Phase transformations in iron oxide–metal systems during intensive plastic deformation

Phase transformations in iron oxide–metal systems during intensive plastic deformation

Materials Science and Engineering A361 (2003) 136–146 Phase transformations in iron oxide–metal systems during intensive plastic deformation V.A. Sha...

303KB Sizes 0 Downloads 16 Views

Materials Science and Engineering A361 (2003) 136–146

Phase transformations in iron oxide–metal systems during intensive plastic deformation V.A. Shabashov∗ , A.V. Litvinov, A.G. Mukoseev, V.V. Sagaradze, D.V. Desyatkov, V.P. Pilyugin, I.V. Sagaradze, N.F. Vildanova Institute of Metal Physics, Ural Branch RAS, 18 S. Kovalesvskaya Street, Ekaterinburg 620219, Russia Received 21 January 2003; received in revised form 10 June 2003

Abstract Schemes of the dynamic dissolution of iron oxides in metallic Me (Fe, Ni, Cr, Ti, Al) matrices have been determined using Mössbauer spectroscopic, X-ray diffraction, and transmission electron microscopic methods. It was shown that phase transformations of the hematite and the magnetite under compression shear (CS) at room temperature were accompanied by appearance of cation-deficient oxides and metallic solid solutions. The dissolution process depended on the ability of the metallic matrices to form solid solutions and chemical compounds with iron and oxygen. New structural forms of oxides, which presumably contained a high oxygen concentration, were found. It was hypothesised that oxide dispersion strengthened alloys could be prepared by CS of a mixture of iron oxides and metals. © 2003 Elsevier B.V. All rights reserved. Keywords: Iron oxides; Deformation; Pressure; Mechanical alloying; Mössbauer spectroscopy

1. Introduction Applications relating to intensive mechanical treatments, such as rolling, ball milling, friction, extrusion, explosion, and compression shear (CS), are focused on nonequilibrium phase transformations of the martensitic type, dynamic dissolution, amorphization, etc. Mechanical alloying or mechanical synthesis, as a method for production of metallic solid solutions and chemical compounds during mechanical activation of a mixture of powders or phases, also refers to nonequilibrium phase transformations and may take place at temperatures close to room temperature or lower. Volume diffusion is hampered or absent at these temperatures. However, intensive cold deformation provides transport of atoms to distances much larger than the interatomic spacing. Although mechanisms of structural and phase transformations during intensive plastic deformation are complicated, certain progress has been achieved recently in understanding the origin of the low-temperature mechanical transport of atoms. This has been largely due to advancement of methods of mechanical activation with



Corresponding author. Tel.: +7-343-249-9338; fax: + 7-343-274-0003. E-mail address: [email protected] (V.A. Shabashov).

0921-5093/$ – see front matter © 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0921-5093(03)00493-3

controlled treatment conditions, such as CS in Bridgman anvils, and the use of resonance methods for analysis of materials at a local level of its nearest atomic neighbors [1–10]. Studies concerned with strain-induced phase transformations in oxides as a basis of a new class of materials, such as oxide dispersion strengthened (ODS) alloys, or permanent magnetic materials, etc. currently present a great scientific and practical interest. Specifically, strain-induced dissolution of the hematite in ball mills was studied [1–5]. General evidence in favor of the decomposition of oxides under CS was adduced [6]. The researchers [1–3,6] concentrated on a physical model of the oxide decomposition, while the authors [4,5] paid special attention to chemical aspects when they analyzed the effect of the deformation medium. Results of CS experiments [7–10] differed considerably from those obtained in ball mills primarily in predominant dissolution of interstitial phases and formation of carbon solid solutions in the metallic matrix of iron and its alloys. Earlier, we showed the possibility of strain-induced dissolution of intermetallics and interstitial phases (such as carbides, etc.) in metal matrices, which was followed by the appearance of substitutional and interstitial solid solutions [7–10]. The experiments revealed, in addition to solid solutions, new structures of ε- and Hägg-carbides [10].

V.A. Shabashov et al. / Materials Science and Engineering A361 (2003) 136–146

A specific feature of the CS method is the possibility to produce a strong plastic deformation in bulky samples without disrupting the structural continuity. One more important condition of CS experiments is the invariance of the affected volume, in which it is possible to control the sample composition and form supersaturated nanocrystalline solid solutions. About 10% or more of the sample material may be carried to the grinding tool and vice versa during mechanical activation in the mills [5]. This carry-over is selective and is facilitated, because dispersed powders have a developed surface. Thermally stable dispersion strengthened alloys, including ODS alloys, usually are prepared by crushing of coarse particles, e.g. Y2 O3 , TiO2 , etc. their compaction, and sintering in metal matrices. This method suggests a large number of impurities and other uncontrollable factors. In the CS method, oxygen ions can be transported from unstable iron oxides to metals without leaving the volume of the sample. It is reasonable to study the strain-induced redistribution of oxygen in iron and matrices of other metals, which is followed by the appearance of solid solutions and new oxides producing a set of properties inherent in ODS alloys. The goal of this study was to analyze the possibility of the strain-induced dissolution of oxides in the iron oxide–metal system and the formation of supersaturated solid solutions of oxygen under CS. In addition, it was interesting to see how the physicochemical properties of metallic matrices, which determine the deformation structure, solubility and chemical affinity for oxide components, influence the phase transformations at hand.

2. Materials and methods Powders of a standard hematite (␣-Fe2 O3 ) having a ¯ with particles 20 ␮m in size corundum-like structure (R3c) on the average and a synthesized Fe3−y O4 magnetite having ¯ the structure of a nonstoichiometric inverse spinel (Fd3m) served as the oxide component. The Fe3−y O4 spinel was produced by mechanical activation of the hematite in a mill and annealing at 1075 K. The initial nonstoichiometric spinel had the formula Fe3−y O4 with v < 0.15 [5]. The metallic matrices were represented by powders of 3d-metals Fe, Ni, Cr and Ti, and p-metal Al. The oxides were mixed with powders of the metallic matrices, which were prepared

137

by filing of bulky samples, to obtain a mixture of 60 mass% Me + 40 mass% ␣-Fe2 O3 (or Fe3−y O4 ). The samples were filed using metal broach files with a hardness >60 HRC. The hardness of the majority of the initial metallic matrices did not exceed 6–7 HRC. Possible contamination was checked in a JCXA-733 X-ray analyzer having the lower impurity detection limit of 0.05%. Then powders were separated according to their particle size in a mechanical sieve. Coarse powders with an average characteristic size of particles equal to ∼200 ␮m were obtained after filing and separation. The iron matrix was a specially prepared 56 Fe isotope containing a certified amount of 0.2% resonance 57 Fe. The natural iron present in the initial hematite and magnetite included ∼3–4% resonance 57 Fe. Use of the special 56 Fe isotope allowed tracing the movement of iron atoms from the initial oxides during phase transformations. A surface-oxidized fine powder of 56 Fe with particles ∼50 ␮m in size was used in addition to the freshly filed coarse powder of 56 Fe. The metallic matrices were chosen considering their different ability to form metallic solid solutions with iron and oxygen present in the initial oxides. One more criterion was the ability of the metallic matrices to oxidize, i.e. their chemical affinity for oxygen. The mechanical synthesis may largely depend on the mechanism of plastic deformation of the metallic matrices and the solid solutions being formed. One of the preconditions for the mechanical synthesis is, in particular, the microhardness ratio [11] and the lattice type [12,13] of the synthesized components. Table 1 specifies the phase composition, the crystalline structure, and the microhardness [14] of metallic matrices with grains of different sizes. Over a large range of compositions and temperatures, Ni and Cr produce a continuous series of solid solutions with Fe and form solid solutions during the CS mechanical synthesis [12,13]. Iron dissolves little in titanium and aluminum and forms intermetallic compounds. Experiments on the mechanical synthesis of the Ti–Fe and Al–Fe systems by the CS method and mechanical activation in ball mills showed that the corresponding solid solutions with a high concentration of intermetallic-unbound iron and a maximum of 1 mass% of intermetallic-unbound iron can be formed in the first [15] and second [16] systems, respectively. Data on equilibrium solubility of oxygen in the metallic matrices Fe, Ni, Cr, Ti and Al [17] put the titanium first and aluminum last, in which solubility is almost zero. Aluminum has an advantage over

Table 1 Crystal structure and microhardness of the metal matrices Metal matrix

Crystalline structure

Variation of grain size d, nm

Variation of microhardness H, GPa

Al Fe Ni Ti Cr

FCC ␣-BCC FCC ␣-HCP ␣-BCC

1500–200 100 000–80 1500–15 50 000–40 1500–70

0.3–1.2 1.8–4.5 1.5–6.2 0.8–6.7 1.8–5.3

138

V.A. Shabashov et al. / Materials Science and Engineering A361 (2003) 136–146

the other metals in the oxidizing ability since it forms a thin surface oxide layer already at room temperature [17]. The mixture was deformed intensively under a shear pressure of 6 GPa in Bridgman anvils, which were made of a sintered WC carbide, at 293 K, the speed ␻∼1 rpm, and the turning angle n×2␲ (n = 5,. . . , 10, n being the number of turns) using the technique described elsewhere [7–10]. A radial flow of the sample’s material from under the anvils was observed in the first 0.5 turn during compression and shear deformation of the sample. Thickness of the sample was stabilized thanks to ‘locking’ of the high-pressure region under further torsional strain. In the CS method, the anvils are displaced radially thanks to the deformation of the structure in the bulk of the sample and, therefore, the sample and the anvils do not exchange their materials. The prepared sample was a bright disk about 80-␮m-thick and 5–6 mm in diameter. The disk was thinned on both sides to 30 ␮m. The true deformation was estimated from the formula ε = ln(ϕr/d), where ϕ is the turning angle, r is the radius (the middle of the peripheral region under study), and d is the thickness of the deformed sample. The anvils were turned through 5–10 revolutions, which corresponded to the true deformation ε = 6,. . . , 7. The strain-synthesized samples were examined using the Mössbauer, electron microscopic, and X-ray analysis methods. The Mössbauer spectra were measured in the geometry for transmission of 14.4-keV gamma-quanta from a 57 Co(Cr) source under a constant acceleration regime. A special software package ‘ms Tools’ [8] providing reconstruction of the density function of the Mössbauer parameters was used to improve the Mössbauer spectrum resolution. The density function of the centers of gravity of the resonance absorption lines, P(V), on the Doppler velocity scale was reconstructed without imposing any a priori constraints on the Mössbauer parameters. The P(V) function was approximated by a linear combination of modified Gaussian lines to determine individual distributions corresponding to positions of iron in the oxides and the alloys. Considering the diversity of the oxide phases, possible distortions caused by nonstoichiometry, and the complexity involved in correlation of the Mössbauer parameters (the isomeric shift IS, the quadrupole splitting QS, the effective magnetic field H, and the integral intensity S), this calculation procedure is the most efficient for both an independent evaluation of the phase composition and subsequent modeling on the basis of a standard approximation of the spectra by a superposition of Lorentzian absorption lines. The microstructure was examined in a JEM-200CX transmission electron microscope. A DRON-4-07 installation was used for the X-ray diffraction analysis.

3. Experimental results In the first place, experiments with initial ␣-Fe2 O3 and Fe3−y O4 samples without metallic matrices were performed

in order to elucidate the effect of a strong CS deformation on the structure of iron oxides. An examination by the Mössbauer spectroscopic method showed, Fig. 1 (the spectrum a and the corresponding distribution P(V)a ), that CS changed little the hematite spectrum: H decreased slightly from 516 to 505 kOe and the line widened considerably. A small concentration (about 5%) of the component B [18], which corresponded to iron cations with the effective valence of + 2.5 in octahedral voids of the Fe3−y O4 inverse spinel, was remarkable. An X-ray diffraction analysis revealed a considerable widening of the line, which confirmed defectness and refinement of the ␣-Fe2 O3 structure, which was followed by appearance of a small concentration of Fe3−y O4 . The spectrum of the initial magnetite (a–a , Fig. 2) reflected a nonstoichiometry of the spinel structure containing a small concentration (5%) of the ␣-Fe2 O3 structure. The Mössbauer spectrum (see b–b , Fig. 2) of the magnetite changed considerably after CS. A group of lines, which was characteristic of the structure of a nonstoichiometric spinel [2,18], appeared. The component C might be due to a strong nonstoichiometry, dispersion of the structure, and the manifestation of relaxation properties [18,19]. Moreover, a wustite doublet and up to 5% metallic iron were prominent in the spectrum. Dissolution of the ␣-Fe2 O3 and Fe3−y O4 oxides was intensified to a large extent when they were deformed in a mixture with Fe, Ni, Cr and Ti, and to a small extent in a mixture with Al (Table 2). Figs. 1 and 2 and Tables 2 and 3 give results of Mössbauer measurements in mixtures of ␣-Fe2 O3 and Fe3−y O4 with 56 Fe before and after CS. Fig. 3 and Table 4 present Mössbauer observations of the shear in mixtures of ␣-Fe2 O3 with Ni, Cr, Ti and Al. The intensity (S) of the ␣-Fe sextet in the integral spectrum of the deformed ␣-Fe2 O3 + 56 Fe mixture was about 36%. Considering a small concentration of 56 Fe in the initial 56 Fe matrix, about 26% of metallic ␣-Fe was reduced from the initial oxide. Moreover, the spectrum contained sextets of a cation-deficient spinel and a doublet of the Fe1−x O structure. Results of the experiments with 56 Fe in different initial states, namely, freshly filed iron and a surface-oxidized fine powder, differed little. In the first case, the hematite turned through a phase transformation to iron in the metallic state (see c–c and d–d in Fig. 2). The spectrum included a doublet of the ␣ + ␥-Fe2 O3 boundary phase [19,20]. The concentration of this component increased after additional refinement of the powder. In the experiment with dispersed iron the process terminated in formation of the wustite (see d–d Fig. 2). Probably, oxide phases on the developed surface of dispersed iron passivated the dissolution of the hematite at room temperature. Results of the Mössbauer phase analysis were confirmed by an X-ray examination (see Fig. 4). A TEM examination confirmed the hematite dissolution, which was accompanied by appearance of a spinel and a wustite. Fig. 5 presents a bright-field image (a) and two dark-field images (b and c). The first image was obtained in reflections of the first ring formed basically by [1 1 0] reflections from ␣-Fe grains 90 nm in size. The second image

V.A. Shabashov et al. / Materials Science and Engineering A361 (2003) 136–146

139

Fig. 1. Mössbauer spectra, a–d, and the corresponding distribution probability functions of the centers of gravity of the resonance lines on the velocity scale P(V), a –d , for ␣-Fe2 O3 (a) and 56 Fe + ␣-Fe2 O3 (b–d). ␣-56 Fe is represented by a freshly filed powder. Treatment: a–a —␣-Fe2 O3 after CS at n = 5 and P = 6 GPa; b–b —initial 56 Fe + ␣-Fe2 O3 mixture; c–c —after CS at n = 10 and P = 6 GPa; d–d —treatment ‘c’ and grinding. Table 2 Parameters of the phase spectra of the initial magnetite and results of the post-CS synthesis of the mixture with 60% filed iron and (2) surface-oxidized dispersed powder No.

Parameters of Mössbauer spectra

Initial mixture of ␣-Fe2 O3 + 60% 56 Fea ␣-Fe2 O3

Synthesis of ␣-Fe2 O3 +60% ␣-Fe2 O3 + Fe3−y O4 (A/B)b

56 Fea

in the form of (1) freshly

56 Fe

Fe1−x O

Fe–mO A /C

␣-Fea

1

IS, mm s−1 (±0.01) QS, mm s−1 (±0.02) H, kOe, (±5) S, % (±2)

0.34 −0.14 516 90

0.32/0.80 −0.12/0.10 505/464 42/2

0.9 0.54 0 8

−0.78/0.25 0.00/−0.21 498/387 6/4

−0.01 0 329 26

2

IS, mm s−1 (±0.01) QS, mm s−1 (±0.02) H, kOe, (±5) S, % (±2)

0.34 −0.14 516 90

0.33/0.79 −0.12/0.20 505/464 60/2

0.91 0.54 0 18

−0.78/0.25 0.00/−0.21 498/387 4/6

– – – 0

The phase analysis is given with respect to the concentration of 57 Fe resonance iron present in the ␣-Fe2 O3 hematite minus the concentration of in the initial 56 Fe matrix. b The ␥, ␣-Fe O and Fe3 + phases at tetrahedral sites of the inverse Fe 2 3 3−y O4 spinel were not resolved and were described by the subspectrum A; the subspectrum B relates to Fe2 + and Fe3 + at octahedral sites of the spinel. a

57 Fe

140

V.A. Shabashov et al. / Materials Science and Engineering A361 (2003) 136–146

Fig. 2. Mössbauer spectra, a–d, and the corresponding distribution probability functions of the centers of gravity of the resonance lines on the velocity scale P(V), a –d , for the Fe3−y O4 spinel (a and b), 56 Fe + Fe3−y O4 (c), and 56 Fe + ␣-Fe2 O3 (d). ␣-56 Fe is represented by a surface-oxidized dispersed powder. Treatment: a–a —initial Fe3−y O4 powder; b–b —Fe3−y O4 after CS at n = 10 and P = 6 GPa; c–c —56 Fe + Fe3−y O4 after CS at n = 10 and P = 6 GPa; d–d —56 Fe+␣-Fe2 O3 after CS at n = 10 and P = 6 GPa.

was obtained in reflections of oxides 3–10 nm in size located inside the first ring. They numbered over two dozens. A similar number of oxide reflections was between [1 1 0] and [2 0 0] ␣-Fe rings. The presence of the wustite in the structure was unquestionable since all its strong reflections and those with interplanar distances over 5 Å were seen. The

hematite and the spinel did not have reflections with so large interplanar distances. The intensity and interplanar distances of the other observed reflections from oxides coincided with tabulated data for the spinel. The dark-field image (Fig. 5b) may contain, in addition to ␣-Fe grains, oxide grains since some of their reflections coincide with ␣-Fe reflections (val-

Table 3 Parameters of the phase spectra of the initial Fe3−y O4 spinel and results of the post-CS synthesis of the mixture with 60% Parameters of Mössbauer spectra

IS, mm s−1 (±0.01) QS, mm s−1 (±0.02) H, kOe, (±5) S, % (±2) a

56 Fea

Initial mixture of Fe3−y O4 + 60%56 Fea

Synthesis of Fe3−y O4 +60%

Fe3−y O4 A/B

␥-Fe2 O3

Fe1−x O

Fe–mO A /C

␣-Fea

0.27/0.64 −0.03/−0.02 493/457 90

0.36 −0.15 504 43

0.91 0.58 0 31

−0.78/0.25 0.00/−0.21 498/387 6/5

−0.02 −0.05 329 5

56 Fe

The phase analysis is given with respect to the concentration of 57 Fe resonance iron present in the Fe3−y O4 spinel minus 57 Fe in the initial 56 Fe matrix.

V.A. Shabashov et al. / Materials Science and Engineering A361 (2003) 136–146

141

Fig. 3. Mössbauer spectra, a–d, and distribution probability functions of the centers of gravity of the resonance lines on the velocity scale P(V), a –d , for (Me + ␣-Fe2 O3 ) mixtures after CS at n = 10 and P = 6 GPa. Me: a–a —Ni; b–b —Cr; c–c —Ti; d–d –Al.

ues of interplanar distances are similar). However, they are not the strongest. The results obtained after deformation of Fe3−y O4 +56 Fe mixture were qualitatively identical to those of ␣-Fe2 O3 + 56 Fe (see c–c and d–d in Fig. 2). However, the phase transformation to the wustite took place in a larger volume. The formation of a nonstoichiometric wustite was accompanied by an increase in the partial contribution of the sextet with 504-kOe fields, which was explained by appearance of the ␥-Fe2 O3 phase (Table 3). An X-ray diffraction analysis showed that the sample did not contain the initial hematite structure. A spectral analysis of the deformed mixture of ␣-Fe2 O3 + Me(Ni, Cr, Ti) reflected a qualitatively similar picture, namely, decomposition of ␣-Fe2 O3 , appearance of spectra of oxides with a smaller oxygen concentration (Fe3−y O4 , Fe1−x O), and formation of spectra of metallic ␣-Fe and Ni–Fe, Cr–Fe and Ti–Fe alloys (see Fig. 3

and Table 4). The volume of iron reduced from the initial ␣-Fe2 O3 structure in the nickel matrix was much smaller (nearly two times) than that in the 56 Fe matrix. The concentration of iron in the Cr–Fe alloy was not inferior to the iron concentration reduced in the 56 Fe matrix. The concentration of iron, which was reduced from the ␣-Fe2 O3 hematite in the Ti matrix to the Ti–Fe metallic solid solution exceeded the concentrations in all the aforementioned cases and was equal to 42%. An exception in the experiments with the metallic matrix was the ␣-Fe2 O3 + Al mixture. The concentration of the Al–Fe structure was 2 at.% or less (relative to Fe), while reduced ␣-Fe was nearly absent. In addition to the said phases, the spectra of the synthesized samples (Figs. 1–3) included sextets A and C with fields 498 and 387 kOe, which had negative isomeric shifts anomalous for oxides of two- and three-valence iron: −0.74 and + 0.25 mm s−1 (relative to ␣-Fe).

142

V.A. Shabashov et al. / Materials Science and Engineering A361 (2003) 136–146

Table 4 Integral intensities (S, %) of the reduced iron components of the alloys and oxides formed after CS of the hematite with metal powders (Me: Cr, Ti and Al)

56 Fe,

Ni,

Components

56 Fe

Ni

Cr

Ti

Al

(FeMe + MeFe) FeOm (A’ + C) ␣ + ␥-Fe2 O3 + Fe3−y O4 Fe1−x O

26 8+5 51 0

6+7 8+7 70 2

13 + 20 4 + 14 37 11

8 + 34 0+5 51 2

1,. . . , 2 0 95 0

Me–Fea

a

Metal solid solutions and intermetallic phases were not separated.

Fig. 4. X-ray diffraction patterns of Me + ␣-Fe2 O3 mixtures (a–d) and the Fe3−y O4 spinel (e). Me: a and b—56 Fe; c—Ni; d—Al. Treatment: a—initial 56 Fe + ␣-Fe2 O3 mixture; b—56 Fe + ␣-Fe2 O3 after CS at n = 10 and P = 6 GPa; c—Ni + ␣-Fe2 O3 after CS at n = 10 and P = 6 GPa; d—Al + ␣-Fe2 O3 after CS at n = 10 and P = 6 GPa; e—initial spinel powder.

tial structure. However, the newly formed oxides are nonstoichiometric (cation-deficient), a fact which points to the release of iron atoms from the oxides. A high stability of the spinel structure under CS over a large deformation interval is not unexpected and can be explained by existence of a wide region of nonstoichiometry for this phase as the oxides are saturated with defects. A comparison of the experimental results obtained for pure Fe3−y O4 and a Fe3−y O4 + 56 Fe mixture (b–b and c–c in Fig. 2) after CS shows that an addition of iron leads to formation of a nonstoichiometric Fe1−x O wustite instead of the regions comprising systems B and C of nonstoichiometric Fe3−y O4 . That is, regions with a distorted cation-deficient nonstoichiometric Fe3−y O4 structure, which appears during CS, transform to the structure of a cation-deficient wustite in the presence of the metallic iron matrix. A positive isomeric shift of the newly formed wustite probably reflects the predominant nucleation of this phase on Fe2 + cations in octahedral voids of the Fe3−y O4 spinel. As can be seen from the spectra and the X-ray data, the reaction also leads to formation of the ␥-Fe2 O3 structure. This phase may be viewed as an extremely nonstoichiometric Fe3−y O4 spinel. A comparison of the experimental results obtained for ␣-Fe2 O3 + 56 Fe and Fe3−y O4 + 56 Fe mixtures, namely, formation of a spinel structure in the first case and a more complete transformation in the second case, as well as an equivalence of the resulting spectra of the synthesized samples (c–c and d–d in Fig. 2), suggests the following scheme of phase transformations, in which cation-deficient oxides are intermediate when iron is reduced from the hematite: ␣-Fe2 O3 → Fe3−y O4 → Fe1−x O → Fe

4. Discussion 4.1. Dissolution of mixtures of α-Fe2 O3 and Fe3−y O4 oxides with 56 Fe The spectrum of the deformed ␣-Fe2 O3 + 56 Fe mixture, which contains oxide components with a smaller oxygen concentration, namely, the Fe3−y O4 and Fe1−x O structures, and also ␣-Fe, reflects the process of iron reduction. Obviously, formation of oxides with a small oxygen concentration from the initial hematite and, finally, appearance of metallic iron is due to release of oxygen atoms from the ini-

(1)

The presence of additional components A and C in the spectra with a negative IS (as compared to oxides of twoand three-valence iron) can be explained by formation of special Fe–mO structures with iron ions having a higher valence and the presence of compression regions preserved in the synthesized samples. It was shown in the foregoing that the reaction includes reduction of iron and formation of oxygen-depleted iron oxides. Excess oxygen anions, which appear during phase transformations, stay in the metallic matrix. This is due to the experimental set-up (see Section 2), by which oxygen is ‘locked’ in a CS-exposed cell and in the material of the sample. The resulting oxygen ‘coats’ probably produce this structure under a high pressure and a

V.A. Shabashov et al. / Materials Science and Engineering A361 (2003) 136–146

143

One may suppose that the mechanical–chemical reaction, which takes place in the Fe/Fe2 O3 mixture during CS in the presence of considerable stresses (caused by a high pressure) acting on the electronic structure, leads to a rearrangement of crystal lattices of iron compounds and formation of ferrites and ferrates of two- and three-valence iron. Phase transformations of the first kind with a considerable pressure hysteresis were detected in some oxides (see, for example [21,22]). It is clear from the H value of the resolved components A and C that iron is in a high-spin state. Usually high-valence iron does not have a large atomic magnetic moment and H [23]. Therefore, observation of this structure may rather point to the appearance of strong covalence effects [24] and phase transformations, leading to a decrease in the specific volume of the structure and, consequently, an increase in the density of s-electrons on the iron core. Considering the structures resolved in the spectra, it is possible to write the hematite transformation in the metallic matrices as follows: ␣-Fe2 O3 + Me→Fe3−y O4 + Fe1−x O + (Fe–O) + Me–O + Me–Fe

(2)

This scheme only reflects the qualitative result of the transformation, but is not the true chemical reaction. If one considers partial contributions of the components in the experiment with the Fe3−y O4 + 56 Fe mixture, the balance of transformation relative to 57 Fe in Fe3−y O4 may be written as 3Fe3−y O4 = 2.2␥Fe2 O3 + 3.1Fe1−x O + 0.5Fe + 1.1(Fe–mO)

Fig. 5. Microstructure of the 56 Fe + ␣-Fe2 O3 sample after CS at n = 10 and P = 6 GPa: a—bright-field image and an electron diffraction pattern; b—dark-field image in the [1 1 0] reflection from ␣-Fe, item one; c—dark-field image of a group of reflections from the wustite structure, item two.

strong deformation. The component C (having a much more positive IS) was resolved earlier in spectra of the Fe3−y O4 spinel [18]. It was explained by local distortions of the structure and relaxation phenomena caused by a deficit of iron cations. The component A was not resolved earlier in spectra of iron oxides.

(3)

In this case, the factor m is equal to 2.1. When the component C is modeled as part of the distorted Fe3−y O4 structure and the component A is resolved separately as the Fe–mO structure, the index m increases. The balance was calculated disregarding the nonstoichiometry, i.e. the difference of v and x from zero. This is allowable, because the deficit of iron cations (according to the chemical analysis, v < 0.15 in the initial spinel) falls within the measurement accuracy of the relative contributions from the spectrum components. Obviously, the oxides are formed exclusively as dissolution products of ␣-Fe2 O3 rather than oxidation of matrix 56 Fe. Possible newly formed 56 Fe oxides should decompose and metallic 56 Fe should be reduced simultaneously. Formation of Fe–O solid solutions was also disregarded, because dissolution of oxygen in metallic iron is very small. The oxygen ‘coats’ of Fe–mO may be localized on different crystallite boundaries, which have a large defect content and are capable of accumulating oxygen. The Fe–mO structure may also be considered as an interstitial phase of oxygen in iron oxides with strong covalence effects and a smaller Fe–O interatomic distance.

144

V.A. Shabashov et al. / Materials Science and Engineering A361 (2003) 136–146

4.2. The effect of properties of the metallic matrix on dissolution of the α-Fe2 O3 hematite The experiments with ␣-Fe2 O3 + Me mixtures showed that replacement of iron by nickel, chromium, titanium or aluminum changes the intensity of hematite dissolution and alters the ratio between the products of phase transformations. When hematite mixtures are exposed to CS, the metallic matrices Ni, Cr and Ti also dissolve ␣-Fe2 O3 , leading to appearance of the magnetite, the wustite and iron structures. Most of the reduced iron passes to metallic Ni–Fe, Cr–Fe or Ti–Fe solid solutions. The aluminum matrix is an exception (see Table 4). One of the criteria of the mechanical synthesis is deformability of mixture components, which depends, in particular, on the ratio between microhardness values of the components. In accordance with [11], the CS mechanical synthesis in Bridgman anvils is retarded if microhardness values of components differ considerably (by more than five times). Microhardness values of the metals used in our experiments allow separating transition metals and aluminum. However, the strength of aluminum increases during deformation (see Table 1). When the hematite is deformed in the aluminum matrix, a small quantity (about 1 mass %) of iron passes to the Al–Fe solid solution and the strength of the structure approaches the strength of the other metals [25]. Therefore, one may think that the degree of the hematite decomposition and strengthening of the mixture are interconnected and, in turn, depend on the ability to form iron solid solutions in the aluminum matrix. A little acceleration in the kinetics of the iron reduction from the hematite in BCC matrices of iron and chromium as compared to the iron reduction kinetics in FCC nickel may be explained considering results of the studies [12,13], which showed that the lattice symmetry of metals affected the mechanical synthesis kinetics. The structures have different lattice constants and are polymorphous in the case of nickel (BCC Fe + FCC Ni) or, probably, isomorphous in the case of chromium (BCC Fe + BCC Cr). It was found [12,13] that the mechanical synthesis of metals with unlike (BCC and FCC) lattices was retarded. However, the extension of this criterion to dissolution of the hematite in titanium and aluminum matrices is obviously insufficient. A comparison of the experimental results concerning dissolution of the hematite and the data on equilibrium dissolution of iron in the corresponding metallic matrices cannot be the only criterion of nonequilibrium dissolution of oxides either. This applies primarily to active dissolution of the hematite in the titanium matrix. Recall that a specific feature of titanium and aluminum consists in a low solubility of iron in these metals and formation of intermetallic compounds. Metallic matrices 56 Fe, Ni and Cr can form solid solutions with iron over the whole range of compositions. An analysis of the data on the dynamic solubility of iron in nickel and chromium matrices during CS shows that these

metals easily dissolve in iron. Dynamic dissolution of iron in titanium takes place in a much wider range of solid solutions [15] and is much larger than dissolution of iron in the aluminum matrix. In accordance with [16], not more than 1 mass% iron can be dissolved in aluminum using mechanical synthesis methods. Despite a limited dynamic solubility of iron in titanium, dissolution of the hematite in this metal is not inferior to or even exceeds hematite dissolution in Fe, Ni and Cr matrices. Therefore, it would be reasonable to analyze the behavior of oxygen, which is the second component of the oxides, in the metallic matrices under discussion. This applies primarily to the oxygen solubility in the metals. Titanium and aluminum represent the most illustrative antipodes in this respect. Titanium can accumulate up to 30% interstitial oxygen. The oxygen solubility in the aluminum matrix is almost zero [17]. Thus, considerably different kinetics of strain-induced dissolution of the hematite in titanium and aluminum matrices may be explained by different dissolution of oxygen in the corresponding matrices. A specific feature of aluminum is that aluminum oxides are formed on the surface of particles of the initial metal. The oxides represent high-strength phases, which probably can retard the strain-induced penetration of oxygen and iron to the metal. The experiments on dissolution of the hematite in matrices of freshly filed and surface-oxidized dispersed 56 Fe powders confirm the passivating effect of the oxide film. Solubility of oxygen in Fe, Ni and Cr matrices is little at room temperature. These elements differ with respect to oxygen ions primarily in their ability to form oxides and other compounds. Formation of nickel oxides impeding the mechanical synthesis is least probable. An analysis of other oxide-containing structures in the spectra of the synthesized samples, namely, the Fe–mO structure, showed that it is formed most readily in the case of iron, nickel and chromium, is absent in the case of aluminum, and is partially present in the case of titanium (Table 4). The result of the aluminum experiment is trivial, that is, hematite dissolution does not take place. The result obtained for titanium can be explained by a predominant release of excess oxygen from the Fe–mO structure to the titanium matrix in the form of interstitial impurities. An increase in the number of sextets A and C in the experiments with 56 Fe, Ni and Cr matrices is related to a predominant transfer of iron (as compared to oxygen) from the hematite to metallic matrices of iron, nickel and chromium, whose iron solubility is unlimited. Since excess oxygen dissolves little in the corresponding metals, it can accumulate on different crystallite boundaries, leading to formation of the Fe–mO structure. Special behavior of A in the titanium experiment confirm the supposition that A corresponds to an oxygen-enriched Fe–mO structure located on crystallite oxide–metal interfaces. A strong intensification of the decomposition of iron oxides in mixtures with metallic matrices suggests that the observed phase transformations have the character of

V.A. Shabashov et al. / Materials Science and Engineering A361 (2003) 136–146

strain-induced dissolution of oxides in metals. It is known that the usual diffusion of interstitial atoms in deformed iron alloys at room temperature does not take place even in the stress field of dislocations [26]. Strain-induced dissolution of oxides at 298 K can be realized only when oxygen atoms pass to interstices. Vacancies and interstitial atoms, which are induced by cold deformation, sharply intensify the decomposition and the transport of atoms from the oxides to the surrounding metallic matrix. Strain-induced dislocations cut the oxides and refine the structure. Refinement of the structure and its saturation with defects provide conditions for both ␣-Fe2 O3 →Fe3 O4 →FeO phase transformations [27] and the transfer of iron and oxygen atoms to solid solutions. Dynamic dissolution of the hematite in metal mixtures includes removal of iron and oxygen ions from ␣-Fe2 O3 , Fe3−y O4 and Fe1−x O oxides to interstices and formation of solid solutions and Fe–mO interstitial oxide phases on defect-loaded crystallite boundaries. Thus, an analysis of the effect of the matrices on dissolution of iron oxides during cold CS suggests that both physical models (differences in the structures induced by a strong cold deformation; the mechanism of interaction between oxide atoms and strain-induced defects; solubility) and chemical criteria (oxidizing ability of the matrix elements; formation of near-surface phases; chemical reactions, etc.) may be used.

5. Conclusion It was shown that strain-induced transformations of the hematite in metallic Me matrices under CS is accompanied by appearance of the following structures: Fe3−y O4, Fe1−x O, Fe–mO, Me–O and Me–Fe. Dissolution of the inverse spinel is facilitated under the same experimental conditions since the energy is not spent for transformation of ␣-Fe2 O3 to Fe3−y O4 . The hematite and the magnetite transform during CS to structures of cation-deficient nonstoichiometric oxides and metallic solid solutions. The iron reduction is accelerated in the presence of a metallic matrix in the mixture and depends on the ability of the metallic matrix to form solid solutions and chemical compounds with iron and oxygen. The reduction process is most intensive in Fe, Cr and Ti matrices. It is retarded in Ni and, especially, Al. The absence of phase transformations in the hematite–aluminum mixture is explained by an extremely low solubility of iron and oxygen in aluminum. Oxides, which appear on the surface of particles of the metallic matrices, also have a passivating effect on strain-induced dissolution of the hematite. CS causes appearance of new structures of iron oxides, which have no analogs under usual conditions. These structures apparently contain a high concentration of oxygen and form boundary solid solutions or interstitial phases of the Fe–mO type with a small spe-

145

cific volume of the structure having a strong covalence effect. The possible transport of oxygen atoms from iron oxides to matrices of other metals, which is accompanied by the appearance of oxygen-supersaturated solid solutions and interstitial phases, can be used to form oxides of the other metals having a set of properties inherent in ODS alloys.

Acknowledgements The authors are thankful to V.A. Barinov, N.I. Noskova, and V.B. Vykhodets for valuable recommendations and help in the work. The study has been supported by the RFFR-Ural (Project No. 01-03-96446) and the RAS Presidium under the program ‘Basic problems of physics and chemistry of nanosized systems and nanomaterials’.

References [1] W.A. Kaszmarek, B.W. Ninham, IEEE Trans. Magn. 30 (2) (1994) 732. [2] M. Hoffmann, S.I. Campbell, W.A. Kaszmarek, Mater. Sci. Forum 228–231 (1996) 607. [3] J. Ding, W.F. Miao, R. Street, P.S. McCormic, Scripta Materiala 35 (11) (1996) 1307. [4] S.I. Campbell, W.A. Kaszmarek, in: S.J. Long F. Grandjan (Eds.), Mössbauer Spectroscopy Applied to Materials and Magnetism, vol. 2, Plenum Press, 1996, p. 273. [5] S.I. Novikov, V.A. Barinov, Fizika i Khimia Obrabotki Materialov 3 (2001) 81. [6] L.F. Vereschagin, E.V. Zubova, K.P. Burkina, G.A. Aparnikov, Dokl. AN SSSR 196 (1971) 817. [7] V.V. Sagaradze, V.A. Shabashov, T.M. Lapina, N.L. Pecherkina, V.P. Pilyugin, Phys. Met. Met. 78 (6) (1994) 619. [8] V.A. Shabashov, A.G. Mukoseev, V.V. Sagaradze, Mater. Sci. Eng. A 307 (2001) 91. [9] A.G. Mukoseev, V.A. Shabashov, V.V. Sagaradze, I.V. Sagaradze, Mater. Sci. Eng. A 316 (2001) 174. [10] V.A. Shabashov, L.G. Korshunov, A.G. Mukoseev, V.V. Sagaradze, V.P. Pilyugin, S.I. Novikov, A.V. Makarov, N.F. Vildanova, Mater. Sci. Eng. A 346 (1–2) (2002) 196. [11] V.V. Neverov, Phys. Met. Met. 73 (1) (1992) 97. [12] A.G. Mukoseev, V.A. Shabashov, V.P. Pilyugin, V.V. Sagaradze, Phys. Met. Met. 85 (5) (1998) 542. [13] V.A. Shabashov, A.G. Mukoseev, D.V. Desyatkov, A.V. Litvinov, Phys. Met. Met., in press. [14] N.I. Noskova, Phys. Met. Met. 86 (2) (1998) 179. [15] M.L. Trudeau, R. Schulz, L. Zaluski, et al., Mater. Sci. Forum 88–90 (1992) 537. [16] A.G. Mukoseev, V.A. Shabashov, I.G. Brodova, V.V. Sagaradze, Abstracts of VIII International Conference on the Mössbauer Spectroscopy and Applications, St. Petersburg, 2002, p. 19. [17] E. Fromm, E. Gebhardt, Gases and Carbon in Metals, Metallurgiya Moscow, 1980, p. 712. [18] H. Annersten, S.S. Hatner, Zeitschrift für Kristallograp. 137 (1973) 321. [19] I.P. Suzdalev, V.N. Buravtsev, V.K. Imshennik, Y.u.V. Maksimov, V.V. Matveev, Izv. AN. Ser. Fiz. 65 (7) (2001) 1028. [20] A.M. Van der Kraan, Phys. Stat. Sol. A 18 (1973) 215.

146

V.A. Shabashov et al. / Materials Science and Engineering A361 (2003) 136–146

[21] K. Parlinski, Eur. Phys. J. B 27 (2002) 283. [22] Y.D. Perfiliev, Zh. Neorgan. Khimii. 47 (5) (2002) 693. [23] V.I. Goldansky, Chemical Applications of Mössbauer Spectroscopy, Mir, Moscow 1970, p. 503. [24] V.I. Nikolaev, V.S. Rusakov, Covalence effects in ferrites-spinels, in: Mössbauer Studies of Ferrites, Izd, MGU Moscow, 1985, p. 122.

[25] V.V. Stolyarov, E.P. Soshnikova, I.G. Brodova, Phys. Met. Met. 93 (6) (2002) 567. [26] A.R. Kuznetsov, V.V. Sagaradze, Phys. Met. Met. 93 (5) (2002) 404. [27] P. Ayyub, M. Multani, M. Barma, V.R. Palkar, R. Vijayaraghavan, J. Phys. C, Solid State Phys. 21 (1988) 2229.