Characterization of Rust Formed on Structural Carbon and Weathering Steels Exposed to Tropical Climate of Thailand

Structural carbon steel (CS) and weathering steel (WS) are widely used in many applications for structural construction, e.g., bridges, buildings, etc., and often are required to be exposed to atmospheric environments. Weathering steel has been developed to provide increased corrosion resistance compared to regular carbon steel by alloying additions of Cu, Cr, and Ni which normally promote the formation of protective layers on the steel surface when exposed to a corrosive atmosphere. It is well known that the formation of rust naturally results from metal-electrolyte corrosion reactions taking place at the free surface of the steel. In natural climates, the atmospheric corrosion behavior of the steel is influenced by many environmental parameters, such as wet-dry cycle pattern, relative humidity (RH), rainfall, temperature, corrodents, etc. In an electrochemical process of corrosion, steel is oxidized while water and oxygen are reduced to form corrosion product, namely rust, consisting of iron oxides. The role of corrosion products is somewhat unclear as many researchers have claimed that some corrosion products can provide protection against further corrosion, while other types may accelerate further corrosion attack.^1-2 

The formation of corrosion products depends on environmental parameters, pollution level, exposure conditions, and steel types.³  There are many types of corrosion products found in carbon steels including hollandite or akaganeite (β-FeOOH), goethite (α-FeOOH), lepidocrocite (γ-FeOOH), magnetite (Fe₃O₄), and maghemite (γ-Fe₂O₃).⁴  Among these, α-FeOOH is the most stable phase and can hinder penetration of corrosive species, moisture, and oxygen, contributing to the protectiveness of the corrosion product.⁵  α-FeOOH is the main component which is randomly formed in the inner rust layer.³  It was evident that for weathering steel exposed to industrial atmosphere where pollution from SO₂ was high, the rust layer contained a considerable amount of α-FeOOH. Weathering steel is more corrosion resistant than mild steel for which the rust has lower amount of α-FeOOH.⁶  It was reported that the rust layer usually consists of β-FeOOH when the steel was exposed to high chloride-containing environment.⁷  Nishimura, et al.,⁸  mentioned that β-FeOOH was unstable and could be reduced by electrochemical polarization. This suggests that the formation of β-FeOOH could accelerate corrosion attack during the initial stages of wet/dry cycling because it can act as a cathode. However, the mechanism of β-FeOOH formation is still unclear, and it can only be detected at low pH.¹  γ-FeOOH is the main component observed in the outer layer of rust, which is usually formed during the initial period of exposure before it transforms into other phases.³  Fe₃O₄ is mostly found on surfaces that have been exposed to high moisture conditions or in crevices where the moisture retention rate is high.⁵  γ-Fe₂O₃ is detected in rust from an environment that contains high oxygen level.⁵ 

In considering the potential protective nature of rust against continued corrosion, α-FeOOH is known to be a protective phase. When it is in an amorphous phase, it is an intermediate that can transform into other phases, which may or may not be protective. Tests that involved a few years exposure, performed by Okada, found that amorphous spinel type iron oxide was present in the rust and that this phase was responsible for corrosion protection.⁹  Kamimura, et al.,²  observed that the ratio of crystalline α-FeOOH and γ-FeOOH (α/γ) increases with exposure time in rural and industrial exposure sites. Nevertheless, this correlation was not observed for a seashore environment when the airborne sea salt level exceeds 0.5 mdd. Instead, β-FeOOH and Fe₃O₄ are formed under exposure to a seashore environment. Kamimura, et al., uses the ratio of α/γ*, which is the ratio of the total mass of α-FeOOH to the total mass of γ-FeOOH, β-FeOOH, and Fe₃O₄ combined, as an index for protection. The corrosion rate is considered to be insignificant in any environment when the ratio of α/γ* is greater than one.² 

The development of corrosion products has also been studied by Misawa, et al.¹⁰  When FeOH⁺ is stable, γ-FeOOH is formed by oxidation of FeOH⁺ in a rapid manner. In slightly acidic solution, α-FeOOH is formed via amorphous ferric hydroxide. If the rust layer is thick, the oxidation rate decreases, and green rust is formed. Afterward, Fe₃O₄ is formed beneath the previous rust layer. γ-FeOOH can be dissolved when pH decreases and α-FeOOH can then be formed by reprecipitation.¹⁰  Yamashita, et al., proposed that the rust layer changes from γ-FeOOH in a few years to amorphous and finally to α-FeOOH in decades.¹¹  Corrosion products require several years to develop stable phases. Stabilization time of corrosion products is shorter in marine environments due to their higher corrosivity and the protectiveness is poorer than those products formed in nonmarine environments.³ 

The outer rust layer is composed of loose aggregates and contains a number of voids and microcracks. The inner layer is composed of densely packed fine particles, which later form into larger aggregates. X-ray diffraction (XRD) analysis shows that the main components of rust layers are α-FeOOH and γ-FeOOH with small amounts of Fe₃O₄. EPMA results show that Cr, Cu, and P are distributed on the inner layer of rust on weathering steel. Transmission electron microscopy (TEM) images show that the size of α-FeOOH particles in rust on weathering steel is much smaller (less than 10 nm) than that on mild steel (hundreds nm). A similar phenomenon is observed in the outer layer of γ-FeOOH.¹¹ 

As mentioned above, types of corrosion products formed under atmospheric environments have been extensively studied in nontropical regions for long-term exposed samples.^{1-2,6-8,10-12}  However, some studies have reported on types and formation of corrosion products after short-term (less than 1 y) exposure in tropical climates.^13-17  Characterization of corrosion products can enable greater understanding of the development of such products and how they may or may not provide protection against further corrosion. The aims of this study are (1) to determine types of corrosion products formed in two types of test specimens: structural carbon steel and weathering steel, after 1-y exposure to tropical climates at five different locations across Thailand, and (2) to compare the results with those from other studies around the world.

Rectangular test specimens with a dimension of 150 mm × 70 mm × 6 mm were prepared from structural CS and WS plates. The compositions of these steels are listed in Table 1. In atmospheric corrosion testing, the exposed steel specimens were installed on an electrically insulated test rack and tilted at 45° from the horizontal plane according to ASTM G50 and JIS Z 2381 standards.^18-19  In order to gain an insight into effects of sample orientation on corrosion behavior, the exposed steel surfaces were set to face both the skyward and the groundward directions. The exposure duration time was one year starting from June 2014. Besides the actual exposure test, climate parameters including RH, temperature, chloride deposition, rainfall, wind speed, and wind direction were monitored using a weather station according to the JIS Z 2382 standard.²⁰  Figure 1 shows the weather station which was set up at each atmospheric corrosion test site for recording the environmental parameters. Chloride (Cl⁻) and sulfur dioxide (SO₂) deposition rates were also determined using a dry gauze method (ISO 9225:2012(E)) and PbO₂ cylinders (ISO 8407:1991(E)), respectively.

After the one-year exposure period, the exposed specimens were taken from the exposure sites, preserved in plastic bags and placed in an ultralow desiccant box until the samples were characterized. In order to avoid any destructive action to rust layers during the sample preparation process, each exposed sample was mounted in epoxy resin prior to being cross sectioned. Optical images were taken using an optical microscope via polarized light mode.

It is widely agreed that the atmospheric corrosion behavior of steel is dependent on the climate parameters. Thus, it is important to know the environmental characteristics of each selected test site location in various regions of Thailand. The environmental data were acquired from five test stations: (1) Bangkok, (2) Chonburi, (3) Phangnga, (4) Rayong, and (5) Khon Kaen. The geographic locations of the exposure sites are shown in Figure 2. Characteristics of these sites are listed in Table 2. In Thailand, there are two major wind directions depending on the season. During the wet season, June–October, the climate is dominated by southwest wind coming from the Indian Ocean. During the dry season, cold wind from China blows from the northeastern direction as indicated by dark arrows in the diagram (Figure 2). Figure 3 shows the corrosion rate observed for each test site after one year exposure with corrosion category given according to ISO 9223 standards.²¹  Corrosion rate was obtained by weight loss method. Only Phangnga in the south of Thailand exhibited the highest corrosive environment of C4 category, whereas other sites were assessed as C2 category. Bangkok and Khon Kaen exhibited similar corrosiveness while Rayong and Chonburi presented comparable corrosion rate. The corrosion rate of the carbon steel was found to be larger than that of the weathering steel at all test sites. Moreover, as shown in Figure 3, the corrosion rate of the CS was significantly higher than that of the WS when exposed to the coastal environment at Phangnga.

Table 3 shows relevant environmental parameters that were monitored during the one-year exposure period. Temperature, RH, and SO₂ are given as average values. It is important to note that Cl and SO₂ values in Table 3 were recorded as the deposition rate of those ions on the corresponding detectors. The value of Cl was based on assumption that all Cl comes from NaCl according to the JIS Z 2382 standard.²⁰  The maximum value of monthly chloride deposition rate is shown in the table to demonstrate the worst-case scenario for corrosion. Phangnga exhibited the highest chloride deposition rate, which was higher in the wet season than in the dry season. Chloride deposition rate can be ranked from highest to lowest in this following order: Phangnga > Chonburi > Rayong > Bangkok > Khon Kaen. Deposition of sulfur dioxide at the Phangnga site was significantly lower than the other test locations. The average temperature was fairly stable at approximately 28°C for all the test sites. Average RH was around 87% during the wet season and 70% during the dry season. Bangkok and Rayong exhibited high sulfur dioxide deposition rate due to their urban and industrial environments, respectively. The Chonburi site showed relatively high chloride deposition rate because the site was only 150 m away from the seashore. Khon Kaen was the mildest site of all showing low chloride and sulfur dioxide deposition rate, however, the variations of RH and temperature from month to month at this test site were the highest. Rain frequency was collected from the ACM sensor shown in Figure 1 when the current increased more than 1 micro-ampere. The ACM sensor uses galvanic current between two metals to determine time of wetness and corrosion rate. More information about the ACM sensor can be found in another study.²² 

Microscopic features of the rust layers were characterized using the same sample cross sections prepared for optical microscopy. A scanning electron microscope (SEM) set at an accelerating voltage of 15 kV was used to characterize rust features, i.e., cracks, porosity, surface topography, and density, in each of the cross sections.

Rust formed on the exposed samples of CS and WS was carefully removed using a ceramic scraper. About 0.3 g of each rust powder and 30 wt% of ZnO was crushed to a uniform size by using a pestle and mortar for 30 m. Powder XRD analysis was conducted with a Cu target. The XRD equipment was equipped with a scintillation detector and a diffracted beam monochromator with 0.8 mm slit size. The setting of the beam was 40 kV and 200 mA. The diffraction angle (2θ) range was between 5° and 80°. The step width was 0.02°. The calibration curve was prepared with ZnO to determine quantitative phase analysis of the rust. Peaks used for XRD quantitative analysis are listed as follows:

Electron Probe Micro Analysis (EPMA) was used to determine elemental distributions in each rust layer. The instrument was set at 15 kV and 50 nA with a field of view of 120 μm × 120 μm. Resolution was 0.4 μm. Dwell time was 30 ms/point.

Spots by Raman Spectroscopy analyses were chosen from bright outer layer rust and a dark inner layer rust for phase determination. The wavelength of the laser was 532 nm with 1% laser power. The objective lens magnification was 10×. A pinhole slit size was 50 μm. Measurement frequency range was 4,000 cm⁻¹ to 100 cm⁻¹. The exposure time was 30 s with 20 repetitions.

Unlike a conventional optical microscope, a polarized optical microscope is capable of differentiating between bright and dark-color rust. Figures 4 and 5 show the polarized optical micrographs of cross-sectioned CS and WS samples exposed skyward and groundward at Bangkok and Phangnga sites, respectively. The rust layer can be seen between the epoxy resin coating (dark area) and steel substrate (bright grey area). It is important to note that the groundward-facing samples were oriented as shown in these figures. From the polarized optical micrographs, thickness of the rust layer was not uniform and areas of localized corrosion attack were observed. In addition, multiple micro cracks occurred in the rust layer, especially in the thicker areas. The difference in thickness of rust formed on CS and WS was minimal. The outermost rust layer was categorized as a bright layer because it was revealed as reddish-brown in color. Underneath the bright rust layer, the dark rust layer was formed adjacent to the steel surface. The thickness of rust layer was directly proportional to corrosion rate: Phangnga > Chonburi = Rayong > Bangkok > Khon Kaen. Rust on groundward-facing specimens was thicker than on those skyward-facing. Bright and dark layers are clearly distinguishable in Figures 4 and 5. For the Bangkok, Rayong, Phangnga, and Chonburi environments, rust clearly consists of the bright and dark layers. But, for Khon Kaen sample, the bright rust layer was observed near the surface of rust and the base metal. Some bright rust appears in a crack of the rust layer. For Phangnga where the rust layers were thick, alternating layers existed between bright rust and dark rust. No difference in morphology of the rust between CS and WS was observed. Overall, bright rust was more prominent in CS than in WS, except for the Phangnga samples.

SEM images show the morphology of cross-sectional outermost and innermost rust layers. The outermost layer was that directly exposed to the atmosphere while the innermost layer was present between the outermost layer and the steel substrate. All samples from the five exposure sites were characterized. Figures 6 and 7 show representative SEM images of CS and WS exposed in Bangkok and Phangnga, respectively. It is important to note that the orientation of the groundward-facing sample is the opposite to that in the previous figures.

In every sample, the outermost layer appeared to be more porous and contained more defects than the inner layer. The outermost layer exhibited a sponge-like feature on the top surface. For the Phangnga samples, the outermost layer seemed to be denser than those from other sites. The innermost layer also seemed to be denser than outermost layers; however, some innermost layer contained cracks and defects. Comparing skyward- and groundward-facing samples, the outermost layer of groundward-facing samples appeared to be denser than that of the skyward in all exposure sites.

Spot Raman spectroscopy was conducted to identify phases present at selected positions in the rust layer. Two spots were characterized on the rust layer of WS exposed skyward at Phangnga, representing the bright outermost and dark innermost rust, as shown in Figure 8. Raman peaks were analyzed with respect to the reference peaks. The results of Raman spectroscopy are tabulated in Table 3. The results revealed that γ-FeOOH was found in the outermost layer (bright color) while α-FeOOH was noticed in the innermost layer (dark color). Unknown peaks (671 cm⁻¹ to 693 cm⁻¹) appeared in all of the WS samples. On the other hand, for the CS samples, the unknown peaks were obtained for the samples exposed at Bangkok, Rayong, and Phangnga. The unknown peaks are preliminary assumed to be magnetite (Fe₃O₄).

Powdered rust collected from CS and WS surfaces by a ceramic scraper was analyzed by XRD. Figure 9 shows XRD diffractograms of all samples. The dashed lines indicate the peaks which were used for quantitative analysis. Quantitative analysis of XRD provided the proportion of each phase present in the rust. Figure 10 shows a quantitative representation of phase composition calculated using internal calibration by ZnO calibration curves. The rust was composed of five phases: amorphous, lepidocrocite (γ-FeOOH), goethite (α-FeOOH), akaganeite (β-FeOOH), and magnetite (Fe₃O₄). The diagram in Figure 10 was truncated at the top and amorphous phase used to complete to 100%. Amorphous phase was the major component found in the rust on both CS and WS, contributing more than 65%. CS exposed at Bangkok had the lowest amount of amorphous phase (65.8%) while CS at Khon Kaen had the highest amount of amorphous phase (82.8%). γ-FeOOH, α-FeOOH, and β-FeOOH had the second, third, and fourth proportions present, respectively. Fe₃O₄ was only obtained for both CS and WS exposed at Phangnga. The γ-FeOOH level was highest for CS exposed at Bangkok (27%) and was the least for CS exposed at Khon Kaen (12.8%). Comparing CS and WS, the level of amorphous phase was lower in CS than in WS except at Khon Kaen. The increased amount of γ-FeOOH was more pronounced in CS than in WS exposed at all the test sites except for Khon Kaen. The presence of β-FeOOH and α-FeOOH was less in CS than in WS for all of the test sites. In the Phangnga climate, the rust on CS contained a higher quantity of Fe₃O₄ compared to WS. For CS, SiO₂ peaks can be clearly observed at Rayong, Chonburi, and Khon Kaen, while for WS, such peaks can only be evidently seen only at Khon Kaen. As illustrated in Figure 9, in the case of Rayong and Chonburi, peaks of SiO₂ and β-FeOOH appear to be overlapping in the diffractogram, therefore they can be misinterpreted. For Khon Kaen, the site was located near to the airport, thus it was possible that SiO₂ from sand and ground dust could deposit on the sample surface. The presence of SiO₂ in the rust layer of CS exposed to urban atmosphere in Cambodia was reported previously by other authors.¹⁶ 

EPMA reveals the distribution of elements in the rust layers. EPMA results for WS exposed in Bangkok are shown in Figure 11. Regarding alloy additions of Cu, Ni, and Cr to enhance corrosion resistance, the results of WS revealed that the rust layer was mostly enriched with Cr followed by Cu. However, Ni distribution was not clearly noticable. Moreover, WS showed concentration gradients of Cu and Cr with both elements seeming to be more concentrated in the inner layer close to the steel substrate. The concentration gradient of Cu and Cr was present in WS at all exposure sites in a similar manner. Figure 12 illustrates the distribution of chloride and sulfur in the rust layer of CS exposed at each test site. It seems that chloride and sulfur are present and concentrated at the bottom of the inner rust layer at Phangnga. At Chonburi, sulfur not only accumulated more at the same location as chloride, but also was distributed throughout the rust layers. At Rayong and Khon Kaen, sulfur was concentrated in the pits. At Bangkok, a high amount of sulfur was detected throughout the rust layers.

Characteristics of rust layers were examined by polarized optical microscopy, SEM, and EPMA. It was found that thicker rust, especially at Phangnga (Figure 5), contained more defects and cracks both in longitudinal and transverse directions than thinner rust layers. The bright rust layer observed in the polarized light microscope appeared to be porous. The dark rust layer was denser and formed as the inner layer. However, the thickness of the dark rust layer was not related to its protectiveness because dark inner layer at Phangnga site where corrosion is the highest is the thickest for all sites. The bright and dark rust layers can be visually distinguished by polarized light microscopy. Alcantara, et al., found that lighter color rust appeared in a high chloride environment while darker color rust appeared in a low chloride environment. However, as exposure time progressed the color of rust became darker.²³  For carbon steel, the rust layer formed has a mottled structure with γ-FeOOH (illuminated under polarized light) and α-FeOOH (dark under polarized light).³  In this present study, the bright layer appeared to be at the outermost region where it was directly exposed to the environment. On the other hand, the dark rust layer was observed near the steel substrate for both groundward-facing and skyward-facing orientations. Rust on groundward-facing samples was clearly thicker than skyward-facing samples due to an absence of rain washout effect. The interlayer of bright rust was more evident in the rust of samples exposed at the Phangnga site. This could be due to the effect of high airborne chloride deposition that causes interlayer formation of bright and dark rust.²⁴  Additionally, the alternating layers between bright and dark rust could be attributed to cyclic changes in climates, particularly the wet-dry cycles, over many periods. The change of climates can cause different phase transformation over many periods of time. It was observed that the bright rust layer in CS is more prominent than that in WS in all sites except for Phangnga. In comparison between CS and WS, the area ratio of bright rust layer to dark rust layer in each site was similar. Image processing was conducted to determine the area ratio of bright to dark rust. A MATLAB algorithm was developed to differentiate between steel substrate, epoxy, and rust, and then to distinguish between the bright and dark rust layers. The process of image processing is demonstrated in Figure 13. Figure 14 shows the percentage of bright rust in the rust layer. The results quantitatively confirmed that bright rust constituted more of the rust layers in CS than in WS.

In a previous study, thick rust was observed to exfoliate.²⁵  One study has found that a critical level of chloride to cause exfoliation of rust was 3 mdd (milligram per dm² per day). However, rust exfoliation was not observed in the current study because the average chloride deposition rates at all test stations of this study were below this critical value. The thickness of rust layers was directly proportional to corrosion rate. Similar findings were found in a previous study.²⁶  The thickness of rust is not uniform such that some areas were attacked more than others resulting in pit formation.

SEM images reveal that the outermost layer seemed to be more porous and contained numerous cracks and defects. The innermost layer was dense while some innermost layers at Bangkok, Rayong, and Chonburi sites contain cracks. Heterogeneity and anomalous appearance of rust layers increase with high salinity.²⁵  However, there was no clear difference of heterogeneity in SEM images of rust at the Bangkok and Phangnga sites. The porous outermost layer allowed easier penetration of moisture and corrodent than the dense inner layer. Thus, the inner rust layer contributed to the protectiveness of the rust.

Certain levels of alloying elements, chloride, and sulfur distribution in the rust layer were determined by EPMA. This variation can be related to corrosion attack. In case of WS, the chloride penetration was easier than in the case of CS. The protectiveness of rust layer of WS is increased as a result of the presence of Ni, Cu, and Cr. In WS, Cu, and Cr were concentrated at the bottom of the rust layer. Ni was not found because its role was to stabilize the passive film.²⁷  Okada, et al., has found that Cu, P, and Cr in the rust layer formed on WS after a few years of exposure while Si was present in the outer layer. Slight amounts of S were observed in both layers. Slight amount of Cl scattered and distributed at isolated positions in both layers. The protectiveness of rust layer of WS was increased as a result of adding Ni, Cu, and Cr elements. In WS, Cu and Cr were concentrated at the bottom of the rust layer. In the case of CS, when rust starts to crack, sulfur and chloride can penetrate through those cracks leading to more severe corrosion. Rain frequency also plays a part in chloride penetration as evident by the high amount of Cl in the rust at Phangnga site. Rusts developed on the samples exposed at Khon Kaen (less rain) did not contain as high amount of Cl as those at Phangnga. At Bangkok, despite low Cl deposition rate, there was some Cl penetration into the rust layer. The most frequent wet-dry cycles ranking from high to low are: Phangnga >> Bangkok > Rayong > Chonburi >> Khon Kaen. Chloride distribution in the rust layer might help the formation of Fe₃O₄ in the Phangnga sample because rusts from other sites do not contain Fe₃O₄. It is important to note the source of the S detected by EPMA mapping shown in Figure 12. For the Phangnga and Chonburi sites (both marine sites), S seems to be distributed in the same regions as Cl. Seawater contains about 2.68 g sulfate ions per one kilogram of seawater.²¹  It is very likely that S in the rust of Phangnga and Chonburi came from salinity. On the other hand, Bangkok (urban), Rayong (industrial), and Khon Kaen (rural) experienced high pollution from combustion engine emissions and industrial pollution. The Khon Kaen site is located near an airport, so pollution from aircraft engines is the likely cause of SO₂. Although Khon Kaen has low SO₂ from the detector, it does not rain as frequently as at other sites, so the washout effect is minimal. It can be seen in Figure 12 that S appears all over the rust while chloride does not. The PbO₂ detector detects SO₂ gas from the atmosphere, not from salinity. This explains why S was detected at Phangnga and Chonburi sites where SO₂ levels indicated by the detector were low.

When corrosion started to occur on CS and WS, γ-FeOOH formed first appearing as orange/brownish rust on the steel surface. The oxidation of FeOH⁺ to γ-FeOOH was very fast.¹⁰  Figure 15 shows a schematic of evolution of rust formed on CS and WS that occurred in this experiment. In the early stage of rust formation, γ-FeOOH was found to be loose and porous, so corrosion continued to occur. As it thickened and more γ-FeOOH formed, γ-FeOOH was partially transformed into amorphous and then α-FeOOH formation was activated via amorphous hydroxide. The β-FeOOH and Fe₃O₄ formed later when the particular environment suits its formation. Yamashita, et al.,¹¹  used infrared transmission microscopy, XRD, and TEM to characterize powdered corrosion product. At the initial stage of atmospheric corrosion, γ-FeOOH is formed primarily. The corrosion rate is high at the initial stage because γ-FeOOH is electrochemically active and acts as an anodic site. As exposure time is extended, dense and thick corrosion product develops. At this stage, γ-FeOOH turns into α-FeOOH via ferric hydroxide. α-FeOOH is said to be stable and electrochemically inactive.¹¹ 

In the current study, it was found that the rust comprises of various phases including amorphous, β-FeOOH, α-FeOOH, γ-FeOOH, and Fe₃O₄. It was possible that γ-Fe₂O₃ was present in the rust layer. However, this phase could not be easily distinguished by XRD because Fe₃O₄ and γ-Fe₂O₃ have very similar crystal structures and the XRD technique sometimes cannot differentiate the two. Both phases can coexist in the inner rust layer, close to the steel surface where oxygen supply is limited.^28-29  From spot Raman spectroscopy, the loose bright orange-colored outermost rust layer was composed of γ-FeOOH. The dense dark innermost rust layer was mainly α-FeOOH. Amorphous phase cannot be located where it exists in the rust layer, although the majority of the rust was amorphous. This observation agrees with previous studies.²⁵  Raman mapping showed that the inner part of rust was α-FeOOH (the central part of the rust layer) and ferrihydrite (Fe₅HO₈.4H₂O). In polarized light microscopy, both of these phases appear dark. For WS, the rust formed showed a similar structure but the specimen exhibited a higher portion of ferrihydrite.²⁶  A review by Morcillo, et al., also claimed that the outer layer consisted of γ-FeOOH and the inner layer was α-FeOOH.³ 

β-FeOOH formation requires halogen ions to stabilize its structure. Chloride content of 2 mol% to 7 mol% has been reported to trigger the formation of such phase.³⁰  It has also been reported that β-FeOOH can be developed when the average RH exceeds 80% and airborne chloride deposition rate is higher than 0.6 mdd.²⁴  However, other results from different studies claimed the critical chloride concentration for β-FeOOH to form was 0.5 mdd.²  In this current study, an airborne chloride deposition rate above 0.6 mdd was detected only at Phangnga during the wet season. However, the phases detected in the present work are different from other previous studies which found β-FeOOH in high chloride environment.^1,7-8  In Thailand, β-FeOOH was present for all sites (see Figure 10) even at Khon Kaen where chloride level was less than 0.1 mdd. However, Fe₃O₄ phase was present only at Phangnga. Fe₃O₄ was formed by ferric species of oxyhydroxides with ferrous species in the solution. β-FeOOH tended to be electrochemically reduced. β-FeOOH was the most likely oxyhydroxide to transform to Fe₃O₄. The potential for transformation to Fe₃O₄ was found to be in the following order: β-FeOOH > γ-FeOOH > α-FeOOH.^31-32  This finding supports the observation that Fe₃O₄ was found only in the Phangnga tests in that the β-FeOOH which formed during exposure in a high chloride environment with high humidity transformed into Fe₃O₄. Asami and Kikuchi also found the similar relationship between β-FeOOH and Fe₃O₄.³³  The reaction equation can be expressed as follows:

Hiller has discovered that rust formed in a high chloride environment contained higher content of Fe₃O₄ than that formed in a lower chloride environment.³⁴  Kamimura, et al., has found β-FeOOH and Fe₃O₄ in the rust layers of samples exposed to marine environment.² 

Most of the other studies have focused on rust characterization on samples with longer exposure times. For example, in a study in Spain on mild steel exposed for 13 y,³⁵  only α-FeOOH, γ-FeOOH, Fe₃O₄/γ-Fe₂O₃, and Fe₅HO₈·4H₂O were found. In Japan, mild steel and weathering steel were exposed for 15 y, 17 y, and 32 y.^33,36-37  Corrosion products consisted mainly of γ-FeOOH, α-FeOOH, β-FeOOH, and Fe₃O₄. Additionally, γ-Fe₂O₃ was observed. β-FeOOH was found at the surface of rust layers from marine sites. Our study should be compared with work on a short-term exposure.

Composition of rust on carbon steel exposed to tropical climates in Thailand consisted of amorphous, α-FeOOH, and γ-FeOOH as the main phases. β-FeOOH was found in rust from every exposure site. Fe₃O₄ was found only in Phangnga samples where chloride deposition rate was the highest among all sites. In South/Central America, there are several studies on corrosion product characterization. Carbon steel was exposed for 14 months in Colombia where chloride and SO₂ deposition rates at Cartagena and Barranquilla were comparable to those for Bangkok and Phangnga, respectively.³⁸  The main compounds found in corrosion product collected from the samples at all test sites were α-FeOOH, γ-FeOOH, and Fe₃O₄, while β-FeOOH was detected only at the station with high chloride deposition rate. In Brazil, 3-, 6-, and 9-month exposure revealed that γ-FeOOH, α-FeOOH, and Fe₃O₄ were the main compositions of the rust layer.³⁹  XRD indicated the presence of Fe₃O₄ on the samples exposed to industrial atmosphere for 6 and 9 months, whereas this phase was not present on the samples exposed at the same site for 2 and 3 months. Moreover, Fe₃O₄ was not detected at all on the samples exposed to urban atmosphere for up to 9 months.¹³  In our study, Fe₃O₄ was detected only after one-y exposure in a marine environment. A study in Panama, whose latitude is similar to Thailand, has also found Fe₃O₄ after exposure in a marine environment.⁴⁰  In Vietnam, characterization by XRD of the corrosion products from three-year exposure tests on carbon steels revealed the presence of γ-FeOOH and α-FeOOH for both urban and marine sites. β-FeOOH was observed only on the samples exposed to an atmosphere with airborne salinity above 0.04 mdd. In terms of annual average chloride deposition rate in the present work only Bangkok and Khon Kaen recorded below 0.04 mdd. Fe₃O₄ has only been observed in unsheltered samples.¹⁴  In Cambodia, carbon steels were exposed for one year at two locations. XRD analyses of the corrosion products indicated γ-FeOOH and α-FeOOH as the main components for all samples. The presence of β-FeOOH was not found due to its low intensity in XRD diffractogram. SiO₂ (which overlaps with β-FeOOH peaks) was present and assumed to be from contamination with small grains of sand. However, surface morphology studies found β-FeOOH.¹⁶  It was possible that β-FeOOH was present as a small amount, so it was not detectable by XRD. However, some other techniques, such as SEM and EPMA, could be more suitable.

Composition of rust on the weathering steel exposed to tropical climates in Thailand is similar to that on the carbon steel. The difference lies in the % content of each phase. Diaz, et al., reported results for weathering steel exposed for one year at two marine atmospheres in Spain.⁴⁰  XRD revealed that the rust developed after one year of exposure at the test station with 0.75 mdd was composed of γ-FeOOH, α-FeOOH, and Fe₃O₄/γ-Fe₂O₃. Surprisingly, β-FeOOH was not detected. Cano, et al.,^26  conducted a study on weathering steel exposed for two year in a rural atmosphere and two marine environments. At rural atmosphere with chloride and SO₂ deposition rates of 2.7 mmd and 0.3 mmd (ISO 9225), respectively, α-FeOOH, γ-FeOOH, and Fe₅HO₈·4H₂O were observed after 2-y exposure. Chloride deposition rate at this site was comparable to Khon Kaen. At intermediate severe marine atmosphere with chloride and SO₂ deposition rates of 29.1 mmd and 0.6 mmd, respectively, the same corrosion product compounds as the rural site was detected. Chloride deposition rate at this site was comparable to Chonburi. It is worth mentioning that chloride deposition rated monitored on the second year at this test site drastically decreased to 11.6 mmd. At severe marine atmosphere with chloride and SO₂ deposition rates of 73.9 mmd and 0.9 mmd, respectively, α-FeOOH, γ-FeOOH, β-FeOOH, Fe₅HO₈·4H₂O, and Fe₃O₄/γ-Fe₂O₃ were found. Chloride at this site is comparable to Phangnga. In another study in Vietnam, β-FeOOH was observed on weathering steel samples exposed at the station with intermediate chloride deposition rate during 3 and 6 months of exposure. After 12 months of exposure, β-FeOOH was found on all samples and γ-Fe₂O₃ seemed to transform to Fe₃O₄ with prolonged exposure of 12 months.^15,17 

The comparison to other studies around the world has shown that there were some differences and similarities in observations of the phases present in corrosion product. The common phases recorded were γ-FeOOH, α-FeOOH, Fe₃O₄, and Fe₂O₃. The results from Thailand, for which Fe₃O₄ was found after exposure at Phangnga and β-FeOOH was found for all sites, are very similar to results of the studies in Panama, Vietnam, and Cambodia whose latitudes are all in the tropical zone. Studies in other countries, outside the tropical region, have found that β-FeOOH can only be detected after exposure to high-chloride environments. β-FeOOH formation in low-chloride environment could be characteristics of this South East Asia region and Panama where RH and temperature are high throughout the year. Further investigation is needed to explain this phenomenon.

The time for rust to stabilize is essential to determine long term protectiveness of the rust and predict long term atmospheric corrosion behavior. For the exposure times in the present work, phase transformations in the rust films were believed to be incomplete and, hence, they may be insufficient to predict long term behavior. The stabilization time is said to decrease as the corrosivity of an environment increases.⁴¹  The stabilization time of corrosion product is shorter in a marine environment due to its high corrosivity. However, the protective nature of the stabilized rust film is said to be poorer than that in a nonmarine environment.³  Characterization of samples after longer-term tests than in the present study may be beneficial towards a better comprehension of rust film development on CS and WS for exposure at urban, marine, and rural sites in Thailand.

• •

  Characterization of rust on CS and WS samples that have been exposed to selected climatic conditions in Thailand for one year was conducted by polarized light microscopy, SEM, XRD, EPMA, and Raman. The results indicated that the bright rust outermost layer was loose and porous and contained many defects. It was identified as γ-FeOOH which could later transform to α-FeOOH or Fe₃O₄ depending on the exposure environment. At unknown locations in the rust layer, β-FeOOH can also transform into Fe₃O₄ under a condition of high chloride deposition combined with high humidity. This transformation was observed only at Phangnga where corrosion rate was the highest due to high chloride and high humidity. The dark innermost layer of rust films was mainly composed of α-FeOOH which was dense and provided some corrosion protection. EPMA results reveal that there was significant amount of sulfur and chloride around any cracks that were present. Sulfur can originate from salinity as well as from atmospheric pollution in a form of sulfate and SO₂. Comparison to other studies has shown that similar results were found for other tropical climate sites in Panama, Vietnam, and Cambodia.

The authors would like to express gratitude to National Metal and Materials Technology Center (MTEC), Thailand and JFE Steel Corp., Japan for financial and technical support. The authors also gratefully acknowledge Dr. John Pearce and Prof. Supapan Seraphin for proofreading and valuable comments.