks are depleted, the number of corrosion pit reaches the maximum and the leaching out of mobile ions and hydration reaction are alleviated [37]. For this reason, the thickening rate of the LL is reduced. With the formation of
5a). Once the calcium sulfate crystals grow out of the cracks and reach at the outer surface of LL, they will act as crystal seeds, favoring the formation of a discontinuous CPL. Indeed, these calcium sulfate crystals block part of the crack, as shown in Fig. 14c. However, considering the large quantities of cracks and the high porosity of the LL, the acid solution can still penetrate into the LL easily and contact the surface of the pristine enamel. For this reason, the corrosion rate of Si54 only declines slightly after three days. For Si60/Si64, its Raman spectra reveals that, because of its high content of silica, the content of the non-bridging oxygen is less and the compactness of silica network is higher than Si54. As shown in Fig. 14d, ion exchange and hydration can only be in progress at the surface. Further corrosion into the interior of Si60/Si64 relies on the hydrolysis of more and more B/AleOeSi and/or SieOeSi bonds (Eq. (2)). Hence, the reaction rate of ion exchange and hydration is reduced. At such a case, as shown in Fig. 14e, the thickening speed of the LL decreases, as
Fig. 11. LSCM image and corresponding 3D model at the same position of a Si60 sample after corrosion in 30% H2SO4 solution at 80 °C for different times: (a) pristine, (b) one day, (c) three day, (d) four day, (e) five day and (f) seven day.
209
Corrosion Science 127 (2017) 201–212
K. Chen et al.
Fig. 12. LSCM image and corresponding 3D model at the same position of a Si64 sample after corrosion in 30% H2SO4 solution at 80 °C for different times: (a) pristine, (b) one day, (c) three day, (d) four day, (e) five day and (f) seven day.
less than that of Si60. Accordingly, only narrow and shallow corrosion pits can be formed. In addition, ion exchange and hydration processes into the interior of Si64 are strongly limited by its compact network. For this reason, as shown in Fig. 8c and Fig. 10b, the thickness of the LL of Si64 is thinner than that of Si60 after corrosion for the same time. 4.2. Self-repairing of corrosion pits In this study, it is also found that, for enamels with high silica content, the corrosion pits formed at the initial stage of corrosion can be self-repaired. As shown in Figs. 11 and 12, in situ observation showed that a large number of corrosion pits formed at surface of the Si60/Si64 enamel bulks within the high-speed corrosion stage (three days and one day, respectively). Thus, the surface became very rough (as the 3D images revealed). However, as the corrosion time increases, the corrosion pits are gradually repaired and the surface becomes flat again. As mentioned earlier, the [SiO4] network of some areas at the enamel surface are relatively loose. So these areas will be preferentially corroded. Consequently, corrosion pits are formed there. According to the corrosion model of glass, proposed by Spierings [38], acid solution at the glass surface can be supersaturated with amorphous silica after the initial corrosion stage. Then, the dissolved silica begins to nucleate at hydroxylated enamel surface, especially at the corrosion pits where the concentration of amorphous silica in acid solution is high, and grow to larger silica spherules [39]. Finally, a GL forms upon the leached layer, repairing the etched enamel surface. Regarding to the repairing speed, since the [SiO4] network of Si64 is more compact than that of Si60, the dissolution degree of Si64 is lower than Si60. There is no doubt that the rate of GL formation is strongly dependent on the degree of enamel dissolution. To be exact, the former is in direct proportion to the later. So, the growth rate of GL on Si64 is lower than that on Si60. For this reason, the degree and the speed of healing of the corrosion pits of Si64 are not as good as Si60.
Fig. 13. Raman spectra of the three model enamel bulks.
GL, finally, the corrosion process of enamel enters a low-speed corrosion stage. The corrosion process of Si64 is similar to that of Si60, but the highspeed corrosion stage is shortened from three days to one day, according to the corrosion kinetics (Fig. 2) and the variation trend of the content of flux cations at surface (Fig. 3). By comparing the Raman spectra of Si60 with that of Si64, it can be found that the intensity of peak at 1070 cm−1 of Si64 is higher than that of Si60, while its intensity of peak at 950 cm−1 is lower than that of Si60. This result indicates that the overall connectivity of tetrahedron silica network of Si64 is higher than that of Si60. Moreover, the results of in situ observation of corrosion show that the number and depth of corrosion pits of Si64 are lower than Si60. Therefore, the loose network area of Si64 is 210
Corrosion Science 127 (2017) 201–212
K. Chen et al.
Fig. 14. Schematic diagram showing the corrosion processes of: (a-c) Si54, and (d-f) Si60/Si64. [4] R. Cordtz, J. Schramm, A. Andreasen, S.S. Eskildsen, S. Mayer, Modeling the distribution of sulfur compounds in a large two stroke diesel engine, Energy Fuels 27 (2013) 1652–1660. [5] R. Cordtz, J. Schramm, R. Rabe, Investigating SO3 formation from the combustion of heavy fuel oil in a four-stroke medium-speed test engine, Energy Fuels 27 (2013) 6279–6286. [6] R. Cordtz, S. Mayer, S.S. Eskildsen, J. Schramm, Modeling the condensation of sulfuric acid and water on the cylinder liner of a large two-stroke marine diesel engine, J. Mar. Sci. Technol. (2017) 1–10. [7] M.T. Kim, S.Y. Chang, O.Y. Oh, J.B. Won, H.W. Park, Failure analysis of enamelcoated carbon steel heating elements of gas–gas heater for flue gas desulfurization system, Eng. Fail. Anal. 14 (2007) 686–693. [8] W. Pongsaksawad, S. Kaewkumsai, S. Sorachot, E. Viyanit, Failure of porcelain coated heating elements used in regenerative air preheaters, Mater. Test. 57 (2015) 185–191. [9] B. Zarenezhad, A. Aminian, A multi-layer feed forward neural network model for accurate prediction of flue gas sulfuric acid dew points in process industries, Appl. Therm. Eng. 30 (2010) 692–696. [10] J.M. Blanco, F. Peña, Increase in the boiler’s performance in terms of the acid dew point temperature: environmental advantages of replacing fuels, Appl. Therm. Eng. 28 (2008) 777–784. [11] X. Cheng, F. Sun, S. Lv, X. Li, A new steel with good low-temperature sulfuric acid dew point corrosion resistance, Mater. Corros. 63 (2011) 598–606. [12] M. Chen, W. Li, M. Shen, S. Zhu, F. Wang, Glass coatings on stainless steels for hightemperature oxidation protection: mechanisms, Corros. Sci. 82 (2014) 316–327. [13] Y. Xiong, S. Zhu, F. Wang, Synergistic corrosion behavior of coated Ti60 alloys with NaCl deposit in moist air at elevated temperature, Corros. Sci. 50 (2008) 15–22. [14] W. Li, S. Zhu, C. Wang, M. Chen, M. Shen, F. Wang, SiO2-Al2O3-glass composite coating on Ti-6Al-4 V alloy: oxidation and interfacial reaction behavior, Corros. Sci. 74 (2013) 367–378. [15] M. Chen, W. Li, M. Shen, S. Zhu, F. Wang, Glass–ceramic coatings on titanium alloys for high temperature oxidation protection: oxidation kinetics and microstructure, Corros. Sci. 74 (2013) 178–186. [16] S.P. Rodtsevich, S.Y. Eliseev, V.V. Tavgen, Low-melting chemically resistant enamel for steel kitchenware, Glass Ceram. 60 (2003) 23–25. [17] S.P. Rodtsevich, V.V. Tavgen’, T.S. Minkevich, Effect of alkali metal oxides on the properties of titanium containing glass enamels, Glass Ceram. 64 (2007) 244–246. [18] V.I. Goleus, T.I. Nagornaya, O.N. Rubanova, T.I. Kozyreva, O.B. Gurzhii, Chemical stability of titanium enamel coatings, Glass Ceram. 69 (2012) 274–275. [19] V.A. Svetlov, Increasing corrosion resistance of titanium enamel coatings by modifying the slip, Glass Ceram. 46 (1989) 293–295. [20] H. Qian, J. Li, W. Jiang, Influence of mill addition of ZrSiO4 on properties of the enamels, Mater. Rev. 27 (2013) 372–374. [21] J. Li, Y. Wang, F. Li, R. Li, W. Jiang, Influence of mill addition of mullite on properties of cast iron enamels, Glass Enamel 41 (2013) 9–12. [22] H. Qian, A study on increasing acid resistance of cast iron enamel by milling addition of SiO2 nano-powder, J. Sh. Ins. Tec. 7 (2007) 45–48. [23] S. Gin, P. Jollivet, M. Fournier, F. Angeli, P. Frugier, T. Charpentier, Origin and consequences of silicate glass passivation by surface layers, Nat. Commun. 6 (2015) 8. [24] C. Cailleteau, F. Angeli, F. Devreux, S. Gin, J. Jestin, P. Jollivet, O. Spalla, Insight into silicate-glass corrosion mechanisms, Nat. Mater. 7 (2008) 978. [25] S. Gin, C. Guittonneau, N. Godon, D. Neff, D. Rebiscoul, M. Cabié, S. Mostefaoui, Nuclear glass durability: new insight into alteration layer properties, J. Phys. Chem.
It should be noted that, in this study, we concentrate on only the corrosion behavior of model enamels in acid solution. In fact, the practical environment that the enamel coatings serve in is much more complex than that of our experiment. Notwithstanding its limitation, this study does suggest that the content of silica has a great influence on the acid resistance of enamel. When using the similar composition to design enamel coating for GGH, to achieve a satisfactory protection, the content of silica in the enamel should be at least 60 wt%. 5. Conclusion From the above study, the following conclusions can be drawn: (1) For enamel with lower content of silica (54.6 wt%), disastrous corrosion happened. The weight loss reached 36.90 mg/cm2 after seven days of corrosion. Ion exchange, network hydrolysis, hydration reacted quickly in it for its low connectivity of silicate network. Consequently, a loose and thick leached layer with tremendous cracks is formed during corrosion, which cannot prevent the further intrusion of acid solution. (2) For enamel bearing higher content of silica, its highly connected silicate tetrahedron network and the fact that corrosion pits are all clogged by a dense gel layer, ensure the durability of the enamel in the acid solution. After seven days of corrosion, the weight loss of Si60 and Si64 were only 0.195 mg/cm2 and 0.093 mg/cm2, respectively. Acknowledgements This project is financially supported by the National Natural Science Foundation of China (No. 51471177), the Youth Innovation Promotion Association CAS (No. 2016178), and by the Fundamental Research Funds for the Central Universities (No. N160205001). References [1] Y. Yang, T. Zhang, Y. Shao, G. Meng, F. Wang, In situ study of dew point corrosion by electrochemical measurement, Corros. Sci. 71 (2013) 62–71. [2] X. Wang, Y. Gao, K. Li, J. Yan, Y. Li, J. Feng, Effect of yttrium on the corrosion behaviour of 09CrCuSb alloy in concentrated sulphuric acid, Corros. Sci. 69 (2013) 369–375. [3] R. Ebara, F. Tanaka, M. Kawasaki, Sulfuric acid dew point corrosion in waste heat boiler tube for copper smelting furnace, Eng. Fail. Anal. 33 (2013) 29–36.
211
Corrosion Science 127 (2017) 201–212
K. Chen et al.
[32] S.A. Brawer, W.B. White, Raman spectroscopic investigation of the structure of silicate glasses. I. The binary alkali silicates, J. Chem. Phys. 63 (1975) 2421–2432. [33] S.A. Brawer, W.B. White, Raman spectroscopic investigation of the structure of silicate glasses (II). Soda-alkaline earth-alumina ternary and quaternary glasses, J. Non-Cryst. Solids 23 (1977) 261–278. [34] H. Tsuda, W.L. Jongebloed, I. Stokroos, J. Arends, Combined raman and SEM study on CaF2 formed on/in enamel by APF treatments, Caries Res. 27 (1993) 445–454. [35] S. Gin, Open scientific questions about nuclear glass corrosion, Procedia Mater. Sci. 7 (2014) 163–171. [36] B. Bunker, Molecular mechanisms for corrosion of silica and silicate glasses, J. NonCryst. Solids 179 (1994) 300–308. [37] W. Vogel, Phase separation in glass, J. Non-Cryst. Solids 25 (1977) 170–214. [38] G.A.C.M. Spierings, Wet chemical etching of silicate glasses in hydrofluoric acid based solutions, J. Mater. Sci. 28 (1993) 6261–6273. [39] T. Geisler, T. Nagel, M.R. Kilburn, A. Janssen, J.P. Icenhower, R.O.C. Fonseca, M. Grange, A.A. Nemchin, The mechanism of borosilicate glass corrosion revisited, Geochim. Cosmochim. Acta 158 (2015) 112–129.
C 115 (2011) 18696–18706. [26] S. Gin, I. Ribet, M. Couillard, Role and properties of the gel formed during nuclear glass alteration: importance of gel formation conditions, J. Nucl. Mater. 298 (2001) 1–10. [27] R. Hellmann, S. Cotte, E. Cadel, S. Malladi, L.S. Karlsson, S. Lozanoperez, A. Seyeux, Nanometre-scale evidence for interfacial dissolution-reprecipitation control of silicate glass corrosion, Nat. Mater. 14 (2015) 307. [28] C. Ken, C. Minghui, W. Qunchang, Z. Shenglong, W. Fuhui, Micro-alloys precipitation in NiO- and CoO-bearing enamel coatings and their effect on adherence of enamel/steel, Int. J. Appl. Glass Sci. (2017), http://dx.doi.org/10.1111/ijag.12284. [29] A. Tournie, P. Ricciardi, Glass corrosion mechanisms: a multiscale analysis, Solid State Ionics 179 (2008) 2142–2154. [30] T. Geisler, A. Janssen, D. Scheiter, T. Stephan, J. Berndt, A. Putnis, Aqueous corrosion of borosilicate glass under acidic conditions: a new corrosion mechanism, J. Non-Cryst. Solids 356 (2010) 1458–1465. [31] L. Dohmen, C. Lenting, R.O.C. Fonseca, T. Nagel, A. Heuser, T. Geisler, R. Denkler, Pattern formation in silicate glass corrosion zones, Int. J. Appl. Glass Sci. 4 (2013) 357–370.
212