In Situ Investigation of the Initial Stages of KCl-Induced Corrosion of a Chromia-Forming Steel at 450°C Using an Environmental Scanning Electron Microscope

Environmental scanning electron microscopy (ESEM) provides an opportunity to maintain a combination of different gaseous environments around a sample while imaging during the experiment in a range of temperatures (from −30°C to 1,500°C).^1-5  Monitoring the interactions between metal surfaces and environmental contaminations, such as gaseous species and alkali salts, at the very early stages of oxidation at high temperatures is one of the practical uses of in situ ESEM experiments.^6-10  Despite the great potential of the technique to view dynamic processes occurring on the materials’ surfaces “live,” the use of this technique for investigating the early stages of oxidation behavior of high-temperature materials is still rare. Working on the oxidation of pure iron, it has been shown that thermogravimetric exposures and in situ ESEM exposures exhibited the same corrosion kinetics and morphology.⁹  In another work by Jonsson, et al.,¹⁰  a low-alloyed steel contaminated with small amounts of KCl(s) and the use of in situ ESEM exposures for understanding the initial stages of oxidation of a low-alloy steel (Fe-2.25Cr-1Mo) was evaluated.

Austenitic FeCrNi stainless steels, such as Sanicro 25^† (UNS S31035),⁽¹⁾ Sanicro 28^† (UNS N08028), and Alloy 310 (UNS S31000), are some of the material choices for power production, for instance in waste-to-energy boilers.^11-13  It is well known that the corrosion resistance of these steels rely on the formation of a protective Cr-rich oxide scale.^14-15  Working with the KCl-induced oxidation behavior of UNS N08028 in an O₂ + H₂O environments at 600°C, Proff, et al.,¹⁶  reported that although a complex oxidation morphology/chemistry was formed on the alloy surface, the Cr-rich oxide scale remained protective during experiments conducted for 168 h. Pettersson, et al.,¹⁷  studied the oxidation behavior of UNS N08028 in an O₂ + H₂O environment at 600°C, and showed that the formation of chromate is responsible for a Cr depletion in the oxide scale.

UNS S31035, which is an austenitic 22Cr25NiWCoCu chromia-forming stainless steel, was primarily designed for advanced pulverized fired steam boilers. This grade of stainless steel is characterized by high oxidation resistance, high creep strength, high structural stability, and good fabricability.^18-19  To the authors’ knowledge, up until now there has been no study regarding KCl-induced oxidation behavior of UNS S31035. Intiso, et al.,^14-15  studied the oxidation behavior of UNS S31035 in dry and wet environments up to 168 h at 600°C to 750°C. The authors reported that the alloy exhibited protective corrosion behavior in O₂ environment at all temperatures by forming a smooth corundum-type base oxide on the surface. However, the scale lost its protective character in O₂ + H₂O atmosphere at temperatures greater than 700°C because of Cr evaporation.

It is important to examine the initial development process and the change in structure and composition of the oxide scales in their earliest stages because these may affect the later oxidation behavior of the alloy. The operation temperature in the super-heater regions of waste-to-energy plants is around 450°C;^20-21  thus, gaining knowledge on the corrosion resistance of steels at temperatures in this temperature regime is of practical importance. While the long-term high-temperature oxidation properties of chromia-forming alloys have been widely investigated in the literature, work on the very early stages of corrosion of this class of stainless steels at low temperatures is less frequent.

The aim of this paper is to show the potential of the in situ ESEM technique for investigating the high-temperature KCl-induced corrosion behavior of a high-alloyed stainless steel, and to shed more light on the effect of KCl on the initial stages of oxidation of the austenitic UNS S31035 at 450°C. Reference exposures were performed in a tube furnace at ambient pressure. Microstructural analyses were performed using focused ion beam (FIB) milling, scanning electron microscopy (SEM) imaging, energy dispersive x-ray analysis (EDX), and scanning transmission electron microscopy (STEM).

The tested alloy is an austenitic stainless steel (UNS S31035) with a nominal chemical composition of (wt%) 22.6 Cr, 25.5 Ni, 3.45 W, 2.99 Cu, 1.57 Co, 0.51 Mn, 0.44 Nb, 0.25 Si, and 0.24 N. The as-received alloy was cut to produce samples with dimensions of 3 × 3 × 2 mm³ using a low-speed saw in order to fit into the small heating stage sample holder used for the ESEM in situ exposures. Each sample was ground using SiC papers, and then polished down to 1 μm using diamond paste. Afterward, the samples were cleaned in acetone and ethanol using an ultrasonic bath. Next, KCl particles were deposited on the sample surface by spraying a saturated solution of 20% distilled water and 80% ethanol with KCl (see Jonsson, et al.,¹⁰  for more details).

In order to validate the in situ ESEM exposures, reference exposures were performed in a standard horizontal tube furnace fitted with a silica tube at 450°C for 1 h with a net flow rate of 200 mL/min (0.2 cm/s) in laboratory air at ambient pressure. The KCl-sprayed samples were mounted on an alumina holder that was put into the tube furnace. After the exposures, the samples were cooled in laboratory air and stored in a desiccator awaiting further analyses.

The in situ ESEM exposures were performed for 1 h at 450°C using an FEI Quanta 200^† field emission gun (FEG) ESEM in the ESEM-mode using laboratory air at a pressure of 4.8 Torr (639.84 Pa). The ESEM has a temperature control unit coupled to a heating stage, which consists of a furnace and a thermocouple, together with a water-cooling system (see Figure 1[a]). During imaging, the emitted low-energy secondary electrons (SEs) interact with the molecules in the chamber to produce (a) an electron cascade effect amplifying the SE-signal and (b) positively charged ions. These ions are drawn toward the negatively charged sample surface where they will neutralize the charging of the sample. A ramp speed of 50°C/min was chosen in this study. While imaging in the ESEM-mode, a high accelerating voltage (30 kV) was used to increase the spatial resolution of imaging and to minimize the primary beam interaction with the gas inside the chamber.

Several KCl(s) particles and their surrounding areas were investigated before and during exposures with the help of a special heat and light insensitive SE detector. In order to be sure that the temperature recorded by the internal thermocouple mounted inside the heating stage showed the real temperature, an external thermocouple was manually attached to the sample to precisely measure the temperature in some trial exposures (see Figure 1[b]). To produce an effective attachment between the external thermocouple and the sample, a small hole was drilled in the sample and then the thermocouple wire was inserted into the hole. The degree of accuracy of the external thermocouple was verified before the experiment at ambient temperature (23°C) and by immersing the external thermocouple in the mixture of ice and water (0°C), and in boiling water (100°C). Furthermore, small droplets of lead were placed beside the sample because the melting point of lead (327°C) is in the range of the experimental temperatures. In the latter case, the droplets were monitored by ESEM as the temperature was increasing and it was noticed that the droplets started to melt when the output temperature of the external thermocouple showed 327°C. Hence, the external thermocouple could reliably be used for calibrating the temperature of the heating stage.

An FEI Versa 3D^† combined FIB/SEM workstation was used to create and investigate cross sections of the oxide scales and subjacent metal of the exposed samples. The site-specific lift-out procedure was performed using the FIB/SEM for subsequent STEM analyses. The ESEM was equipped with an Oxford Inca^† EDX system with a silicon drift detector and the elemental composition of the oxide scales’ cross sections were determined using the microscope high-vacuum mode. To further investigate the morphology and chemical composition of the thin base oxide formed on most of the surface, STEM analysis was performed in an FEI Titan 80-300^† TEM/STEM instrument operating at 300 kV. The microscope was equipped with an EDX detector.

Several initial trials were made in order to evaluate the overall reaction rates for the oxidation processes and to accurately calibrate the temperature of the ESEM hot stage. As mentioned in the Experimental Procedures, the ESEM hot stage is equipped with a temperature unit; however, it was required to make sure that the temperature at the location of the sample was calibrated. Figure 2 shows the temperature calibration curve, which exhibited a rather large difference between the real temperature, the temperature that was recorded by the external thermocouple, and the temperature that was shown by the instrument, i.e., the internal thermocouple. Note that because the external thermocouple exhibited an accurate temperature measurement (verified by means of boiling water, lead, etc.), it is referred as “real temperature” in Figure 2, and the data obtained from the internal thermocouple was calibrated accordingly. After the calibration procedure was successfully performed, the in situ corrosion experiments could be conducted.

As the pressure in the ESEM was reduced to 4.8 Torr (639.84 Pa) and the temperature increased, some of the KCl particles that were sprayed on the sample surfaces disappeared. It was found that in a pure H₂O atmosphere, the salt particles vanished within minutes or even seconds at 450°C, making corrosion studies impossible. When lab air, which has a H₂O content of around 1% to 3% depending on the outside temperature, was used as the oxidizing atmosphere in the ESEM chamber, the residence time for KCl(s) was much longer and large oxide crusts formed at the KCl particles (see below for more details). Hence, lab air was used in this study. It was found that after the initial corrosion attack, a relatively steady state was reached after approximately 15 min. Thus, the exposure time was limited to 1 h.

Figure 3 shows a sequence of ESEM-SE images acquired in situ in the ESEM from UNS S31035 with added KCl during heating and after 4, 6, and 60 min of exposure in lab air at 450°C. The typical size of the KCl particles was in the range of 10 μm to 30 μm. After 4 min of exposure, some of the smaller KCl particles disappeared leaving some oxide traces, while fast growing oxide crusts formed at some of the bigger salt particles. After 6 min, most of the salt particles were consumed, leaving behind oxide crusts and small oxide aggregates at their original positions. After the initial corrosion, the oxide growth rate decreased abruptly and a slow-growing “steady state” was achieved until the exposure was terminated (60 min). It should be noted that at the end of the exposure, all of the KCl was consumed; the fast growing oxide at the former salt particles had grown into porous oxide crusts, while the thin base oxide covering most of the surface in between the KCl particles had grown slightly thicker.

After exposure, the sample was moved to the FIB/SEM for further analyses. An SEM-backscattered electron (BSE) plan view image of the general morphology of the in situ corroded UNS S31035 is shown in Figure 4(a). Thick oxide crusts remained on the surface as skeletons of former KCl particles. A smooth base oxide with slightly thicker nodules was formed between the former salt particles. Figure 4(b) displays a higher magnification SEM-SE image showing the typical morphology of small oxide nodules with the size of 1 μm to 3 μm.

Figure 5 shows SEM-SE images of two cross sections of a large oxide crust and the thin and smooth base oxide scale prepared by FIB milling. In the case of the large oxide crust (Figure 5[a]), no Pt layer was applied, as the oxide scale was much thicker than the depth of ion-beam damage, while for preparing cross section of the thin base oxide, two distinct Pt layers (ion-assisted and electron assisted Pt layers) were applied on the corroded surface prior to the FIB milling procedure. Figure 5(a) shows that the KCl particle was replaced by a shell-like feature that roughly mimicked the shape of the former KCl particle. Additionally, thick porous oxide crusts were formed at the corners of the former salt particle. The image also shows the presence of numerous small nodules on the sample surface (see also Figure 4[b]). Figure 5(b) shows the cross section of the thin base oxide scale (20 nm to 50 nm thick) and the nodules (reaching ~250 nm).

Figure 6 shows an SEM-BSE micrograph at a tilt angle of 45°, together with the corresponding EDX elemental maps. As described earlier, the oxide crusts were present in the corners of the former KCl particle (see the bright regions in the oxygen map in Figure 6). The center of the porous crust was rich in Cr (with some Fe) and O, which suggest the formation of chromium-iron oxide, with the highest amount of Cr at the center. The brighter (BSE image) area at the outer part of the crust was enriched in Ni, Cr, Fe, and O, indicating the presence of spinel-type oxide. The oxide shell was enriched in Fe, Cr, Ni, and O, which can be associated with spinel-type oxide. There is no clear indication of an oxidation affected zone (OAZ) in the top part of the metal. The small nodules were enriched in K, O, and Cr, suggesting they were made up of potassium chromate. At some nodules, there were enrichments in K and Cl, indicating some traces of small KCl regions. In addition, there was Cl present in some regions at the oxide crusts. These were not associated with K and are therefore believed to be metal chlorides.

Figure 7 shows two STEM-high-angle annular dark-field (HAADF) micrographs acquired from a cross section of the base oxide and one nodule. A continuous base oxide layer with the thickness of 20 nm to 100 nm formed on the alloy surface, also below the nodule. The images show the presence of an OAZ (100 nm to 250 nm) that was formed below the oxide layer, in the top part of the alloy. However, no internal oxidation was observed. The STEM-EDX results of the FIB-prepared cross section are presented in Figure 8. The nodule consisted of Cr, K, and O, again confirming the presence of potassium chromate. The thin base oxide extended below the nodule, featuring an Fe-rich top layer and a Cr-rich bottom oxide layer. The OAZ was enriched in Ni, W, and Co and depleted in Cr and Fe. The detection of regions enriched in Cl right below the thin base oxide layer and at the scale/alloy interface is notable. The existence of voids within the potassium chromate nodule was probably a result of decomposition caused by the electron beam during the STEM analysis.

In order to validate the results obtained from the in situ ESEM exposure, a reference exposure was performed using a standard tube furnace. This ex situ exposure was conducted at the same temperature as the in situ analysis, i.e., at 450°C in lab air. The key difference was the higher pressure (1 atm = 760 Torr = 101.325 kPa) and the use of a flowing gas (flow rate 0.5 cm/s). Figures 9(a) and (b) show SEM-BSE micrographs of UNS S31035 after 1 h ex situ and in situ exposures, respectively. A thin and smooth base oxide, together with some slightly thicker potassium chromate nodules, was formed in both cases (compare Figures 9[a] and [b]). The main difference was the less reacted salt particles on the surface of the ex situ exposed sample compared to that of the in situ one. Accordingly, most of the KCl remained after 60 min and the oxide crusts and shells were not as well developed as in the in situ exposures.

The corrosiveness of alkali salts, such as KCl and NaCl, toward high-alloyed stainless steels at high temperatures causes fast kinetics of the initial reactions on the surface,^16,22-25  suggesting that it would be possible to study the early corrosion process of stainless steels by in situ ESEM microscopy. A drawback of the in situ ESEM exposures is that it is generally not possible to introduce highly corrosive gases into the ESEM chamber, as this would lead to severe corrosion of microscope parts, especially at high temperatures. Therefore, only mild atmospheres can be used, such as O₂/H₂O mixtures. Using these types of gases, it is possible to study the corrosion dynamics of low-alloyed steels, or even pure Fe, within a reasonable time, typically 8 h. For corrosion resistant materials, such as stainless steels, a more corrosive environment is needed in order to form oxide scales that are not extremely thin all over the surface, making in situ ESEM imaging difficult. Applying KCl particles to the surface, prior to the exposure, creates a locally highly corrosive environment, even for stainless steels. These types of steels are also intended for corrosive environments, such as when salt deposits are present on the material surface, making the results relevant also from a technological point of view.^26-32 

It is clear from the results obtained in this work that the in situ ESEM exposure and imaging technique is unique, providing insights into the growth dynamics of complicated phenomena, such as the very early stages of KCl-induced corrosion of high-alloyed steels. By continuously recording images during the exposure, the oxide scale growth can be monitored far away from (in between), close to, and even on the KCl particles.

The main differences between ESEM oxidation exposures and standard tube furnace exposures are the lower pressure, the presence of the electron beam on the surface, the less controlled temperature, and lack of flow rate control in the ESEM. The lower pressure limits the maximum temperature that can be used when salt-sprayed samples are exposed, as the salt will be consumed rapidly. For KCl, the maximum usable temperature is around 450°C in lab air. When 500°C was used, the KCl particles vanished almost completely during ramping up of the temperature, making corrosion studies impossible. Also, the gaseous environment can affect the consumption rate of the salt. Thus, in this case, lab air could be used at 450°C, but if pure water vapor was used, the KCl particles vanished almost immediately at the same temperature. The KCl consumption is thought to be a result of the Equations (1) and (2):

Thus, in an environment containing more water vapor, the KCl consumption speed is expected to be higher, in line with these observations.

The consumption of KCl particles inside the ESEM chambers is also a result of the corrosion attack of the steel (see Mortazavi, et al.,³³  for an estimated consumption rate comparison).

As mentioned, there are two other differences between using an ESEM and an ordinary tube furnace for exposures. First, the flow rate cannot be fixed, as in a tube furnace setup, but rather is controlled by heat convection around the sample and the vacuum pump. Second, the presence of the electron beam on the surface during ESEM exposures is also a difference compared to standard exposures. In this work, the relevance of the ESEM exposures could be validated by the results from furnace exposures. The samples exhibited similar oxide scale morphology and features when exposed in the ESEM and in the tube furnace, with the main difference being the higher amount of KCl remaining after 1 h exposure in the tube furnace resulting from less salt consumption.

This is in line with earlier work where the ESEM conditions have not been seen to have a marked effect on the oxide scale thickness, morphology, and features formed, both with KCl¹⁰  and without KCl present.⁹  However, in atmospheric corrosion, the ESEM results differ from standard setups, mainly because of the presence of the electron beam.⁴ 

In general, far away from the KCl particles, a thin base oxide formed together with potassium chromate nodules (see the K, Cr, and O maps in Figures 6 and 8). Most oxide growth occurred at the KCl particles, forming hollow shells and porous crusts (Figures 4[a] and 5[a]). Because of the large corrosion crusts and the base oxide formation, an OAZ was found in the top sub-micrometer part of the metal (Figures 7 and 8). Thus, the presence of KCl in an O₂/H₂O gaseous environment was quite corrosive toward the UNS S31035 after 1 h at 450°C at a reduced pressure of 4.8 Torr (639.84 Pa). Next, the observed regions and features are discussed in more detail.

In general, the main oxide formation occurred in the KCl-particle regions. After approximately 15 min, the KCl particles were consumed and/or volatilized, leaving behind porous oxide crusts (Figures 3 and 6), which formed at former KCl particle edges and consisted of a mixture of Fe-Cr oxide, together with a rim of spinel oxide consisting of Ni, Fe, Cr, and O. The general appearance in the KCl regions is very similar to that found on UNS N08028^16-17  and Type 304 (UNS S30400)³⁴  exposed with KCl(s) at 600°C. On the other hand, KCl particles induced much more severe corrosion on a low-alloyed steel (UNS K21590, Fe 2.25Cr) at 400°C, forming a thick porous oxide scale.^10,35 

When UNS S31035 was exposed to lab air at 450°C in the ESEM, a thin and smooth double-layered base oxide developed in between the KCl particles all over the surface of the alloy. The base oxide consisted of a top layer mainly rich in iron oxide and a continuous bottom layer mainly rich in chromia, with a total thickness of 20 nm to 100 nm (see the STEM-HAADF micrographs and STEM/EDX maps in Figures 7 and 8). A similar double-layered base oxide has been reported to form on UNS S31035 in dry O₂ and wet O₂ environments at 600°C.^14-15  Intiso, et al.,¹⁴  also performed convergent beam electron diffraction analysis on the thin oxide layer and suggested that the layer could be either alternating FeCr₂O₄ (67% Cr) and (Fe,Cr)₂O₃ (80% Cr) structures or alternating FeCr₂O₄ (67% Cr) and metastable (Fe,Cr)-spinel (80% Cr).

STEM-EDX analysis showed the development of an OAZ that was depleted in Cr and Fe (Figure 8). This is probably a result of the formation of the Cr/Fe-rich corrosion products, as the Cr volatilization is limited at 450°C in O₂ + H₂O environment.³⁶ 

The chlorination of the alloy was evidenced by detecting Cl beneath the continuous thin Cr-rich base oxide layer (Figure 8). In a study conducted by Jonsson, et al.,³⁷  on an austenitic stainless steel AISI 310S (UNS S31008), similar metal chlorides were detected below the chromium-rich oxide after 1 h of exposure in a 5% O₂ + 95% N₂ atmosphere and the presence of 500 ppm HCl at 500°C. It was suggested that the formation of metal chlorides is linked to the inward diffusion of chloride ions through the oxide scale and according to the following equation:

Pan, et al.,³⁸  described the increased corrosion rate of two multiphase Fe-Ni-Al and Fe-Ni-Al-Cr alloys in air at 650°C in the presence of KCl as resulting from the formation of potassium chromate. They reported the development of a non-protective Cr₂O₃ scale because of the reaction between chlorine and the protective oxide scale, based on the following equation:

The enrichment of Cl at the oxide scale/alloy interface (see the Cl map in Figure 8) in the present work shows that such mechanisms could also be the case for KCl-contaminated UNS S31035 at low temperature (450°C) and short exposure time (1 h) in O₂ + H₂O environments. In the presence of KCl, metal chlorides are suggested to be formed according to Equation (5). The rate of Equation (5) is proposed to be accelerated because of the formation of potassium chromate, according to Equation (4), which further deteriorates the protective character of the chromia scale:

Thus, the reaction of KCl with the protective chromium oxide resulted in the acceleration of the corrosion process mainly resulting from the formation of potassium chromate. Although metal chlorides are present at the metal/oxide interface, predictions on the long-term oxidation behavior of UNS S31035 in the studied environment requires further prolonged exposures.

The slightly thicker (250 nm) nodules in the base oxide were rich in Cr, K, and O, thus being associated with potassium chromate (Figures 6 and 8). Such nodules have been observed previously both on chromia^16-17,34,39  and alumina forming⁴⁰  steels at 600°C, and also at 400°C on chromia forming steels.³⁹  These chromate particles are not stable, and generally disappear after longer exposure times (~24 h). As explained earlier (see Equation [2]), it is believed that their formation can be detrimental for the corrosion properties, as it uses Cr that could have been used to maintain a chromia-rich scale.⁴⁰ 

• This work shows the possibility to obtain dynamic information about KCl-induced corrosion of highly alloyed steels at 450°C by in situ ESEM oxidation exposures. The results are validated by reference tube furnace exposures.

• It is shown that an accurate temperature calibration of the ESEM hot stage is required for performing reliable in situ ESEM exposures.

• The technique of using pre-sprayed KCl samples works well up to 450°C in lab air, while a lower temperature must be used in pure water vapor because of extensive salt consumption.

• The presence of KCl in an O₂/H₂O environment is locally very corrosive toward the studied stainless steel (UNS S31035) in a matter of minutes even at this relatively low temperature.

• Two types of oxide morphologies were obtained; far away from KCl particles a thin (~50 nm) base oxide formed, while large porous oxide formations were created at the KCl particles.

• The base oxide consists of a double oxide layer where the top layer is rich in iron oxide and the bottom layer is rich in chromia. There is evidence of metal chloride formation beneath the base oxide.

• Potassium chromate particles formed in the base oxide regions by depleting the oxide scale in Cr, which is considered to be detrimental for the corrosion resistance.

• Because of the formation of the corrosion products, a thin OAZ developed in the top part of the metal.

This work was performed in collaboration with the Swedish High Temperature Corrosion (HTC) center. A grant from Knut and Alice Wallenberg Foundation for acquiring the FEG-SEM instrument is gratefully acknowledged.