Metastable Chromium-Rich Oxide Formed during Plasma Spraying of High-Alloy Steel

Oxidation of Metals, Vol. 54, Nos. 5/6, 2000 Metastable Chromium-Rich Oxide Formed during Plasma Spraying of High-Alloy Steel J. Leitner,*§ J. Dubský,† J. Had,* F. Hanousek,‡ B. Kolman,† and K. Volenı́k† Received October 18, 1999; revised June 21, 2000 During plasma spraying of metals in air, rapid-oxidation reactions occur, in most cases. In oxidation products of Cr-rich steels, Fe–Cr spinel oxide is often found as a dominant oxide phase. A thermodynamic analysis of a system composed of Fe–13%Cr alloy and water vapor or air showed that the oxidation product in a wide range of high temperatures is Fe3AwCrw O4 (w>2). This tetragonally distorted spinel oxide is not stable at room temperature. Water vapor and air were considered as limits of the gaseous-phase composition in atmospheric spraying by a water-stabilized plasma gun, where the composition of the plasma plume is modified by air entrainment. The equilibrium calculations enabled determination of the effects of temperature and gas-to-solid ratio on w. To show the existence of chromium-rich, tetragonally distorted spinel oxide experimentally, a typical product resulting from oxidation of 13%Cr–steel particles during their flight in the plasma plume was studied after rapid solidification. This was made possible by trapping and quenching the flying particles in liquid nitrogen at a distance from the plasma-gun nozzle corresponding to the nozzle–substrate distance in conventional plasma spraying. The results obtained by X-ray diffraction, Mössbauer spectroscopy, and X-ray fluorescence analysis showed that this oxide, in which w≈2.4, constituted the dominant phase in the oxidation product. KEY WORDS: Plasma spraying; chromium steel; oxidation; Fe–Cr spinel oxide. INTRODUCTION Plasma spraying of high-alloy steels in the ambient atmosphere is accompanied by oxidation reactions resulting in the presence of some *Institute of Chemical Technology, Technická 5, 166 28 Praha 6, Czech Republic. †Institute of Plasma Physics ASCR, Za Slovankou 3, 182 21 Praha 8, Czech Republic. ‡Institute of Inorganic Chemistry ASCR, 250 68 Řež u Prahy, Czech Republic. §To whom all correspondence should be addressed. 549 0030-770X兾00兾1200-0549$18.00兾0  2000 Plenum Publishing Corporation 550 Leitner et al. amounts of oxides in plasma deposits. It has been shown1–3 that in the oxidation products of some plasma-sprayed conventional Cr– and Cr–Ni– steels, Fe–Cr spinel oxides prevail. Chromium-deficient FeCr2O4 has been found mostly in plasma deposits.1,3 FeCr2O4 enriched in chromium has been identified after the rapid particle solidification2 in the early stage of oxidation in flight of molten steel powder particles in the plasma plume. The existence of the chromium-rich oxide, in which the Fe :Cr ratio may even fall to zero, giving rise to Cr3O4 , has been reported as early as in 19554 (‘‘distorted spinel’’) and during the following decade.5–7 However, this oxide is rarely observed, because it is not stable at room temperature. In equilibrium calculations reported,1,8 the composition of spinel-type solid solution Fe3AwCrwO4 has been taken into consideration only in the range 0⁄w⁄2. The present considerations relate to plasma spraying by a water-stabilized plasma gun. The plasma consists of dissociated and ionized water vapor. Strong air entrainment is typical of spraying in the ambient atmosphere. The composition of the plasma plume changes rapidly with distance from the nozzle. This is accompanied by a rapid decrease of plasma temperature. The metallic powder is injected into the plasma plume at a certain distance from the nozzle. It can be assumed that the temperature of molten metallic droplets and of the gas in close vicinity does not exceed the boiling point of steel. In the plume, the equilibrium concentrations of charged species are very low at this temperature and that is the reason, to a good approximation, the gaseous phase can be taken for being composed of neutrals. In general, the Fe–Cr–O system is of great technological interest. It has been studied extensively, particularly with respect to melting, refining, and corrosion of Cr–steels and to the refractory properties of oxides. The thermodynamic properties and phase relationships in the Fe–Cr–O system have been critically assessed by Pelton et al.9 and more recently by Taylor and Dinsdale.10 It follows from these reports that the spinel phase is the principal solid oxide for a wide range of oxidation conditions up to ≈2400 K. The spinel phase can be considered as a solid solution of stoichiometric oxides—Fe3O4 , FeCr2O4 , and Cr3O4 . Taylor and Dinsdale10 have modeled the solid solution using the compound-energy model 11 with four sublattices. The model predicts composition of the spinel phase in the range of xCr G0.67–0.87 in equilibrium with both the metallic alloy phase (solid or liquid) and the corundum-type phase (solid solution of Fe2O3 and Cr2O3) in the temperature range 1200–2000 K. The chromium content in the spinel is xCr GnCr 兾(nCrCnFe )GnCr 兾3; nCr Gw in the formula Fe3AwCrwO4 . The existence of the Cr-rich spinel phase in the temperature range 1873–2098 K Metastable Chromium-Rich Oxide Formed during Plasma Spraying 551 under strongly reducing conditions has been confirmed experimentally by Toker, Darken, and Muan.12 The aims of the present paper are as follows: thermodynamic analysis of the systems Fe–Cr–O–H and Fe–Cr–O–N under the conditions of plasma spraying of Cr–steels; assessment of thermodynamic requirements to form the Cr-rich, distorted-spinel phase; and experimental evidence of the Crrich, distorted-spinel oxide in plasma-sprayed Cr–steel, verifying the results given in Ref. 2. CALCULATION METHOD AND INPUT THERMODYNAMIC DATA A nonstoichiometric method, based on minimization of the total Gibbs energy of the system on a set of points satisfying the material-balance conditions, was used in equilibrium calculations.13 The substances involved in the calculations are listed in Table I. The solid Fe–Cr alloy phases in bcc and fcc structures as well as the metallic liquid phase were described as substitutional solutions. The Redlich–Kister equation14 was used to express the mixing properties. Thermodynamic data of pure elements were taken from SGTE database.15 The Redlich–Kister parameters are due to Andersson and Sundman.16 The solubility of oxygen in the metallic melt was neglected. The oxide melt was modeled as an ideal solution of liquid FeO and CrO1.5 . The data of pure oxides were taken from the JANAF Table.17 A simplified compound-energy model was used for the spinel phase. Three sublattices were considered within the model. The first sublattice is occupied by divalent cations Fe2C and Cr 2C (tetrahedral sites in normal spinel), the second by trivalent Fe3C and Cr 3C cations (octahedral sites in normal spinel), and the third is filled with O2− anions. The spinel stoichiometry can be expressed as [(Fe2+)1Ay (Cr 2C)y ]tetr[(Fe3C)2Az(Cr 3+)z]octO4 Table I. Substances Considered in Equilibrium Calculations Phase Gaseous Liquid metal alloy Oxide melt BCC metal alloy FCC metal alloy Spinel Corundum Wüstite Components Fe, FeO, Cr, CrO, CrO2 , CrO3 , O2 , H2 , H2O, N2 Fe, Cr FeO, CrO1.5 Fe, Cr Fe, Cr Fe3O4 , FeCr2O4 , Cr3O4 Fe2O3 , Cr2O3 Fe0.947O 552 Leitner et al. Within this simplified model, the deviations from metal-to-oxygen stoichiometry and cation distribution in tetrahedral and octahedral sites are not possible. As components, Fe3O4 , FeCr2O4 , and Cr3O4 were chosen. This model enabled variation of the total chromium content xCr in the range of 〈0, 1〉; nevertheless, the ratio of divalent and trivalent cations was limited, the condition y⁄z being valid in all cases. Thermodynamic data of Fe3O4 and FeCr2O4 were taken from Barin;18 Cr3O4 was taken from Taylor and Dinsdale.19 The accepted Gibbs energy values of FeCr2O4 have been recently confirmed by Hino et al.20 The data of Cr3O4 were somewhat modified as described later. To verify the reliability of this model, some calculations were carried out for comparison with the experimental results given by Katsura et al.21 and by Toker et al.12 During these calculations, the Gibbs energy of Cr3O4 was set by 0.5% more negative and the interaction parameter LI I of divalent cations Fe2C and Cr 2C in the tetrahedral-site sublattice was assessed as LI I兾RTG−1 to obtain better agreement with the published data. The parameter LI I I of interaction between trivalent cations Fe3C and Cr 3C in the octahedral-site sublattice was placed at zero, in accordance with Petric and Jacob.22 The same model was applied for the corundumlike phase where two sublattices were considered, but mixing was only limited to one of them. The stoichiometry of the corundumlike phase can be expressed as (Fe1Ay Cry )2O3 . Ideal mixing in the cation sublattice was supposed. Thermodynamic data of Fe2O3 and Cr2O3 were taken from Barin.18 EXPERIMENTAL PROCEDURES Steel ČSN 17 021 (Czech equivalent of AISI 410) was plasma sprayed. The feedstock powder contained (percentage by mass) 13.2 Cr, 1.5 Mn, 0.2 Ni, 0.8 Si. The particle dimensions of feedstock powder were 100– 140 µm. Spraying was conducted by a 160 kW water-stabilized plasma gun WSP  PAL 160. To use spraying conditions different from the cases reported,2 shrouding by protective gas was not applied. To obtain the product of the first (in-flight) oxidation stage, not modified by the second oxidation stage (this starts at the moment of the droplet impact against the substrate), the flying droplets were trapped and quenched in liquid nitrogen. The starting spraying distance (nozzle—liquid nitrogen level) was 370 mm. During the short (4 sec) spraying run, the liquid nitrogen level fell by ≈20 mm. The feeding distance (nozzle—powder injector) was 70 mm. Metastable Chromium-Rich Oxide Formed during Plasma Spraying 553 Fig. 1. Typical oxide crust after metallic phase dissolution. To separate the oxidation product from the steel-powder particles, the metallic phase was dissolved in a solution of iodine in methanol. The procedure has been described elsewhere.1 The character of oxide crusts and their fragments after metallic-phase dissolution is shown in Figs. 1 and 2. The differences of local oxide-layer thickness in various sites on the surface of a single spherule can be seen in Fig. 2. To obtain powder suitable for Xray and spectroscopic measurements, oxide crusts and their fragments were milled in a corundum mortar. The lattice parameters were determined by X-ray diffraction (XRD). To estimate xCr or w, extrapolated values23 were employed. Another method applied for xCr and w determination was Mössbauer spectroscopy, again taking into account the data reported earlier.23 As a direct method enabling measurement of xCr 兾xFe , X-ray fluorescence analysis of the separated oxidation product was used. Approximate results can only be expected because of possible traces of undissolved steel and alien oxide phases in the sample. 554 Leitner et al. Fig. 2. Fragments of oxide crust showing local thickness and details of surface morphology. RESULTS AND DISCUSSION Equilibrium Calculations The equilibrium compositions of the above-mentioned systems were calculated in the temperature range 1500–2500 K at atmospheric pressure for various input compositions. The initial amount of the metallic phase was set at 10 mol, its composition being Fe–13% (by mass) Cr. The initial amounts of the gaseous phase (water vapor or air) were 1–6 mol. Higher amounts of the gaseous phase were not taken into consideration because they would imply consumption of ≈20% of the metallic phase or more. Inflight oxidation of metallic particles is a typical surface reaction, which means that only consumption of a small percentage of the metallic phase can be expected. Figures 3 and 4 show the variation of total chromium content in the spinel phase (xCr ) with temperature. The chromium content in this phase, stable up to ≈2400 K, mostly increases with increasing temperature and decreasing initial amount of gaseous phase (ngo ). At low temperatures, this Metastable Chromium-Rich Oxide Formed during Plasma Spraying 555 Fig. 3. Variation of chromium content in the spinel phase with temperature for a set of initial amounts of gaseous phase (water vapor) in the range of nog G1–6 mol; the initial amount of the metallic phase was 10 mol. phase is stable only at high ngo . From the calculations it follows that oxidation may result in a wide range of spinel-oxide compositions. In water vapor, xCrF0.67 at lower temperatures, which corresponds to cubic spinel. A continuous increase above xCr G0.67 occurs at higher temperatures, accompanied by tetragonal distortion of the lattice. At low ngo , the corundum phase is formed instead of spinel and its composition is close to pure Cr2O3 . At some initial conditions, corundum and spinel phases coexist in equilibrium. The oxide melt, the composition of which is shifted toward FeO, coexists with Fe-rich metallic melt and spinel at high temperatures. Sample Analyses As a rule, the products of oxidation accompanying plasma spraying of high-alloy steels contain some minor phases in addition to the dominant spinel oxide.1–3 To show the composition of chromium-rich, distorted spinel resulting from the in-flight reaction of steel ČSN 17 021, a sample was chosen containing minimum minor phases. Three criteria were used for identifying the chromium-rich, distorted spinel: (1) tetragonal distortion of the cubic spinel lattice, (2) peak separation of the typical doublet in the Mössbauer spectrum, and, (3) xCr 兾xFeH2. 556 Leitner et al. Fig. 4. Variation of chromium content in the spinel phase with temperature for a set of initial amounts of gaseous phase (air) in the range of nog G1–6 mol; the initial amount of the metallic phase was 10 mol. X-Ray Diffraction (XRD) Lattice parameters were determined by XRD: aG8.624B10−10 m and cG7.996B10−10 m. Because of a small number of discernible diffraction lines and to their shapes, the order of magnitude of standard deviations was as high as 0.01B10−10 m. Nevertheless, strong tetragonal distortion of the spinel lattice is obvious. In Ref. (23), the lattice parameters have been plotted against xC r . Some extrapolation of the plots was necessary, giving the approximate value of xCr ≈0.80, which corresponds to w≈2.4. Mössbauer Spectroscopy The oxide yielded a Mössbauer spectrum of the same character as the spectrum shown,2 where acetylene was employed as a shrouding gas. In the strongly quadrupole-split doublet constituting the spectrum, the peak separation was 2.45 mm兾sec. An extrapolation of the Mössbauer data23 is difficult, nevertheless, w ¤ 2.4 can be estimated from this value. X-Ray Fluorescence Analysis . This method gave xCr 兾xFe G3.05; wG 2.3. From the results of the measurements, it follows that the dominant oxide phase in the plasma-sprayed steel sample was tetragonally distorted Metastable Chromium-Rich Oxide Formed during Plasma Spraying 557 spinel oxide, Fe3AwCrw O4 , where the mean value of w was ≈2.4. This value must be taken as an average, because in individual steel spherules, some scatter of w is to be expected due to differing reaction conditions along various trajectories in the plasma plume. Effectively, the presence of alien phases in the sample was negligible. In particular, no traces of the corundum phase were found. It is possible that besides the crystalline oxide, formation of a glassy, chromium-rich oxide phase may result from quenching in liquid nitrogen. However, no diffuse peak originating from a noncrystalline phase was found in the XRD pattern. Even if a glassy oxide phase were present, the conclusions drawn from X-ray fluorescence analysis would only be affected.The XRD and Mössbauer peaks give an unambiguous evidence of the distorted spinel oxide, irrespective of the possible presence of a glassy phase. CONCLUSIONS Equilibrium calculations show that tetragonally distorted spinel oxide Fe3AwCrw O4 (wH2) is stable in a range of high temperatures in air, as well as in water vapor. These conditions correspond to the situation typical of plasma spraying, where chromium-steel particles are oxidized in flight. It can be assumed that the oxidation product is liquid during the major part of the reaction time. As mentioned above, the plasma temperature falls rapidly with distance from the plasma gun nozzle. With respect to the high melting point of the oxide, this may solidify in flight before contact with liquid nitrogen. However, oxide solidification at the first moments after trapping in liquid nitrogen is more probable. In any case, rapid quenching in liquid nitrogen occurs, which maintains the distorted-spinel structure at low temperatures. On the other hand, if molten-steel droplets impact against a substrate to constitute a coating, which is allowed to cool down slowly, further oxidation and diffusion reactions tend to increase the Fe :Cr ratio in the spinel oxide. Finally, as shown,1,3 the oxide is converted into slightly nonstoichiometric, chromium-deficient cubic FeCr2O4 , which is stable at room temperature. The character of the dominant tetragonally distorted spinel oxide was effectively the same as in cases where shrouding was employed.2 The influence of shrouding only seems to consist in decreasing the oxidation rate, not in changing the composition or structure of the oxidation products. ACKNOWLEDGMENTS This work was supported by the Grant Agency of the Czech Republic, Grant GACR 106兾99兾0298. 558 Leitner et al. REFERENCES 1. K. Volenı́k, J. Leitner, F. Hanousek, J. Dubský, and B. Kolman, J. Thermal Spray Technol. 6, 327 (1997). 2. K. Volenı́k, F. Hanousek, P. Chráska, J. Ilavský, and K. Neufuss, Mater. Sci. Eng. A272, 199 (1999). 3. P. Chráska, A. Denoirjean, P. Fauchais, O. Lagnoux, K. Neufuss, O. Schneeweiss, and K. Volenı́k, Proc. 14th Intern. Symp. Plasma Chemistry, Vol. 4, pp. 2019–2024. M. Hrabousky, M. Konrad, and V. Kopecky, Eds. (Institute of Plasma Physics ASCR, Prague, 1999). 4. D. C. Hilty, W. D. Forgeng, and R. L. Folkman, J. Metals AIME Trans. 203, 253 (1955). 5. W. Koch, J. Bruch, and H. Rohde, Arch. Eisenhüttenwes. 31, 279 (1960). 6. A. Hoffmann, Arch. Eisenhüttenwes. 36, 155 (1965). 7. A. Adachi and N. Iwamoto, Trans. Iron Steel Inst. Jpn 6, 188 (1966). 8. J. Leitner and K. Volenı́k, High Temp. Mater. Sci. 35, 139 (1996). 9. A. D. Pelton, H. Schmalzried, and J. Sticher, J. Phys. Chem. Solids 40, 1103 (1979). 10. J. R. Taylor and A. T. Dinsdale, Z. Metallkd. 84, 335 (1993). 11. M. Hillert, B. Jansson, and B. Sundman, Z. Metallkd. 79, 81 (1988). 12. N. Y. Toker, L. S. Darken, and A. Muan, Metall. Trans. B 22B, 689 (1991). 13. P. Voňka and J. Leitner, CALPHAD 19, 25 (1995). 14. O. Redlich and A. T. Kister, Ind. Eng. Chem. 40, 345 (1948). 15. A. T. Dinsdale, CALPHAD 15, 317 (1991). 16. J. O. Andersson and B. Sundman, CALPHAD 11, 83 (1987). 17. M. W. Chase et al., J. Phys. Chem. Ref. Data 14 (Suppl. No. 1, JANAF Thermochem. Tables, 3rd edn.) 1 (1985). 18. I. Barin, Thermochemical Data of Pure Substances, 2nd edn. (VCH, Weinheim, 1993). 19. J. R. Taylor and A. T. Dinsdale, Z. Metallkd. 81, 354 (1990). 20. M. Hino, K. Higuchi, T. Nagasaka, and S. Ban-ya, ISIJ Intern. 34, 739 (1994). 21. T. Katsura, M. Wakihara, S. Hara, and T. Sugihara, J. Solid State Chem. 13, 107 (1975). 22. A Petric and K. T. Jacob, J. Amer. Ceram. Soc. 65, 117 (1982). 23. N. Iwamoto and A. Adachi, Trans. Iron Steel Inst. Jpn 9, 59 (1969).