influence of the oxidation-reduction process on the

© 2011 ISIJ 630

ISIJ International, Vol. 51 (2011), No. 4, pp. 630–637

Influence of the Oxidation-Reduction Process on the Surface and Sub-surface Microstructure of Intercritically Annealed CMnSi TRIP Steel

Young Feng GONG and Bruno Charles De COOMAN

Graduate Institute of Ferrous Technology, Pohang University of Science and Technology, Pohang, South Korea.E-mail: [email protected]

(Received on August 11, 2010; accepted on December 21, 2010)

The surface and sub-surface of an intercritically annealed Si-bearing CMnSi TRIP steel were investigatedby high resolution transmission electron microscopy after pre-oxidation in a N2+0.3 vol-% O2 atmosphereand subsequent reduction in a N2+10 vol-% H2 atmosphere. After the initial pre-oxidation the surface con-sisted of areas covered with a thick Fe3O4 oxide with an open porous structure and areas covered with athin compact Fe3O4 film connected to the steel by xMnO.SiO2 oxide bridges connecting the oxide film tothe matrix. Both porous and compact oxide contained embedded grains of MnO. A grain boundary net-work of amorphous SiO2 was formed in the steel subsurface under the porous Fe3O4 after the pre-oxida-tion treatment. No internal oxidation was observed below the compact Fe3O4 layer. Whereas, the Fe3O4

was fully reduced to metallic Fe after reduction annealing, the selective oxides MnO, c-xMnO.SiO2 (x>1)and a-xMnO.SiO2 (x<0.9) were unaffected by the reduction annealing. As no amorphous a-SiO2 or a-xMnO.SiO2 (x<0.9) film forming oxides were present at the surface of the Si-bearing TRIP steel after thepre-oxidation and reduction annealing, the treatment is believed to lead to an improved wettability of thesurface by liquid Zn during hot dip galvanizing and Zn-coatings which are free of bare spot defects. Zn-coating adhesion problems may however arise from the formation of large pores at the Fe3O4/steel inter-face, which are still present after the oxide reduction.

KEY WORDS: TRIP steel; oxidation-reduction; hot dip galvanizing.

1. Introduction

The galvanizability of Transformation-Induced Plasticity(TRIP) steels by means of hot dip galvanizing is a key factorto determine the suitability of TRIP steel for automotiveapplications as thinner TRIP steel sheet used to producelighter car-bodies must have a high resistance to perforationcorrosion. The low dew point N2 + (5–10) vol-% H2 gasatmospheres used in the annealing section of continuous HotDip Galvanizing (HDG) lines lead to the reduction of theiron oxides formed during cold rolling. It also results in theexternal selective oxidation of the key alloying elements Siand Mn. The presence of film forming surface oxides,especially amorphous a-xMnO.SiO2 (x<0.9) and a-SiO2,leads to the deterioration of the wettability of the steel bymolten Zn, prevents the formation of the Fe2Al5–xZnx

inhibition layer and leads to bare spot defects.1–3)

The selective oxidation at the surface and in thesubsurface of DP, IF and TRIP steels is well documented,and CMnSi TRIP steels with Si contents in the range of 0.5–1.5 mass-% are usually considered more difficult to processwhen using a conventional HDG line operation practice.3–8)

Figure 1 illustrates schematically the three methodswhich have been proposed to improve the difficultiesencountered during the hot dip galvanizing of CMnSi TRIP

steel. In the first method, a high dew point atmosphere isused which causes the internal selective oxidation of Si andMn to SiO2 and xMnO.SiO2, respectively.6,7) In the secondmethod the pre-deposition of a thin layer of Fe, Ni or Cu

Fig. 1. Schematic showing the three methods proposed to avoidZn-coating defect formation during hot dip galvanizing ofSi-alloyed TRIP steel.

ISIJ International, Vol. 51 (2011), No. 4

631 © 2011 ISIJ

prevents selective oxidation altogether.9–11) In the thirdmethod, the steel surface is pre-oxidized and the selectiveoxidation results in isolated, rather than continuous, oxideparticles which are embedded in a thicker Fe-oxide surfacelayer.12–15) This layer is then reduced to metallic Fe in areduction step. The oxidation-reduction process prevents theenrichment of Si and Mn as a thin surface oxide film ofxMnO.SiO2 or SiO2 during the intercritical annealing pro-cess. As a result the formation of coating defects such asbare spots is inhibited. Bordignon et al.14) have described theoxidation-reduction process in detail. They have shown thatthe presence of selective oxides of Si and Mn created duringcontinuous annealing deteriorated the Zn-coating adhesionafter hot-dip galvanizing, and proved that the problem couldbe addressed by using an annealing atmosphere withincreased oxidizing potential. Nakanishi et al.12) have alsoproposed an oxidization-reduction method to improve thequality of Zn-coatings on Si-bearing steel. Details of theiroxidation-reduction process have not been published.

No information is currently available on the morphology,

size, distribution or composition of the oxides of the alloy-ing elements and the type of iron oxides scale formed on aCMnSi TRIP steel during the pre-oxidation annealing pro-cess and after the reduction. Therefore, High ResolutionTransmission Electron Microscopy (HR-TEM) of FocusedIon Beam (FIB) milled cross-sectional samples was used tocarry out the first comprehensive analysis of the composi-tion and micro-structure of the compounds formed at thesurface and in the near surface region of conventionalCMnSi TRIP steel during an oxidation-reduction annealingtreatment.

2. Experimental

The material used in the present work was industriallyproduced, cold rolled, full hard CMnSi TRIP steel, with thefollowing composition: 0.11 mass-% C, 1.53 mass-% Mn,and 1.46 mass-% Si. The sample surface was mirror-polished with a 0.5 μm diamond suspension prior to anneal-ing in order to avoid effects solely related to the roughness

Fig. 2. (a) Examples of EDS spectra for MnO, SiO2, and the compound oxides 2MnO.SiO2 and MnO.SiO2. (b) Examplesof EDS spectra for Fe3O4 and reduced Fe.

© 2011 ISIJ 632


of the strip surface. The continuous annealing simulationwas carried out in two types of gas atmospheres. The pre-oxidizing atmosphere was a N2+0.3 vol-% O2 gas mixture.The choice of the a gas atmosphere with 0.3 vol-% of oxy-gen is based on the work of Bordignon et al.,13–15) whoshowed that the oxide formed during the pre-oxidation waseffectively removed during a short exposure to a reduction-promoting N2+10 vol-% H2 gas mixture. The dew point was–40°C. The sample heating rate was +5°C/min. Theintercritical annealing temperature was 860°C and thesoaking time was 162 seconds. After the pre-oxidationstage, the samples were quenched to room temperature ata rate of –15°C/s with high purity He and prepared for HR-TEM analysis.

For the observation of the microstructure after the fulloxidation-reduction process, the samples were heated at arate of +5°C/min to the intercritical annealing temperatureof 860°C, in the pre-oxidizing N2+0.3 vol-% O2 gas mixture,which was replaced by the reduction-promoting N2+10 vol-%H2 gas mixture when the sample reached the soaking tem-perature. The dew point was –40°C. The sample was keptat the soaking temperature for 162 seconds and quenched toroom temperature at a rate of –15°C/s with high purity Hegas.

The samples were investigated in a ZEISS ULTRA 55Scanning Electron Microscope (SEM) equipped with afield-emission electron source for the observation of theflat-on surface microstructure. Cross-sectional TEMsamples were prepared by Focused Ion Beam (FIB) millingto determine the composition, crystal structure, morphology,and spatial distribution of the oxides in a JEOL JEM-2100FFE-TEM operated at 200 kV. The determination of theoxides composition was done by means of a nano-beamINCA Energy Dispersive Spectroscopy (EDS) attachment.Figure 2 shows examples of typical EDS spectra. The EDS

spectra are for MnO, SiO2, 2MnO.SiO2 and MnO.SiO2. Atypical quantitative analysis for the compound oxides is alsolisted. In addition, Fig. 2 illustrates the difference betweenthe EDS spectrum for the Fe3O4 layer after the pre-oxidationand the reduced Fe layer obtained after the reductiontreatment.

3. Results

A typical flat-on view of the surface of a CMnSi TRIPsteel sample intercritically annealed in the N2+0.3 vol-% O2

oxidation-promoting atmosphere is shown in Fig. 3. Theoxides were present on the surface as either thick porousoxides (e.g. zone A) or thin compact oxides layers (e.g. zoneB). In addition, small oxide ridges were present at thesurface. These may be caused by a slightly higher oxidationrate where the grain boundaries intersecting the surface. Thesurface areas covered with the porous oxides were clearlyseparated from the compact oxide areas.

Figure 4 shows a cross-sectional view of the poroussurface oxide formed in zone A of Fig. 3. Note that the thinlayer with a dark contrast at the surface of the sample is aprotective Pt coating used to protect the thin surface oxidefilms during the FIB specimen preparation. Figure 4(a)

Fig. 3. SEM flat-on image of CMnSi TRIP steel after intercriticalannealing in a 0.3%O2+N2 atmosphere showing oxides withporous (A) and compact (B) morphologies. Small oxideridges, indicated with arrows, are present at the surface.These may be caused by grain boundary oxidation.

Fig. 4. (a) Cross-sectional TEM micrograph of the porous oxidearea labeled A in Fig. 2, showing the upper Fe3O4 zone(Zone I) and the lower selective internal oxidation zone(Zone II). Zone II is characterized by a grain boundary net-work of amorphous a-SiO2 oxides. Small voids and oxidesare present at the interface between Zone I and II. (b) Lat-tice image and electron diffraction pattern (inset) for Fe3O4

in area a of (a). (c) Lattice image and electron diffractionpattern (inset) for amorphous a-xMnO.SiO2 (x=0.08) inarea b of (a). (d) Lattice image of the interface betweenamorphous a-SiO2 and the ferrite matrix in area c of (a).Inset: diffraction pattern for amorphous a-SiO2.


633 © 2011 ISIJ

shows that the oxide consists of two distinct zones: a surfaceoxidation zone containing mostly porous Fe3O4 (Zone I) andan internal selective oxidation zone consisting mostly of agrain boundary network of amorphous a-SiO2 (Zone II).Voids, 200–300 nm in size, were present within thepolycrystalline Fe3O4 surface oxide layer. The surface oxidelayer in zone I was identified as Fe3O4 by EDS and electrondiffraction. The layer thickness was in the range of 1300–2000 nm. In the subsurface zone II, the steel was notoxidized and consisted of ferrite, as shown by the electrondiffraction pattern in the inset of Fig. 4(a).

Between the outer porous Fe3O4 layer (zone I) and theselective oxidation zone (zone II), a thin transition layer canclearly be observed in the micrograph. Small voids with adiameter of 100–150 nm and mixed Si and Mn oxides werepresent at the interface between zone I and zone II.

Figure 4(b) shows a detailed view of the Fe3O4 oxide nextto one of the larger voids in the area labeled a in Fig. 4(a).The Fe3O4 lattice image and the corresponding diffractionpattern are shown in Fig. 4(b). The surface oxide layer caneasily be identified as Fe3O4 by the periodic contrast fromthe clearly visible (200)Fe3O4 lattice planes. The (200)Fe3O4

inter-planar spacing was measured to be 0.225 nm.A high resolution lattice image of the oxides present at

the interface between zone I and zone II (area b in Fig. 4(a))is shown in Fig. 4(c). The electron diffraction pattern clearlyshows that it is amorphous. EDS analysis of this interfaceoxide showed that it is a Mn–Si compound oxide with acomposition that can be written as xMnO.SiO2 with x=0.08.

Figure 4(d) shows the interface between the steel matrixand grain boundary oxide network in the subsurface arealabeled c in Fig. 4(a). The grain boundary oxide networkwas found to consist entirely of amorphous a-SiO2.

Figure 5 shows a cross-sectional view of the compactoxide zone B in Fig. 3. In contrast to zone A, the surfaceconsists of a thin compact Fe3O4 oxide film, which does notcontain pores. The Fe3O4 film has a thickness in the rangeof 250–450 nm (Fig. 5(a)). There is no evidence for a a-SiO2

grain boundary oxide network in the subsurface. In addition,large voids are present between the compact Fe3O4 surfaceoxide film and the steel matrix. Both are connected by smallbridging oxides. EDS and electron diffraction patternanalysis indicated that the steel matrix was ferrite.

An enlargement of the area labeled b in Fig. 5(a) is shownin Fig. 5(c). The dotted line on the micrograph indicates theposition of the interface between xMnO.SiO2 (x<0.9) andthe steel surface. The transition from the Fe3O4 layer to aninternal void is shown in Fig. 5(d). The upper part of thevoid consists of Fe1–xMnxO. Figure 5(e) shows the interfacebetween the oxide layer and the steel matrix for area d inFig. 5(a). EDS analysis and electron diffraction revealed thatthis thin oxide layer was amorphous a-SiO2 oxides.

A typical flat-on overview of the surface of the CMnSiTRIP steel after the intercritical oxidation-reductionannealing is shown in Fig. 6. After the reduction annealing,the surface consists of thicker porous areas (zone A) and athin pore-free areas (zone B).

Figure 7 shows a cross-sectional view of the porous zone

Fig. 5. (a) Cross-sectional TEM micrograph of the compact oxidearea labeled B in Fig. 2, showing a void-free Fe3O4 surfacelayer and the presence of a large void at the oxide/metalinterface. (b) Lattice image of Fe3O4 and correspondingelectron diffraction pattern (inset) in area a of (a). (c)Lattice image of xMnO.SiO2 (x<0.9) oxide in area b of (a).(d) Lattice image of the upper edge of the interfacial void inarea c in (a). (e) Lattice image of the a-SiO2/matrix inter-face in area d of (a).

Fig. 6. SEM flat-on image of TRIP steel after intercritical oxidation-reduction annealing, showing areas with porous (A) andcompact (B) morphologies after the reduction of the Fe3O4

surface oxide layer.

© 2011 ISIJ 634


labeled A in Fig. 6. After the reduction treatment, the Fe3O4

oxide appears to be fully reduced to metallic iron. IsolatedMnO oxide particles are embedded within the pure Fe layer.The compound oxides of Mn and Si are still present at theinterface. The grain boundary network of amorphous a-SiO2

has also not been affected by the reduction and is stillclearly present.

Figure 8 gives a more detailed analysis of the arealabeled A in Fig. 7. The Fe3O4 is clearly reduced to metalliciron and isolated MnO oxides particles are embedded in theFe film. The voids have thin edges which contain compoundoxides of Mn and Si (Fig. 8(a)). Figure 8(b) shows thetransition from the reduced metallic iron to un-reducedxMnO.SiO2 (x>1) to voids, corresponding to the arealabeled a in Fig. 8(a). The dotted white line indicates theposition of the interface between the metallic iron andxMnO.SiO2 (x>1) oxide.

The lattice image of an embedded MnO particle locatedin the area labeled b in Fig. 8(a), which was not reducedduring the reduction stage, is shown in Fig. 8(c) togetherwith its corresponding electron diffraction pattern. TheMnO oxide particle could be identified as MnO by the clearperiodic contrast of the MnO lattice planes. The MnO

inter-planar spacing was measured to be 0.240 nm.A lattice image of the ferritic Fe in area c of Fig. 8(a) is

shown in Fig. 8(d), together with the corresponding electrondiffraction pattern. The inter-planar spacing was measuredto be 0.101 nm, which corresponds to the ferrite α

interplanar spacing.Figure 8(e) shows the lattice image of the oxides close to

the voids in the area labeled d in Fig. 8(a). EDS analysisconfirmed that this was crystalline c-xMnO.SiO2 (x>1). Alattice image of the transition from un-reduced Fe3O4 to

amorphous a-SiO2 oxide is shown in Fig. 8(f), whichcorresponds to the area labeled e in Fig. 8(a).

Figure 9 shows a more detailed analysis of the areaslabeled B and C in Fig. 7. Figure 9(a) is an enlargement ofthe top surface of the reduced Fe3O4 oxide. The narrow areawith a bright contrast is the actual surface after theoxidation-reduction annealing. EDS analysis and latticeimaging (Fig. 9(b)) show that this surface was metallic Feand that no oxides of Mn and/or Si were present at thesurface. It clearly shows that the combination of pre-oxidation and reduction during intercritical annealing ofTRIP steel makes it possible to avoid the formation ofamorphous a-SiO2 and/or a-xMnO.SiO2 (x<0.9) oxideswhich are known to be the cause of most HDG Zn-coatingdefects in Si-bearing TRIP steels.2,5–7)

The internal grain boundary oxide network is still presentafter the oxidation-reduction annealing, as shown in Fig.9(c), which corresponds to the area labeled C of Fig. 7. Thelattice image of the amorphous a-SiO2 oxides and itscorresponding diffraction pattern are shown in Fig. 9(d).

Fig. 7. Cross-sectional TEM micrograph corresponding to theporous area A in Fig. 5. The Fe3O4 surface was reduced tometallic iron except for the embedded MnO particles. Thevoids, the compound oxides of Mn and Si and the amor-phous a-SiO2 grain boundary network are not altered bythe reduction stage of the oxidation-reduction annealing.

(111) (111)

(103)

Fig. 8. (a) Cross-sectional TEM micrograph of the CMnSi TRIPsteel after the intercritical oxidation-reduction annealing forthe porous area labeled A in Fig. 5. (b) Lattice imagecorresponding of area a in (a), showing that the interiorvoid surface consists of amorphous a-xMnO.SiO2 (x>1). (c)Lattice image and corresponding electron diffraction pat-tern (inset) for the embedded MnO particle in area b of (a).(d) Lattice image and corresponding electron diffractionpattern (inset) for ferrite in area c of (a). (e) Lattice imagefor crystalline c-xMnO.SiO2 (x>1) present between twovoids in area d of (a). (f) Lattice image of amorphous Fe3O4

and a-SiO2 on the void surface in area e of (a).


635 © 2011 ISIJ

Figure 10 shows a cross-sectional view of the area B inFig. 6 after the intercritical oxidation-reduction annealing.The voids originally formed between the Fe3O4 oxide layerand steel matrix are still present after the reduction. Thecompound oxides of Mn and Si at the void surfaces are alsostill present, as shown in Fig. 10(a). An enlargement of thearea labeled A in Fig. 10(a) is shown in Fig. 10(b). It showsthat the initial Fe3O4 oxide layer is clearly reduced tometallic iron. Some isolated MnO particles are embedded inthis thin pure Fe surface layer as indicated in Fig. 10(b).

Figure 10(c) shows the lattice image and the ferriteelectron diffraction pattern corresponding to the pure ironarea a in Fig. 10(b). The lattice image of an isolated MnOoxide embedded in the pure iron surface layer is shown inFig. 10(d), which is an enlargement of the area labeled b inFig. 10(b). The MnO inter-planar spacing is measured tobe 0.240 nm. EDS analysis also indicated that the oxide wasMnO. The MnO and pure Fe at the void surface in areas cand d in Fig. 10(b) are shown in the Figs. 10(e) and 10(f).It shows that initial Fe3O4 bridges between interfacial voidsreduced to the pure Fe and detached from the upper layer.The detailed lattice image of the oxides at the edge of thevoid in area d in Fig. 10(b) is shown in Fig. 10(f) and itshows the transition from the amorphous oxide a-xMnO.SiO2 to the a-SiO2.

4. Discussion

Whereas much work has been done on the oxidation ofiron and steel in conditions of dry oxidation16,17) and inatmospheres containing water vapor,18) no detailed researchresults are available, with the exception of the work byBordignon et al.,14) about the iron oxides formation and theirsubsequent reduction during the oxidation-reduction processcarried out in conditions corresponding to those of a contin-uous annealing furnace.

The oxidation of pure Fe, certainly in terms of the funda-mental point defect mechanisms controlling the mass trans-port in the oxides, is now well established. Of particularrelevance to the discussion of the present results is the fun-damental work by Dieckmann and co-workers on transportin Fe3O4.19,20) Their work has shown that, at high oxygenactivities, the mass transport in Fe3O4 required for thegrowth of this oxide is controlled by the outward diffusion

Fig. 9. (a) Cross-sectional TEM micrograph of CMnSi TRIP steelafter intercritical oxidation-reduction annealing in area B ofFig. 6. (b)TEM lattice image for the ferrite in area a of (a).(c) Cross-sectional TEM micrograph of CMnSi TRIP steelafter intercritical oxidation-reduction annealing showingthe grain boundary network of amorphous a-SiO2 in area Cof Fig. 6. (d) Lattice image and electron diffraction pattern(inset) of the amorphous a-SiO2 oxide in area b of (c).

(111)

Fig. 10. (a) TEM cross-sectional micrograph of CMnSi TRIP steelafter intercritical oxidation-reduction annealing for thearea corresponding to area B in Fig. 5, showing that theFe3O4 iron oxide layer is fully reduced to ferrite. Embed-ded MnO particles are still present in this layer. (b) Magni-fied view of the area labeled A in (a). (c) Lattice image andcorresponding electron diffraction pattern (inset) for thepure ferrite surface layer in area a of (b). (d) Lattice imageand corresponding electron diffraction pattern (inset) forthe embedded MnO particle in area b of (b). (e) TEMmicrograph of MnO present at the upper side of the void atthe reduced oxide/metal interface. (f) TEM micrograph ofFe present at the lower side of the void in area d of (b). (g)TEM micrograph of the transition from a-xMnO.SiO2 toamorphous a-SiO2 in area e of (b).

© 2011 ISIJ 636


of metal cations and the inward movement of metal cationvacancies. Fe3–dO4 is therefore a more appropriate formulafor magnetite, where d is the deviation from stoichiometryreflecting the presence of the point defects. Their worksuggests that in the present case the outward transport of Fe-cations in Fe3O4 dominates and no oxygen anion diffusiontakes place. Oxygen is incorporated at the outer oxidesurface and creates cation vacancies which diffuse into themetal. These vacancies are removed at the oxide/metalinterface by the nucleation of pores. The oxidation kineticsis decreased by the presence of these vacancies, as a largenumber of vacancies will eventually form pores which canjoin up to prevent further Fe diffusion. This will result in athin compact Fe3O4 oxide layer connected to the substratemetal via small interfacial bridging oxides. This mechanismhas one important consequence: the direct oxidation of Feby oxygen is essentially blocked once a continuous compactFe3O4 oxide layer is formed. This is clearly not the case inmany thermal oxidation conditions encountered in practice.Oxygen can apparently continue to have a direct access tothe metal after the formation of the oxide layer. It has beensuggested that in these cases oxygen can still move inwardfrom the gas phase because dislocations, grain boundaries ormicro-cracks in the oxide provide access to the metal sur-face to oxygen-bearing gas molecules.20)

The situation of TRIP steel is more complex as it involvesthe presence of other oxidizable elements, i.e. C, Mn and Si.In addition, both ferrite and C-enriched austenite are presentduring the oxidation-reduction process. During theintercritical annealing of TRIP steel, carbon is initiallypresent as cementite and as solute carbon in the intercriticalaustenite phase later on in the process.

In general, carbon is more easily oxidizable than Fe, butits behavior will depend on the carbon content, the temper-ature and whether the carbon is present as cementite or insolution in the carbon-enriched austenite phase. Carbon canreact with adsorbed oxygen from the atmosphere or withdissolved oxygen to form gaseous CO or CO2. This surfacedecarburization process facilitates the carbon diffusiontowards the surface. During the heating, carbon is present asFe3C, which gradually forms the carbon-enriched austenite.It is well known that the oxidation of Fe3C leads to the rapidformation of a superficial protective Fe3O4 layer which pre-vents further oxidation of the cementite and allows Fe3C togo into solution:21)

Fe3C + 2O2 → Fe3O4 + C

Note that a similar situation results even if iron oxide isformed initially and carbon remains in solution in the aus-tenite phase shielded from direct oxidation by the iron oxidefilm. Wolf and Grabke22) have clearly shown that carbondoes not dissolve in the lattice or in the grain boundaries ofFe3O4. This was confirmed by Manenc and Vagnard23) whoobserved that decarburization in Fe–C alloys occurred onlyin places where the oxide was broken. In areas where theoxide was strongly adherent to the metal, they observed car-bon enrichments.

The presence of carbon leads to the formation of poresand micro-cracks in the oxide layer. Engell24) has shown thatthe oxidation of both carbon and iron takes place in Fe–Calloys. The carbon is oxidized to CO and CO2 gas by reduc-

tion of the iron oxide at the oxide/metal interface:

Fe3O4 + 2C → 3Fe + 2CO2

The formation of gases causes the cracking of the oxideand this exposes more metal to oxygen. As a consequence,the locally higher carbon content of the austenite leads tomore oxidation because in these areas the fast diffusion ofmolecular oxygen is possible through a porous oxide. Thepresence of pores also enhances the decarburization and thisis in agreement with well-known observation that extensivedecarburization is often associated with the presence ofporous, less-adherent oxides. A similar observation has beenreported by Pandey25) in the case of the decarburizationoccurring during oxidation of Mn-free Ni. As the carboncannot diffuse through the compact oxide layer, CO2 gas isformed at the interface.

The schematic of Fig. 11 gives an explanation for thesurface and interface phenomena observed during theintercritical oxidation-reduction annealing of CMnSi TRIPsteel. In this figure the pore formation by CO2 is taken as aworking assumption, which remains to be proven experi-mentally.

In the initial state, the microstructure consists of ferriteand pearlite. As the steel is heated to the intercritical holdingtemperature, the pearlite constituent transforms to a carbon-rich austenite phase. During the pre-oxidation, the surfaceferrite and austenite phase oxidize differently due to theirdifference in carbon content.

Fig. 11. Schematic showing the difference in oxidation-reductionmechanisms at the surface and in the subsurface of ferrite(left) and carbon-enriched austenite areas (right).


637 © 2011 ISIJ

In the ferrite no decarburization occurs, and the oxidationis by growth of the Fe3O4 layer only, which grows by Fe cat-ions diffusion through the oxide layer. This is compensatedby the formation of vacancies at the oxide/metal interface.The compact Fe3O4 layer shields the metal surface fromdirect contact with the gas atmosphere and the oxygen activ-ity at the Fe3O4/metal becomes so low that Mn and Si, theircontent being locally increased by the oxidation of Fe canreduce the Fe3O4 and form the more stable xMnO.SiO2.Simultaneously with this interfacial reduction, the vacanciesform voids which join up to form large interfacial pores. Inthe reduction stage, the Fe3O4 is reduced to very pure Fe byhydrogen. The larger pores, the isolated embedded MnOparticles and the xMnO.SiO2 are unaffected by the process.

In the austenite phase both the oxidation of iron anddecarburization occur. The oxidation of iron is by thegrowth of a Fe3O4 layer. This initially compact layer shieldsthe carbon from decarburization by the gas atmosphere, andwhen the oxygen activity at the Fe3O4/metal becomes lowenough, carbon, its content being locally increased by theoxidation of Fe, will reduce the Fe3O4 and form CO2 gaswhich create micro-voids and cracks in the oxide layer. Thisexposes the Fe-depleted metal surface to the gas atmosphereand has three important consequences: the oxide layer willappear more voluminous, decarburization will proceed at afaster pace and internal selective oxidation can take place.

In the reduction stage, the porous Fe3O4 is very efficientlyreduced to very pure Fe by hydrogen. The micro-cracks, theembedded MnO particles and the xMnO.SiO2 and SiO2

selective oxides are unaffected by the reduction process.Whereas the oxygen activity is too low and the annealing

times too short for Fe2O3 formation, it is not known whymagnetite rather that Wüstite (FeO) is formed in the pre-oxidation stage. The presence of isolated MnO particlesembedded in Mn-free Fe3O4 surface oxide does howeversupport the oxide analysis made in the present work, as it iswell documented that MnO forms the continuous solidsolution Manganowüstite (Fe1–xMnx)1–ΔO with FeO.26)

The oxidation-reduction process is clearly effective inproducing a very pure Fe surface on CMnSi TRIP steelwithin very short times. This is expected to lead to asubstantial reduction in the number of bare spot defects inthe Zn-coating. Having said this, the formation of largeinterfacial pores by vacancy condensation ordecarburization may eventually result in spallation of thecoating in forming operations.

5. Conclusions

The pre-oxidation of a CMnSi TRIP steel duringintercritical annealing in a N2+0.3%O2 atmosphere and thesubsequent reduction of the surface oxide by a N2+10%H2

atmosphere with a dew point of –40°C were investigated indetail by means of HR-TEM.

The pre-oxidized surface consists of austenite grains

covered with a porous Fe3O4 oxide and ferrite grainscovered with a thin compact Fe3O4 layer. The porous Fe3O4

oxide has large voids due to decarburization. The openoxide structure enables selective oxidation to take place andthis lead to the formation of a grain boundary network ofamorphous a-SiO2 oxides in the metal.

The thin compact Fe3O4 layer is separated from thematrix by large oxide/matrix interfacial voids. The surfaceoxide is connected to the substrate by thin crystalline c-xMnO.SiO2(x>1) and amorphous a-xMnO.SiO2(x<0.9)bridging oxides.

The reduction step of the oxidation-reduction processdoes not affect the MnO, c-xMnO.SiO2(x>1) and a-xMnO.SiO2(x<0.9) oxides and the voids formed during thepre-oxidation process are also still present after reduction.The oxidation-reduction process results in the formation ofa thin surface layer of very pure Fe. This should lead to areduction of the Zn-coating defects on Si-bearing TRIPsteels in HDG lines. The increased risk of coating spallationduring forming operation, resulting from the presence oflarge voids, should however be evaluated.

REFERENCES

1) B. C. De Cooman: Curr. Opin. Solid State Mater. Sci., 8 (2004), 285.2) J. Mahieu, S. Claessens and B. C. De Cooman: Metall. Mater. Trans.

A, 32, (2001), 2905.3) A. R. Marder: Prog. Mater. Sci., 45 (2000), 193.4) X. Vanden Eynde, J. P. Servais and M. Lamberigts: Surf. Int. Anal.,

35 (2003), 1004.5) J. Mahieu, S. Claessens, B. C. De Cooman and F. Goodwin: Proc. 6th

Int. Conf. on Zinc and Zinc Alloy Coated Steel Sheet (GALVATECH2004), ed. by M. A. Baker, Association for Iron and Steel Technol-ogy, Warrendale, PA, (2004), 529.

6) Y. F. Gong, Han S. Kim and B. C. De Cooman: ISIJ Int., 49 (2009),557.

7) Y. F. Gong, Han S. Kim and B. C. De Cooman: ISIJ Int., 48 (2008),1745.

8) X. S. Li, S.-I. Baek, C.-S. Oh, S.-J. Kim and Y.-W. Kim: Scr. Mater.,57 (2007), 113.

9) M. Lamberigts, J.-P. Servais: ECSC Final Report 7210 EC/203, (1994).10) K. Nishimura, T. Ohara, R. Mizoguchi, S. Fuji: European Patent EP

0480122 (B1), (1992).11) T. Matsumoto: CAMP-ISIJ, 4 (1991), 1632.12) R. Nakanishi, H. Irie, M. Nakamura, K. Okamoto and M. Shimizu:

European Patent EP 1829 983 (A1), (2007).13) L. Bordignon and X. Vanden Eynde: European Patent EP 1999 287

(A1), (2009).14) L. Bordignon, X. Vanden Eynde and R. Fransen: Rev. Métall., July–

August (2004), No. 7–8, 559.15) L. Bordignon, R. Fransen, R. Leuschner, W. Warnecke and G.

Angeli: EUR Report Number 22014, (2006).16) D. Geneve, D. Rouxel, P. Pigeat, B. Weber and M. Confente: Appl.

Surf. Sci., 254 (2008), 5348.17) D. Geneve, D. Rouxel, P. Pigeat, B. Weber and M. Confente: Mater.

Sci. Eng. A, 435–436, (2006), 1.18) S. R. J. Saunders, M. Montiero and F. Rizzo, Prog. Mater. Sci., 53

(2008), 775.19) R. Dieckmann and H. Schmalzreid: Ber. BunsenGes. Phys. Chem., 81

(1977), 344.20) S.Tinkler, R. Dieckmann: J. Mater. Sci., 27, (1992), 3799.21) A. K. Galwey, D. M. Jamieson, B. R. Wheeler and R. Reed: Corros.

Sci., 14 (1974), 527.22) I. Wolf and H. J. Grabke: Solid State Commun., 54 (1985), No. 1, 5.23) J. Manenc and G. Vagnard: Corros. Sci., 9 (1969), 857.24) H. J. Engell: Z. Electrochemie, 63 (1959), 835.25) M. C. Pandey: Oxid. Met., 48 (1997), No.1–2, 129.26) P. Franke and R. Dieckmann: J. Phys. Chem. Solids, 51, No. 1, 45.

influence of the oxidation-reduction process on the

Documents