Surface appearance

The lack of reproducibility seen in Δm/A measurements is related to the non-uniform appearance of the specimen surfaces. As is shown subsequently, the alloys formed oxide scales consisting of multiple layers. The outer layers, identified by XRD and local Raman analysis as iron oxides, did not develop uniform surface cov-erage. At the oxide/gas interface, an Fe2O3 layer (grey colour) and Fe2O3 whiskers (red colour) formed preferentially at the bottom of the specimens, that is, down-stream in the gas flow. Figure 3.2(a) shows a typical example. However, the surface coverage by Fe2O3 varied from specimen to specimen, and even from one face to the other of a given specimen. The fact that the surface morphology was influenced by

Figure 3.2 – Photographs showing specimen surface appearance after reaction at 650 °C. Fe–2.25Cr in Ar–20CO₂ for (a) 240 h and (b) 336 h; (c) Fe–9Cr–

10Ni, 120 h in Ar–20CO₂–5H₂O. Black arrows indicate direction of gas flow.

the direction of the gas flow was confirmed by running an experiment in an hori-zontal furnace (see Fig. 3.2(b)). In addition, partial spallation of the external oxide layers (Fig. 3.2(c)) occurred to highly variable extents, in terms of both surface and thickness (number of layers) of spalled oxide. Spallation was most important in the case of the 9Cr ternaries.

The surface oxides had similar microstructures after reaction of all alloys in all gases. Typical surface micrographs are shown in Fig. 3.3. Oxide phases were

identi-(a) Fe₃O₄ and Fe₂O₃ layers (b) Fe₂O₃ layer and whiskers (c) Fe₂O₃ whiskers Figure 3.3 – Optical micrographs showing microstructure of surface oxides grown on Fe–9Cr after 120 h exposure to Ar–20CO₂ at 650 °C.

fied individually by local Raman analysis. The Fe3O4layer consisted of large, faceted crystals (Fig. 3.3(a)). Hematite was present as a fine-grained layer (Fig. 3.3(b)), sometimes covered by whiskers (Fig. 3.3(c)).

3.1.3 Oxidation products

During exposure to dry and wet CO2, the low chromium alloys produced external oxide scales consisting of multiple layers. Optical micrographs are shown in Fig. 3.4 in the case of the 9Cr alloys. All scales were separated in two parts. The outer part contained one or several single-phase layers; it was rather compact but presented signs of coarse mechanical fracture: large cracks both parallel and normal to the alloy surface, and large cavities. In some cases, part of the outer scale was not visible in cross-section, either because it was not formed at this location or because it spalled during cooling (see Fig. 3.2). The inner scale consisted of a fine grained multiphase layer, with a fine porosity. In the case of Fe–2.25Cr (Fig. 3.5) and Fe–

9Cr (Fig. 3.4(a,d)), the inner scale porosity was extensive; pores were oriented in a direction parallel to the alloy surface and at times coalesced into one or several rows. The number of these rows increased with gas H2O content. Porosity in the inner scale of the 9Cr ternary alloys was much less pronounced and more uniformly

(a) Fe–9Cr (b) Fe–9Cr–10Ni (c) Fe–9Cr–20Ni

(d) Fe–9Cr (e) Fe–9Cr–10Ni (f) Fe–9Cr–20Ni

Figure 3.4 – Optical micrographs of oxide scales formed after 120 h reaction in (a-c) Ar–20CO₂ and (d-f) Ar–20CO₂–20H₂O at 650 °C.

Figure 3.5 – Optical micrographs showing porosity in the inner layer of the scale formed on Fe–2.25Cr after 120 h exposure to Ar–20CO₂ at 650 °C.

dispersed, as seen in Fig. 3.4.

The inner layer of the scale grown on Fe–9Cr contained two phases, as is evident on the SEM view of the etched cross-section in Fig. 3.6(a). In the case of the 9Cr

(a) Fe–9Cr (b) Fe–9Cr–10Ni (c) Fe–9Cr–20Ni

Figure 3.6 – Metal/oxide interface after 120 h reaction in Ar–20CO₂ at 650 °C.

(a) SEM SE view of specimen etched with Murakami’s reagent; (b) and (c) OM views of as-polished cross-sections.

ternaries, the inner layer contained two or more oxide phases, some metal islands and cracks normal to the alloy surface (Fig. 3.6(b,c)). Metal islands were concentrated near the interface between inner and outer scale of the Fe–9Cr–20Ni alloy. The Fe–

9Cr alloy was prone to internal oxidation during exposure to dry and wet CO2. The volume fraction of oxide precipitates was seen to be very large, but examination by SEM confirmed that a two-phase metal + oxide mixture was indeed formed (see for example Fig. 3.6(a)). Internal oxidation was observed in none of the other alloys. The Fe–9Cr–10Ni alloy did form oxide protrusions in the metal substrate (Fig. 3.6(b)), but these were more or less in contact with the oxide scale.

Chemical analysis by SEM–EDS of an oxide scale formed after reaction of Fe–

9Cr in Ar–20CO2showed that the outer layers were iron oxides, while the inner layer also contained chromium. Diffractograms from a series of planes parallel to the alloy surface were obtained by alternately grinding off a controlled thickness of oxide and performing XRD analysis. Signals produced by different oxide layers were recorded on a given diffractogram, because neither the oxide layers nor the grinding planes were perfectly parallel to the alloy surface. However, all phases could be identified by examining the results of 7 diffractograms, given the depth at which they were recorded, and in light of the cross-section micrograph obtained from the opposite face of the specimen. Selected diffractograms obtained this way after exposure to Ar–20CO2 are shown in Fig. 3.7. The procedure was repeated after reaction of Fe–9Cr in Ar–20CO2–5H2O, yielding the same results. The three layers composing the outer scale were, from the scale/gas interface, Fe2O3, Fe3O4 and FeO. The inner layer was identified as a mixture of FeO and mixed Fe–Cr spinel. The signal from

Figure 3.7 – Diffractograms recorded in the oxide scale grown on Fe–9Cr after 240 h reaction in Ar–20CO₂ at 650 °C.

the latter phase was seen to match FeCr2O4 reference patterns better than Fe3O4

ones; however, the exact composition could not be determined because the lattice parameter of the Fe-rich and Cr-rich spinel solid solutions vary only slightly with composition. Simarly, the internal oxidation zone was identified as an α-(Fe,Cr) + (Fe, Cr)3O4 mixture, with no further specification of the composition possible.

Chromia was not detected at all.

The oxides formed after reaction of an Fe–9Cr–10Ni specimen in Ar–20CO2– 20H2O were identified by local Raman analysis (Fig. 3.8). Comparison with the work of McCarty and Boehme [152] showed that the outer layers were Fe2O3 and Fe3O4. The Raman signal recorded in the inner layer did not enable the phases of this region to be distinguished from one another. Wustite is Raman active with main bands at∼ 280 and ∼ 400 cm⁻¹[153]; these were not observed. The signal was characteristic of spinel oxides, with a main band at ∼ 675 cm⁻¹. While this figure is too high to correspond to pure Fe3O4 [152], it is also too low for the NiCr2O4

and NiFe2O4 spinels, according to the data tabulated in Ref. [154]. Given that it contained at least two phases (see Fig. 3.6(b,c)) the inner layer may be described as a mixture of several spinel phases of general composition (Fe, Cr, Ni)3O4. As is seen in Fig. 3.4, the scales grown on Fe–9Cr–10Ni and Fe–9Cr–20Ni in Ar–20CO2

Figure 3.8 – Raman spectra recorded in the oxide scale grown on Fe–9Cr–10Ni after 120 h reaction in Ar–20CO₂–20H₂O at 650 °C.

and Ar–20CO2–20H2O all had a similar appearence. Thus the phase constitution derived from Raman analysis on one specimen is thought to be representative, to some degree, of the scales grown on both alloys in all gases.