ISIJ International
Online ISSN : 1347-5460
Print ISSN : 0915-1559
ISSN-L : 0915-1559
Regular Article
Influence of Annealing Atmosphere on Oxide Removability in a 3.2% Si Non-Grain Oriented Electrical Steel
Vladimir Vinicio Basabe
Author information

2017 Volume 57 Issue 1 Pages 148-154


In this work, the effects of gas composition, elapsed time of reaction and temperature during annealing on scale removability were investigated for a (3.2 wt.% Si) non-grain oriented electrical steel, and the results were discussed from the viewpoint of oxide morphology.

The annealing tests were carried out under conditions similar to those of industrial annealing operations. For this purpose, steel samples with their original as-received tertiary scale were annealed over the temperature range 900–1000°C in O2–CO2–H2O–Ar–N2 and N2–H2 gas mixtures and in pure N2. After annealing, oxide/steel samples were cooled in air and some were water quenched to study the effect of thermal shock on oxide scale removability. During the annealing tests, four types of oxide scales were observed: oxide scale without idiomorphic growth (Type I), oxide scale with idiomorphic growth (Type II), neutral scale (Type III) and reduced scale (Type IV). The experiments showed that the annealing atmosphere, annealing time, temperature and cooling media influence the morphology and removability of oxide. In general, the experiments indicated that a reducing atmosphere during annealing and water cooling at the end of annealing are the ideal conditions for oxide removal.

1. Introduction

In hot strip mills, the heating of slabs is necessary to soften the steel before rolling. During the reheating of the slab, a primary scale of around 2–3 mm in thickness is formed when the slab is reheated to about 1200–50°C. After the first descaling, thin secondary and tertiary scales grow fast in the subsequent stages of the hot rolling process. In the final stage of coiling, tertiary scales formed during hot rolling undergo a cooling process and a phase change as wüstite is decomposed into magnetite and iron.

After hot rolling, hot rolled coils are then directed for further processing. In general, the main steps of the final production process in non-grain oriented electrical steels are trimming of the strip, first annealing & descaling, cold rolling and final annealing & coating. During the first annealing operation, the oxide scales formed on the surface of hot rolled coils react with the annealing atmosphere. The resultant surface condition after annealing can cause surface defects. To prevent these defects, the oxide scales formed during annealing are removed by sand blasting and pickling. Although sand blasting and pickling are common and reliable practices, the loss of material due to scale defects still occurs in the production of non-grain oriented electrical steel. Therefore, studies on scale formation during annealing in conditions similar to those of industrial processes are of great interest to better understand scale removability and surface quality.

In the analysis of oxide scale removal, it is necessary not only to analyse scale removability as a function of sand blasting & pickling processes but also as a process that is related to the control of the scale during annealing. In the present work, the approach to scale removability is dynamic, in that several factors and their relationships were analysed with the purpose of defining better oxide scale removal conditions and heating practices.

2. Experimental

2.1. Annealing Test

The chemical composition of the investigated steel is given in Table 1. The steel was supplied as hot rolled coil. For the annealing test, the samples were not polished i.e. the samples were annealed with their original as-received tertiary scale as it occurs in industrial annealing conditions. The dimensions of the samples were 20 mm wide, 25 mm long and 2.3 mm thick. The samples were machined with a 1 mm hole at the top for wire suspension.

Table 1. Chemical composition of the investigated steel (mass%).

The oxidation rate increases with gas velocity until a critical gas velocity is reached. The critical gas velocity for air1) is 4.2 cm/s, 2.54 cm/s for carbon dioxide and 11.68 cm/s for steam.2) In this study, a gas velocity of 5 cm/s was used to ensure high oxidising conditions.

During annealing operations strip speeds are in the range of about 15–30 m/min which in turn results in short annealing times. Thus, a short standard annealing cycle of 166 s was chosen for the annealing simulation. In addition, an extended annealing cycle of 406 s was selected to observe the influence of time, Fig. 1.

Fig. 1.

Annealing cycles used to simulate annealing in laboratory conditions.

A vertical tube furnace was used to heat the samples in the desired atmosphere to either 900, or 950 or 1000°C. The selected atmosphere was obtained by controlling the flow of air, CO2, N2 and Ar gases. The water vapour was added by passing the gas mixtures through a sealed water-heated container. In the case of a hydrogen-containing atmosphere, a premixed N2–3H2 gas mixture was used to create a reducing atmosphere.

For the annealing test, the steel samples were introduced to the furnace that was at the desired soaking temperature. After annealing, the oxide/steel samples were extracted from the furnace and cooled in air. Some samples were water quench after annealing to investigate the effect of water-cooling on scale removability. The detailed description of the annealing conditions and cooling media are given in Tables 2 and 3. These gas mixtures were selected to study annealing in: a) products of combustion (propane), b) products of combustion and nitrogen, c) nitrogen, and d) nitrogen-hydrogen gas mixture.

Table 2. Gas mixtures for annealing test.
OxidisingG1 = 0.5O2-11.2CO2-15.6H2O-0.9Ar-N2
G2 = G1+60%N2=0.2O2-4.5CO2-6.2H2O-0.4Ar-N2
G3 = G1+80%N2=0.1O2-2.2CO2-3.1H2O-0.2Ar-N2
NeutralG4 = 100N2
ReducingG5 = N2-3H2
Table 3. Annealing conditions and cooling media.
Temperature (°C)Gas mixtureStandardExtended
Air cooledWater cooledAir cooledWater cooled

Conditions tested are indicated with an asterisk (*)

2.2. X-ray Diffraction Analysis

Hot rolled coils and annealed samples were analysed by X-ray diffraction (XRD) to reveal the oxide phase composition before and after annealing. The analysis was carried out using a Bruker D8 diffractometer. The X-ray source was a conventional cobalt target X-ray tube set to 30 KV and 40 mA. The diffraction patterns were collected from 20–120° in reflection mode. The qualitative and quantitative analyses were performed by Rietveld analysis and Bruker Topas software package for Rietveld refinement.

2.3. Microscopic Observations

Oxide scales tend to fracture during hot mounting, masking the original oxide morphology. Therefore, for this study, the samples were cold mounted and polished with SiC paper to a 1200 grit surface finish. Diamond paste of 0.25 μm was used in the final stage. After polishing, the samples were carbon coated for scanning electron microscopy. In addition, an optical microscope was used to observe oxide/steel cross-sections. Low magnifications (e.g. X100) and high magnifications (e.g. X8000) were used at ten locations in the oxide scales to reveal their morphology in different annealing conditions.

3. Results

3.1. Industrial Hot Rolled Coil

The typical morphology of tertiary oxide scale formed on hot rolled coils is depicted in Fig. 2. The dark areas in the oxide are physical defects such as pores and micro-cracks formed during oxide growth or subsequent cooling. The grey dotted areas below the oxide scale are internal oxidation.

Fig. 2.

Oxide/steel cross-section of hot rolled coil showing internal oxidation (red oval).

An example of the mole fraction of the oxides formed on hot rolled coil is illustrated in Fig. 3. Notice that the X-ray diffraction analysis corresponds to the entire surface (i.e. oxide and steel substrate) then the ferrite mole fraction of the steel substrate is included in the X-ray plot. The X-ray diffraction pattern showed that oxide on hot rolled coil is formed mostly of hematite, magnetite, fayalite and a negligible amount of wüstite.

Fig. 3.

Surface diffraction pattern of hot rolled coil.

3.2. Annealing in Laboratory Conditions

3.2.1. Oxide Morphology

The results of the quantitative phase analysis by X-ray are illustrated in Fig. 4. In the chart, the limit between oxidising conditions and reducing conditions corresponds to the neutral nitrogen atmosphere G4, Fig. 4(d). The oxidising conditions increase from G4 to the left of the chart while reducing conditions increase from G4 to the right. Here, it is important to emphasise that laboratory nitrogen is not 100% pure, and some impurities (about 2 ppm of O2) are always present, and thus some oxidation of iron is expected to occur. Consequently, although the term neutral atmosphere is applicable here, we can not assume that it is totally inert in terms of its oxidising potential.

Fig. 4.

Surface phase composition diagrams showing the influence of annealing atmosphere and temperature. The scales were annealed in the standard cycle over the temperature range 900–1000°C in an oxidising atmosphere (G1, G2, G3), neutral atmosphere (G4) and reducing atmosphere (G5). The conditions in which the different types of scales I, II, III or IV were observed are indicated in the chart.

The correspondent microscopic observations are given in Figs. 5, 6, 7, 8. The oxide/steel cross-sections showed that depending on temperature and atmosphere four types of oxide scales can be observed during annealing operations.

Fig. 5.

Optical micrographs of oxide scales annealed at 900°C in a standard cycle in: a) G1, scale Type I, b) G2, scale Type I, c) G3, scale Type I.

Fig. 6.

Optical micrographs of oxide scales annealed at 950°C in a standard cycle in: a) G1, oxidising atmosphere, idiomorphic growth (red circle), scale Type II, b) G2, oxidising atmosphere, idiomorphic growth (red circle), scale Type II, c) G3, oxidising atmosphere, scale Type I, d) G4, neutral atmosphere, scale Type III, e) G5, reducing atmosphere, some residual scale (black arrows) remained after oxide reduction by the addition of hydrogen, reduced scale is indicated by white arrows, scale Type IV.

Fig. 7.

Optical micrographs of oxide scales annealed at 1000°C in a standard cycle in: a) G1, idiomorphic growth (red circle), scale Type II, b) G2, idiomorphic growth (red circle), scale Type II, c) G3, scale Type I.

Fig. 8.

Optical micrographs of oxide scales annealed in an extended cycle at: a) 950°C in G1, idiomorphic growth (red circle), scale Type II, b) 1000°C in G1, idiomorphic growth (red circle), scale Type II, c) 950°C in G5, residual scale (black arrow), reduced scale (white arrow), scale Type IV.

Type I - This scale corresponds to hot rolled coil annealed in oxidising atmospheres at 900°C in G1, G2 and G3, at 950°C in G3 and at 1000°C in G3, Figs. 4, 5, 6(c), 7(c). In this type, the oxide scale didn’t develop idiomorphic growth i.e. the oxidation rate was more or less uniform on the initial oxide scale surface.

Type II - It was observed in oxidising atmospheres at the temperatures of 950°C in G1 and G2 and at 1000°C in G1 and G2, Figs. 4, 6(a), 6(b), 7(a), 7(b), 8(a), 8(b). In this type, some favoured grains overgrew others as larger idiomorphic crystals resulting in an outermost layer of hematite with an irregular surface.

An increase in the elapsed time of reaction from 166 s (standard cycle) to 406 s (extended cycle) did not produce significant changes in the morphology of the scales i.e. idiomorphic growth was observed as well in the extended cycle (compare Figs. 6(a), 8(a)) and (Figs. 7(a), 8(b)).

Type III - This type corresponds to scale annealed in a neutral nitrogen atmosphere (G4) that had a similar morphology to the one of hot rolled coil, i.e. no idiomorphic growth and absence or negligible amount of wüstite phase, Figs. 4 and 6(d).

Type IV - It was obtained in a reducing atmosphere (G5) and was characterized by a thin oxide (i.e. the residual scale after reduction). The oxide was mostly reduced in the standard cycle (166 s), and an increase in time from 166 s to 406 s did not have a pronounced effect as the oxide was already reduced after 166 s of annealing (compare Figs. 6(e), 6(f) and 8(c)).

3.2.2. Effect of Water Quenching after Annealing

Oxide scales annealed in oxidising and neutral conditions were partially removed during quenching in water, Figs. 9 and 10(a), 10(b). Conversely, oxide scales annealed in reducing conditions were nearly completely removed by the combined effect of a reducing atmosphere and water quenching, Figs. 10(c), 10(d).

Fig. 9.

Optical micrographs of oxide/steel sample annealed at 950°C in a standard cycle in G1. The oxide scale was partially removed by thermal shock during water quenching.

Fig. 10.

Optical micrographs of quenched oxide scales after annealing at 950°C in: a) G1, extended cycle, oxide partially removed by thermal shock, internal oxidation (oval), b) G4, standard cycle, oxide partially removed by thermal shock, internal oxidation (oval), c) G5, standard cycle, oxide removed by reduction and thermal shock, residual oxide (black arrow), d) G5, extended cycle, oxide removed by reduction and thermal shock, residual oxide (black arrows), reduced scale (white arrow).

4. Discussion

In the present work annealing under conditions similar to those of industrial operations was investigated and results are summarized in Table 4.

Table 4. Classification of scale.
Scale TypeClassification by processAtmosphereTemp. (°C)Remarks
TertiaryHot rollingAir<1000Uniform layer
Type IAnnealing in low oxidising conditionsG1900Uniform layer
Type IIAnnealing in high oxidising conditionsG1950, 1000Idiomorphic growth
G2950, 1000
Type IIIAnnealing in neutral conditionsG4950Uniform layer
Type IVAnnealing in reducing conditionsG5950Residual oxide

4.1. Oxide Phase Composition

4.1.1. Tertiary Scales

Oxide scales formed during hot rolling are composed of wüstite, magnetite and a hematite.3) The wüstite phase decomposes to magnetite and alpha-iron when cooling to temperatures below 570°C.4,5) The decomposition of wüstite depends on the cooling temperature; the fastest decomposition of wüstite is observed at 400°C.5) The cooling rate also influences the decomposition of wüstite. A high cooling rate partially or completely prevents the decomposition of wüstite.4,5,6) In the present work, slow coil cooling appears to have decomposed wüstite, which in turn resulted in the absence or at most a negligible amount of wüstite, Fig. 3.

4.1.2. Scales Formed during Annealing

Oxidising Atmosphere - The results of the present investigation (see Fig. 4) do not correspond to oxide phase composition diagrams in which wüstite is the predominant phase at high temperature (above 570°C). This disagreement is due to the fact that annealing was carried out in a short time (166 s) on already oxidised samples (initial hot rolled coil with its original as-received tertiary scale) that had an initial composition of hematite and magnetite i.e. the short time of annealing was not enough for a thermodynamic equilibrium to be reached. Moreover, in a high silicon content steel, (>3%Si), a passive oxide layer of silica (SiO2) is formed on the steel substrate that retards oxidation.7) Thus, the presence of a passivation layer may have also contributed to the absence and/or negligible amount of wüstite, i.e. low oxidation rates were not sufficient to form wüstite in a short period of time. Here, it is important to mention that silica is generally X-ray amorphous (either really amorphous or nano-crystalline) and it can’t be detected by X-ray diffraction analysis. Consequently, silica was excluded in the oxide phase composition of the present work. Nevertheless, the presence of a passivation silica film has been reported as a very thin film that it is difficult to detect; X-ray photoelectron spectroscopy (XPS) has been used to detect this very thin film of silica.7) For the above reasons, it is fair to assume that the amount of silica if any in the present work was negligible (SiO2%<<1) and it doesn’t affect the net observations of surface phase composition of Fig. 4. In general, it was observed that in all oxidising conditions (scales Type I and II) a rise in temperature increased the mole fraction of hematite while diminished the fraction of magnetite, Figs. 4(a), 4(b), 4(c).

Neutral Atmosphere - In this condition, the oxide scale (Type III) had less hematite (see Fig. 4(d)) than the ones formed in G3. This indicates that even a small increase in oxidation potential from G4 to G3 produced scale morphology changes.

Reducing Atmosphere - In G5 the content of iron oxides diminished while the iron content increased, Fig. 4(d). The rise in iron was due to: a) a thinner oxide (after oxide reduction) that allowed a deeper penetration of the X-ray beam in the steel substrate i.e. larger contribution of the steel substrate to the diffraction pattern than in the case of thicker oxide scales, b) the presence of reduced scale i.e. metallic iron.

In general, fayalite is concentrated close to the oxide/steel interface. Consequently, a thin enriched fayalite oxide was left after oxide reduction in G5. In this situation, the X-ray pattern is obtained from a zone richer in fayalite i.e. the relatively high content of fayalite in the X-ray pattern was due to the chosen area for the diffraction rather than to a real increase in fayalite.

4.2. Idiomorphic Growth

Sachs and Tuck8) attributed the formation of the irregular outer surface to the formation of idiomorphic crystals at the surface. They hypothesized that during the initial oxidation in dilute atmospheres, the rate-controlling mechanism is the surface reaction, hence the adsorption of oxygen atoms or absorption of ions would occur faster at preferred planes of the oxide crystal, leading to idiomorphic crystals at the surface and a scale with an irregular outer surface.

The idiomorphic growth observed during annealing in oxidising atmospheres (G1 and G2) corresponds partially with observations of many authors that have reported idiomorphic growth of wüstite in dilute atmospheres.6,9,10,11) The present work differs in the sense that idiomorphic growth took place in a high oxide (hematite) instead of in a lower oxide (wüstite). In general, previous studies in dilute atmospheres were carried out on polished samples where wüstite was the first phase to appear on the surface of the steel at high temperature (above 570°C). As pointed out previously, the initial nature of the hot rolled coil surface before annealing (covered by magnetite and hematite) coupled with a passivation layer appears to be the main reason for the absence of wüstite after annealing. In this situation, the adsorption of oxygen occurred faster at preferred hematite planes rather than at wüstite planes.

Hematite can be formed by iron ions moving outward, or oxygen ions moving inward in the hematite layer.12) In the outward migration of iron ions in the hematite layer, the iron ions migrate via iron ion vacancies together with electrons. The reaction for the formation of hematite at the gas/hematite interface is:   

2F e 3+ +6 e - + 3 2 O 2 =F e 2 O 3 (1)

In the case of oxygen ions migrating inward in the hematite layer, the iron ions and electrons (in excess of requirements for reduction of hematite to magnetite) react with oxygen moving via oxygen vacancies. The reactions for the oxygen ionization and formation of hematite at the hematite/magnetite interface are:   

1 2 O 2 +2 e - = O 2- (2)
2F e 3+ +3 O 2- =F e 2 O 3 (3)

There is disagreement among researchers regarding the formation of Fe2O3. In contrast to the above model, the formation of hematite has been considered only as an oxygen ion migration process.13) The later work of Schwenk and Rahmel14) concluded that because of experimental difficulties, it is impossible to clearly establish the reactions in which Fe2O3 is really formed. In the present work, it seems that hematite was formed at the hematite/gas interface where certain crystallographic planes absorbed oxygen faster than others resulting in an idiomorphic growth.

4.3. Transition from Type I to Type II

The transition from Type I to Type II occurred with an increase in oxidising potential and/or temperature, Table 4.

The transition with oxidizing potential at 950°C and 1000°C took place when the free oxygen and water vapour were raised from G3 to G2 and G1. Idiomorphic growth was not observed in the oxidising atmosphere (G3). Probably in this situation the oxidation potential/oxidation rates were too low for idiomorphic growth to take place.

The transition in G1 and G2 occurred as well with an increase in temperature from 900 to temperatures ≥950°C. At this point, the increase in temperature resulted in higher diffusion/oxidation rates that allowed idiomorphic growth.

4.4. Oxide Reduction by Hydrogen

In general, there are two models that illustrate the reduction steps of hematite into iron:

a) The first model assumes a transformation in two stages. In this model, Fe2O3 is reduced to Fe3O4, and then reduced to metallic iron.15)

b) In the second model, the reduction of hematite into iron takes place in a three stage process. The initial Fe2O3 is reduced to Fe3O4, then reduced to FeO, and finally to metallic iron.16)

In this work, the absence/negligible amount of wüstite in the residual scale obtained in G5 may suggest that reduction took place accordingly to the first model through the reactions:   

3F e 2 O 3 + H 2 =2F e 3 O 4 + H 2 O (4)
F e 3 O 4 +4 H 2 =3Fe+4 H 2 O (5)

4.5. Scale Removability

4.5.1. Annealing in Oxidising and Neutral Atmospheres

The oxide/steel interface is characterised by a sharp change in physical properties such as linear coefficient of thermal expansion and thermal conductivity, which in turn generate a thermal wave from the oxide surface to the metal substrate during cooling.17) During quenching/cooling the thermal wave generate contractions that result in a horizontal undercutting, (crack), of the oxide scale.17,18)

In this study, the localization and propagation of the crack always occurred in the scale that was partially removed, Fig. 9. This corresponds with previous observations of oxide grown in dilute atmospheres in which it was observed that the fracture strength of the scale is lower than the fracture strength of the scale/steel interface.10,19)

In general, the removability of oxide scales annealed in oxidising conditions was unsatisfactory. Some scale detached during quenching but isolated oxide islands remained adhered to the metal substrate.

4.5.2. Annealing in a Reducing Atmosphere

The X-ray diffraction patterns showed that the residual scale after quenching was composed of a mixture of phases, Fig. 11. This shows the complexity of the oxide scale present in hot rolled coils in which a mix of phases is present even in the case of thin/negligible residual scales.

Fig. 11.

Surface diffraction patterns of oxide/steel samples annealed at 950°C in N2–3H2 for: a) 166 s, standard cycle, b) 406 s, extended cycle. After annealing, the samples were water quenched.

The results of annealing in a reducing atmosphere are summarized in Fig. 12. For comparison purposes some oxides that were annealed n a reducing atmosphere but air cooled instead of water quenched are included. Here, it can be seen that water quenched samples had a slightly lower oxide content than air cooled samples. This small difference indicates that mostly of the oxide was removed by the reduction atmosphere during annealing rather than during quenching.

Fig. 12.

Surface phase composition diagram showing the influence of annealing time and cooling media. The scales were annealed in standard and extended cycles at 950°C in a reducing atmosphere (G5).

The removal of oxide scale was satisfactory in samples annealed in a reducing atmosphere and water quenched. This strongly suggests that optimum scale removability is achieved by the combined effect of a reducing atmosphere and water cooling.

5. Conclusions

The morphology and removability of tertiary oxide scales annealed under conditions similar to those of industrial annealing operations in a (3.2 wt.% Si) non-grain oriented electrical steel were investigated and the results are summarized as follows:

(1) Depending of the annealing atmosphere and temperature, oxide scales can be classified as four types: oxide scale without idiomorphic growth (Type I), oxide scale with idiomorphic growth (Type II), neutral scale (Type III) and reduced scale (Type IV).

(2) In oxidising conditions, the effect of annealing time on scale morphology and removability was small i.e. idiomorphic growth was present as well in the extended cycle and after quenching similar residual scales were observed.

(3) In reducing atmospheres, the oxide was mostly reduced in the standard annealing cycle. Thus, an increase in time in the extended cycle had a small effect on scale removability as the oxide was already reduced.

(4) During quenching the stresses arising from temperature change resulted in a horizontal undercutting (crack) inside the scale close to the oxide/steel interface, indicating that the fracture strength of the scale was lower than the fracture strength of the oxide/steel interface.

(5) The best conditions for removing oxide scale are achieved with a reducing atmosphere and water quenching. This indicates that industrial annealing should be carried out in reducing atmospheres with a water cooling section at the furnace exit.

(6) Equilibrium phase diagrams are not appropriate for predicting oxide scale phase composition in short annealing times. Diagrams applicable to tertiary scales that are undergoing transformation during annealing are required that will eventually permit the development of more accurate models of annealing.

© 2017 by The Iron and Steel Institute of Japan