Atomistic Study of Surfaces and Interfaces of Fe-Cr and Fe-Cr-Al Alloys

Transcription

1 Materials Science Forum Online: ISSN: , Vol. 762, pp doi: / Trans Tech Publications, Switzerland Atomistic Study of Surfaces and Interfaces of Fe-Cr and Fe-Cr-Al Alloys K. Kokko 1,2,a, S. Granroth 1,2,b, M.H. Heinonen 1,2,c, R.E. Perälä 1,2,d, T. Kilpi 1,2,e, E. Kukk 1,2,f, M.P.J. Punkkinen 1,2,g, E. Nurmi 1,2,h, M. Ropo 3,i, A. Kuronen 4,j, and L. Vitos 5,6,7,k 1 Department of Physis and Astronomy, University of Turku, FI Turku, Finland 2 Turku University Centre for Materials and Surfaces, Turku, Finland 3 Department of Physics, Tampere University of Technology, PL 692, FI Tampere, Finland 4 Accelerator Laboratory, P.O. Box 43, FIN University of Helsinki, Finland 5 Applied Materials Physics, Department of Materials Science and Engineering, Royal Institute of Technology, SE Stockholm, Sweden 6 Department of Physics and Astronomy, Division of Materials Theory, Uppsala University, Box 516, SE Uppsala, Sweden 7 Research Institute for Solid State Physics and Optics, Wigner Research Centre for Physics, P.O. Box 49, H-1525 Budapest, Hungary a kalevi.kokko@utu.fi, b sasuma@utu.fi, c markku.heinonen@utu.fi, d riiper@utu.fi, e tmlamb@utu.fi, f edwin.kukk@utu.fi, g marpunk@utu.fi, h eero.nurmi@utu.fi, i matti.ropo@gmail.com, j antti.kuronen@helsinki.fi, k levente.vitos@physics.uu.se Keywords: Fe, Cr, Al, alloys, oxidation, interface, Auger electron spectroscopy, hard x-ray photoelectron spectroscopy, first principles calculations, Monte Carlo simulations Abstract. Surface and interface properties of Fe-Cr, Fe-Al, and Fe-Cr-Al are studied using Exact Muffin-Tin Orbitals and Monte Carlo methods and with x-ray photoelectron and Auger electron techniques. Surface composition is investigated as a function of oxidation (heating) time. Hard x ray photoelectron spectroscopy (HAXPES) is used to scan non destructively the compositions below the surface. It is found that Cr boosts the Al segregation to the surface. Introduction The fundamental physical difference between Fe and Cr plays a decisive role in the thermodynamics of Fe-Cr based alloys: Cr favours an antiferromagnetic coupling to its neighbours whereas Fe prefers a ferromagnetic one. Small amount of Cr, coupled antiferromagnetically to the host matrix, can be dissolved in Fe. However, with increasing Cr content the average distance between, in this case, ferromagnetically coupled Cr atoms decreases inducing increasing of the Cr- Cr pair energy; the energy of a Cr-Cr dimer in Fe increases 240 mev when Cr atoms are brought from a third-neighbour position to a first-neighbour position [1]. Therefore, with increasing Cr content the energy of the homogeneous Fe-Cr alloy is pushed up inducing: peculiar Fe-Cr mixing enthalpy [2], bulk Cr threshold ( 10 at.% Cr in bulk) for a Cr containing surface [3], and α-α phase separation [4]. Chromium in bulk affects also surface properties: within 8 23 at.% Cr in bulk the Cr containing surface has lower surface energy than the pure Fe surface [1] and adding 10 at.% Cr to Fe-Al, the Al concentration can be dropped from 15 to 3 at.% without weakening the protective alumina surface scale [5,6]. In the present work some effects of substitutional Cr in Fe and Fe-Al are investigated using photoelectron and Auger electron measurements and computer simulations. All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of Trans Tech Publications, (# , Pennsylvania State University, University Park, USA-17/09/16,22:38:06)

2 Materials Science Forum Vol Methods The alloy samples were made by induction melting under argon flow from elemental components of purity better than %. The bulk concentrations were checked using Energy Dispersive X-ray analysis. An Fe layer was grown on top of the HAXPES samples (homogeneous Fe-Al and Fe-Cr- Al) by vapour deposition to make the detection of Al diffusion towards the surface easier. More details of the sample preparation and measurements can be found in Refs. [6,7]. The Fe/Cr bilayer was grown by electron beam physical vapour deposition from elemental Fe and Cr on Si substrate. The Fe layers were cleaned by sputtering in the ultra high vacuum HAXPES analyzer chamber to start the investigations with fresh and unoxidized samples. The photoemission spectra were collected using both laboratory x-ray photoelectron spectroscopy (XPS) (PHI ESCA 5400 Electron Spectrometer, Perkin Elmer) with monochromatic Al K radiation and synchrotron radiation excited HAXPES with high kinetic energy experimental station at KMC-1 beamline at Helmholtz-Zentrum Berlin, Bessy II. The adjustable photon energy range (2000 ev ) makes it possible to study photoelectrons with high kinetic energy which increases their inelastic mean free path making the technique bulk sensitive as a comparison to surface sensitive laboratory XPS or soft x-ray range synchrotron radiation. Thus bulk sensitive investigations of atomic concentrations and chemical state are possible without altering the original sample structure or chemistry by sputtering. Also depth-profiling can be done by exploiting different sampling depths by adjusting the photon energy of radiation. Auger electron spectroscopy (AES) was carried out with PHI 610 Auger System. The electronic structure calculations, based on the density functional theory [7,8], were performed using the Exact Muffin-Tin Orbitals method [1,6,9,10,11]. The chemical potential differences of an A-B-C alloy were calculated as μ A -μ B = (1/N)( U/ x A ) x C =constant, where μ, U, N, and x are the chemical potential, internal energy, number of atoms, and atomic fractions of the components, respectively. Larger systems were studied using classical potentials. The two-band embedded atom method potential for Fe-Cr alloy [12] describes the alloy phase diagram well. Particularly, the negative mixing enthalpy at small Cr concentrations is reproduced. A hybrid Monte Carlo (MC) molecular dynamics (MD) method was used to study the concentration profile of an FeCr layer structure. Atomic movements were modelled using the standard MD algorithm with Andersen thermostat [13]. In order to reach equilibrium also in the chemical ordering degree of freedom atom type swaps were applied using the Metropolis algorithm. Simulation system consisted of unit cells ( Å, atoms). In the initial configuration all the chromium was embedded in iron as a layer that corresponded to 10% and 50% of the total number of atoms. The Fe/Cr interface was in the (001) direction of the bcc lattice. The length of the simulation ranged from 8000 to steps with one step consisting of 10 MD steps and swap trials of all possible atom pairs having different types. Periodic boundary conditions were applied and simulations were performed in 300 and 700 K. Results and Discussion Diffusion and segregation in Fe-Cr. The concentration profiles of the double layer sample (Fe/Cr/Si) were monitored using HAXPES at different temperatures as a function of photon energy, time, and oxidation state of the surface. The results are shown in Figure 1. First the quality of the double layer structure was checked by scanning the binding energy range of Cr 2s, 2p and 1s core levels before any annealing was performed. Initially no Cr was found using 2300 ev, 4000 ev and 7300 ev photon energies that correspond to inelastic mean free path values from 2 to 7 nm and thus probing depths of approximately 6-20 nm depending on the core level studied [14]. This indicates the proper double layer structure of the sample, pure Fe layer on top of a Cr layer. Then the sample was annealed at 500 o C for 5, 15 and 50 min. The spectra were measured after each annealing. The gradual Cr diffusion to the surface through the Fe layer was observed with annealing time (Fig. 1a). Even after 5 min annealing at 500 o C at pressure of about mbar the sample was oxidized and the ratio of metallic and oxidized Cr was investigated. The described fast oxidation well demonstrates the difficulties that relate to the studies of the initial oxidation. The appearance of the

3 730 Physical and Numerical Simulation of Materials Processing VII pronounced feature on the high binding energy side of the metallic Cr ( 577 ev binding energy) makes it possible to distinguish the oxidized Cr from the unoxidized one. The inset presented in Fig. 1a nicely demonstrates the versatility of HAXPES by presenting a very rarely, and because of Figure 1a. HAXPES 2s and 2p spectra of Fe/Cr/Si. Inset: Cr 1s spectra before (dashed) and after 50 min annealing at 500 o C. Figure 1b. 2p HAXPES spectra of Fe/Cr/Si after annealing 50 min at 500 o C. the high binding energy with other techniques also difficultly, observed Cr 1s spectra that enable also the surface sensitive HAXPES investigations by low kinetic energy of the photoelectrons. As Fig. 1b shows, with increasing photon energy (i.e. with increasing probing depth) the intensity of the 2p peak of metallic Cr (oxidized Cr) increases (decreases). This shows that oxygen does not penetrate deep into the metal and therefore, oxidizing the surface at low temperatures and consequently observing the binding energy shifts and intensities of different components, can be used as a tool to resolve the surface concentrations of Fe-Cr alloys. More detailed analysis with peak fitting procedure is still needed before more profound conclusions about the oxidation states of Cr can be done. However, by comparing the spectra of bulk Cr and Cr 2 O 3 reference samples to the spectra of Fe/Cr/Si (Fig. 1b), initial conclusions about the oxidation of Cr can be drawn. Figure 1 shows that with annealing time the amount of Cr at the surface (oxidized Cr) increases faster than the amount of Cr in the Fe-layer indicating that there is an upper limit of dissolved Cr in bulk Fe. This is in agreement with previous first-principles calculations [3] and with the present MC-MD calculations for Fe/Cr interface. According to the simulations, in the thermodynamic ground state, part of the chromium of the initial 100% Cr layer is dissolved into the Fe side until the Cr concentration reaches the solubility limit: 6 at.% at 300 K and 9 at.% at 700 K (Fig. 2). The MC-MD simulations are in line with the results of the first-principles calculations and experimental findings. Investigation of the interface between the Cr rich region and Fe rich region revealed that clear faceting was observed with (110) facets. Initial oxidation of Fe-Al and Fe-Cr-Al investigated by x-ray photoemission. The initial oxidation of Fe10Cr10Al and Fe13Al alloys were studied using laboratory XPS. Before oxidation sample surfaces were cleaned by sputtering. To restore the original crystal structure samples were then heated to 700 o C for 5 min. This cycle was repeated until no carbon or oxygen was detectable in the XPS spectra. Oxygen pressure was raised up to 10 6 Torr during sample oxidation. After 1000 Langmuir oxygen exposure at room temperature all metallic components are oxidized suggesting slow atomic diffusion at this temperature range. At 700 o C only Al is oxidized. This can be related to the high oxygen affinity of Al and high atomic diffusion at this temperature. The oxidation rate of Fe10Cr10Al is initially higher than that of Fe13Al, but drops soon below that. This suggests that the protective Al-oxide scale forms rapidly in Fe10Cr10Al whereas Fe13Al is not able to produce any protective Al-oxide scale [9].

4 Materials Science Forum Vol Figure 2. Cr concentration profiles of Fe/Cr interface, MC/MD simulations (curve with points), initial Cr profile (continuous curve). Upper and lower panels: total Cr concentration 50 and 10 at.%, respectively. Vertical scale: atomic fractions. Segregation and oxidation of Al on the surface of Fe10Al, Fe5Cr10Al, and Fe10Cr10Al were investigated using HAXPES. After 60 min annealing in total at 450 o C the Al 1s spectrum was measured using 2500 ev (Fig. 3a), 4000 ev (Fig 3b), and 7500 ev photon energies. The spectra can be fitted with four components as follows: metallic Al, Al rich alloy (binding energy shift ev), amorphous Al oxide (shift +2.1 ev), and γ-alumina oxide (shift ev). Thus the oxide thickness, chemical state of Al and formation of Al rich layer close to the surface can be studied in a detailed way by comparing the intensities of different components of Al 1s spectra observed with different photon energies. The segregation of Al to the surface of Fe10Cr10Al is stronger than that of Fe10Al and Fe5Cr10Al showing that Cr enhances the surface segregation of Al, i.e. Cr improves the high-temperature oxidation resistance of Fe-Al alloys. In these very low oxygen dose HAXPES studies Al was the only oxidized component. Figure 3a. Al 1s HAXPES spectrum of Fe10Cr10Al after annealing at 450 o C for 60 min with photon energy of 2500 ev. Figure 3b. Using the photon energy 4000 ev the bulk component contribution is increased compared to the 2500 ev case.

5 732 Physical and Numerical Simulation of Materials Processing VII Oxidation of Fe-Al and Fe-Cr-Al in ambient atmosphere, Auger electron spectroscopy. The depth profiles of Fe13Al and Fe10Cr10Al (Fig. 4) show that under ambient atmosphere at 1000 o C Figure 4a. AES surface depth profiles of the oxide scale of Fe10Cr10Al oxidized at 1000 o C and 1 atm for 5 min [6]. Figure 4b. AES surface depth profiles of the oxide scale of Fe13Al oxidized at 1000 o C and 1 atm for 1 min [6]. only Fe10Cr10Al forms the protective Al-oxide scale [6] whereas Fe is able to migrate through the oxide formed on Fe13Al. The improved oxidation resistance in Cr-containing Fe-Al alloys can be related to the atomic chemical potentials. According to the electronic structure calculations Cr boosts Al segregation to the surface [15]. A relation, c Cr = /(c Al 0.014) (Fig. 5, dotted curve), for the Cr and Al bulk concentrations specifying the lower limit for the high-temperature corrosion resistance in Fe-Cr-Al was derived from the chemical potentials of Fe and Al [9]. The dotted curve in Fig. 5, based on our first-principles calculations, properly separates the good and poor oxidation resistant Fe-Cr-Al alloys in the experimental surface oxidation map where the 5 g/(m 2 h) oxidation rate curve shows the experimental borderline between corrosion nonresistant/resistant alloys. Figure 5. Fit to experimental oxidation rate of Fe-Cr-Al [6], (continuous curve) and calculated border (c Al,c Cr ) between corrosion resistant and non resistant alloys, (dotted curve). Experimental map of surface oxides of Fe-Cr-Al above 1000 o C (triangles [16], circles, Fig. 4), filled (open) symbols refer to corrosion resistant (non resistant) alloys. Conclusions Annealing the Fe covered Cr at 500 o C induces Cr diffusion through the Fe layer to the surface and under oxygen atmosphere considerable amount of Cr 2 O 3 can be found at the surface. This can be related to the calculated solubility limit of Cr in bulk Fe and the Cr enrichment on the surface of high-cr Fe-Cr alloys. The present first-principles calculations for Fe-Al alloys predict that addition of Cr to Fe-Al alloys enhances Al segregation to the surface. This is a robust phenomenon since it is mainly due to the change in chemical potentials in the bulk side and therefore the phenomenon is

6 Materials Science Forum Vol expected to be effective in different kind of surfaces in various conditions. By relating the computationally found phenomenon with the experimental map of high-temperature surface oxidation of Fe-Cr-Al the experimental and theoretical boundaries between oxidation non resistant and resistant Fe-Cr-Al alloys are found to be in good agreement. References [1] M. Ropo, K. Kokko, E. Airiskallio, M.P.J. Punkkinen, S. Hogmark, J. Kollár, B. Johansson and L. Vitos, Journal of Phys.: Condens. Matter. 23 (2011) [2] P. Olsson, I.A. Abrikosov, L. Vitos and J. Wallenius, J. Nuclear Mater. 321 (2003) 84. [3] M. Ropo, K. Kokko, M.P.J. Punkkinen, S. Hogmark, J. Kollár, B. Johansson and L. Vitos, Phys. Rev. B 76 (2007) (R). [4] J.K. Sahu, U. Krupp, R.N. Ghosh and H.-J. Christ, Mater. Sci. Eng. A 508 (2009) 1. [5] Y. Niu, S. Wang, F. Gao, Z.G. Zhang and F. Gesmundo, Corrosion Sci. 50 (2008 ) 345. [6] E. Airiskallio, E. Nurmi, M. H. Heinonen, I.J. Väyrynen, K. Kokko, M. Ropo, M.P.J. Punkkinen, H. Pitkänen, M. Alatalo, J. Kollár, B. Johansson and L. Vitos, Corrosion Sci. 52 (2010) [7] P. Hohenberg and W. Kohn, Phys. Rev. 136 (1964 ) 64. [8] W. Kohn and L.J. Sham. Phys. Rev. 140 (1965 ) [9] M.H. Heinonen, K. Kokko, M.P.J. Punkkinen, E. Nurmi, J. Kollár and L. Vitos, Oxidation of Metals 76 (2011) 331. [10] L. Vitos, I.A. Abrikosov, and B. Johansson, Phys. Rev. Lett. 87 (2001) [11] L. Vitos, Computational Quantum Mechanics for Materials Engineers: The EMTO Method and Applications. Engineering Materials and Processes Series. Springer-Verlag, London, [12] P. Olsson, J. Wallenius, C. Domain, K. Nordlund and L. Malerba, Phys. Rev. B 72 (2005) [13] H.C. Andersen, J. Chemical Phys. 72 (1980) [14] C.-O.A. Olsson, S. Malmgren, M. Gorgoi, and K. Edström, Electrochemical and Solid-State Letters, 14 (2011) C1. [15] E. Airiskallio, E. Nurmi, M.H. Heinonen, I.J. Väyrynen, K. Kokko, M. Ropo, M.P.J. Punkkinen, H. Pitkänen, M. Alatalo, J. Kollár, B. Johansson and L. Vitos, Phys. Rev. B 81 (2010) [16] P. Tomaszewicz and G. R. Wallwork, Rev. High Temp. Mater. 4 (1978) 75.