Microstructures and phase formation in rapidly solidified Sm–Fe and Sm–Fe–Ti–C alloys

Microstructures and phase formation in rapidly solidified Sm–Fe and Sm–Fe–Ti–C alloys

Journal of Magnetism and Magnetic Materials 188 (1998) 353—360 Microstructures and phase formation in rapidly solidified Sm—Fe and Sm—Fe—Ti—C alloys ...

624KB Sizes 0 Downloads 43 Views

Journal of Magnetism and Magnetic Materials 188 (1998) 353—360

Microstructures and phase formation in rapidly solidified Sm—Fe and Sm—Fe—Ti—C alloys J.E. Shield!,*, C.P. Li", D.J. Branagan# ! Department of Materials Science and Engineering, University of Utah, Salt Lake City, UT 84112, USA " Ames Laboratory, Iowa State University, Ames, IA 50011, USA # Idaho National Engineering and Environmental Laboratory, Idaho Falls, ID 83415, USA Received 23 March 1998

Abstract Partially ordered Sm Fe , closer to the SmFe structure, formed upon melt spinning of Sm Fe . A strong 2 17 7 11 89 dependence of grain size on the wheel speed was also observed, with wheel speeds less than 40 m/s generating grain sizes above the single domain limit for Sm Fe N . The addition of Ti and C resulted in an order of magnitude grain 2 17 x refinement compared to the binary alloys. The Ti and C also effectively inhibited grain growth during annealing. The nitrided Sm—Fe—Ti—C alloys also displayed a rapid decrease in coercivity with decreasing grain size, which was attributed to increased exchange interactions. ( 1998 Elsevier Science B.V. All rights reserved. PACS: 75.50.W; 81.40.R Keywords: Microstructure; Interstitial permanent magnets; Rapid solidification; Magnetic properties

1. Introduction The interstitial compound Sm Fe N has re2 17 x cently gained a significant amount of attention because of its excellent magnetic properties, including a high saturation magnetization of 1.54 T, anisotropy field of 14 T, and a moderately high Curie temperature of 470°C [1,2]. The formation of the binary Sm—Fe precursor compound has been suc-

* Corresponding author. Tel.: 801/581-3179; fax: 801/5814816; e-mail: [email protected].

cessfully accomplished by mechanical alloying, conventional ingot casting, and melt spinning [1,3—7]. Early work on the melt spinning of Sm—Fe alloys revealed that the phase formation was dependent on the Sm content, with higher Sm content ('17 at%) resulting in the formation of an amorphous structure [8]. In addition, the formation of the disordered SmFe (TbCu -type) struc7 7 ture was preferred over the ordered Sm Fe 2 17 (Th Zn -type) structure near the Sm Fe 2 17 2 17 stoichiometry. The difference between the two structures is that the transition metal dumbbells

0304-8853/98/$19.00 ( 1998 Elsevier Science B.V. All rights reserved. PII: S 0 3 0 4 - 8 8 5 3 ( 9 8 ) 0 0 2 0 8 - X

354

J.E. Shield et al. / Journal of Magnetism and Magnetic Materials 188 (1998) 353—360

possess long-range order in the Sm Fe structure, 2 17 whereas in the SmFe structure they randomly 7 occupy Sm sites. The resulting structural relationships between the two structures are a " 2~17 J3a and c "3c . In addition, another 1~7 2~17 1~7 structure, isostructural with the ThMn tetrag12 onal structure, can be derived from these structures. Evidence of all three of these structures, Sm Fe , SmFe and SmFe , has been observed 2 17 7 12 close to the Sm Fe stoichiometry, with the exact 2 17 structural state strongly dependent on the Sm content [5]. At hyperstoichiometric compositions, the more ordered Sm Fe structure tends to form. As 2 17 the Sm content is decreased, the amount of ordering decreases and the structure becomes more like the SmFe compound. At even lower Sm content, 7 the structure can be described as a tetragonal compound isostructural with ThMn . Both the SmFe 12 7 and ThMn -type compounds are metastable, al12 though Ti additions stabilize the ThMn -type 12 SmFe Ti structure [9]. In addition to the forma11 tion of these phases during rapid solidification, the SmFe phase was also formed, regardless of wheel 2 speed. The formation of these phases was also observed to be relatively independent of wheel speed, and interestingly very little amorphous phase formation occurred near the Sm Fe stoichio2 17 metry, even at wheel speeds up to 60 m/s. While these previous investigations explored the phase formation and magnetic properties of rapidly solidified Sm—Fe alloys and their nitrides, this present paper reports the first microstructural analysis of melt spun Sm—Fe alloys.

for TEM were prepared by ion beam thinning of ribbons with a Gatan 600 Duomill operating at 4.5 kV. Samples for magnetic measurements were prepared by forming the Sm Fe N interstitial com2 17 x pound by grinding the ribbon in a mortar and pestle followed by a heat treatment at 550°C for 4 h in flowing N . The introduction of interstitial nitro2 gen was verified by X-ray diffraction. The magnetic properties of nitrided samples were measured with a PARC vibrating sample magnetometer, which has a maximum field of 2.3 T, after pulse magnetization at 6.5 T. No demagnetization corrections were employed when evaluating the data.

3. Results and discussion X-ray diffraction patterns of the Sm Fe sam11 89 ples revealed peaks that were indexed to the SmFe 7 phase for all wheel speeds (Fig. 1). There was no evidence of long-range order (i.e., the Sm Fe 2 17 phase), denoted by, for example, the presence of the M0 2 21 4N peak at 31.6° 2h. This phase was the dominant phase at all solidification rates. Samples melt spun at 10, 15 and 20 m/s also contained a small amount of the SmFe phase, while the sample melt 2 spun at 30 m/s was phase pure SmFe , by X-ray 7 diffraction. It is quite interesting that no primary Fe formed in any of the samples, at least to the

2. Experimental procedures Samples of nominal compositions Sm Fe and 11 89 (Sm Fe ) Ti C were alloyed by arc melting 2 17 100~2x x x appropriate quantities of Sm of 99.5% purity, Fe of 99.9% purity, Ti of 99.9% purity and electrodegrade C. The arc-melted ingots were then remelted by RF induction heating and rapidly solidified with a single roller melt spinning apparatus. Tangential wheel velocities of 10, 15, 20, 30 and 40 m/s were utilized in this study. As-solidified material was characterized by X-ray and electron diffraction and transmission electron microscopy (TEM). Samples

Fig. 1. X-ray diffraction patterns of Sm Fe melt spun from 11 89 10 to 40 m/s. Phases present included SmFe , SmFe and a-Fe. 7 2

J.E. Shield et al. / Journal of Magnetism and Magnetic Materials 188 (1998) 353—360

355

detection limits of X-ray diffraction. This is especially surprising for the material melt spun at 10 m/s, which provides a relatively low solidification rate. To bypass the formation of primary Fe, an undercooling of greater than 200 K from the equilibrium melting point is necessary. Apparently non-equilibrium effects contribute to the suppression of solid Fe, as it is unlikely that an undercooling of significant magnitude is achieved at 10 m/s. The exception to this was the sample melt spun at 40 m/s; in this case, a small fraction of a-Fe was present. However, this is thought to be due to chemistry affects arising from a slightly greater loss of Sm during processing in this sample compared to the others. The solidification behavior and hence phase formation can vary significantly across the ribbon cross section. For example, low solidification rates can lead to the development of columnar growth, which manifests itself through preferred orientation and can be detected by X-ray diffraction. X-ray diffraction results of unground ribbon aligned either wheel or free side up indicate random orientation similar to the powder samples (Fig. 2). No major differences were observed in the peak intensities of the free and wheel side. This indicates that grains are randomly oriented. While the X-ray diffraction results did not reveal long-range ordering of the transition metal dumbbells, electron diffraction revealed the presence of some degree of ordering in the Sm Fe samples 11 89 regardless of wheel speed. The [11 2 11 0] zone axes

of samples melt spun at 30 and 40 m/s are shown in Fig. 3. The reflections labeled ‘A’ and ‘B’ were indexed according to the Sm Fe structure as the 2 17 (0 0 0 3) and (3 0 31 0) reflections, respectively. The less intense reflections between the incident spot and the (3 0 31 0) reflection can then be indexed as the (1 0 11 0) and (2 0 21 0) reflections. Considering the SmFe structure, the reflections ‘A’ and ‘B’ 7 correspond to the (0 0 0 1) and (1 1 21 0) reflections, respectively; however, the weak reflections are not allowed for the SmFe structure. Thus, even at high 7 wheel speeds a limited amount of chemical order on the Sm Fe lattice exists. 2 17

Fig. 2. X-ray diffraction scans of the (a) wheel and (b) free side of ribbon melt spun at 10 m/s. The similar diffraction patterns indicate a random distribution of grains.

Fig. 3. Selected area electron diffraction pattern of the [1 2 1 0] zone axis of Sm Fe melt spun at (a) 30 m/s and (b) 40 m/s 11 89 displaying long-range order on the Sm Fe lattice. 2 17

356

J.E. Shield et al. / Journal of Magnetism and Magnetic Materials 188 (1998) 353—360

The microstructures of the rapidly solidified Sm Fe alloys were examined by TEM. The 11 89 grain size systematically decreased with increasing wheel velocity and ranged from 0.8 lm at 15 m/s to 0.3 lm at 40 m/s (Fig. 4). In addition, a highly mottled contrast was observed within the individual grains, which is especially evident in the sample melt spun at 15 m/s (Fig. 4a). This mottled contrast arises from antiphase boundaries and the large amount of chemical disorder in the sample. The relatively broad X-ray diffraction peaks of Fig. 2 also reflect the structural disorder and are not due to grain size effects. Upon annealing at 800°C for 15 min, both electron and X-ray diffraction patterns revealed an increased amount of or-

der. In SAD patterns, the superlattice reflections increased in intensity (Fig. 5a). In addition, the increased order was manifested in the microstructure as well. The mottled contrast present in the as-solidified material disappeared and gave way to planar defects perpendicular to the c-axis (Fig. 5b). These planar defects are essentially antiphase boundaries arising from faults in ordering along the c-axis. Alloys of the form (Sm Fe ) Ti C with 11 89 100v2x x x x"0, 2 and 6 were formed by melt spinning at 15 m/s. With x"2, a decreased amount of SmFe 2 was observed when compared to x"0 (Fig. 6). With x"6, no SmFe was present; however, a sig2 nificant amount of a-Fe formed. With both the

Fig. 4. Bright field images of Sm Fe melt spun at (a) 15 m/s, (b) 30 m/s and (c) 40 m/s. 11 89

J.E. Shield et al. / Journal of Magnetism and Magnetic Materials 188 (1998) 353—360

357

Fig. 6. X-ray diffraction patterns of (Sm Fe ) Ti C 11 89 100~2x x x melt spun at 15 m/s with x equal to 0, 2 and 6.

Fig. 7. X-ray diffraction patterns of (Sm Fe ) Ti C melt 11 89 94 3 3 spun at 20, 30 and 40 m/s. Fig. 5. (a) SAD pattern of Sm Fe melt spun at 30 m/s and 11 89 annealed at 800°C for 15 min. The superlattice reflections have noticeably increased in intensity (compare with Fig. 3a). (b) Bright field image of Sm Fe melt spun at 30 m/s and an11 89 nealed at 800°C for 15 min.

x"2 and 6 alloys, the SmFe phase was again the 7 primary phase formed, with no evidence of ordering from the X-ray diffraction pattern. In addition, the lattice parameters for both x"2 and 6 were consistent with the hexagonal SmFe structure 7 with no change in the c/a ratio as x increased. A c/a ratio of J3/2 is indicative of a transition to the ThMn -type tetragonal structure that has been 12 observed in both hypostoichiometric Sm—Fe alloys [5] and SmFe Ti [9] compounds. Alloys with 11 x"3 were melt spun at 20, 30 and 40 m/s. A de-

creasing amount of SmFe was observed, and at 2 40 m/s it was not detectable from the X-ray diffraction patterns (Fig. 7). The SmFe phase again did 7 not display long-range ordering detectable by Xray diffraction. Finally, no evidence of an amorphous structure was observed, even for the sample melt spun at 40 m/s. As did the binary Sm Fe alloys, the grain size 11 89 of the (Sm Fe ) Ti C alloys showed a strong 11 89 94 3 3 dependence on wheel speed. However, the addition of Ti and C significantly refined the grain size when compared to the binary Sm Fe . Here, the grain 11 89 size ranged from 170 nm at 20 m/s to 25 nm at 40 m/s (Fig. 8). In addition, the grain morphology of the TiC-added alloys became more equiaxed and less faceted than what was observed in the binary

358

J.E. Shield et al. / Journal of Magnetism and Magnetic Materials 188 (1998) 353—360

Fig. 8. Bright field images of (Sm Fe ) Ti C melt spun at (a) 20 m/s, (b) 30 m/s, and (c) 40 m/s. 11 89 94 3 3

alloys. The fine grain size precluded single-grain SAD patterns, making it difficult to analyze the effect of Ti and C on the long-range order. Very fine grains, presumably of TiC, were also observed at some triple junctions. The Ti and C additions also inhibited grain growth, as samples melt spun at 30 m/s and annealed at 700°C for 15 min retained a grain size of approximately 100 nm. Sm Fe N samples for magnetization measure11 89 d ments were produced by heat treating the samples at 550°C for 4 h in flowing N . The X-ray dif2 fraction patterns revealed the presence of the disordered SmFe , similar to the as-solidified struc7 ture with no evidence of chemical order. In addition, the shift in the peak positions was consistent

with the introduction of interstitial N, and resultant lattice parameters indicated nearly full nitrogenation (i.e., x+2.5) [10]. Also present was a-Fe, with no evidence of SmFe . In order to ascertain the 2 origin of phase transfer, a sample was annealed at 550°C for 4 h in argon. In this case, the phase constituency consisted of the SmFe and SmFe 7 2 phases. In a nitrogen atmosphere, the SmFe reacts 2 to form a-Fe and SmN. The decomposition of SmFe observed here is consistent with previous 2 results [11]. The magnetic properties of binary Sm Fe N 11 89 d were relatively constant for alloys melt spun at 15, 20 and 30 m/s, with relatively low coercivities (+1 kOe) and remanent ratios (+0.31). The sample

J.E. Shield et al. / Journal of Magnetism and Magnetic Materials 188 (1998) 353—360

Fig. 9. Demagnetization curves of (Sm Fe ) Ti C N . 11 89 94 3 3 d

melt spun at 40 m/s, on the other hand, displayed a higher coercivity (4 kOe) and remanent ratio (0.5 — the maximum for non-interacting, isotropic structures). The differences in the magnetic properties can be attributed to the grain size. The single domain size can be estimated from the Kittel formula, D "17.6c/k M2, which for Sm Fe N results 2 17 d # 0 4 in D +360 nm with c"3.9(10)~2 J/m2 and k M " # 0 4 1.54 T [12,13]. Thus, the samples melt spun at 15, 20 and 30 m/s had grain sizes above the single domain limit, while the sample melt spun at 40 m/s had an average grain size just below the single domain limit, and an associated higher coercivity. The multidomain grains allowed relatively uninhibited domain wall motion, resulting in easy demagnetization. However, the chemical disorder did affect domain wall motion, at least to a limited degree, as more chemically ordered structures, obtained by annealing at 800°C, had a coercivity of only 600 Oe. The (Sm Fe ) Ti C alloys were nitrided with 11 89 94 3 3 the same treatment given to the binary Sm Fe 11 89 alloys. Also, the X-ray diffraction results indicated similar results, with lattice parameters consistent with full nitrogenation. The grain sizes of these alloys were all below the single domain limit. With these samples, the magnetic properties decreased with decreasing grain size. The sample melt spun at 20 m/s had a coercivity of 3.2 kOe and a remanent ratio of 0.49, while the sample melt spun at 40 m/s, which had an average grain size on the order of

359

Fig. 10. Magnetic properties H and M /M as a function of c r s grain size in Sm Fe N and (Sm Fe ) Ti C N . 11 89 d 11 89 94 3 3 d

25 nm, had a coercivity of only 800 Oe and a remanent ratio of 0.08. The shape of the hysteresis loop in the second quadrant also became more wasp-waisted as the grain size decreased (Fig. 9). The rapid decrease in coercivity and the loop shape can be attributed to increased exchange interactions arising from the finer grain size. In addition, annealing treatments of the (Sm Fe ) Ti C 11 89 94 3 3 melt spun at 30 m/s resulted in slightly higher coercivity. This can be attributed to the greater degree of order resulting from the annealing treatment and the fact that the Sm Fe structure has a higher 2 17 anisotropy than SmFe [5,6]. The overall relation7 ship between the grain size and magnetic properties of the as-spun Sm Fe N and (Sm Fe ) 11 89 d 11 89 94 Ti C N alloys is shown in Fig. 10. 3 3 d 4. Conclusions In conclusion, melt-spun binary Sm Fe alloys 11 89 tended to form the disordered SmFe structure, 7 although a limited amount of long-range order was observed. The grain size ranged from approximately 0.3 to 0.8 lm and depended on wheel speed. The nitrided samples with grain sizes above the single domain limit displayed poorer magnetic properties. X-ray diffraction patterns revealed that the Ti- and C-containing alloys formed the disordered SmFe structure. Furthermore, there was 7

360

J.E. Shield et al. / Journal of Magnetism and Magnetic Materials 188 (1998) 353—360

no evidence of the tetragonal ThMn -type 12 structure, which can be stabilized by Ti additions. In this case the Ti does not substitute for Fe into the structure, but forms the TiC compound. The addition of Ti and C to the binary Sm Fe 11 89 alloy also effectively refined the grain size and inhibited grain growth during subsequent annealing treatments. The alloying additions also decreased the amount of SmFe that formed during 2 solidification. The magnetic properties of the nitrided samples decreased with decreasing grain size due to increased intergranular exchange interactions. This work indicates that selective alloying can be used to alter the microstructure without affecting the intrinsic magnetic properties of the base compound.

Acknowledgements J.E.S. acknowledges support from the University of Utah Research Committee. Work at Ames Laboratory was supported by the Director of Energy Research, Office of Basic Sciences, U.S. Department of Energy under contract no. W-7405ENG-82. Work at INEEL was supported by DOE Idaho Operations Office Contract DE-AC0794ID13223. The authors are extremely grateful to

R.W. McCallum for useful discussions and to J. Baczuk for alloy preparation. References [1] M. Katter, J. Wecker, L. Schultz, R. Grossinger, J. Magn. Magn. Mater. 92 (1994) L14. [2] J.M.D. Coey, Hong Sun, J. Magn. Magn. Mater. 87 (1991) L251. [3] X.C. Kou, R. Grossinger, T.H. Jacobs, K.J.H. Buschow, J. Magn. Magn. Mater. 88 (1990). [4] K. Schnitzke, L. Schultz, J. Wecker, M. Katter, Appl. Phys. Lett. 56 (1990) 2252. [5] M. Katter, J. Wecker, L. Schultz, J. Appl. Phys. 70 (1991) 3188. [6] F.E. Pinkerton, C.D. Fuerst, Appl. Phys. Lett. 60 (1992) 2558. [7] H.T. Kim, Q.F. Xiao, Z.D. Ahang, D.Y. Geng, Y.B. Kim, T.K. Kim, H.W. Kwon, J. Magn. Magn. Mater. 173 (1997) 295. [8] Y. Xingbo, T. Miyazaki, T. Izumi, H. Saito, M. Takahashi, IEEE Trans. Magn. 23 (1987) 3104. [9] K. Ohashi, Y. Tawara, R. Osugi, M. Shimao, J. Appl. Phys. 64 (1988) 5714. [10] M.Q. Huang, Y. Zheng, K. Miller, J.M. Elbicki, S.G. Sankar, W.E. Wallace, R. Obermyer, J. Magn. Magn. Mater. 102 (1991) 91. [11] C. Ishizaka, T. Yoneyama, A. Fukuno, IEEE Trans. Magn. 29 (1993) 2833. [12] J.F. Hu, T. Dragon, M.L. Sartorelli, H. Kronmu¨ller, Phys. Stat. Sol. (a) 136 (1993) 207. [13] K. Kobayashi, X. Rao, J.M.D. Coey, D. Givord, J. Appl. Phys. 80 (1996) 6385.