Grain Boundary Oxidation and the Chromium- Depletion Theory of Intercrystalline Corrosion of Austenitic Stainless Steels

High temperature microscopy during oxidation of austenitic stainless steels shows that an increased rate of oxidation is found in a narrow zone along grain boundaries where chromium carbides have been precipitated. This lends support to the chromium-depletion theory of intercrystatline corrosion of these steels. Approximate solution of the appropriate diffusion problem shows good agreement with the experimental results. Rbum~L’examen au microscope et a haute temperature du comportement d’aciers inoxydables austenitiques rtvele une vitesse d’oxydation accrue dans une zone ttroite, le long des limites de grains oti les carbures de chrome ont et& precipites. Ceci corrobore Ia theorie de la corrosion intercristalline de ces aciers par appauvrissement en brome. Les resultants experimentaux concordent de man&e satisfaisante avec solution approchee du probltme de diffusion. Zusammenfassung-Mit einem Hochtemperaturmikroskop konnte festgestellt werden, dag die Oxydation von austenitischen Stahlen in einer engen Zone in der Umgebung der Korngrenzen mit erhohter Geschwindigkeit verlluft, werm auf den Korngrenzen Chromkarbide ausgeschieden sind. Dieser Befund stiitzt die Chromverarmungstheorie der interkristallinen Korrosion dieser Legierungen. Naherungslosungen der Diffusionsgleichung ergaben gute Ubereinstimmung mit den experimentellen Ergebnissen. Pe@epaT c llOnlOlL(bIO BbICOKOTenlIIepaTypHOI-0 MKKpOCKOIIa npIl OKIlCJIeHlIM ayCTeIiHTHbIx HepW-lFlelOK(I~X CTaJIefi nOKaaaH0, YTO ITOBbIUleHHaFi CKOpOCTb OKlICJIeHIlR 06Hapj-WlBaeTCR B yaKoR aoxe BAonb rparrmw aepea, me ocamRamTcR KapBanbI xpoMa. MO&HO npeunonoiKMTb, 9TO MeHCKpHCTaJIjIlITHaR KOppO3IlH BTHX CTaJIefi npO~ICXO~IlT no TeOpJIH lICTOIQeHMR xponra. flpn6JIWKeHHOe peILIelIIle COOTBeTCTByIOUJeti ~II~~y3LIOBHO~ npo6nenlbI XOpOUIO cornacyeTcR aKcnepmieHTa~bmnm fiaHHbmn. INTRODUCTION SEVERAL theories have been proposed to account for intercrystalhne corrosion of chromium-nickel stainless steels1 In this paper arguments are put forward, based upon high-temperature microscopy, strengthening the case for the so-called chromiumdepletion theory. It is shown here that this high-temperature method differentiates more easily between the several theories than the usual methods in which the phenomena are studied in an electrolyte solution. *Manuscript received April 1966. 413 414 P. J. GELUNGS and M. A. DE JONGH Basic phenomena2 The general conclusions reached with respect to intercrystalline corrosion of the chromium-nickel austenitic stainless steels can be summarized as follows: (i) In several media (e.g. HNOS, H,SO,, etc.) these steels show a selective attack along the grain boundaries leading to disintegration of the metal after heat treatment in the region 5004350°C. (ii) After this heat treatment microscopic investigation shows a continuous network of chromium carbides along the grain boundaries. (iii) Heat treatment at about 1050°C followed by rapid quenching destroys the sensitivity to intercrystalline corrosion and microscopic examination shows that all carbides have then gone into solution. (iv) Alloying with titanium or niobium leads to the formation of very stable titaniumor niobium-carbides dispersed throughout in the steel; these stabilized steels are insensitive to heat treatment in the above-mentioned region of 500-850°C. (v) Finally, at lower carbon contents the sensitivity is in general lower under otherwise identical circumstances. Theories Of the several theories proposed to explain these phenomena the two main ones are : (a) chromium-depletion theory; (b) electrochemical theory. For a more extensive account of these theories see ref. (1). (a) Chromium-depletion theory. A chromium-nickel austenitic stainless steel quenched in water after heat treatment at 1050-I 100°C is a supersaturated solution of carbon in the Fe-Cr-Ni-austenitic mixed crystal. Annealing at temperatures of about 500-850°C or cooling slowly after heating to higher temperatures lead to the precipitation of carbides. This takes place mainly at the grain boundaries, particularly at the lower temperatures. Because ot this local precipitation the chromium and carbon have to be transported to the grain boundary by diffusion. The diffusion coefficient of carbon is much larger (by a factor of about IO*) than that of chromium. M This is related to the fact that carbon is present interstitially, chromium substitutionally. As a consequence the concentration of chromium in the neighbourhood of the grain boundary after precipitation of the chromium carbides (of composition Cr,,C,) is decreased below that in the bulk of the crystallites. Now the corrosion resistance of these steels is due to their passivity for which, under average conditions, a minimum chromium-content of about 15 % is necessary.2 In the region adjacent to the grain boundary after carbide precipitation, the concentration may be decreased below this value. This then leads to a destruction of the passivity of these zones and hence to intercrystalline corrosion. (b) Electrochemical theory. The precipitation of chromium carbides is again supposed to be the primary cause of the intercrystalline corrosion but, as pointed out by Stickler and Vinckler,6*7 these carbides are more noble than the metal itself and are electronic conductors. This causes galvanic action in which the carbides are the local Grain boundary oxidation and the chromium-depletion theory 415 cathodes and the metal is the anode. Due to the small surface area of the carbides the attack is limited to a narrow zone directly adjacent to the grain boundaries. It is clear that a differentiation between these mechanisms by means of experiments in an electrolyte solution is not possible. Several authors have investigated the composition of the electrolyte after intercrystalline corrosion and have found a decreased chromium content of the solution compared to the metal. a~9 Investigation of fractured surfaces of stainless steels with carbide precipitates also gave indications of a decreased chromium content.‘O The chromium-depletion theory, however, predicts that the oxidation rate at high temperatures will also be greater along the grain boundaries due to the precipitation of chromium carbide which thus lowers the Cr content of the alloy at these areas. Figure 1 shows how the rate of oxidation”s12 increases with decreasing chromiumcontent. % Cr FIG. 1. Rate of oxidation of Fe-Cr-alloys as a function of Cr-content. Observation of the oxidation of stainless steel in which chromium carbides have been precipitated makes it possible to make a choice between these two theories, since the electrochemical theory does not predict an increased oxidation rate along the grain boundaries. Apparatus EXPERIMENTAL All experiments were performed by means of the Vacutherm vacuum heating chamber mounted on a Melabor microscope13 (C. Reichert, Vienna). The specimen was heated in a high vacuum and viewed through a quartz window. The temperature of the specimen was measured by means of a thermocouple inserted in a small hole close to the surface under investigation. The heating chamber could be evacuated to 1O-s Torr and if necessary gases could be admitted during heating. The heating chamber was flushed with argon before evacuation. The argon was of type N48 with as maximum amounts of impurities: H,O 5 ppm, N2 10 ppm, 0,l ppm. 416 P.J. GELLINGS and M. A. DE JONGH Materials Table 1 gives the compositions of the four types of austenitic stainless steel used in this investigation. TABLE 1. PERCENTCOMPOSITIONOFALLOYS AISI AISI AISI AK1 304 347 321 316 C 0.042 0.035 0.048 0.049 Cr 17.7 18.3 18.1 18.6 Ni 9.3 9.6 9.2 9.7 MO 1.85 Nb 0.22 Ti 0.26 The folIowing heat treatments were used: (a) as received. (b) as (a), followed by 20 h at 650°C. (c) 2 h at 105O”C, quenched in water. (d) as (c), followed by 20 h at 650°C. (e) as Cc), followed by 1 h between 850-900°C. (f) as (e), followed by 20 h at 650°C. With regard to these heat treatments the following comments can be made: b, d, f-annealing at 650°C caused maximum precipitation of chromium carbides on the grain boundaries. c, d, e, f-heating at 1050°C causes dissolution of all carbides and quenching in water gave a rate of cooling fast enough to hold all carbon in supersaturated solution. e, f-the stabilized materials (347 and 321) were subjected to a stabilizing heat treatment between 850-900°C to give optimum conditions for precipitation of titanium or niobium carbides. After heat treatment the specimens were ground wet on Carborundum paper (grades 32MOO-60C) and polished with 7 p diamond paste. They were then polished successively with 0.3 p and 0.05 p aluminium oxide. This was followed by final polishing with clean water. The thermal etching was performed in the Vacutherm at the temperature of investigation (65O’C). Due to the extreme sensitivity to slight surface imperfections (scratches and even the deformed zonel* under former scratches) the finest grade of Al,4 had to be used, Furthermore it is necessary to remove all Al,4 which may be embedded in soft spots and particultirly on grain boundaries as otherwise enhanced oxidation may rest&l5 This was the reason for the final polishing with water alone. Oxidation The traces of H,O and 0, present in the argon and leaking into the heating chamber were sufficient to cause a high enough rate of oxidation in most cases if the pressure was kept between about 3 + 1O-5 and IO-* Torr. At higher pressures the oxidation was FIG. 2. A.I.S.I. type 301 after 20 min at 65o’C’. Magnification 300. FIG. 3. A.I.S.I. type 301, heat treatment d, photographed after 60 min at 650-C. Magnification ): 300. FIG. 4. A.I.S.I. type 321, heat treatment e, photographed after 20 min at 650°C. Magnification x 300. FIG. 5. A.I.S.I. type 321, heat treatment f, photographed after 20 min at 650°C. Magnification x 300. FIG. 6. A.I.S.I. type 347, heat treatment a, photographed after IO min at 650 C. Magnification 300. FIG. 7. A.I.S.I. type 347, heat treatment a, photographed after 40 min at 650%. Magnification x 300. Grain boundary oxidation and the chromium-depletion theory 417 too fast, but in some cases where the oxidation was very slow the pressure was raised above lOA Torr for a short time. After reaching the desired vacuum at the start of an experiment the temperature was raised as quickly as possible while maintaining the vacuum. The thermal etching at 650°C took approximately 2-3 min. RESULTS It is well known that the colour changes during the thickening of the oxide layer (interference films) can be used to determine the thickness of the oxide.11s16 These colour changes occur in the order: yellow-brown-blue-grey-yellow (second order) which correspond to thicknesses of about 460-600-720-820-920 A. By observing the colour changes on different specimens it was possible to distinguish three different types of oxidation behaviour. I. Grain boundary oxidation. In this case the grain boundary and its direct surroundings show the thickest oxide layer from the commencement of oxidation. After 5 min they are visible as sharp black lines with a dark brown border on both sides of width approximately 2 x 2 p. After 10 min these borders become blue and somewhat wider while the grain boundary proper remains visible as a black line in the centre. After about 25 min the grains themselves start to change their browncolourinto blue, beginning from the grain boundary. Simultaneously the grain boundary changes its colours from grey to second order yellow. Two stages in this oxidation are shown in Figs. 2 and 3. II. General surface oxidation. Here the whole surface shows a discoloration and the grain boundaries are only visible as lines of demarcation between the differently coloured grains (Figs. 4 and 5) which oxidize at somewhat different rates due to their difference in orientation.” The general change in colour is from brown to blue and grey to second order yellow after about 25 min. In most cases the oxidation was slow in starting and the pressure had to be raised to initiate the oxidation. III. Intermediate oxidation. In this case the oxidation starts at the grain boundaries and scratches, but spreads rather quickly over the whole surface. After about 15 min the grain boundaries have disappeared altogether as is shown in Figs. 6 and 7. In Table 2 the observations on the different materials after heat treatment have been summarized. TABLET. OBSERVA-IIONSONOXIDATION BEHAVIOUR Heat treatment AI.3 AISI AISI AISI 304 347 321 316 I III(H) II I II II LII II II I II Imrr) I III 11OJQ II 11 -