Continuous mapping of structure–property relations in Fe1−xNix metallic alloys fabricated by combinatorial synthesis

Intermetallics 9 (2001) 541–545 www.elsevier.com/locate/intermet

Continuous mapping of structure–property relations in Fe1
xNix metallic alloys fabricated by combinatorial synthesis Young K. Yooa, Tsuyoshi Ohnishia, Gang Wanga, Fred Duewera, X.-D. Xianga,*, Yong S. Chub, Derrick C. Mancinib, Yi-Qun Lic, Robert C. O’Handleyd a

Environmental Energy Technologies Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA b Advanced Photon Source, Argonne National Laboratory, Argonne, IL 60439, USA c Intematix Corporation, Moraga, CA 94556, USA d Department of Materials Science and Engineering, Massachusetts Institute of Technology, Cambridge, MA 021391, USA Received 2 April 2001; accepted 3 April 2001

Abstract A combinational approach has been taken to the linkage of composition and magnetic properties of Fe–Ni alloys. This linkage has been addressed by preparing the entire range of the binary alloys on a single material chip, in the form of a ‘continuous phase diagram’. # 2001 Elsevier Science Ltd. All rights reserved. Keywords: B. Magnetic properties; B. Phase diagram

The study of structural and physical property relationships has been always at the center of materials science. Traditionally the mapping of structural and physical properties as functions of material composition was accomplished through the synthesis and analysis of samples with discrete compositions prepared one at a time. This process generally consumes extensive time and manpower. It is also limited in its compositional resolution, which may subsequently miss important opportunities in material discovery. Recently, the combinatorial materials science approach has shown a promise to accelerate material discovery and optimization [1–11]. To date the central theme and focus of this approach have been to combine parallel synthesis with rapid characterization techniques to map out physical or chemical properties as functions of composition for a given system. The question remains whether it is feasible to map the complex relationship between lattice structure and physical properties as a function of composition for a multi-component system in one combinatorial

* Corresponding author at Intematix Corporation, 351 Rheem Boulevard, Moraga, CA 94556, USA. Tel.: +1-925-631-9005; fax: +1-925-631-7892. E-mail address: [email protected] (X.-D. Xiang).

experiment. Addressing this issue, we present here the first detailed mapping of crystal structure and physical properties of the well-documented Fe1
xNix binary alloy system on a single material chip with linear composition x changing from zero to one. This system is of interest because it shows a BCC to FCC transformation with increasing x as well as rich changes in the magnetic properties with x. Long before the recent activity in combinatorial materials science, there have been efforts to partially address this problem. For example, diffusion couples and triples have been used to map the structural phase diagram of binary and ternary systems [12]. This approach relies on the existence of a small diffused-region (in the order of tens of microns) at the interface of multi-component systems after heat treatment. The key to the success of this approach is the capability of analytical tools to determine accurately the local composition and crystal structure with high spatial resolution (ideally much less than 1 mm). Electron-microscope-based techniques are often adequate for this application [13]. However, tools with similar resolution are not always available to characterize many other physical properties. As a consequence, the approach remains difficult to apply in mapping out structural and physical property relations.

0966-9795/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved. PII: S0966-9795(01)00030-9

542

Y.K. Yoo et al. / Intermetallics 9 (2001) 541–545

The ‘‘composition spread’’ technique represents a different approach, which generates a map of varied composition in a large 2-dimensional space. The earlier effort to use this approach can be traced back to Kennedy’s work in 1965 [14]. Recently, van Dover and his colleagues at Bell Laboratories have improved the technique and applied it to solving important industrial materials problems [7]. Usually, ‘‘composition spread‘‘ techniques rely on the geometric arrangement of target and substrate to generate composition gradients. As a result, the gradients were not linear and vary with different precursors and deposition conditions. Time-consuming chemical analysis and coordinate transformation are usually required to generate a map (or phase diagram) conforming to standard binary or ternary representation. We have recently adopted a technique relying on moving precision masks during deposition to generate linear composition profiles. We have used this method to map physical properties of several oxide binary and ternary phase diagrams [9,10,15,16]. Our approach focuses more on achieving a well-defined local composition (eliminating the need for routine stoichiometry analysis) in an area large enough for physical property characterization and small enough to match available desired substrates (often critical for growing high quality crystalline films). No previous study has demonstrated reliable mapping of both structural and physical properties in a continuous phase diagram (CPD). In this study, we chose a well-documented metallic alloy system to demonstrate for the first time that present techniques can be used to reliably map the crystal structure and physical property relationship as a function of composition. Fe/Ni multi-layered films were deposited on sapphire (0001) substrates ( 1 cm long) at room temperature using an ion-beam-sputtering, combinatorial deposition system. During the deposition, computer controlled moving masks spatially control the thickness profile of each layer. The system has a deposition chamber with base pressure of 10
10 Torr and a load-lock chamber to exchange the sample substrate without breaking the vacuum. Ar+ ions with an acceleration voltage of 1000 eV are used as an incident beam and a typical deposition rate of 0.3 A˚/s for Ni and Fe targets can be achieved. Moving a shutter with a constant speed across the surface of the substrate during the deposition generated a linear thickness gradient. Three sets of gradient layers of Ni (0–418 A˚) and Fe (519–0 A˚) were deposited by moving the shutter from one edge to the other edge of the substrate. As-grown multi-layered films were annealed at 600 C for 3 h in evacuated ampoules to obtain full compositional mixing normal to the film plane without oxidation and to achieve alloy phase formation. The samples were air quenched after annealing. Auger depth-profile analysis of annealed sample shows uniform profiles of Ni and Fe concentration across the thickness of the film, in contrast with step-wise layer-profiles for

the as-deposited sample. This suggests that the layered films are thoroughly mixed within the thickness of the film. Without a capping layer, a slight oxidation is expected on the top, but the characterizing probes penetrate enough into the film to measure essentially bulk properties of the sample. The crystal structure of the sample was measured at Argonne National Laboratory using X-ray diffraction with a narrow beam at the SRICAT 2-BM bending magnet beamline of the Advanced Photon Source. The beam size used was 50 mm along the compositional spread direction. Magnetic properties were measured with a scanning Hall probe and scanning magneto-optical Kerr effect (SMOKE) measurements with spatial resolution of less than 200 mm. The CPD chip exhibited two dominant crystal structures. The Ni-rich alloys form a face-centered-cubic (FCC) structure (g-phase) and the Fe-rich alloys form a body-centered-cubic (BCC) structure (a-phase). The structural phase transition between the a-phase and the g-phase occurs around x=0.12 (with a rather broad two-phase overlap range from x=0.08 to 0.18). Both structures were found to be ‘‘rotationally’’ epitaxial to the underlying sapphire (0001) substrate. For the gphase, its [111] and [110] axes were found to be aligned with the [0001] and 1100] axes of the substrate, respectively. Due to the 6-fold rotational symmetry of the substrate lattice, all six rotationally equivalent domains were found. A typical transverse (
) scan result is shown in Fig. 1. While the in-plane mosaic spread of the film was relatively poor (about 5 in FWHM), the surfacenormal mosaic spread was excellent (narrower than 0.1 in some regions of the film). The a-phase alloy exhibited different but related rotational epitaxy. Its [110] direction was found to be parallel with the [111] axis of the g-phase. As we will show below, the alignment of the [110] axis of the Fe-rich a-phase to the [111] axis of Nirich g-phase is a consequence of the near coincidental pffiffiffi matching such that aa = 2 pffiffiffi of their lattice constants  ag = 3. In addition, the [110] axis of the Fe-rich alloy was found to be parallel with the [1120] axis of the substrate. Because of the two-fold rotational symmetry of the g-phase structure in the plane normal to the [110]

Fig. 1. Azimuthal (f) scan of the (200)-diffraction peak from FCCFe0.66Ni0.36 in the binary CPD.

Y.K. Yoo et al. / Intermetallics 9 (2001) 541–545

axis, a total of 12 equivalent rotational domains were found on the substrate with 6-fold rotational symmetry. The epitaxial growth is important here for a number of reasons. First, it made it possible to study off-specular diffraction, which allowed the determination of the crystal structures (in this case, distinguishing g-phase and a-phase). Second, well-oriented films of high quality allow reliable characterization of physical properties. While the Ni-rich and Fe-rich alloys retain their respective epitaxial orientations, the lattice constant of the Fe1
xNix system varies non-linearly throughout the CPD chip. Fig. 2a shows the specular diffraction pattern of the sample. The contour map and colored intensity profiles were generated by a series of linear scans with the momentum transfer (q) perpendicular to the substrate (0001) plane. The y? scan revealed only the (111) diffraction peak from the Ni-rich g-phase and the (110) peak from the Fe-rich a-phase due to the excellent texture of the sample. Because of the anisotropic stresses in the thin-film sample, the cubic structures of both phases are slightly distorted so that the lattice parameters obtained from the diffraction peak normal and parallel to the substrate crystal plane are slightly different (by a few percent). A detailed discussion of the crystal struc-

Fig. 2. (a) X-ray intensity map of the surface-normal d-spacing obtained by radial (–2) scans along the surface-normal direction. The observed d-spacings are converted to lattice constants. The superimposed curve in red is the bulk discrete composition results calculated from previously reported lattice constants at 15 C of both BCC (aphase) and FCC (g-phase) Fe–Ni alloys [20,23]. (b) Saturation magnetization (Ms) obtained with scanning Hall probe, and relative Kerr rotation measured by SMOKE with longitudinal configuration at a wavelength of 6328 A˚ (He–Ne laser).

543

ture and the residual strain in the film will be published elsewhere [17]. For simplicity, however, we converted the measured d-spacings perpendicular to the substrate crystal plane into BCC and using pffiffiffiFCC lattice parameters pffiffiffi the relationship, aBCC ¼ 2d110 and aFCC ¼ 3d111 . The regions on the left of 0% and the right of 100% of atomic Ni concentration correspond to the pure Fe and Ni phases, respectively. (The Ni concentration was assumed from the growth conditions and confirmed by the Auger analysis.) In these regions, the positions of the diffraction peak remain constant and the measured lattice constants are consistent with the reported bulk lattice constants of pure Fe and Ni. The solid lines in Fig. 2a are the previously reported lattice constants from the discrete composition samples [18,22]. Our data from the CPD chip reproduce all the important features in the lattice constant-composition curve. For example, the phase transition from BCC to FCC in bulk samples does not occur abruptly, similar to our observation. Moreover, the local maxima and the inflection point observed in bulk sample studies are also reproduced in our CPD data. Our data, however, do present some minor quantitative discrepancies. For example the phase transition starts at slightly lower Ni concentration in our CPD sample compared to the previous data obtained from bulk samples. This can be attributed to the stabilizing effect of the gphase at lower Ni concentration because the hexagonal symmetry of the substrate is likely to lower the interfacial energy of the FCC structure of g-phase. In addition, the lattice constants measured from our CPD chip are consistently smaller in the range of x=0.15–0.9. Moreover, our data exhibit some deviation from the linear decrease of lattice constant over x=0.5–0.9, probably due to the larger degrees of strain densities present in this region of the film. The extent of strain in the CPD chip is evident from the broadening of the diffraction peak [19]. It must be noted that strain influence between neighboring regions is expected. However, the range of such strain field is largely governed by grain size, which is well below the resolution of measurement probes. Fig. 2a clearly shows the evolution of the range of lattice constants, equivalent to the peak width as a function of Ni concentration. In particular, the peak width at x=0.6–0.8 is 2–3 times greater than the other regions of the Ni concentration and 4 times larger than that of the pure Ni phase. It is in this region that the lattice constant of CPD compositions deviates most from that of the bulk study. The relative saturation magnetization (Ms) and Kerr rotation measured by scanning Hall probe and SMOKE, respectively, are plotted in Fig. 2(b). The value of Ms is relative because the Hall probe detects the perpendicular, (0001) plane of the substrate, component of magnetic field near the edge of the magnetically saturated film. A dip in Ms at around x=0.3 was observed; the composition of the dip agrees well with the result of Yensen [20] and Masumoto [21]. This dip is generally

544

Y.K. Yoo et al. / Intermetallics 9 (2001) 541–545

associated with the invar behavior near (but not at) the FCC–BCC transition [22–24]. Such dip is due to magnetic phase transition to a mixed phase consisting of both ferromagnetic and antiferromagnetic regions [25]. The shape of magnetization-versus-composition curve is known to depend markedly not only on the fabrication and heat treatment of alloys, but also on applied field intensity [26]. In our case, the shape of the Ms dip measured with applied field intensity of 150 Oe is very similar to that reported by Bozorth measured under similar field intensity [26]. Relative Kerr rotation shows a quite different composition dependence than does Ms. A very steep dip was observed at around x=0.14, coincident with the largely discontinuous phase boundary [Fig. 2(a) and (b)] determined by the X-ray study. In general the magneto-optic Kerr effect depends on Ms with slightly different power dependence in different materials. For example, longitudinal Kerr rotation angle K was found to vary with Ms as K  (Ms)1.7
2.1 within FCC phase of different compositions [27,28]. The steep dip at x=0.14 suggests an anomaly in Kerr rotation at the structural phase transition. This is not surprising because the Kerr effect arises from the spin-orbit coupling and can be quite sensitive to the symmetries of electronic structure and crystal structure [29]. In particular, the Kerr effect arises from interband transitions that couple states of different angular momentum. The dip in the Kerr rotation coincident with the structural transformation suggests that the same states involved in the Kerr effect may also be driving the structural transformation in a Jahn–Teller sense. Our observation of anomalous Kerr rotation at the first-order structural phase transition in Fe–Ni system suggests that this anomaly may be used to study the structural phase transitions in other systems. Our data from the CPD chip demonstrate good overall agreement with the previous data taken from Fe–Ni alloys of the discrete compositions. The minor discrepancies between the present and previous data reveal opportunities for further study because they point to possible anomalies in the structure and properties of the composition-spread thin films. The knowledge of overall behavior and detailed discrepancies due to different substrate or annealing effects of structure–physical property relationship for a given material system are extremely useful for the academic research and industrial applications. For example, magnetic alloys such as Fe–Ni alloys are used widely in information industry in a thin film format. The knowledge from bulk sample study alone is often insufficient for device development. Our approach demonstrated here provided a systematic and highly efficient means to shorten the materials/ devices development cycle. In conclusion, our first demonstration of mapping the detailed structural and physical property relationship using thin film CPDs has generated convincing evidence

that the central issue of materials science can be addressed with considerably increased efficiency. The techniques presented here should allow rapid investigation of many important material systems and accelerate materials discovery based on a quick understanding of the relationship between structural and physical properties of new material system. One important application would be to map the structure–property relations in ferromagnetic shape memory alloys (FSMAs); some of these have been shown to exhibit a 6% magnetic-fieldinduced strain due to twin-boundary motion in their martensitic phase.

Acknowledgements This work was partially supported by DARPA, Contract No. MIPR98-0765 through the US Department of Energy under Contract No. DE-AC03-76SF00098. The use of Advanced Photon Source at Argon National Laboratory was supported by US Department of Energy under contract No. W-31-109-Eng-38.

References [1] Xiang X-D, Sun X, Bricen˜o G, Lou Y, Wang K-A, Chang H, et al. Science 1995;268:1738–40. [2] Briceno G, Chang H, Sun X, Schultz PG, Xiang X-D. Science 1995;270:273. [3] Wang J, Yoo YK, Gao C, Takeuchi I, Sun X, Xiang X-D, et al. Science 1998;279:1712. [4] Sun X, Gao C, Wang J, Xiang X-D. Appl Phys Lett 1997; 70:3353. [5] Danielson E, Golden JH, Mcfarland EW, Reaves CM, Weinberg WH, Wu XD. Nature 1997;389:944. [6] van Dover RB, Schneemeyer LF, Fleming RM. Nature 1998;392:162. [7] Danielson E, Devenney M, Giaquinta DM, Golden JH, Haushalter RC, Mcfarland EW, et al. Science 1998;279:837. [8] Sun X, Xiang X-D. Appl Phys Lett 1998;72:525. [9] Chang H, Gao C, Takeuchi I, Yoo Y, Wang J, Schultz PG, et al. Appl Phys Lett 1998;72:2185. [10] Chang H, Tacheuchi I, Xiang X-D. Appl Phys Lett 1999;74:1165. [11] Xiang X-D. Annu Rev Mater Sci 1999;29:149. [12] Hasebe M, Nishizawa T. Application of phase diagrams in metallugy and ceramics, NBS Special Pub. 496, 1978. vol. 2, p. 911–54. Washington, DC. [13] Romig Jr. AD. Bull Alloy Phase Diagr 1987;8:308. [14] Kennedy K, Stefansky T, Davy G, Zackay CF, Parker ER. J Appl Phys 1965;36:3808. [15] Yoo YK, Duewer F, Yang H, Yi D, Li JW, Xiang X-D. Nature 2000;406:704. [16] Yoo YK, Duewer F, Fukumura T, Yang H, Yi D, Hasegawa T, et al. Phys Rev B 2001;63:224421. [17] Chu YS, Yoo Y, Almer JD, Mancini DC, Ohnish T, Xiang X-D. To be published.. [18] Pearson WB. A handbook of lattice spacings and structures of metal alloys. New York: Pergamon, 1958 (chapter 11). Jette E, Foote F. Trans Am Inst Mining Met Engrs 1936; 120, 259. Owen EA, Yates EL, Sulley AH, Proc Phys Soc (London) 1937; 49: 315. Bradley AJ, Jay AH, Taylor A. Phil Mag 1037; 23: 545.

Y.K. Yoo et al. / Intermetallics 9 (2001) 541–545 [19] [20] [21] [22]

Warren BE, Averbach BL. J Appl Phys 1950;21:595. Yensen TD. J Am Inst Elec Engrs 1920;39:396. Masumoto H. Sci Repts Tohoku Imp Univ 1929;18:195. O’Handley RC. Modern magnetic materials. New York: John Wiley & Sons, 2000. [23] Dumpich G, Ka¨stner J, Kirschbaum U, Mu¨hlbauer H, Liang J, Lu¨beck Th, et al. Phys Rev B 1992;46:9258. [24] Schumann FO, Willis RF, Goodman KG, Tobin JG. Phys Rev Lett 1997;79:5166.

545

[25] Freeland JW, Grigorov IL, Walker JC. Phys Rev B 1998;57:80. [26] Bozorth RM. Ferromagnetism. New York:Van Nostrand, 1951 (chapter 5). [27] Sato T, Simatsu T, Miyahara T, Miyazaki T. IEEE Trans J Mag Japan 1991;6:869. [28] Miyahara T, Takahashi M. Jpn J Appl Phys 1976;15:291. [29] Liu ZQ, Bader SD. Rev Sci Inst 2000;71:1243.