Hot Corrosion of Na₂ SO ₄-Coated Ti₃AlC₂ in Air at 700-1000°C

Fully dense polycrystalline material used in this work was fabricated by a novel solid-liquid reaction synthesis method using commercially available elemental powders of Ti, Al, and graphite with near stoichiometry as starting materials. A detailed description of the fabrication process has been published elsewhere.¹⁹ The specimens with dimensions of mm³ were obtained by cutting a hot-pressed slab using electrodischarge machining (EDM). The as-cut flat specimens were ground to 1000 grit finish, ultrasonically degreased in acetone, followed by blow drying. Thereafter, saturated solution was brushed onto the surfaces of the specimens that were placed on a hot plate kept at 200°C. The amount of salt was controlled to be mg/cm².

Weight changes of coated with mg/cm² in flowing air at various temperatures were continuously measured in a Setsys 16/18 thermoanalyzer (Setaram, France) using thermogravimetric analysis (TGA). The flat specimens were suspended with a Pt wire and heated at a rate of 40°C/min to the tested temperature. At each temperature, duplicate runs were carried out to check the reproduction of the test. After TGA tests, the specimens were washed in hot water to remove soluble salts and then characterized from a physicochemical point of view. Phases in the specimens after hot corrosion were identified using X-ray diffraction (XRD) and Raman spectroscopy. XRD was carried out in a D/max 2500PC diffractometer (Rigaku, Japan) with Cu Kα radiation. The tube voltage was 50 kV and the tube current was 200 mA. XRD patterns were collected from 5 to 90° in 0.02° steps with a scanning rate of 0.4° (2θ)/min. Raman spectra were recorded at room temperature in a LabRam HR 800 Raman spectrometer using a He-Ne laser with 632.8 nm wavelength as excitation source. For cross-sectional microstructure observations, the flat specimens were carefully sectioned and polished. Surface and cross-sectional morphology observations of the hot-corroded specimens were conducted in a JSM-6301F field emission scanning electron microscope with a resolution of 1.5 nm (JEOL, Japan). The SEM is equipped with a LINK ISIS 300 energy-dispersive spectroscopy (EDS) system (Oxford Instruments, UK). Prior to the scanning electron microscopy (SEM) examinations, a thin layer of gold was sputtered on the surface and cross section of the hot-corroded specimens.

The hot corrosion kinetics was evaluated in the form of continuous weight change with exposure time using TGA. As mentioned in the Experimental section, duplicate runs were conducted independently to check the reproduction of the tests. The results show that specific weight gain rates are approximate, indicating that the tests have good reproduction.

The hot corrosion kinetics of coated with in air at 700-1000°C is presented in Fig. 1a. It is seen that the specific weight gain increases with time and the rates are much larger, especially at high temperatures, than those of the oxidation of in air,^{3 20} implying that the presence of on accelerated the oxidation rate. In other words, hot corrosion of occurred in the presence of The hot corrosion rates are relatively low at 700 and 800°C (which are below the melting point of pure (884°C)), whereas above the melting point of the weight gains increase rapidly at 900 and 1000°C, indicating that suffered severe hot corrosion by

Zoom In Zoom Out Reset image size

At 700°C, the hot corrosion rate was low and obeyed a parabolic rate law. However, at 800°C the hot corrosion kinetics was generally linear. At 900°C where was present in fused form, the hot corrosion kinetics showed a typical tri-stage feature for hot corrosion by fused salts, as shown in Fig. 1b. In stage I, the so-called gestation period, the weight gain increased at a relatively slow rate (7.20 mg cm⁻² h⁻¹). After stage I, the hot corrosion came into a high-speed corrosion period as marked by stage II in which the weight gain increased at a high-speed rate (22.3 mg cm⁻² h⁻¹). Due to the formation of oxides on the surface during the high-speed corrosion period, the hot corrosion rate remarkably slowed and came into stage III, i.e., the steady-state hot corrosion stage in which the hot corrosion rate was low (0.60 mg cm⁻² h⁻¹). The low weight gain rate in the gestation period (stage I) is probably due to formation of a thin oxide layer as the sample was coated with on the hot plate kept at 200°C. cannot contact until the thin layer of oxide formed during the coating of is dissolved by it. Once directly contacts it results in a high-speed hot corrosion featured by the large weight gain rate. The flux of oxygen through the sodium sulfate melt in either the gestation stage, high-speed corrosion period, or the steady-state corrosion stage cannot account for such high weight gain rates. According to Fick's first law and the approximation of Luthra,¹⁷ the flux of oxygen can be estimated by

where J is the flux of oxygen, D is diffusivity of oxygen in a sodium sulfate melt, is the concentration of oxygen at the melt/gas interface, is the concentration of oxygen at the melt/carbide interface, and x is the melt thickness. D can be taken as a typical liquid diffusivity cm² s⁻¹).²¹ The solubility of oxygen in melt at 900°C is mol cm⁻³ atm⁻¹.²² According to Henry's law, is estimated to be mol cm⁻³. is small relative to and can be ignored. Finally, x is calculated from the cross section to be about 14 μm. Substituting these values into Eq. 1 gives an oxygen uptake of 0.16 mg cm⁻² h⁻¹ oxygen. At higher temperature of 1000°C, no gestation period was observed. It started from a high-speed corrosion period and lasted for about 3 h, followed by a parabolic steady-state hot corrosion as shown in Fig. 1c. A weight loss in the initial stage occurred, which is similar to the interaction of TiC powders with in 1:1 molar ratio in flowing oxygen at 1200 K reported by Mobin and Malik.²³ The initial weight loss was presumably due to the reduction of the salt to and/or and subsequent explusion of and gases.

During exposure of sodium sulfate to air at high temperatures, especially above the melting point of (884°C), it evaporates molecularly;²⁴ consequently, weight loss results. So the effect of the evaporation of sodium sulfate on the hot corrosion kinetics should be taken into account. Uhlig and co-workers²⁴ studied vaporization kinetics of from 900 to 1200°C in 150 Torr They reported that the kinetics was linear in all cases and that the rate constants at 900 and 1000°C were and mg cm⁻² h⁻¹, respectively. As the weight gains after 20 h hot corrosion at 900 and 1000°C are in the range of tens of micrograms, the weight loss resulting from evaporation of sodium sulfate can be neglected with respect to the low evaporation kinetics of

Analogous to the hot corrosion kinetics at 800°C, the steady-stage hot corrosion kinetics at 900°C is also linear, whereas at 1000°C the steady-stage hot corrosion kinetics obeys typical parabolic rate law. The change in kinetics from linear at lower temperatures (800 and 900°C) to parabolic at higher temperature (1000°C) is discussed in detail in the following sections.

Phases in the specimens are the remains after hot corrosion, which can provide important information on what happened during the hot corrosion. So phase identification was carried out after obtaining the hot corrosion kinetics. The phases identified by XRD diffraction are summarized in Table I.

Table I. 

Phases identified by XRD in the specimens hot corroded in air at 700-1000°C for 20 h.a
Phases	Relative intensities at different temperatures
	700°C	800°C	900°C	1000°C
	vs	vs	⋯	⋯
(anatase)	vw	⋯	⋯	⋯
	w	vs	s	w
	⋯	s	vs	vs
	⋯	m	s	vs
	⋯	w	w	vw
	⋯	w	⋯	⋯
	⋯	vw	w	w
	⋯	⋯	w	m
a and

For the sample hot corroded at 700°C, the dominant phase was with two modifications of (anatase and rutile), which is similar to the oxidation in air.²⁰ For the sample hot corroded at 800°C, the intensity of the diffractions from (rutile) was greater than that of was also detected. In addition to these two oxides, Na-containing titanium dioxide and alumina including and were identified, which indicates that interaction of with occurred.²⁵ At lower temperatures of 700 and 800°C, substrate was detected by XRD. However at 900 and 1000°C, the substrate was not detected by XRD. At 900 and 1000°C, a species containing more Na, i.e., was detected, and its intensity increased with temperature. In the temperature range 800-1000°C, the intensity of the diffractions from (rutile) decreased with temperature while those for Na-containing species of and increased. It should be noted that in the samples hot corroded at 900 and 1000°C, substrate was not detected, suggesting that only the outer scales were detected by XRD.

Raman spectroscopy was also used to tail the changes in phase and structure evolution of oxides formed on the surface of during hot corrosion by with temperature. At 700°C, characteristic Raman bands for (rutile) and especially that centered at 146.7 cm⁻¹ for (anatase) are obviously observed. As temperature increased to above 800°C, the Raman bands showed similarity, suggesting that the oxides formed on the surface of during hot corrosion by has similar crystal structure.

The typical surface morphologies of -coated hot corroded in air at 700-1000°C for 20 h showed distinctly different characteristics. At 700°C, the surface of the sample was smooth and its microstructure was also relatively smooth. For the sample hot corroded at 800°C, the surface became rough with numerous nodules, as shown in Fig. 2a. A high-magnification micrograph of the nodules indicates that they consist of coin-like grains such as those shown in Fig. 2b. Elemental composition analysis by EDS indicates that the coin-like grains contain about 2.5 atom % Na, which is inconsistent with the XRD result in which Na-containing species are identified (Table I). As temperature increased to 900°C, nodules consisting of columnar grains are obvious. Micropores and microcracks are also observed on the nodules. Figure 2c shows the surface morphology of the specimen hot corroded at 900°C for 20 h. The micropores and microcracks are regarded as the evolution of gas phase during hot corrosion. Mobin and Malik²³ studied the high-temperature interaction of titanium carbide (TiC) powders with in flowing air at 1200 K. They found that high-temperature interaction of titanium carbide powders with initially resulted in rapid weight losses followed by rapid weight gains. The weight losses are due to the evolution of gaseous phases during the decomposition of carbide in the presence of The weight gains are attributed to the formation of sodium metal oxide and metal sulfide by the mutual interaction of metal oxide, and In the present work, contains a large amount of carbon whose oxidation would result in gaseous CO and/or these reactions are thermodynamically favorable²⁶

The expulsion of gaseous phases through melt left micropores and microcacks on the surface. At 1000°C, the morphology of the surface of the sample became smooth again and the grains of the oxides grew bigger (Fig. 2d). Microcracks are also observed on the surface. Elemental composition analysis of the outer grains by EDS demonstrates that the sodium content increases to 6-9 atom % Na depending on different regions.

Zoom In Zoom Out Reset image size

Information obtained from the surfaces of the samples is important to understand the hot corrosion behavior of however, it is insufficient. The microstructure information and especially the elemental distribution from cross section is more important than that obtained from the surface. So, the cross-sectional microstructures of the samples after hot corrosion were also carefully investigated by SEM/EDS.

Because the hot corrosion at 700°C is slight; the scale is rather thin, which is in agreement with the hot corrosion kinetics (Fig. 1a). The hot corrosion rate is similar to that of oxidation in air,²⁴ so hot corrosion at 700°C is not described in detail here.

For the sample hot corroded at 800°C, the corrosion scale is relatively thin. For those hot corroded at higher temperatures of 900 and 1000°C where fused was present, the thickness of the scales increase markedly, revealing that suffered severe attack by at high temperatures, which agrees with the high corrosion rates at those temperatures (Fig. 1a). Figure 3a shows the secondary electron image of the cross section of -coated hot corroded at 800°C for 20 h. At 900°C, the scale consists of a porous thin outer layer, which is consistent with the surface morphology, and a relatively dense thick inner layer. In the middle of the thick inner layer, an Al-rich area is observed. At 1000°C, an obvious gap is observed between the thin outer layer and the thick inner layer in the scale of the sample hot corroded. The gap is regarded as the place where soluble salts are present during the hot corrosion. Immediately behind the gap, an Al-rich area is present. Figures 3b and c present backscattered electron images of the specimens hot corroded at 900 and 1000°C, respectively.

Zoom In Zoom Out Reset image size

We have known that at 800°C, the steady-stage hot corrosion kinetics is linear, which suggests that the rate-limiting step is not diffusion. Maps of element distribution and line scan along the white line as shown in Fig. 3a-c are in good agreement. For the sake of brevity, only the line scan results are shown in Fig. 4. From those results, it is clear that an obvious sulfur-rich area is present at the substrate/scale interface. For the specimens hot corroded at 900 and 1000°C, sulfur-rich areas are also observed at the substrate/scale interfaces. The presence of sulfur-rich bands at the substrate/scale interfaces demonstrates that chemical reaction of with occurred during hot corrosion. Another feature of the sulfur-rich bands is that the sulfur-containing species can be soluble in water because clear traces of dissolution are obvious at the substrate/scale interfaces where gouges are present. Figure 5 shows the high-magnification micrograph at the substrate/scale interface as indicated by the white arrow in Fig. 3c. Dot analysis results of elemental composition at the interfaces are summarized in Table II. It is seen from Table II that the sulfur content decreases with increasing temperature. The higher sulfur content at the substrate/scale interface of the sample hot corroded at 800°C suggests that the hot corrosion was still in an active stage.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In the previous works on the oxidation of in air, an inner continuous layer was formed at high temperatures and the amount of increased with temperature, enabling excellent oxidation resistance.³ Whereas, for the hot corrosion of the same material at all temperatures investigated in the present study, no continuous layers were observed even at the highest temperature of 1000°C. The difference in the latter case from the former is that there is a layer of coating on during exposure to hot air. Therefore, the effect of should be discussed.

Until now several hot corrosion mechanisms including sulfidation,²⁷ acid-basic dissolution,²⁸ and electrochemistry²⁹ have been proposed. No matter what the mechanisms are, the character of needs to be understood first. is an electrolyte which follows the following equilibrium⁶

Any reaction that is favorable for the reaction moving to the right would result in the dissociation of

Because fused is an ionic conductor, the hot corrosion mechanism should generally be electrochemistry. In other words, hot corrosion itself is an electrochemical process that includes anodic oxidation, cathodic reduction, and ion diffusion. In the hot corrosion of anodic oxidation mainly consists of the anodic dissolution of Ti and Al

Accompanying the anodic dissolution, strong reductivity results in the reduction of Some anions are formed and combine with cations of Ti and Al to form oxides, and the others increase the basicity in the fused anions and their reduction product could migrate inward through nonprotective scale, e.g., in the scale/ceramic interface where the reductivity is high enough for the formation of sulfides. In this study, sulfur-rich bands are present in the scale/ceramic interfaces (Fig. 3 and 4), implying that the reductivity is much higher there.

The cathodic reaction process consists of the reduction of anions and oxygen. The weight gain during hot corrosion comes from the oxygen in air. In fused sulfate there is an equilibrium

Oxygen in atmosphere can be dissolved in melt according to the following equation

Apart from oxygen in molecular form in the melt, its main presence form is Because the solubility of oxygen in fused salt is small²¹ and its diffusion rate slow,²² diffusion of molecular oxygen in the fused salt to the ceramic surface cannot account for such high weight gain rate. The concentration of ions according to Eq. 8 increases with increasing temperature, and then the concentration of increases, which consequently results in the increase in the corrosion rate of alloys. migrated to the surface of the scale and was reduced, which contributed to the formation of new oxides. Because fused salt is a strong electrolyte, the diffusion of ion migrating to the surface of the scale is sufficient to support such high weight gain due to corrosion.

As anions were reduced, basicity increased in the fused salt and the basic dissolution of oxides, especially occurred^{30 31}

After basic dissolution of occurs, the formed ions migrate to the place where the basicity is low. In the place where can stably exist, it precipitated again. In the specimens hot corroded at 900°C, the precipitated existed in the middle of the scale; while in the sample hot corroded at 1000°C it was present in the nearly outer layer. The fact that the Al-rich layer moved outward at 1000°C implies that the basicity near the ceramic/scale interface increases with increasing temperature. Due to the dissolution of and reprecipitation, the compactness of scales decreased, which is favorable for the diffusion of ions.

Because the scale formed at 900°C is relatively thin and the compactness is low, the kinetics in steady stage is linear. It is thus reasonable to conclude that the rate-limiting step at this temperature is the electrochemical reaction between the fused salt and rather than the diffusion of ions. At 1000°C, the precipitated layer is dense, and the kinetics in steady stage is parabolic, which suggests that the rate-limiting step at this temperature is the diffusion of ions.

At low temperature of 800°C, although this temperature is below the melting point of the morphology of the hot-corroded specimen is rough, implying that fluxing took place during hot corrosion. Given the fact that the kinetics in steady-stage corrosion is linear, it is reasonable to conclude that the rate-limiting step is the electrochemical reaction of substrate with the salt.