Coking and Dusting of Fe–Ni Alloys in CO–H2–H

ORI GIN AL PA PER Coking and Dusting of Fe–Ni Alloys in CO–H2–H2O Gas Mixtures Jianqiang Zhang Æ David J. Young Received: 10 January 2008 / Published online: 10 July 2008 � Springer Science+Business Media, LLC 2008 Abstract Metal dusting of Fe–Ni alloys was investigated in a CO–H2–H2O–Ar gas corresponding to aC = 19.6 at 650 �C. Thermogravimetric analysis showed that increasing the nickel content in the alloy decreased the initial rate of carbon uptake. A uniform Fe3C scale formed on pure iron, a layer with mixed structures of Fe3C, c and a-Fe developed on ferritic Fe–5Ni, and small amounts of Fe3C developed at the surface of an austenite layer grown on two-phase (a + c) Fe–10Ni. At nickel levels above 10%, no carbide appeared. These observations are shown to be broadly consistent with local equilibrium according to the Fe–Ni–C phase diagram. How- ever, the failure of higher nickel austenitic alloys to form the (Fe,Ni)3C expected at high carbon activities indicates a barrier to nucleation and growth of this phase. Graphite deposition was catalysed by (Fe,Ni)3C on ferritics and by the metal itself on austenitics. The rates of carbon deposition on Fe–60Ni corresponded to the existence of three parallel and independent paths: the synthesis gas, the Boudouard and the carbon methanation reactions. Keywords Fe–Ni � Carburization � Carbide � Carbon filaments � Graphite Introduction Metal dusting is a catastrophic form of high temperature corrosion in which carbon-supersaturated gas degrades iron-, nickel- and cobalt-base alloys into a dust of carbon and particulate metal or metal carbides [1–4]. Two mechanisms J. Zhang � D. J. Young (&) School of Materials Science and Engineering, University of New South Wales, Sydney, NSW 2052, Australia e-mail: d.young@unsw.edu.au J. Zhang e-mail: j.q.zhang@unsw.edu.au 123 Oxid Met (2008) 70:189–211 DOI 10.1007/s11085-008-9115-0 have been proposed according to whether or not metastable carbide forms. For ferritic alloys, this mechanism [2–6] involves the super-saturation of iron with carbon, subsequent formation of cementite at the surface, graphite deposition on the cementite, and carbide decomposition or disintegration. These processes produce fine metal or carbide particles which then strongly catalyse further carbon deposition. A separate mechanism that does not involve carbide formation has been formulated for the metal dusting of nickel and cobalt-based alloys [2, 5–9]. As for iron-based alloys, the process begins with carbon saturation of the alloy. The subsequent graphite precipitation is thought to result in the disintegration of the saturated metal matrix into small metal particles, which catalyse further carbon deposition. On the basis of the above description, it is expected that alloying of iron with nickel should lead to a transformation from one mechanism to the other. This has in fact been confirmed experimentally [2, 10–12]. However, there is disagreement as to the level of nickel required to change the mechanism. Nava Paz and Grabke [2] found no surface cementite on Fe–Ni samples containing 10 and 20 wt% Ni in CO–H2–H2O gases with aC = 5 and 15.5 at 650 �C. Grabke et al. [11, 12], however, reported that a carbide M3C (M = Fe, Ni) was detected by metallography and X-ray analysis on Fe–Ni alloys containing up to 30% Ni, in CO–H2–H2O gases with aC = 1.8–7.5 at 650 �C. Pippel et al. [10] using transmission electron microscopy found that carbide formation took place up to a nickel content of 5 wt% and concluded that none formed at higher nickel levels. At a Ni concentration of 10 wt% and above, direct disintegration of the metal by graphite precipitation took place in the same way as for pure Ni. The variation of dusting rate with Ni/Fe ratio is important in assessing the behaviour of heat resisting steels and nickel base alloys. Grabke et al. [11, 12] showed very clearly that carbon deposition (coking) rates decreased monotonically with increases in nickel levels from 10 to 80 wt% in binary Fe–Ni alloys. In their catalysis research on Fe–Ni alloys, Park et al. [13] found the rate of carbon uptake decreased with increased nickel concentration to a minimum at 60%, but then increased as nickel levels were further increased to 70%. At still higher nickel levels, the rate decreased again. However, it was not clear whether these catalysts were alloys or bimetallic mixtures. This unusual result was not reflected in the metal dusting results of Grabke et al. [11, 12], where mass gains due to carbon uptake on alloys containing more than 80 wt% nickel were too small to be measured with the technique used. There is a need for data at the high Ni/Fe ratios which are found in Inconels and other nickel base alloys. Gas composition is found to affect iron and nickel dusting rates and the resulting carbon deposit morphology. Zhang et al. [7, 14] investigated gas composition effects for both metals, and concluded that the carbon uptake rate increased with pCO, reaching a maximum at about 70 vol%. Three morphologies of carbon were found: an approximately uniform layer, graphite particle clusters, and graphite filaments, the relative amounts varying with gas composition. In the case of nickel, internal graphite precipitation was found along grain boundaries. This internal precipitation was also possible for Fe–Ni alloys containing more than 70% nickel, according to Grabke et al. [11]. 190 Oxid Met (2008) 70:189–211 123 The metal dusting resistance of Fe–Ni alloys is of practical importance since, with the addition of chromium, this system forms the basis of a variety of heat resisting alloys. The aim of the present work was to determine more precisely the effect of nickel concentration on the dusting of binary Fe–Ni alloys. In particular, the behaviour at high Ni/Fe ratios typical of Inconels was examined, along with that of alloys with compositions in and near the a-c two phase region. Experimental Test alloys were prepared from high purity constituent metals, 99.99% Ni and 99.99% Fe, by argon arc melting. Eight alloys were prepared in this manner: Fe– 5Ni, Fe–10Ni, Fe–20Ni, Fe–40Ni, Fe–60Ni, Fe–70Ni, Fe–80Ni, and Fe–90Ni (all compositions in weight percentage). The alloys were then annealed for 24 h at 1,000 �C in flowing varigon gas (5% H2 in Ar), and cut into rectangular samples, 14 9 7 9 1.5 mm. All samples were ground on SiC paper to 1200 grit, polished to 3 lm, and ultrasonically cleaned in acetone before reaction. Dusting reactions were carried out in a vertical tube furnace at 650 �C. The kinetics of carbon uptake were monitored by thermogravimetric analysis (TGA) using a microbalance from which the sample was suspended by silica hooks. After purging the reactor with argon, the desired reaction gas of CO–H2–H2O–Ar was introduced. The H2, CO and Ar flowrates were set using mass flow controllers, and the H2 plus CO mixture passed through saturated LiCl water solution at a temperature adjusted to obtain the required water vapour content [15]. The total gas flowrate was fixed at 400 mL/min to establish a linear gas flow rate of 0.57 m/min in the reaction zone. The total gas pressure was maintained at 1 atm. To investigate the effect of alloy composition on Fe–Ni metal dusting, the reaction gas composition was fixed at 25%CO–25%H2–0.5%H2O–49.5%Ar with a carbon activity of aC = 19.6, calculated according to the reaction: CO þ H2 ¼ H2O þ C ð1Þ as aC ¼ K1 pH2 pCO pH2O ð2Þ where K1 is the equilibrium constant for Reaction (1). To investigate the effects of H2 and CO potentials on reaction, the partial pressure of one component was fixed, and the other was varied by adjusting the fraction of diluent Ar in the gas. The detailed gas compositions and their calculated carbon activities are listed in Tables 1 and 2. The Fe–60Ni alloy was used in this investigation. For all gases used, partial pressures of oxygen were low enough to avoid any iron and nickel oxide formation. The exit gas compositions were analysed for CO, CO2 and CH4, using a nondispersive infrared photometer. The measurement was applied for conditions of a fixed inlet pCO = 0.15 atm, but variable pH2 during the reaction of Fe–70Ni alloy. This low pCO value was selected because of gas analyzer limitations. Oxid Met (2008) 70:189–211 191 123 Reacted samples were carefully removed and subjected to subsequent analyses by scanning electron microscopy (SEM) for surface coke examination, optical metallography for cross-section analysis, X-ray diffraction (XRD) for phase analysis, and transmission electron microscopy (TEM) together with energy dispersive spectroscopy (EDS) for reaction product analysis. The TEM cross- section was prepared using an FEI Nova Nanolab 200 Dualbeam FIB. Before ion milling, a platinum coating was applied locally in-situ to protect the surface. TEM analysis was conducted using a Philips CM200 field emission gun TEM. Results Effect of Alloy Composition Figure 1a shows weight gain kinetics for the alloys in a 25%CO–25%H2– 0.5%H2O–49.5%Ar gas. The initial carbon uptake rates calculated from Fig. 1a are presented in Fig. 1b as a function of alloy composition. In general, increasing the alloy nickel concentration decreased the initial carbon uptake rate. This decrease was approximately linear with nickel concentration up to about 60–70%. Only slight decreases were achieved by further nickel additions. The carbon uptake rate of pure nickel was very small, only about 1% that of pure iron. The results of SEM examination of the carbon deposited on iron are shown in Fig. 2. The low magnification image in Fig. 2a shows a coke layer on the surface, with a thickness of about 25 lm estimated by measuring the section revealed by a crack. Examination of a fracture surface revealed an internal structure of roughly aligned graphite filaments (Fig. 2b). The diameter of the filaments was normally 0.2–0.3 lm, but some fine filaments with diameter about 30 nm were also found, mixed with the predominantly coarse ones (Fig. 2c). Table 1 Gas composition— set I Experimental No. pCO (atm) pH2 (atm) pH2O (atm) pAr (atm) aC 1 0.25 0 0.0025 0.7475 – 2 0.25 0.05 0.003 0.697 6.5 3 0.25 0.25 0.005 0.495 19.6 4 0.25 0.5 0.0075 0.2425 26.2 5 0.25 0.7 0.0095 0.0405 28.9 Table 2 Gas composition— set II Experimental No. pCO (atm) pH2 (atm) pH2O (atm) pAr (atm) aC 1 0.05 0.25 0.003 0.697 6.5 2 0.15 0.25 0.004 0.596 14.7 3 0.25 0.25 0.005 0.497 19.6 4 0.5 0.25 0.0075 0.2425 26.2 5 0.7 0.25 0.0095 0.0405 28.9 192 Oxid Met (2008) 70:189–211 123 Carbon filaments on the surface of the Fe–Ni alloys varied in diameter with alloy composition. As the alloy Ni/Fe ratio was increased, the fraction of coarse filaments was reduced and fine filaments became predominant (Fig. 3). When the nickel content was 40% or more, only fine filaments were observed. The diameter of the filaments was about 20 nm and did not change with further increases in nickel content (Figs. 3d, e, and f). In addition, carbon particle clusters were found on the surface of Fe–Ni alloys and pure nickel (Fig. 4) but not on pure iron. The particle sizes were rather similar 0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40 45 50 Reaction time, h W ei gh t g ai n, m g/ cm 2 Fe Fe-5NiFe-10Ni Fe-40Ni Fe-20Ni Fe-60Ni Fe-70Ni Fe-80Ni Ni Fe-90Ni 0 5 10 15 20 25 30 35 40 45 0 0.5 1 1.5 2 2.5 3 3.5 0 10 20 30 40 50 60 70 80 90 100 N C (s) D C 1/ 2 , 1 0- 8 c m s- 1/ 2 In iti al ra te , m g/ cm 2 h Ni percentage, % (a) (b) 2/1)( CD s CN at aC = 1 α γα+γ Fig. 1 Weight gain kinetics for Fe–Ni alloys in a 25%CO–25%H2–0.5%H2O–49.5%Ar gas at 650 �C (aC = 19.6): (a) weight gain versus reaction time and (b) carbon uptake rate and calculated NC (s)DC 1/2 for austenite in equilibrium with graphite Oxid Met (2008) 70:189–211 193 123 on all alloys, in the range of 0.1–1.5 lm. Some filaments were also found interspersed with the particle clusters (Figs. 4a and d). Examination of the coke by TEM showed details of both graphite particle clusters and filaments (Fig. 5). For iron rich alloys, e.g. Fe–5Ni, 10Ni and 20Ni, coarse filaments with metal-rich particles at their tips or along their length were observed (Figs. 5a–c). For Fe–60Ni, fine carbon nanotubes were observed (Fig. 5d). On all alloys, graphite particle clusters were clearly present, and contained embedded metal-rich particles. Particle compositions analysed by EDS for their metal content are shown in Table 3. For Fe–5Ni, two types of particles were detected: one with low nickel concentrations of 1–3% and the other with high nickel levels of 10–18%, much higher than the nominal alloy content of 5%. Figure 6 shows cementite particles in the coke from Fe–5Ni identified by selective area diffraction. Low nickel content particles were also detected in the coke on Fe–10Ni, but in only a few cases. The majority of particles formed from that alloy had Ni/Fe ratios similar to that of the substrate. Alloys containing more than 10% nickel developed only one type of particle, with compositions similar to those of the parent alloys (Table 3). Figure 7 shows metallographic cross-sections of iron and low nickel alloys after etching with alkaline sodium picrate solution to reveal the presence of Fe3C. A thick coke layer is visible on the surface of all samples. For pure iron, a rather uniform Fe3C layer with a thickness of about 2 lm was observed at the metal surface Fig. 2 Surface morphologies of Fe after 7 h reaction: (a) low magnification image showing coke layer; (b) enlargement of one coke fracture; and (c) high magnification image of filaments, showing metal-rich particles at their tips 194 Oxid Met (2008) 70:189–211 123 beneath the coke (Fig. 7a). The Fe–5Ni alloy developed a surface layer about 10 lm thick with a mixed structure of Fe3C and ferrite and/or austenite (Fig. 7b). Further etching with 2% nital revealed a clear boundary between the product layer and the parent alloy, but was not able to distinguish ferrite and austenite in the layer (Fig. 8a, b). XRD surface analysis detected ferrite and austenite (see Table 4). Further analysis of this layer by TEM was performed on a sample milled parallel to the surface in the middle of the layer (Fig. 9a). The two phases were identified as cementite and surrounding ferrite (Fig. 9b). On the Fe–10Ni alloy, only a small amount of Fe3C was found, and no continuous layer developed (Fig. 7c). Etching with 2% nital revealed (Fig. 8) that a Fig. 3 Comparison of filaments formed after carburization reaction: (a) Fe; (b) Fe–10Ni; (c) Fe–20Ni; (d) Fe–40Ni; (e) Fe–80Ni; and (f) Ni Oxid Met (2008) 70:189–211 195 123 single-phase metal layer had developed to a depth of about 40 lm beneath the surface of this initially two-phase (a + c) alloy. No Fe3C was observed at the surface of alloys containing 20% or more nickel (Fig. 7d). Figure 10 shows cross-sections of alloys with nickel contents C60% after 50 h carburization. Increasing the nickel content reduced the coke thickness. In some samples, relatively large metal inclusions were occluded into the coke (Fig. 10b–d). In the case of pure nickel, internal carbon precipitation was observed at grain boundaries as has been reported elsewhere [7]. This phenomenon was not observed in any of the alloys. Surface reaction products were removed from the alloys and analysed by X-ray diffraction. The results in Table 4 show that in all cases, graphite was present. The only detectable metal-containing phase in the coke formed on pure iron was Fe3C. A mixture of Fe3C, ferrite, and austenite was detected in the coke on Fe–5Ni. Cementite was also detectable, but in only trace amounts in the Fe–10Ni coke, along with larger amounts of austenite. Coke formed on higher nickel alloys contained austenite as the only metal phase. Effect of Gas Composition The effect of gas composition on metal dusting of Fe–60Ni was investigated by fixing either pH2 or pCO but varying other components as shown in Tables 1 and 2. Fig. 4 Graphite particle clusters formed in (a) Ni; (b) Fe–60Ni; (c) Fe–40Ni and (d) Fe–5Ni 196 Oxid Met (2008) 70:189–211 123 Figure 11 shows the weight uptake rate changes accompanying alterations in gas composition during the course of a single experiment. The rates of carbon uptake are presented in Fig. 12 as functions of gas composition. In both cases, the rate of carbon uptake increased with pCO or pH2 up to 0.25 atm. In the case of fixed pH2 ; the rate increased further at higher pCO values. However, in the other case of fixed pCO, further increases in pH2 produced only small changes in the rate. The exhaust gas which exited the reactor was analysed for CO, CO2 and CH4 under reaction conditions of a fixed pCO = 0.15 atm, but varied pH2 : Results are Fig. 5 TEM bright field views of coke morphologies: (a) Fe–5Ni; (b) Fe–10Ni; (c) Fe–20Ni and (d) Fe– 60Ni Oxid Met (2008) 70:189–211 197 123 shown in Table 5. With low pH2 ; no CH4 was detected. Increasing pH2 increased the amount of CH4 but reduced the level of CO2, which was found in all exit gases. Solid carbon was deposited in cooler parts of the reactor as well as on the reacting alloys, preventing quantitative correlation of gas composition changes with measured coking rates. Discussion Carbon uptake was in all cases intimately associated with metal disintegration (dusting). Coke formed on nickel and austenitic Fe–Ni alloys consisted of filaments, clusters of particles, and rather more dense material at the metal surface. As seen in Fig. 5, filaments carry metal nanoparticles at their tips, while the clustered particles carry metal at their cores. Coke filaments grown on ferritic materials similarly carry Fe3C particles (Fig. 6). Denser coke formed at the metal surface under these reaction conditions is shown elsewhere [9, 16] to contain large numbers of cementite nanoparticles in the case of iron, and of nickel when that metal is reacted. Thus the extent of coke formation is strongly correlated with the extent of dusting, Table 3 Compositions (wt%) of particles in the coke Alloy Fe Ni Fe–5Ni 97–99 1–3 82–90 10–18 Fe–10Ni 97–100 0–3 87–94 6–13 Fe–20Ni 77–83 17–23 Fe–40Ni 51–65 35–49 Fe–60Ni 37–43 56–63 Fig. 6 TEM bright field images of cementite particles in the coke of Fe–5Ni alloy (inserts show diffraction patterns of marked cementite particles) 198 Oxid Met (2008) 70:189–211 123 in agreement with earlier observations [11, 17] based on chemical analysis of the complete deposit. Effect of Alloy Composition Metal dusting of iron and nickel proceeds according to different mechanisms, although both involve gas adsorption, carbon dissolution and supersaturation of the near-surface metal. Pure iron forms iron carbide but nickel does not. This observation has been confirmed many times [2–12] and is again corroborated in this work (Figs. 7a and 10e). Furthermore, the present finding that cementite forms on low nickel content alloys, but not at higher levels is in qualitative agreement with the initial reports of Grabke et al. [2, 11]. The transition between the two regions can be understood in terms of the Fe–Ni–C phase diagram shown in Fig. 13. It is assumed that local equilibrium with metastable (Fe,Ni)3C is attained, and a steady-state diffusion description applies. Because the diffusion coefficient of interstitial carbon is much higher than that of substitutional iron–nickel interdiffusion, a reasonable approximation is that only carbon diffuses, and hence the mole fraction ratio NNi/NFe remains unchanged within the reaction zone. The diffusion path for an Fe–5Ni alloy is represented in this way by the line AB in Fig. 13. It corresponds to a Fig. 7 Cross-sections of (a) Fe after 7