*2.5. Characterization*

The morphology of the ceramic coating and carbon steel substrate was characterized by scanning electron microscopy (SEM; JSM-6700F, JEOL, Tokyo, Japan) equipped with an energy dispersive X-ray spectroscopy (EDS, NORAN, Thermo Fisher, Waltham, MA, USA). The phase-transition occurred at different temperature was characterized by X-ray diffraction (XRD; X'Pert Pro, Philips, Amsterdam, The Netherlands) operated from 5◦ to 90◦ for 5.25 min. The weight change and thermal change during heating process were detected by TG-DTA (TG-DTA; STA449, Netzsch, Nuremberg, Germany). The TG-DTA test was operated from room temperature to 1250 ◦C at a heating rate of 10 ◦C/min. The test was conducted under flowing air.

#### **3. Results and Discussion**

#### *3.1. Performance of Coating*

The anti-oxidation performances of the coatings with different proportion of SiO2@Al were investigated by the weight loss during the heating process in muffle furnace. As shown in Figure 3, when temperatures ranged from 1050 to 1150 ◦C, a higher proportion of SiO2@Al led to a better anti-oxidation performance. However, when the proportion exceeded 10%, the enhancement of the anti-oxidation performance was not obvious. When temperatures ranged from 1200 to 1250 ◦C, the anti-oxidation performance of coating improved with the increasing amount of SiO2@Al until 10% and reduced with further SiO2@Al addition, especially at 1250 ◦C. Therefore, in view of the anti-oxidation performance and the cost, the ceramic coating with 10% of SiO2@Al was chosen as the appropriate one to conduct the following investigations.

**Figure 3.** Anti-oxidation performances of the coatings with different proportion of SiO2@Al.

The non-isothermal kinetic for sample protected by ASMA, sample protected by AS, and bare sample are shown in Figure 4. The apparent activation energy, (*E*a), was calculated by the Arrhenius equation. As shown in Equation (6), if the concentration of reactant (*B*) was constant, the reaction rate (*v*) would have a positive correlation with the reaction rate constants (k), which is the function of temperature described by Equation (7) (Arrhenius equation).

$$v = B \times \mathbf{k} \tag{6}$$

$$\mathbf{k} = \mathbf{A} \times \mathbf{e}^{(E\_\mathbf{a}/\mathcal{R}T)} \tag{7}$$

The relationship between reaction rate (*v*) and apparent activation energy (*E*a) could be deduced with Equations (6) and (7) by a logarithmic operation. The ultimate equation was described as Equations (8) and (9).

$$\text{lnk} = -\frac{E\_d}{\text{RT}} + \text{lnA} \tag{8}$$

$$
\ln \upsilon = -\frac{E\_a}{RT} + \ln \text{A} + \ln \text{B} \tag{9}
$$

where, A and B were constant. R was 8.314 J·mol−1·K−1.

Therefore, the dependent variable (ln*v*) had a linear relationship with the variable (1/*T*). As described in Equations (3) and (4), the reaction rate (*v*) could be calculated by the weight loss per unit area and the duration time of samples heated to a certain temperature. Based on the kinetic results at different temperatures as shown in Figure 4, a linear relationship was observed between ln*v* and 1/*T*. The slope of each fitting was equal to (−*E*a/R) as described in Equation (8). It was a remarkable fact that the result of the sample protected by AS was not able to be fitted by a single linear equation because there was an inflection point at about 1150 ◦C. Therefore, the kinetic of the sample protected by AS should be divided into two parts (AS1 and AS2) by 1150 ◦C.

As shown in Figure 4, the apparent activation energy (*E*a) of bare sample was 120.11 kJ/mol, while that of the sample protected by ASMA was 326.50 kJ/mol. The apparent activation energy (*E*a) of the sample protected by AS between 1025 and 1150 ◦C was 106.14 kJ/mol, which was approximate to the bare sample. The apparent activation energy (*E*a) of the sample protected by AS between 1150 and 1250 ◦C was 224.05 kJ/mol, which was between that of the sample protected by ASMA and the bare sample. Generally, higher apparent activation energy resulted in a better anti-oxidation performance of coating. Therefore, ASMA exhibited an excellent anti-oxidation performance between 1025 and 1250 ◦C, while AS only worked above 1150 ◦C. Additionally, the anti-oxidation performance of ASMA was much better than AS due to its larger apparent activation energy (*E*a). To further prove the enhancement of anti-oxidation of ASMA, the non-isothermal experiment was conducted by a continuous thermo-balance. As shown in Figure 5, compared with AS, ASMA showed an obviously better anti-oxidation performance during the temperature-rise period.

**Figure 4.** Results of non-isothermal kinetics.

**Figure 5.** Results of continuous thermo-balance from 200 to 1250 ◦C with a rate of 10 ◦C/min.

Since the serious oxidation of carbon steel always occurred over 1250 ◦C, the isothermal kinetic was conducted at 1250 ◦C. It was widely accepted that the oxidation of carbon steel followed the parabolic law [3,23,24], which meant that the oxidation process was controlled by the diffusion. The reaction constant could be calculated by Equation (10) [2].

$$\left(\Delta\omega\right)^{2} = \mathrm{H} + \mathrm{k}\_{\mathrm{P}} \times t\_{i} \tag{10}$$

Here, *ti* (s) is the duration of oxidation, Δω is the weight gain per unit area (mg/cm2), H is a constant, and kp (mg<sup>2</sup>·cm<sup>−</sup>4·s<sup>−</sup>1) is the reaction rate constant.

In this study, the linear relationship between Δω and *ti*1/2 was used to evaluate kp [2]. As shown in Figure 6, the reaction rate constant (kp) of the bare sample, the sample protected by AS, and the sample protected by ASMA was 2.09, 0.096, and 0.046 mg<sup>2</sup>·cm<sup>−</sup>4·s<sup>−</sup>1, respectively. It can be concluded that ASMA possessed a better anti-oxidation performance than AS due to its smaller reaction rate constant at 1250 ◦C.

**Figure 6.** Results of isothermal kinetics.

#### *3.2. Morphology of Coating*

The microstructure was an intuitive evidence for evaluating the compactness of coating which controlled the inner diffusion rate of oxygen. Therefore, anti-oxidation performance of coating was better with a more compact structure. The outer-surface microstructures of sample protected by the ASMA and sample protected by the AS were investigated after the thermal treatments of different temperature (900, 1150, and 1250 ◦C). As described in Figure 7, ASMA formed a compact structure at 900 ◦C and the compactness improved with the increasing of temperature until 1250 ◦C. The amount of liquid phase increased with the rising temperature, the porosity reduced and thus the sintering of coating was enhanced at the same time. However, the sample protected by AS possessed a porous structure composed of isolated particles at 900 ◦C, while the compactness improved at 1150 ◦C. The above results suggested that SiO2@Al could accelerate the sintering of the ceramic coating and enhance the compactness.

**Figure 7.** SEM images of outer-surfaces belonging to sample protected by ASMA and sample protected by AS at different temperatures.

#### *3.3. Protection Mechanism*

Figure 8 showed the TG-DTA data of the ASMA and AS slurry. As shown in Figure 8a, there was an endothermic peak at about 662.85 ◦C, an exothermic peak at 977.7 ◦C, and an exothermic peak at 1153.91 ◦C for ASMA slurry, while there was only a gradual exothermic trend during heating process for AS slurry. In order to interpret the possible reactions during the heating process, the phase transitions were investigated by XRD method at 500, 800, 1100, and 1250 ◦C, which were near the endothermic or exothermic temperature. Compared with Figure 9a,b, it was found that no specific phase formed between 500 and 800 ◦C, thus the endothermic peak at 662.82 ◦C could be attributed to the melting of Al powders in SiO2@Al. Phases of Si and Al3.21Si0.47 were newly formed within ASMA when temperature increased from 800 to 1100 ◦C, which indicated that an exothermic reaction occurred as described in Equation (11) and it gave rise to the exothermic peak at 977.7 ◦C [25]. Compared with Figure 9c,d, it was remarkable that mullite and cristobalite formed within ASMA when temperature ranged from 1100 to 1250 ◦C. Therefore, the exothermic peak at 1153.91 ◦C might result from the exothermic reaction expressed as Equations (12) and (13) [26,27]. In contrast, there was no new phase formed within AS during the whole heating process except for cristobalite. To sum up, the Al powders in SiO2@Al melted at 662.85 ◦C and reacted with SiO2 on its surface, which generated local high temperature and the sintering of ceramic raw materials, thus contributing to its excellent anti-oxidation performance, which was consistent with the non-isothermal kinetics. Besides, mullite and cristobalite formed at 1150 ◦C could further enhance the coating compactness and improve the anti-oxidation performance above 1150 ◦C [19]. However, ceramic raw materials just sintered above 1150 ◦C within AS, and thus showed an unobvious anti-oxidation effect below this temperature.

$$4\text{Al} + 3\text{SiO}\_2 \rightarrow 3\text{Si} + 2\text{Al}\_2\text{O}\_3 \tag{11}$$

SiO2 (amorphous) → SiO2 (cristobalite) (12)

$$\text{3Al}\_2\text{O}\_3 + 2\text{SiO}\_2 \to \text{3Al}\_2\text{O}\_3 \cdot 2\text{SiO}\_2 \tag{13}$$

**Figure 8.** TG-DTA curves of (**a**) ASMA and (**b**) AS.

**Figure 9.** Comparisons of XRD result of the sample protected by ASMA and AS slurry at dffierent temperatures: (**a**) 500 ◦C, (**b**) 800 ◦C, (**c**) 1100 ◦C, (**d**) 1250 ◦C.

According to the SEM-EDS results of coating cross-sections exhibited in Figure 10, both the coatings were able to prevent the carbon steel substrate from oxidization at 1250 ◦C. Compared with Figure 11a,b, ASMA was composed of a hercynite layer, a mullite layer, and a ceramic layer with high aluminum content, while AS consisted of a Fe3O4 layer, an olivine layer composed of FeO and 2FeO·SiO2, a ceramic layer composed of Al2O3 and cristobalite. Equations (14) or (15) possibly occurred to form the hercynite (Al2O3·FeO) within ASMA. As reported in the literature [28], the structure of hercynite could be expressed as (Fe2+1−*<sup>x</sup>*Al3+*x*)tet(Al3+2−*<sup>x</sup>*Fe2+*x*)octO4 where 0 < *x* < 0.25, while that of Fe3O4 could be expressed as (Fe3+)tet(Fe3+Fe2+)octO4 [29]. Referring to Freer and O'Reilly [30], the diffusion activation energy of Fe2+ at octahedral site and tetrahedral site were 0.2(±0.05) eV and 0.71

(±0.09) eV, which indicated that the diffusion of iron ions within hercynite was slower than Fe3O4 due to the lower fraction of Fe2+ occupied the octahedral site. Therefore, hercynite also played a significant role for enhancing the anti-oxidation ability of ASMA. However, hercynite was easily oxidized as shown in Equation (16) under oxidation condition. As mentioned in Section 3.1, the compactness of ASMA was greatly improved with the addition of SiO2@Al, which effectively prohibited the inner diffusion of oxygen. As a result, the hercynite layer could be stable within the ceramic coating. The result agreed well with the finding in isothermal kinetic experiment.

$$3\text{Al}\_2\text{O}\_3 \cdot 2\text{SiO}\_2 + 3\text{FeO} \rightarrow 3\text{Al}\_2\text{O}\_3 \cdot \text{FeO} + 6\text{SiO}\_2\tag{14}$$

$$\text{Al}\_2\text{O}\_3 + \text{FeO} \rightarrow \text{Al}\_2\text{O}\_3 \cdot \text{FeO} \tag{15}$$

$$4\text{Al}\_2\text{O}\_3 \cdot \text{FeO} + \text{O}\_2 \rightarrow 4\text{Al}\_2\text{O}\_3 + 2\text{Fe}\_2\text{O}\_3\tag{16}$$

**Figure 10.** SEM-EDS images of cross-sections belonging to (**a**) the sample protected by ASMA and (**b**) the sample protected by AS at 1250 ◦C.

**Figure 11.** SEM images with results of EDS measurements performed on selected areas of cross-sections belonging to (**a**) sample protected by ASMA and (**b**) sample protected by AS at 1250 ◦C.
