Corrosion of Fe-Cr-Si Alloys in Oxidizing and Sulphidizing-Oxidizing Atmospheres

: To clarify the mechanism of the third-element effect in sulphur-containing and sulphur-free oxidation environments, the corrosion behaviours of four kinds of Fe-xCr-ySi (x = 5, 10 at.% and y = 5, 10 at.%) alloys were studied at 600 ◦ C in a H 2 -CO 2 and a H 2 -CO 2 -H 2 S gaseous mixture with the same oxygen partial pressure. The results showed that, in the pure oxidizing atmosphere, thin and slowly growing protective oxide layers were formed on the alloys surfaces. Conversely, all alloys formed a corrosion product layer with an outer layer of FeS and an inner layer of a mix of oxides and sulphides in the oxidizing-sulphidizing atmosphere, which meant that adding Cr into the alloy as the third element had less of an effect on improving the alloy in the harsh sulphidizing-oxidizing environment. The oxidation and sulphidation mechanism as well as the effects of chromium and silicon on the corrosion resistance of the alloys was discussed.


Introduction
High temperature sulphurization corrosion often occurs to sectors in the process of oil refining, crude oil distillation and coal gasification [1][2][3]. In oxidizing-sulphidizing mixtures with low-oxygen and high-sulphur activities, the service life of most commercial alloys is not satisfactory, as they tend to form sulphide or mixed oxide-sulphide scales with large growth rates [4][5][6][7][8][9]. Recently, a large effort has been devoted to developing alloys with acceptable corrosion resistances in these aggressive atmospheres [10][11][12].
Alloying is an important means to improving the corrosion resistance of steel [13]. Among numerous alloy elements, silicon is widely studied because it helps to form a protective layer on the surface of steel [14][15][16][17]. Wang et al. [18] found that the addition of Si improved the corrosion resistance of a Fe-25Cr alloy in H 2 -H 2 O-H 2 S. Unfortunately, it was still difficult for the Fe-Si alloy to form a protective SiO 2 layer in an oxidizingsulphidizing atmosphere, even if the silicon content was as high as 13 at.% [14,19]. In order to promote the selective oxidation of silicon, a third element with activity between Fe and Si can be added to the Fe-Si binary alloy. As a result, the Si content required for the formation of SiO 2 can be significantly reduced. It has been proven that the addition of Cr can boost the oxidation resistance of Fe-Si alloys, as continuous and protective scales can be formed [20][21][22]. Such a beneficial effect of Cr has been illustrated on the basis of the theory of the third-element effect (TEE) proposed by Wagner [23] and expanded by Gesmundo et al. [24]. In their theory, elemental Cr in the alloy acted as a oxygen absorber to prevent the internal oxidation of silicon.
It is worth noting that the TEE is mainly applied to the degradation of alloys in atmospheres containing only one oxidant, especially oxygen [25][26][27]. So far, however, there has been little discussion about the mechanism of the TEE in a vulcanization-oxidation mixed atmosphere. Whether Cr can promote the selective oxidation of Si in a special sulphur-containing high temperature environment so as to form a protective oxide layer on the substrate is still a problem. This paper sought to remedy these problems of insufficient data by analysing the oxidation-sulphidation behaviour of Fe-Cr-Si alloys with a proper chromium and silicon concentration [14] in a H 2 -CO 2 -H 2 S mixture (referred to as O/S) at 600 • C. For comparison, the same alloys were also oxidized in the same temperature in H 2 -CO 2 mixtures (O) at the same oxygen partial pressure.

Experimental Procedures
The vacuum induction melting method was used to prepare four kinds of Fe-Cr-Si alloys (named Fe-5Cr-5Si, Fe10Cr-5-Si, Fe-5Cr-10Si and Fe-10Cr-10Si, respectively). The raw materials of the samples (Fe, Cr and Si) were purchased from Sinopharm Chemical Reagent, Shanghai, China. The main melting process parameters were as follows: melting vacuum of 0.2 Pa, melting temperature of 1350~1400 • C, refining temperature of 1500 • C and refining time of 10 min. After proper annealing (1000 • C, 48 h, 0.2 Pa), the alloy ingots were equilibrated, and the residual mechanical stresses were removed. The actual composition of the alloy was measured by using an ICP spectrometer (ICP 7000, GLMY, Wuxi, China, test conditions: plasma power 1400 w; pump speed 30 rpm; cooling gas flow 13.00 L/min; auxiliary gas flow 1.50 L/min; atomizer flow 0.80 L/min and additional air flow 0.2 L/min).
An electric spark cutting machine (DK7735, Weihai CNC, Taizhou, China) was used to prepare 2 cm × 1 cm × 1 cm samples, which were successively abraded down to a 2000-grit SiC grit and cleaned. In order to obtain the metallograph of the alloy, the polished sample was etched in 4% nitric acid-alcohol solution at room temperature for 20 s.
Considering the hot corrosion and mechanical properties, the service temperature of steel without nickel is about 600 • C. The oxidizing experiments were carried out in a microbalance B-92, SETARAM (Lyon, France) by inflating an H 2 -CO 2 gas mixture at 600 • C for 24 h. For thermodynamic equilibrium, the gas mixtures were passed through a porous platinum catalyst before contacting the samples. The discontinuous oxidation-sulphidation experiments were implemented in a three-section horizontal furnace (Model number: STF553456C, Lindberg/blue M, Waltham, MA, USA) at 600 • C for 24 h. The samples were suspended in a quartz tube and exposed to the gaseous mixture of H 2 -CO 2 -H 2 S. The specimens were weighed periodically after they were removed from the reaction tube. A new sample was used for each oxidation-sulphidation test in order to avoid errors resulting from partial scale spalling during cooling. The weight of the samples was weighed by an electronic balance (BP211D, Sartorius, Gottingen, Germany, with precision of 0.01 mg). The reproducibility of both kinds of experiments was confirmed by testing at least three times for the same corrosion test. Table 1   After corrosion, scanning electron microscopy (SEM-FEI INSPECT F50, FEI, Hillsboro, OR, USA, test conditions were set as: Acc.V: 15.0 KV; beam size: 1 µm; B.C.: 20 nA) and an energy dispersive X-ray microanalysis attachment to the SEM (EDAX-OXFORD X-Max, Oxford Instruments, Oxford, UK) were utilized to identify the scale morphologies and compositions as well as the distribution of the elements. The oxidized samples were also observed with optical microscopy (OM-LSM 510, Carl Zeiss, Oberkochen, Germany). As a means of identifying the corrosion products within the oxidation-sulphidation scale, an XRD (D8, Bruker, Billerica, MA, USA, Cu Ka radiation at 40 kV, scanning speed: 4 • /min) test was carried out by alternately grinding off some of the corrosion products and performing XRD measurement on the ground surface. All the phase diagrams were calculated using the software of HSC Chemistry 6.0 (Outotec, Helsinki, Finland).

Prepared Sample
The actual and nominal compositions of the alloys are shown in Table 2. In agreement with the Fe-rich corner of the Fe-Cr-Si phase diagram ( Figure 1) [28], the Fe-5Cr-xSi alloys were composed of single α-Fe phase (Figure 2a), while the Fe-10Cr-xSi alloys were composed of α-Fe matrixes and dispersed intermetallic compound particles of Cr 9 Fe 9 Si 2 ( Figure 2b). After corrosion, scanning electron microscopy (SEM-FEI INSPECT F50, FEI, Hillsboro, OR, USA, test conditions were set as: Acc.V: 15.0 KV; beam size: 1 μm; B.C.: 20 nA) and an energy dispersive X-ray microanalysis attachment to the SEM (EDAX-OXFORD X-Max, Oxford Instruments, Oxford, UK) were utilized to identify the scale morphologies and compositions as well as the distribution of the elements. The oxidized samples were also observed with optical microscopy (OM-LSM 510, Carl Zeiss, Oberkochen, Germany). As a means of identifying the corrosion products within the oxidation-sulphidation scale, an XRD (D8, Bruker, Billerica, MA, USA, Cu Ka radiation at 40 kV, scanning speed: 4°/min) test was carried out by alternately grinding off some of the corrosion products and performing XRD measurement on the ground surface. All the phase diagrams were calculated using the software of HSC Chemistry 6.0 (Outotec, Helsinki, Finland).

Prepared Sample
The actual and nominal compositions of the alloys are shown in Table 2. In agreement with the Fe-rich corner of the Fe-Cr-Si phase diagram ( Figure 1) [28], the Fe-5Cr-xSi alloys were composed of single α-Fe phase (Figure 2a

Corrosion Kinetics
The corrosion kinetics of the four Fe-Cr-Si alloys in the oxidizing and oxidizingsulphidizing atmospheres are shown in Figure 3. After corrosion for 24 h in the oxidizing atmosphere, all the alloys except for Fe-5Cr-5Si showed small mass gains (less than 0.03 mg/cm 2 ) and tended to remain unchanged at the end of the experiment. The parabolic rate constant (k p ) of the samples were 2.78 × 10 −5 (Fe-5Cr-10Si), 5.98 × 10 −6 (Fe-10Cr-5Si) and 1.82 × 10 −5 (Fe-10Cr-10Si) mg 2 /(cm 4 ·h), respectively. The Fe-5Cr-5Si alloy showed the greatest weight gain among the four alloys, which was about 0.07 mg/cm 2 . Using linearity to fit the oxidation kinetics curve of Fe-5Cr-5Si, the oxidation rate was 0.00258 mg/(cm 2 ·h). All the alloys gained more weight in the oxidizing-sulphidizing atmosphere than in the pure oxidizing atmosphere. The largest mass gain was observed in Fe-5Cr-5Si, which was about 33 mg/cm 2 , while the smallest mass gain was observed in Fe-10Cr-10Si, which was about 10 mg/cm 2 . As a whole, Fe-xCr-5Si exhibited greater weight gains than Fe-xCr-10Si.

Corrosion Kinetics
The corrosion kinetics of the four Fe-Cr-Si alloys in the oxidizing and oxidizing-sulphidizing atmospheres are shown in Figure 3. After corrosion for 24 h in the oxidizing atmosphere, all the alloys except for Fe-5Cr-5Si showed small mass gains (less than 0.03 mg/cm 2 ) and tended to remain unchanged at the end of the experiment. The parabolic rate constant (kp) of the samples were 2.78 × 10 −5 (Fe-5Cr-10Si), 5.98 × 10 −6 (Fe-10Cr-5Si) and 1.82 × 10 −5 (Fe-10Cr-10Si) mg 2 /(cm 4 ·h), respectively. The Fe-5Cr-5Si alloy showed the greatest weight gain among the four alloys, which was about 0.07 mg/cm 2 . Using linearity to fit the oxidation kinetics curve of Fe-5Cr-5Si, the oxidation rate was 0.00258 mg/(cm 2 ·h). All the alloys gained more weight in the oxidizing-sulphidizing atmosphere than in the pure oxidizing atmosphere. The largest mass gain was observed in Fe-5Cr-5Si, which was about 33 mg/cm 2 , while the smallest mass gain was observed in Fe-10Cr-10Si, which was about 10 mg/cm 2 . As a whole, Fe-xCr-5Si exhibited greater weight gains than Fe-xCr-10Si.      The scales that formed on the Fe-10Cr-5Si alloy (Figure 5a,c) were protective, and the surfaces of alloys were smooth and continuous. In addition, the grain boundaries of the alloy could be clearly observed under an optical microscope (Figure 5b). According to the XRD spectrum (Figure 5d), the scales were composed of Cr2O3. Both Fe-5Cr-10Si and Fe-10Cr-10Si formed similar scales. The formation of SiO2 on the four alloys could be detected by XRD. However, SiO2 might have formed during this experiment, according to the research of Bamba [20].   (Figure 6b) showed that its element distribution was not uniform. Further, the element depth profiles for O, Si, S and Cr in the internal layer ( Figure 6c) showed that, contrary to oxygen, the general trend of the S content was decreasing towards the surface. In the internal layer, there were also silicon and chromium oxide particles. The XRD patterns of the whole scales are shown in Figure  6d. Pure FeS made up the outermost layer of the multilayer structure. The outer part of the internal layer was composed of pure FeS, and FeCr2S4 was formed in the inner part of the internal layer. The scales that formed on the Fe-10Cr-5Si alloy (Figure 5a,c) were protective, and the surfaces of alloys were smooth and continuous. In addition, the grain boundaries of the alloy could be clearly observed under an optical microscope (Figure 5b). According to the XRD spectrum (Figure 5d), the scales were composed of Cr 2 O 3 . Both Fe-5Cr-10Si and Fe-10Cr-10Si formed similar scales. The formation of SiO 2 on the four alloys could be detected by XRD. However, SiO 2 might have formed during this experiment, according to the research of Bamba [20].

Microstructures of the Scales on Fe-Cr-Si Alloys Corroded in the Oxidizing-Sulphidizing Atmosphere
The cross-sectional structures, element depth profiles and XRD spectra of the scales formed on the four Fe-Cr-Si alloys corroded in the H 2 -CO 2 -H 2 S mixture for 24 h are shown in Figures 6-9. Multilayered outer scales and a mixed compound inner layer formed on the Fe-5Cr-5Si alloy (Figure 6a). Both layers had fissures parallel to the original alloy surface. In addition, the interfaces of the internal layer and base alloy were uneven. The high-resolution SEM photo (A + B model) of the internal layer (Figure 6b) showed that its element distribution was not uniform. Further, the element depth profiles for O, Si, S and Cr in the internal layer ( Figure 6c) showed that, contrary to oxygen, the general trend of the S content was decreasing towards the surface. In the internal layer, there were also In the same way as the scales of Fe-5Cr-5Si, scales formed on the surface of Fe-10Cr-5Si that had an external layer with a multilayer structure and an internal mixed layer (Figure 7a,b). However, the interface between the scales and the base alloy was flat. There were similar trends in the element profiles of O, Si, S and Cr in the internal layer, but the concentration gradients of S and O were larger than those of Fe-5Cr-5Si. The XRD ( Figure  7d) pattern of the scales showed that the external layer was composed of pure FeS. the outer part of the internal layer was composed of FeS and FeCr2S4 and the inner part of the internal layer consisted of FeS and FeCr2O4. In the same way as the scales of Fe-5Cr-5Si, scales formed on the surface of Fe-10Cr-5Si that had an external layer with a multilayer structure and an internal mixed layer (Figure 7a,b). However, the interface between the scales and the base alloy was flat. There were similar trends in the element profiles of O, Si, S and Cr in the internal layer, but the concentration gradients of S and O were larger than those of Fe-5Cr-5Si. The XRD (Figure 7d  The scales on Fe-5Cr-10Si were also divided into two sub-layers, where one was a thick external multi-layered FeS layer, and the other one was an internal mixed layer (Figure 8a). As can be seen from the enlarged image of the internal layer (Figure 8b), the alloy particles were distributed throughout the layer. In addition, there was a layer rich in Si that formed under the scale, and the continuity of the layer was destroyed by corrosion products. The XRD spectra (Figure 8d) showed that the corrosion products in the internal layer were FeS, FeCr2S4, Fe2SiO4 and Fe3Si, and the layer under the internal layer was composed of Fe3Si. The scales on Fe-5Cr-10Si were also divided into two sub-layers, where one was a thick external multi-layered FeS layer, and the other one was an internal mixed layer ( Figure 8a). As can be seen from the enlarged image of the internal layer (Figure 8b), the alloy particles were distributed throughout the layer. In addition, there was a layer rich in Si that formed under the scale, and the continuity of the layer was destroyed by corrosion products. The XRD spectra (Figure 8d) showed that the corrosion products in the internal layer were FeS, FeCr 2 S 4 , Fe 2 SiO 4 and Fe 3 Si, and the layer under the internal layer was composed of Fe 3 Si. The corrosion of Fe-10Cr-10Si produced an external layer of sulphides and a thin internal layer rich in oxides followed by a discontinuous intermetallic compound layer of Fe3Si (Figure 9a). Furthermore, the external layer was divided into two sub-layers ( Figure  9b). According to the XRD pattern of the external layer, the corrosion products of the external layer were FeS and FeCr2S4 (Figure 9c). The chemical composition (at.%) of the different parts of the scales analysed by EDS are shown in Table 3. The corrosion of Fe-10Cr-10Si produced an external layer of sulphides and a thin internal layer rich in oxides followed by a discontinuous intermetallic compound layer of Fe 3 Si (Figure 9a). Furthermore, the external layer was divided into two sub-layers (Figure 9b). According to the XRD pattern of the external layer, the corrosion products of the external layer were FeS and FeCr 2 S 4 (Figure 9c). The chemical composition (at.%) of the different parts of the scales analysed by EDS are shown in Table 3.

Discussion
The

Discussion
The  Figure 10b. In addition, the formation lines of the spinels are also shown in Figure 10b. Furthermore, the partial pressures of the oxygen and sulphur used in the present work are marked by black squares in Figure 10. also shown in Figure 10b. Furthermore, the partial pressures of the oxygen and sulphur used in the present work are marked by black squares in Figure 10. The oxygen pressure of the oxidizing atmosphere fell inside the fields of stability of Fe3O4, Fe2SiO4 and FeCr2O4, while the oxygen and sulphur pressures of the oxidizing-sulphidizing atmosphere lay in the fields of stability of FeS, SiO2 and Cr2O3. However, the oxygen pressure in the oxidizing-sulphidizing mixture was still sufficient enough to form Fe3O4. As a result, depending on the shape of the diffusion path of oxygen, FeO and Fe3O4 still formed if the alloy was covered by an external FeS layer. After crossing the external FeS layer, the actual diffusion path ran along the lines for the FeS-FeCr2S4 and/or the FeS-FeCr2O4 equilibrium, since FeS was generally observed in the mixture with spinels in the internal layer of the scales.

Oxidation Mechanism in the H 2 -CO 2 Mixture
The XRD and EDX analyses of the H 2 -CO 2 mixture revealed that the oxidation products of the four kinds of alloys were Fe 3 O 4 , Cr 2 O 3 and FeCr 2 O 4 . The oxidation products ware consistent with the thermodynamic predictions. The spinel crystal (FeCr 2 O 4 ) was formed according to the reaction of Cr 2 O 3 with iron and oxygen [29], the reaction of which could be formulated as: 2Fe Except for Fe-5Cr-5Si, which formed nodules along the grain boundaries, all the alloys formed continuous and protective scales. The nodules were composed of non-protective oxides, such as Fe 3 O 4 . This difference between the scale structures formed on the four kinds of alloys could be illustrated by the oxidation theory which is about the transition between different oxidation models of ternary alloys under a high oxygen pressure [30]. In addition, the oxidation behaviours of Fe-Cr-Si alloys with various contents of chromium and silicon oxidized in 1atm and 10 −20 atm O 2 at 700 • C have been studied. Furthermore, the critical concentrations of chromium and silicon required for different oxidation models were calculated by Guo [21]. In this study, it was found that, when oxidized in 10 −20 atm O 2 at 600 • C, the contents of Cr and Si in the Fe-5Cr-5Si alloy lay in the field that formed the internal oxidation zone of Cr + Si beneath the external FeO x scales but was also quite close to the formation of external mixtures of SiO 2 and Cr 2 O 3 . The scale structure of Fe-5Cr-5Si oxidized at 600 • C was similar to that oxidized at 700 • C. Accordingly, the Fe-5Cr-5Si alloy's composition fell in the same position as Fe-Cr-Si alloys oxidized at 600 • C under 10 −24 atm O 2 . Therefore, the addition of another 5 at.% of Cr and/or Si to the Fe-5Cr-5Si alloy moved its location in the kinetic diagram from the non-protective zone to the protective zone.

Corrosion Mechanism in the H 2 -CO 2 -H 2 S Mixture
The corrosion products, i.e., FeS, SiO 2 , FeCr 2 S 4 and FeCr 2 O 4 , formed in the O/S atmosphere were consistent with the prediction based on the thermodynamic analysis. In addition, sulphides of chromium and silicon were not formed, mostly because the gasphase oxygen pressure adopted in this work was many orders of magnitude larger than the values required for the SiS-SiO 2 and CrS-Cr 2 O 3 equilibriums, as show in Figure 10.
Several Fe-based binary alloys such as Fe-Y, Fe-Si and Fe-Ce [11,17,31] will form an external layer of FeS after being corroded in an oxidation-sulphidation atmosphere. Such layers are generally divided into various sub-layers, which have been shown to be poorly adherent to the inner scale region [32]. The formation mechanism of FeS external layers has been discussed previously [20]. This theory also applied to the formation of the FeS external layers on the Fe-Cr-Si alloys in this experiment. The formation of oxides (especially silicon oxide and chromium oxide) in the inner layer suggests that oxygen-bearing molecules were able to penetrate through the outer multi-layer of FeS. They may have spread through discontinuities in the external FeS scales such as in pores or micro cracks [33].
In O/S atmospheres, the corrosion kinetics curve of Fe-Cr-Si alloys gradually changes from linear to parabolic as Cr and Si contents increase. As a consequence, adding chromium and silicon to Fe can slow its corrosion rate. As well as the FeCr 2 S 4 layer on Fe-10Cr-10Si, the scale rich in oxides of chromium and silicon could explain this beneficial effect. However, the network of FeS in the inner layer also provides a pathway for iron diffusion, which cannot reduce the growth rate of the external FeS layer [34].
There were intermetallic particles of Fe 3 Si formed within the internal layer on Fe-5Cr-10Si and even a discontinuous layer of Fe 3 Si formed in contact with the base alloy on both the Fe-5Cr-10Si and Fe-10Cr-10Si alloys. It has been reported that, in Fe-Si alloys, the diffusion rate of Fe increases with the increase in the content of Si [30]. Based on the formation of the Fe 3 Si layer and FeCr 2 S 4 layer on Fe-xCr-10Si, it is reasonable to conclude that the increase in silicon content also accumulates the outward diffusion of both Fe and Cr in Fe-Cr-Si alloy.
In summary, the increase in the content of chromium and silicon in Fe-Cr-Si alloy reduces its corrosion rate in the oxidizing-sulphidizing atmosphere with respect to Fe-Si alloys [21]. This result appears to be connected with the formation of increasing amounts of chromium and silicon compounds such as FeCr 2 O 4 , Cr 2 O 3 and SiO 2, which restrict the outward diffusion of iron through the internal mixed layer of the scales [30].

The Effect of Sulphur
According to the thermodynamic phase diagram of Fe-O-S shown in Figure 10b, FeS was the only stable iron compound in contact with the present oxidizing-sulphidizing atmosphere. It is reasonable to predict that an FeS layer will be the external layer of the scale. As a matter of fact, for these systems, water or CO 2 molecules were apparently able to penetrate through the external sulphide region, forming oxides of iron, chromium and silicon and spinels in the inner scale region, generally mixed with some FeS. Based on the structure of the scale, the diffusion path inside the Fe-O-S phase diagram ( Figure 10) shifted from the lower left to the equilibrium line between FeCr 2 S 4 and FeS and then to the equilibrium line between FeCr 2 O 4 and FeS before reaching the metal field.
Compared to the oxidizing and oxidizing-sulphidizing atmospheres, the Fe-Cr-Si alloys exhibited different corrosion kinetics. There was a dramatic rise in the corrosion rates for all the alloys when there was a large sulphur partial pressure in the gas, resulting in overall mass changes increased by three to five orders of magnitude after 24 h corrosion. As a result, both the external and mixed layers had large quantities of FeS. Compared to iron oxides, FeS had a much higher defect concentration [33], which provided a fast diffusion path for ions. According to Mrowec [35] and Meier [36], under the same temperature, the pure sulphidation of iron is much faster than its pure oxidation due to the higher diffusion coefficient of Fe in FeS. Despite Cr and Si oxides forming in the internal mixed layer, the FeS still had a network structure, providing a fast diffusion path, leading to a fast mass gain rate. However, the presence of some Cr and Si oxides in the inner scale region was still responsible, as noted above, for the limited reduction in the corrosion rates of the alloys with respect to the behaviour of pure iron in the same environment.

The Effect of Chromium
It has been reported that Fe-Si alloys fail to form a protective layer at 600 • C, even when the silicon content is up to 13 at.% in both the oxidizing atmosphere and the oxidizingsulphidizing atmosphere [14], while the addition of 5 at.% Cr to Fe-5Si alloy induces the formation of an almost continuous protective scale in the oxidizing atmosphere. The beneficial effect of 5 at.% Cr can be interpreted by the theory of the TEE proposed by Wagner [22]. Chromate oxide scales reduce the oxygen pressure at the alloy surface, which leads to silicon oxide formation [21]. Nevertheless, in an oxidizing-sulphidizing atmosphere, the addition of chromium between 5 and 10 at.% to Fe-5Si and Fe-10Si was not sufficient to prevent FeS formation, as shown in Figures 6-9. Due to the fast growth of FeS, which has a high concentration of defects, ions were able to diffuse quickly inward and outward during corrosion. Thus, even in Fe-10Cr-10Si, the chromium could not be oxidized but reacted with FeS according to the following reaction [37]: 2FeS + 4Cr + 3S 2 = 2FeCr 2 S 4 (2) This reaction led to the formation of an FeCr 2 S 4 layer underneath the outer FeS layer. Furthermore, the chromium could not promote the selective oxidation of silicon. It had little beneficial effect on improving its oxidation-sulphide resistance when chromium was added to the Fe-Si alloys at 5 at.% to 10 at.%.

Conclusions
Based on the above results and discussion, the following conclusions can be drawn:

1.
Cr promoted the formation of a chromium-rich oxide layer on the Fe-Si alloys in a H 2 -CO 2 atmosphere, thus significantly improving the corrosion resistance of the alloys, especially when the content was more than 5 at.%.

2.
Introducing H 2 S gas into the H 2 -CO 2 atmosphere made the Fe-Cr-Si alloys form loose and easy-to-peel FeS outer layers and FeCr 2 S 4 inner layers instead of a dense oxide layer, thus significantly intensifying the corrosion of the alloy.

3.
Attributable to the failure of Cr in promoting the selective oxidation of Si, the introduction of Cr to the Fe-Si alloys had little beneficial effect in improving the corrosion resistance of iron-based alloys in a H 2 -CO 2 -H 2 S atmosphere.