Corrosion of Chromium in Glass Melts

Journal of The Electrochemical Society, 153 �3� B121-B127 �2006� B121

Downloa

Corrosion of Chromium in Glass MeltsA. Carton, C. Rapin,z R. Podor, and P. Berthod

Laboratoire de Chimie du Solide Minéral, UMR 7555, Faculté des Sciences et Techniques,Vandoeuvre-Les-Nancy 54 506, France

The corrosion of pure chromium was studied in four molten glasses, using both scanning electron microscopy and electrochemicalmethods to characterize the metal/glass interactions. It is shown that direct immersion of chromium into glass does not allowobtaining a protective oxide scale, even if the glass contains oxidizing species such as FeIII, ZnII, etc. In these conditions, thecorrosion mechanisms vary with temperature and glass composition. When the metal is oxidized in hot air preliminary to glassdiving, the passive state is reached and related to the presence of a Cr2O3 continuous layer at the metal/glass interface. It ismaintained up to a temperature called “depassivation temperature.” This temperature is close to 1160°C in a borosilicate glass andis shifted to the higher values when the glass is enriched into oxidizing species �FeIII�. Coupling both SEM and electrochemicaldata shows that the Cr2O3 layer continuity is ensured by microgalvanic couples occurring between parts where the oxide scale isin contact with the metal substrate and other parts where the oxide scale is peeled off. When the glass temperature is higher thanthe depassivation temperature, a chromium boride CrB layer develops at the metal/glass interface.© 2006 The Electrochemical Society. �DOI: 10.1149/1.2165745� All rights reserved.

Manuscript submitted December 8, 2004; revised manuscript received September 27, 2005.Available electronically February 1, 2006.

0013-4651/2006/153�3�/B121/7/$20.00 © The Electrochemical Society

Most of the alloys and superalloys used in the glass-making in-dustry are chromia-forming materials.1-3 Generally, when a givenmaterial is used for a specific application, the process is neverchanged because it is known that a variation of one parameter candrastically change the metal behavior in molten glass, particularly itscorrosion resistance.4 One of the most important parameters is theglass temperature. Increasing the glass temperature increases its flu-idity and the making of the glass becomes easier. On the contrary,the mechanical properties of the alloy are decreased as well as itscorrosion resistance.5-7

Many works concerning the study of the pure metals behavior inglass melts have been realized since the late 1950s. The most recentworks concern Pt,4,8-11 Mo,9,10,12,13 W,14 Ta,10,15 Nb,16 Ti,17-19 Zr,20

Ni,2,21 Fe, Co, and Cr,2,22 which are proposed to be used in theglass-making industry as raw materials or coatings on ceramics. Theresults of these studies indicate that the behavior of pure metalsimmersed in molten glasses �in terms of corrosion rates and corro-sion layers� varies with glass composition, melt temperature, and themetal potential �in the case of use of the material as a fusion elec-trode�. Di Martino et al.2 have shown that the glass working tem-perature increase can induce important changes in the metal behav-ior versus glass corrosion. As an example, these authors have shownthat when pure chromium rods are dived in a C-type borosilicatemolten glass, there is a corrosion mechanism modification when theglass temperature is increased.

Most of the alloys in contact with molten glasses that are used inthe glass industry are chromia-forming materials. As a consequence,the knowledge of pure chromium corrosion in glass melts will pro-vide useful information concerning the comprehension of chromia-forming alloys in molten glasses. The aims of the present study areto determine the corrosion processes of chromium in four differentborosilicate glasses and to characterize the corrosion mechanismchanges that can occur when the glass temperature is increased.Both immersion tests and in situ electrochemical measurements inglass melts will be carried on.

Materials

Chromium plates: Pure Cr pieces �99.99% purity, Alfa Aesar�were melted using a HF furnace and chromium plates are cut fromthe ingot.

Chromium rods: 5-mm-diam Cr rods �99.5% purity, Goodfellow�are used.

Glasses: Three different glasses were provided by industrial part-ners in the form of 5-mm-diam pellets. Their compositions are re-

z E-mail: [email protected]

address. Redistribution subject to ECS term169.230.243.252ded on 2014-11-24 to IP

ported in Table I. The G-S+ glass is obtained by mixing in an agatemortar 1 wt % Fe2O3 with 99 wt % of the G-S glass and firing themixture at T = 1400°C during 48 h.

Experimental

Immersion tests.— The immersion test procedure was defined inorder to exhibit the metal corrosion by molten glass and to preservethe corrosion layer sequence at the metal/glass interface afterimmersion.15 Metallic samples were quickly introduced in moltenglass directly from room temperature to the working temperature.After the run duration, the sample is removed from the furnace andcooled in air at room temperature. Then, metallographic cross sec-tions perpendicular to the metallic sample are prepared. The layerthickness measurements and determination of corrosion phase com-positions are obtained from the sections observation by optical mi-croscope, scanning electron microscopy �SEM�, and electron probemicroanalysis �EPMA�.

Electrochemical techniques.— The electrochemical methods ap-plied to the study of metal corrosion in glass melts have been ex-tensively described by several authors.2,23-26 The experimental de-vice consists in a furnace in which a Pt-10%Rh crucible �12 cmdiam, H = 13 cm� filled with 1.5 kg of glass is placed in air atmo-sphere. A three-electrode system is composed of a working electrode�5 mm diam, H = 20 mm Cr rod polished to the 2400 grid SiCpaper�, two counter electrodes �25 � 2 � 10 mm platinum plate�,and a self-constructed yttria-stabilized comparison electrode �classi-cally used in molten glass2,26,27�. This apparatus allows the measure-ment of electrochemical parameters such as corrosion potential�noted Ecorr� and polarization resistance �noted Rp� of chromium inthe molten glass. A PAR M263A potentiostat was used to performelectrochemical measurements. Polarization resistances are recordedby polarizing the working electrode from Ecorr − 15 mV to Ecorr+ 15 mV with a 600 mV/h scan rate. Polarization curves are re-corded from Ecorr − 15 mV to 0.4 V with a 600 mV/h scan rate.

In this study, all potentials are given versus yttria-stabilized elec-trodes. Details of this apparatus are given elsewhere.2 Specific de-tails concerning the tests performed as a function of temperature aregiven in the Results section.

SEM.— The SEM investigations were performed with a PhilipsXL30 instrument �the Netherlands� using a 15 kV accelerating volt-age. Element analyses were performed by energy dispersive spec-trometry in order to identify the main constitutive elements of thesamples, using a Kevex Sigma detector �U.S.� equipped with a No-ran analysis system �U.S.�.

) unless CC License in place (see abstract). ecsdl.org/site/terms_uses of use (see

http://ecsdl.org/site/terms_use

B122 Journal of The Electrochemical Society, 153 �3� B121-B127 �2006�B122

Downloa

EPMA.— Precise quantitative determinations of the compositionof the different phases were made using a CAMECA SX50 electronmicroprobe �France�. The determination of the boron or nitrogencontent is realized from the electron probe microanalyses. The pres-ence of boron or nitrogen in the phases is verified by the acquisitionof a complete spectrum using a CAMECA SX100 electron micro-probe �France� equipped with a wavelength dispersion spectrometerspecific for light elements �Be and B�. The weight content percent-age of chromium in the CrBx or CrNx compounds is measured andthe complement to the 100% is assumed to be the correspondingboron �or nitrogen� weight percentage. Atomic ratios are then calcu-lated and the stoichiometry of the compound is determined.

Immersion Tests Results

The immersion tests of pure chromium in glass melts were car-ried out as a function of both temperature and glass composition�Table I�. The metal/glass interface morphologies obtained from thecross sections are reported on Fig. 1a-i. The compositions of thecorrosion products are directly reported on the photomicrographs.

Table I. Glass compositions expressed in oxide weight percent.

Glass SiO2 B2O3 Al2O3 LiO2 Na2O

G-S 64.5 4.5 3.4 / 16.0G-S+ 63.9 4.4 3.4 / 15.8G-Zn 58.8 18.2 4.3 2.6 7.0G-Fe 46.5 18.5 10.0 / 20.0

Figure 1. Metal/glass interface morphologies in three different glasses obta�g, h, i� G-Zn glass.


One can observe that the morphologies of the metal/glass interfacesdepend on the run temperature and on the oxidizing species presentin the molten glass. The processes implicated in the chromium cor-rosion can be explained by referring to the redox couple potentialscale determined in the G-S glass at T = 1050°C.2 This scale can beextrapolated to the temperature domain explored in this study and tothe different used glasses because there is only weak variations ofthe redox couple potential values with temperature and meltcomposition.27,28

G-S glass.— A Cr2O3 layer is spontaneously formed by chro-mium when immersed in the G-S molten glass at T = 1030°C.2 Thismorphology is observed up to 1050°C �Fig. 1a�. When increasingthe immersion temperature, this layer does not remain stable and aCrB scale replaces it. Indeed, the solubility of chromia in moltenglasses increases with increasing the molten glass temperature29,30

and the corrosion of chromium yields to Crn+ species that are dis-solved in the molten glass. In the 1050–1200°C temperature do-main, both Cr2O3 and CrB can be formed and observed at the metal/

O CaO ZnO Fe2O3 ZrO2 MgO SO3

2 7.2 / 0.15 / 3.0 0.12 7.1 / 1.15 / 3.0 0.1

5.2 3.2 / 0.7 / // / 5.0 / / /

t three different temperatures. �a, b, c� G-S glass, �d, e, f� G-Fe glass, and

K2

1.1.//

ined a



B123Journal of The Electrochemical Society, 153 �3� B121-B127 �2006� B123

Downloa

glass interface �Fig. 1b�. When T is higher or equal to 1300°C, onlyboride species are stable �Fig. 1c�. The metal/glass interface mor-phology changes can be related to a change in the redox couples�Table II2,19,26� involved in the chromium corrosion: BIII/B0 reactswith CrII/Cr0 instead of O2/O2−. This phenomenon have been pre-viously reported in the case of Ti,18,19 Nb,16 Ta,15 as well as Zr.20

One can note that silicides can also be formed with CrB, because theredox potentials of both BIII/B0 and SiIV/Si0 are very close in theredox potential scale established in molten glasses.3 All these resultsclearly indicate that changes in the corrosion mechanisms of chro-mium in molten G-S glass occur when the melt temperature in-creases.

G-Fe glass.— In this glass, both O2 and iron species dissolved inthe molten glass can be the oxidizing species according to the redoxpotential values �Table II�. When T = 1200°C, a noncontinuousCr2O3 layer develops at the metal/glass interface while no iron nod-ules are formed �Fig. 1d�. At this temperature, the oxidizing specie isO2. At higher temperature, iron is reduced to the metallic state at thechromium plate contact. The high-temperature diffusion of iron intochromium yields to the formation of Fe-Cr solid solutions �Fig. 1eand f�. When the substrate is completely modified into the Fe-Crsolid solution, the corrosion rates that are determined do not corre-spond to that of pure chromium and it cannot be compared withthose determined at lower temperature. In this case, the corrosionmechanism changes are mostly correlated to substrate compositionmodifications involved by a reaction between the corrosion productsand the studied metal as previously observed in the case oftitanium.18,19

G-Zn glass.— When T = 1200°C, a layer composed by a mix-ture of Cr2O3 and CrB develops at the metal/glass interface �Fig.1g�. The first oxidizing system present in the glass that is involved inthe chromium corrosion is ZnII/Zn0. The reduction of the Zn�II�species into metallic zinc yields to the formation of the Cr2O3 crys-

Figure 2. Polarization resistance curves of pure chromium in the G-S glassat four temperatures.

Table II. Formal potentials of selected redox couples.2,19,26

Redoxcouples

Potential�V�

O0/O2− +0.1CrVI/CrIII +0.03FeIII/FeII −0.41ZnII/Zn0 −0.76CrIII/CrII −0.79FeII/Fe0 −1.05BIII/B0 �−1.2SiIV/Si0 −1.2CrII/Cr0 �−1.2


tals. The boiling temperature of Zn is equal to 904°C, that is to saylower than the experimental temperature. The metallic zinc that isformed during the reaction do not reacts with chromium and evapo-rates after its formation. When the glass is locally impoverished intozinc, there is a change in the oxidizing system involved in the chro-mium corrosion. B�III� present in the glass becomes the oxidizingspecie and the CrB scale develops. At higher temperature, the oxi-dizing system change probably occurs more rapidly and only theCrB crystals are observed at the metal-glass interface �Fig. 1h and i�.The oxidation of the chromium substrate results in the formation ofCr�II� and/or Cr�III� species dissolved in the glass melt.

Corrosion rates.— The corrosion rates of chromium in the dif-ferent glasses were determined from direct thickness measurementsand are summarized in Table III. When plotted in an Arrheniusdiagram, the variations of the corrosion rates versus 1/T are notlinear, indicating that the corrosion layer modifications observed bylooking at the metal-glass morphologies effectively correspond tocorrosion mechanism changes.

Electrochemistry Results

The results of the immersion tests indicate that chromium can bestabilized in the passive or in the active state, depending on thephysico-chemical conditions �glass composition, temperature�. Theeffect of temperature seems to be particularly important because itcan induce changes in the corrosion mechanisms involved in thechromium degradation. The presence of a Cr2O3 scale at the metal-glass interface is a characteristic of the passive state while its ab-sence is characteristic of an active corrosion process. Electrochemi-cal techniques allow the in situ follow-up of the chromium behaviorin molten glasses by continuously measuring corrosion potentials,polarization resistance and recording anodic polarization curves.

Study at a fixed temperature.— The anodic polarization curves�recorded from Ecorr − 20 mV to 0.4 V� of chromium correspondingto the immersion of pure chromium in a G-S glass were recorded ata four different temperatures. The corresponding curves are reportedon Fig. 2. The corrosion potential and polarization resistance valuesdetermined at each temperature are summarized in Table IV. Thecorrosion potential values near −1.3 V indicate that the metal pieceis in the active state when it is directly immersed in molten glass,

Figure 3. Polarization curves of pure chromium in the G-S+ glass at threetemperatures.

Table III. Corrosion rates of chromium determined in the threeglasses, at the three temperatures „�m h−1

….

Glass G-S G-Zn G-Fe

1200°C 50.4 87.9 23.91300°C 28.4 72.8 75.31400°C 34.8 86.3 58




Downloa

whatever the run temperature is. This behavior is confirmed by thelow polarization resistance values that are approximately equal to10 � cm2. The anodic polarization curves of chromium recordedin the G-S glass indicate the ability of the metal to be stabilized inthe passive state when immersed in molten glass. The curves ob-tained at T = 1050, 1100, and 1180°C are characteristic of an activebut passivable material. The initial potential is low �−1.3 V, i.e.,oxidizing species are SiIV and/or BIII�, and an anodic peak is fol-lowed by a wide passivation plateau with relatively low currentdensities �about 0.4–1 mA/cm2�. It means that the corrosion prod-ucts obtained by increasing potential are protective. Preliminarystudy2 has shown that in the passivation domain, chromium iscovered with a Cr2O3 oxide scale. The last part of the polarizationcurve, when E � 0.0 V, shows an increase of the current densitywith the potential increase. This domain is attributed to a transpas-sivation zone. The potential is enough high to oxide the CrIII

of Cr2O3 to CrVI. For E � +0.3 V, the increase of current is as-signed to the oxidation of silicate network to oxygen. The twoshoulders present near −1050 and −500 mV are the result ofiron species reoxidation. FeIII ions of the glass are reduced intoFeII and Fe0 at the electrode when its potentials is low enough:FeIII → FeII for E � −410 mV and FeII → Fe0 for E � −1050 mV.

Different parameters, such as the anodic peak, the current densityon the passivation plateau and the length of the passivation plateauare reported on Table IV. The current densities on the anodic peakincrease with increasing temperature, while the anodic peak maxi-mum is shifted to the higher potentials. This results in the decreaseof the passivation plateau length with increasing temperature.

The same set of experiments was performed in the G-S glassenriched with 1 wt % Fe2O3 �noted G-S+�. Iron was added to theglass in order to determine the effect of an oxidant addition. Theanodic polarization curves obtained in this glass are reported on Fig.

Figure 4. Polarization curves recorded using a Pt electrode in the G-S+ glassat three temperatures.

Table IV. Rp and Ecorr values determined for pure chromium immeCharacteristic values of the anodic peak and passivation plateau.

T�°C�

Rp�� cm2�

Ecorr�mV�

Ano

I�mA/cm2�

G-S 1050 10.0 −1300 6.21100 12.1 −1320 8.21180 12.6 −1320 15.11300 7.4 −1285

G-S+ 1050 15.1 −1320 5.01100 7.8 −1320 8.91180 8.6 −1270 17.1


3. They are similar to those obtained in the G-S glass, indicating thatlimited additions of oxidizing species do not seem to strongly influ-ence the metal behavior in the glass melt. The polarization curveswere recorded using a Pt electrode in the G-S+ glass �Fig. 4�. Thevalue of the diffusion current on the reduction plateau increases withincreasing temperature. Nevertheless, this increase is not sufficientto obtain a current that is higher than the critic current for passiva-tion recorded on the polarization curves. The benefic effect of anoxidant addition was also observed when studying the behavior ofCo-base alloys in industrial glasses where the Fe2O3 content is ashigher as 2 wt %. In this glass, the chromia-former alloys are spon-taneously in the passive state.31

At T = 1300°C, the curve that is recorded in the G-S glass ischaracteristic of a non-passivable metal. The corrosion current lin-early increases when increasing the metal potential. This indicatesthat there is a limit temperature above which chromium cannot bepassived in the G-S glass.

Study as a function of temperature.— A specific experimen-tal method was developed in order to precise the temperature effectupon the passive state stability.

To obtain a passive state, a treatment was realized on chromiumelectrode before immersion in molten glass. This method, calledpreoxidation, consists to develop a homogenous Cr2O3 layer on thesurface electrode by a hot air oxidation. It is realized in the furnacejust above the glass crucible. After preoxidation, long lasting poten-tial recordings indicate that when Cr2O3 layer is stabilized in theglass, the alloy can be maintained in the passive state for more than75 h at 1050°C �Fig. 5�.2

In this study, a chromium electrode was first preoxidized andthen immersed in molten glass at T = 1000°C for 2 h. During 2 h,the stability of the passive state at 1000°C is confirmed by electro-chemical recordings of Ecorr and Rp. Then, the glass temperature isregularly increased from 1000 to 1200°C with a 5°C/h rate and

Figure 5. Ecorr and Rp recordings on pure Cr electrode immersed in moltenG-S glass at T = 1050°C as a function of run duration.

in the G-S and G-S+ glasses at T = 1050, 1100, 1180, and 1300°C.

ak Passivation plateau

E�mV�

I�mA/cm2�

Ebeginning�mV�

Eend�mV�

�E�mV�

−1100 0.3 −800 −40 840−1055 0.2 −800 0 800−960 0.7 −630 15 645

−1130 0.1 −840 10 850−1080 0.7 −875 25 900−950 1.2 −650 15 665

rsed

dic pe




Downloa

from 1200 to 1300°C with a 10°C/h rate �in order to limit thedissolution of the alumina rods of the electrodes at high tempera-ture�. The potentials are recorded during the whole heat treatment,while the polarization resistances are recorded once an hour. Theglass temperature is monitored with a Pt/Pt-Rh10 thermocouple di-rectly immersed in the glass. The evolution of Ecorr and Rp arereported as a function of the glass temperature on Fig. 6a and b. Theglasses used are both G-S and G-S+ glasses.

The curves that were recorded systematically present three dis-tinct parts noted I, II, and III on Fig. 6a. The data are interpretedconsidering that the redox potential of a redox couple does not varymore than 0.2 V in the considered temperature range �1000–1300°C� for a given glass composition. In the first part of the curve,the electrode potential is higher than −100 mV. The oxidizingspecie is the oxygen solved in the glass melt. This potential value, aswell as the polarization resistance value that is equal to 50 � cm2,are characteristic of a passive state. Di Martino et al.2 have estab-lished that the transpassive state �chromium oxidation up to the oxi-dation degree VI� is only observed when the metal potential ishigher than 0 V.

In the second part of the curve, there is a change in the oxidizingsystem: the oxidant is Fe�III� species present in the glass. Chromiumremains in the passive state and its corrosion potential is −600 mV.That is to say in the passivation domain that was previously de-termined by the polarization curves. The polarization resistance val-ues are approximately equal to 600 � cm2: they correspond tolow corrosion rates �approximately 1 mm yr−1 according to Di Mar-tino et al.2�.

When the experimental temperature increases, the corrosion po-tential and the polarization resistance of chromium are dropped to


the low values. In the third part of the curve, which corresponds tothe highest temperatures, the values of the different parameters re-corded are lowered to −1300 mV for the corrosion potential andclose to 10 � cm2 for the polarization resistance. These values arecharacteristic of a material in the active state. In this case, the majoroxidizing species are Si�IV� and B�III�.

The corrosion potential and polarization resistance decreases in-dicate that there is a transition from the passive to the active state. Inthe G-S glass, the transition occurs at T = 1160°C. This temperatureis called the depassivation temperature.

The same experiment was performed in the G-S+ glass. The re-sults are reported in Fig. 6b. The recorded curve presents the samegeneral feature as the one obtained in the G-S glass. The depassiva-tion temperature that is determined from this curve is 1210°C. Thepresence of oxidizing species in the molten glass yields a depassi-vation temperature increase of 50°C.

Observation of the metal/glass interfaces.— In order to com-plete the study, chromium rods were immersed in G-S molten glassfollowing the same heating sequence as the one described previ-ously. The samples are taken out from the glass at, respectively,1030, 1100, 1180, and 1300°C. The photomicrographs of the metalglass morphologies are reported in Fig. 7a-d. The gray zone that islocated between chromium and chromium oxide is a chromium ni-tride �the presence of nitrogen was confirmed by WDS using aSX100 electron microprobe� with Cr2N composition. Their forma-tion is probably due to the reaction between nitrogen solved in themolten glass and chromium. This phenomenon is also observed dur-ing the oxidation of pure chromium in air between 800 and1000°C.32

Figure 6. Ecorr and Rp follow-up of pre-oxidized chromium rods immersedin the G-S glass �a� and in the G-S+ glass�b�, as a function of glasstemperature.




Downloa

When T = 1030°C, the chromia layer that is formed is compactand adherent all along the metal/glass interface �Fig. 7a�. Whensome peeling off is locally observed, there is a development of asecond chromia layer that is always in contact with the metal, en-suring the continuity of the protective oxide scale. This phenomenoncan be explained by the apparition of microgalvanic couples be-tween the peeled off and covered zones that yield the healing of theCr2O3 layer, that is, the preservation of the passive state. This phe-nomenon has already been evidenced.3

When T = 1100°C, the morphology of the chromia layer formedat the metal/glass interface clearly shows an increase of the oxidepeeling off �Fig. 7b�. However, the galvanic coupling phenomenonis always active and ensures the healing of the oxide scale. Chro-mium remains in the passive state.

When T = 1180°C, the chromia layer thickness is equal to15 �m and the layer formed is continuous �Fig. 7c�. However, thislayer presents numerous and large peeling off that is probably in theform of blisters. The thickness and continuity of the Cr2O3 layerdoes not allow the molten glass to penetrate between the metal andthe oxide scale. The zones that are still in contact with the metal arepassive while those that are peeled off are active. This morphologyprobably illustrates the limit case from which the ratio between thepassive and the active surfaces is favoring the spontaneous healingof the chromia layer by galvanic coupling.

When T = 1300°C, no oxide scale is formed at the metal/glassinterface. Chromium borides �with the CrB composition� showing acrenel-like morphology develop �Fig. 7d�. The metal is in the activestate.

The variations of the corrosion potential just before the transitionfrom the passive to the active state are reported in Fig. 8. The mea-


sured variations show potential breakdown followed by potentialincreases. They can be related to the successive chromia layer fail-ures and recoveries due to the healing by galvanic coupling. There isa competition between the failure of the Cr2O3 layer due to thestrains generated by the layer growth and the healing of this oxidescale by galvanic couplings.

Conclusion

When the chromium piece is directly dived into a glass melt, achromia layer can develop between the metal and the glass if the

Figure 7. Metal/glass morphologies ob-served after the immersion of preoxidizedCr rods in molten glass at T = 1000°Cand respectively heated up to 1030 �a�,1100 �b�, 1180 �c�, and 1300°C �d�.

Figure 8. Zoom of the Ecorr and Rp vs temperature near the depassivationtemperature.




Downloa

melt temperature is low enough. This chromia layer can be suffi-ciently compact and thick to ensure the protection of the metal fromfurther corrosion �as in the case of chromium in the G-S glass atT = 1025°C2�. The presence of an oxidizing specie in the glass meltcan enhance the ability of chromium to form a Cr2O3 layer, but thelatter may be insufficiently compact to protect the metal from meltcorrosion. Only a decrease of the corrosion rate is observed �as inthe case of chromium immersed in the G-Fe glass at T = 1200°C�.When the melt temperature is high, the chromium piece is inthe active state when immersed directly in the melt and reactswith oxidizing species present in the glass �formation of Cr-Fesolid solution in the case of chromium immersed in the G-Fe glassat T = 1300 and 1400°C� or with the glass itself �formation ofborides and/or silicides�. Electrochemical measurements performedat different temperatures confirm these observations �Ecorr and Rpvalues recorded in G-S and G-S+ glasses at different temperaturesare characteristic of the active state�. They also indicate thatchromium can be passived up to a limited value of the sampletemperature.

The oxidation treatment of the chromium rods in hot air beforeimmersion in molten glass �called preoxidation� allows the protec-tion of this metal from melt corrosion, by ensuring the developmentof a compact chromia layer at the chromium/glass interface. Thisprotection remains active up to the “depassivation temperature.”When the melt temperature is higher than the “depassivation tem-perature,” the metal reacts with the glass and corrodes, yieldingthe same corrosion features as those observed after the immersionof chromium without the preoxidation treatment in the melt. Asan example, the Cr/glass interface morphology obtained atT = 1300°C after the direct diving of the metal in the melt �seeFig. 1c� is similar to the metal/glass feature reported in Fig. 7d,where the chromium rod was preoxidized before immersion in theglass at T = 1000°C and heated up to 1300°C.

The electrochemical methods used in the metal/glass interactioncharacterization give information concerning the whole sample. Themetal/glass interface observations by scanning electron microscopygive information concerning the contact between the metal and theglass. There is a good accordance between the two data sets. Bothappeared herein as complementary and they clearly show the tem-perature dependence of the Cr behavior in molten glasses, in a ther-modynamic and kinetic point of view. Coupling all these data allowsproposing a corrosion mechanism of the chromium in molten glass.

Universite Henri Poincare assisted in meeting the publication costs of
this article.

References1. P. Steinmetz and C. Rapin, Mater. Sci. Forum, 369–372, 865 �2001�.2. J. Di Martino, C. Rapin, P. Berthod, R. Podor, and P. Steinmetz, Corros. Sci., 46,

1849 �2004�.3. J. Di Martino, C. Rapin, P. Berthod, R. Podor, and P. Steinmetz, Corros. Sci., 46,

1865 �2004�.4. C. Rapin, M. Vilasi, R. Podor, A. Carton, B. Gaillard-Allemand, P. Berthod, and P.

Steinmetz, Mater. Sci. Forum, 461–464, 1125 �2004�.5. J. Di Martino, S. Michon, L. Aranda, P. Berthod, R. Podor, and C. Rapin, Ann.

Chim. Sci. Mat., 28, S231 �2003�.6. S. Michon, P. Berthod, L. Aranda, C. Rapin, R. Podor, and P. Steinmetz,

CALPHAD: Comput. Coupling Phase Diagrams Thermochem., 27, 289 �2003�.7. P. Berthod, S. Michon, J. Di Martino, S. Mathieu, S. Noel, R. Podor, and C. Rapin,

CALPHAD: Comput. Coupling Phase Diagrams Thermochem., 27, 279 �2003�.8. E. M. Bakrim, K. El Mortajine, and J. P. Hilger, Ann. Chim. (Paris), 16, 519

�1991�.9. J. K. Higgins, J. Electrochem. Soc., 140, 3436 �1993�.

10. S. K. Sundaram, Mater. Res. Soc. Symp. Proc., 500, 309 �1998�.11. B. Fischer, D. Freund, A. Behrends, D. F. Lupton, and J. Merker, Erzmetall, 51,

429 �1998�.12. S. K. Sundaram, S. Kamakshi, and E. Robert, Mater. Res. Soc. Symp. Proc., 322,

273 �1994�.13. S. K. Sundaram, S. Kamakshi, and E. Robert, J. Am. Ceram. Soc., 79, 1851 �1996�.14. G. Lautenschlaeger, K. Schneider, A. Morstein, L. Dursch, and S. Meaman, Jpn.

Kokai Tokkyo Koho, pat., 5 �2003�.15. R. Podor, C. Rapin, N. David, and S. Mathieu, J. Electrochem. Soc., 151, B661

�2004�.16. R. Podor, N. David, C. Rapin, and P. Berthod, Ceram. Trans., 141, 207 �2004�.17. R. Kirsch, Glass Sci. Technol. (Offenbach, Ger.), 74, 166 �2001�.18. R. Podor, C. Rapin, N. David, and P. Berthod, Glass Sci. Technol. (Offenbach,

Ger.), 77, 36 �2004�.19. C. Rapin, R. Podor, S. Michon, P. Berthod, and S. Mathieu, Glass Sci. Technol.

(Offenbach, Ger.), 77, 88 �2004�.20. R. Podor, N. David, C. Rapin, M. Vilasi, and P. Berthod, J. Solid State Chem.,

submitted for publication.21. B. Gaillard-Allemand, A. Littner, R. Podor, C. Rapin, P. Steinmetz, and M. Vilasi,

Glass Odyssey, 6th ESG Conference, Montpellier �France� �2002�.22. J. Di Martino, C. Rapin, P. Berthod, R. Podor, and P. Steinmetz, 15th International

Corrosion Congress, Frontiers in Corrosion Science and Technology, Grenada�Spain� �2002� �CD-ROM�.

23. B. Tremillon, in Electrochimie Analytique et Réactions en Solution, Tome 1, Mas-son, Paris �1993�.

24. F. Baucke, in Sensors: A Comprehensive Survey, W. Göpel, Editor, Vol. 3�2�, p.1155, VCH Publishers �1992�.

25. F. Baucke, Glastech. Ber., 61�4�, 87 �1988�.26. D. Lizarazu, P. Steinmetz, and J. L. Bernard, Mater. Sci. Forum, 251–254, 709

�1997�.27. O. Pinet and C. Di Nardo, Ceram. Trans., 109, 367 �2000�.28. C. Russel, and G. Von der Gonna, J. Non-Cryst. Solids, 260, 147 �1999�.29. S. Abdelouhab, M. Toplis, R. Podor, C. Rapin, and P. Berthod, J. Non-Cryst.

Solids, Submitted.30. C. W. Kim, K. Choi, J. K. Park, S. W. Shin, and M. J. Song, J. Am. Ceram. Soc.,

84, 2987 �2001�.31. S. Michon, Thesis, University H. Poincaré �2004�.32. H. Buscail, Y. P. Jacob, M. F. Stroosnidjer, E. Caudron, R. Cueff, F. Rabaste, and S.

Perrier, Mater. Sci. Forum, 461–464, 93 �2004�.



Documents

Corrosion of Chromium in Glass Melts