Local Structure and Glass Transition in Zr-Based Binary Amorphous Alloys

Transcription

1 Materials Transactions, Vol. 46, No. 1 (25) pp to 2286 #25 The Japan Institute of Metals EXPRESS REGULAR ARTICLE Local Structure and Glass Transition in Zr-Based Binary Amorphous Alloys Tetsu Ichitsubo 1; *, Eiichiro Matsubara 1, Junji Saida 2 and Ho-Sou Chen 3 1 Department of Materials Science and Engineering, Kyoto University, Kyoto , Japan 2 Center for Interdisciplinary Research, Tohoku University, Sendai , Japan 3 Bell Laboratories, Lucent Technology, New Jersey 8833, USA The physical significance of the glass transition observed by differential scanning calorimetry (DSC) in the metallic glasses was considered through the measurements of the heating-rate,, dependence of the glass transition temperature,, and the crystallization temperature, T x,in the Zr 7 Cu 3 and Zr 7 Ni 3 amorphous alloys and X-ray study of their structures in as-quenched and crystallized states. Zr 7 Cu 3 exhibits the glass transition before crystallization, but Zr 7 Ni 3 is immediately crystallized at heating rates of conventional time scale in the DSC measurement. The heating rate c at the intersection of the two linear curves of and T x against log provides us with a significant measure to determine the glass-forming ability or thermal stability of the metallic glasses. By heating at larger than c, the crystallization is suppressed and the glass transition is clearly observed even in Zr 7 Ni 3. The thermal stability of the Zr 7 Cu 3 amorphous alloy is caused by retardation of crystallization due to the amorphous structure that is different from the Zr 2 Cu crystalline phase. In contrast, the thermal instability of Zr 7 Ni 3 is attributed to the structural similarity to the Zr 2 Ni crystalline phase. Thus, suppressing the crystallization is shown to be a key to enhance the thermal stability of the present amorphous alloys. (Received August 3, 25; Accepted September 8, 25; Published October 15, 25) Keywords: glass transition, metallic glasses, zirconium-based amorphous, differential scanning calorimetry, superheating, X-ray diffraction method, structual analysis, supercooled liquid 1. Introduction *Corresponding author, tichi@mtl.kyoto-u.ac.jp Amorphous state can be obtained in many alloys by rapidly quenching from the melts at a quenching rate more than 1 6 C/min, but in some alloys, it can be prepared even at a much slower quenching rate, e.g. 1 1 C/min. 1 5) These amorphous alloys with good glass-forming ability (GFA) usually show the glass-liquid transition and hence they are called metallic glasses, in conformity with the definition that a glass is an amorphous solid which exhibits a glass transition. 4) Initially, the metallic glasses have been found only in noble-metal based alloys, such as, Pd Ni P, Pt Ni P, etc. 3) Through 199s, they have been discovered in other multicomponent systems without containing either a noble metal or a metalloid element. 6) Their good GFA and reversible glass-liquid transition have attracted considerable attention, in comparison with oxide and polymer glasses. As it is mentioned above, the metallic glasses are characterized by the presence of the glass transition and their GFA is evaluated with some empirical parameters, such as =T l or =T m and a temperature span of a supercooled liquid region, T x ¼ T x, where is the glass transition temperature, T l the liquidus temperature, T m the melting temperature and T x the crystallization temperature. A fundamental question raised here is concerned about the thermal stability of metallic glasses: What kinds of factor determine whether the glass transition exists or not? That is, what is the physical significance of observed by differential scanning calorimetry (DSC) in the metallic glasses? We believe that considering this question would link to understanding of the thermal stability of metallic glasses, and recently reported the feature of the glass transition of a less stable metallic glass based on the DSC measurements. 7) In this paper, based on the experimental results of the heatingrate dependence of T x and and the X-ray structural analyses in both amorphous and crystallized states, we intend to discuss an essential characteristic of the glass transition and the thermal stability in metallic glasses. In this study, we have chosen two binary Zr 7 Cu 3 and Zr 7 Ni 3 amorphous alloys. The Zr 7 Cu 3 amorphous alloy shows the glass transition prior to crystallization on heating and the Zr 7 Ni 3 amorphous alloy is immediately crystallized. 7 9) In the sense that the glass transition exists, Zr 7 Cu 3 is like a metallic glass rather than a normal (marginal) amorphous alloy, and Zr 7 Ni 3 belongs to the latter. 2. Experimental Alloy ingots having the nominal compositions were prepared by the arc-melting method using 99.9 mass%zr, mass%cu and 99.9 mass%ni. Zr 7 Cu 3 and Zr 7 Ni 3 amorphous ribbons about.3 mm thick and 2 mm wide were prepared by a single roller melt-spinning technique in argon atmosphere with a roller spinning at 4 min 1 and quenching from 12 C for Zr 7 Cu 3 and from 13 C for Zr 7 Ni 3. The densities of the amorphous ribbons measured by Archimedes method are 7. g/cm 3 for Zr 7 Cu 3 and 7.3 g/cm 3 for Zr 7 Ni 3, respectively. Differential scanning calorimetry (DSC) measurements were carried out using a standard commercial instrument (Perkin Elmer Diamond DSC) with an about 5 1 mg sample. Heat flow was measured during heating at a constant heating rate up to ¼ 5 C/min, which is the highest attained in the present instrument. Structures of the as-prepared amorphous alloys were determined by high energy X-ray diffraction at SPring-8 synchrotron radiation on beam line BL4B2. Monochromatic incident X-rays of kev from a Si(111) monochromator were used as incident beams. Scattering from the samples was detected by a portable Ge solid state detector. Observed intensities were corrected for absorption, polarization and

2 Local Structure and Glass Transition in Zr-Based Binary Amorphous Alloys 2283 Compton scattering, and converted to electron units per atom by the generalized Krogh Moe Norman method, 1) using the tabulated X-ray atomic scattering factors and anomalous dispersion terms. 11) An interference function QiðQÞ was computed from the resultant coherent scattering intensity, I eu ðqþ: QiðQÞ ¼QðI eu ðqþ hf 2 iþ=hf i 2 ; ð1þ hf i¼ X2 j¼1 x j f j ; hf 2 i¼ X2 j¼1 x j f 2 j ; where Q ¼ 4 sin =, 2 is the scattering angle, is the wavelength, and x j and f j are the atomic fraction and X-ray atomic scattering factor of the jth element. A radial distribution function (RDF) is calculated by Fourier transformation of QiðQÞ: 4r 2 ðrþ ¼4r 2 o þ 2r ZQ max QiðQÞ sin QrdQ; where ðrþ is the number density function, o the average number density and Q max the maximum Q-value in the present measurement. The advantage in the high energy diffraction method is a large Q max value of about 27 nm 1 in the present measurements which reduces a truncation error in Fourier transformation so that the spatial resolution of RDF is greatly improved. Coordination numbers and atomic distances at the first peak of RDF were evaluated by fitting the experimental QiðQÞ through the least squares method. 12) Crystalline phases in the crystallized samples during the DSC measurements were investigated by the ordinary 2 X-ray diffraction with MoK radiation. ð2þ ð3þ a Exothermic b Exothermic 2 2 Zr 7 Cu 3 5 C/min 4 C/min 3 C/min 2 C/min 1 C/min 4 C/min 2 C/min 1 C/min Zr 7 Ni 3 5 C/min 4 C/min 3 C/min 2 C/min 1 C/min 4 C/min 2 C/min 1 C/min Temperature ( C) 4 Temperature ( C) 5 4 C/min 5 C/min Fig. 1 DSC profiles at various heating rates for (a) Zr 7 Cu 3 and (b) Zr 7 Ni 3 amorphous alloys Results Figures 1(a) and (b) show the DSC profiles at various. These profiles are obtained by subtracting the DSC profile of the second run for the crystallized sample in the first run from that of the first run for the amorphous sample. was evaluated as an onset temperature of an endothermic deviation from the nearly linear variation with temperature in Fig. 1. Crystallization accompanies a sharp exothermic peak, and T x was determined as an onset temperature of the peak. In the thermally stable Zr 7 Cu 3 amorphous alloy, is observed at every. On the other hand, in the less stable Zr 7 Ni 3 amorphous alloy, clearly appears only at very high rates above about 2 C/min. Figure 2 shows the interference functions QiðQÞ of the two amorphous alloys. Dotted curves correspond to the functions fitted for determination of coordination numbers and atomic distances in Table 1. RDFs calculated through Fourier transformation of QiðQÞ are shown in Fig. 3. Positions of the nearest neighbor pairs are indicated in the figure. The first peak of RDF for Zr 7 Cu 3 consists of three atom pairs, i.e., Cu Cu at :26 :2 nm, Zr Cu at :284 :1 nm and Zr Zr at :319 :1 nm. These atomic distances are almost equal to those calculated from Goldschmidt radii of Zr (.16 nm) and Cu (.128 nm). As is seen in Table 1, occupancies of Zr and Cu around Zr calculated from coordination numbers for Zr Zr and Zr Cu pairs are close to the concentrations of Zr and Cu in the original glassy matrix, respectively. These structural parameters in Zr 7 Cu 3 indicate that the amorphous alloy is a random structure. This appears in the shape of the first peak of RDF for Zr 7 Cu 3 in Fig. 3(a). In contrast, the first peak of the Zr 7 Ni 3 amorphous alloy shows a clear split in Fig. 3(b). The peak at larger r corresponds to Zr Zr pairs and that at lower r accords with Zr Ni and Ni Ni pairs. In Zr 7 Ni 3, the atomic distance of Zr Ni pairs (.27 nm) is about 5% shorter than the distance (.284 nm) calculated from Goldschmidt radii of Zr and Ni (.124 nm). At the same time, the atomic distances of Zr Zr and Ni Ni pairs are similar to those calculated from the atomic radii. Only Zr Ni pairs show differently from the others. The occupancies of Zr and Ni around Zr in Table 1 are quite different from their concentrations. These structural parameters in Zr 7 Ni 3 indicate that some chemical short range order (CSRO) clusters are formed in the amorphous state because of the presence of a strong chemical bond between Zr and Ni atoms. From this structural viewpoint, the Zr 7 Ni 3 amorphous alloy is obviously different from the Zr 7 Cu 3 amorphous alloy consisting of randomly arranged atoms. Figure 4 shows two examples of X-ray diffraction profiles at ¼ 2 and 5 C/min in the crystallized Zr 7 Cu 3 and Zr 7 Ni 3 alloys. All the samples were prepared by quenching

3 2284 T. Ichitsubo, E. Matsubara, J. Saida and H.-S. Chen Qi(Q) (nm -1 ) Qi(Q) (nm -1 ) 5 5 (a) Zr 7 Cu 3 (b) Zr 7 Ni Q (nm -1 ) exp. cal. exp. cal. Fig. 2 Interference functions QiðQÞ of (a) Zr 7 Cu 3 and (b) Zr 7 Ni 3 amorphous alloys. Solid and dotted curves correspond to the experimental profile and the profile fitted for determination of coordination numbers and atomic distances in Table 1. 2π 2 rρ(r) (nm -2 ) 2π 2 rρ(r) (nm -2 ) Zr-Cu Cu-Cu Zr-Ni Ni-Ni Zr-Zr Zr-Zr (a) Zr 7 Cu 3 (b) Zr 7 Ni r (nm) Fig. 3 Radial distribution functions of (a) Zr 7 Cu 3 and (b) Zr 7 Ni 3 amorphous alloys. Table 1 Atomic distances and coordination numbers in Zr 7 Cu 3 and Zr 7 Ni 3 amorphous alloys are compared with those calculated from crystalline data of Zr 2 Cu and Zr 2 Ni. Occupancies of Zr and Cu in the amorphous alloys are evaluated from the coordination numbers. r is the variations of atomic distances due to crystallization. Zr 7 Cu 3 r ij N ij Occu. Zr 2 Cu r ij N ij r Zr Zr :319 :1 7:9 :2 71% Zr Zr :9% Zr Cu :284 :1 3:2 :1 29% Zr Cu % Total Cu Zr :284 :1 7:5 :2 74% Cu Zr % Cu Cu :26 :2 2:6 :6 26% Cu Cu % Total Zr 7 Ni 3 r ij N ij Zr 2 Ni r ij N ij Zr Zr :319 :1 9:1 :1 78% Zr Zr % Zr Ni :27 :1 2:6 :1 22% Zr Ni % Total Ni Zr :27 :1 6:1 :2 75% Ni Zr % Ni Ni :253 :7 2: :6 25% Ni Ni % Total from the temperature where the exothermic peak of crystallization appears (i.e., rapidly cooling from the peak temperature in the DSC profile). Crystalline phases formed in Zr 7 Cu 3 and Zr 7 Ni 3 crystallized in the DSC measurements at every are tetragonal Zr 2 Cu (I4/mmm, 139) and Zr 2 Ni (I4/mcm, 14), respectively. Coordination numbers and atomic distances calculated from the crystalline data are also tabulated in Table 1 for comparison, where the atomic distances of the Zr Zr pairs are averages weighted by coordination numbers based on the crystalline structures. Structures in these crystalline phases are quite different each other because of the strong chemical bond between Zr and Ni. The Cu Cu distance is more than 2% longer than the Ni Ni distance and the Zr Cu distance is also about 5% longer than the Zr Ni distance. 4. Discussion The reduced glass transition and crystallization temperatures T rg and T rx, which are obtained by dividing the experimental and T x by the liquidus temperatures 1 C for Zr 7 Cu 3 and 11 C for Zr 7 Ni 3, are plotted as a function of log in Fig. 5. The T rg and T rx values are fitted with an empirical linear relation proposed by Lasocka: 13)

4 Local Structure and Glass Transition in Zr-Based Binary Amorphous Alloys 2285 Intensity (counts/s) a Zr 7 Cu 3 5 C/min Zr 7 Cu 3 2 C/min θ (deg) T rx,rg T x,g =T l ¼ a x,g þ b x,g log : T rx largely increases with increase in. On the other hand, T rg shows a much smaller change in every. This evidently shows that the glass transition and crystallization are controlled by different kinetic processes. Since the crystallization is controlled mainly by single-atom diffusion, the crystallization is largely retarded by a delay of the diffusion at a high heating rate. A small positive slope for T rg could be attributed to the structural relaxation of the rapidly quenched amorphous ribbons. The heating rates c and reduced temperature T rc at the intersection of the two and T x curves are 1. C/min and.46 (315 C) for Zr 7 Cu 3 and 17 C/min and.47 (374 C) for Zr 7 Ni 3, respectively. By heating the amorphous alloys at higher than c, we can observe the glass transition. The crystallization takes place prior to the glass transition at lower than c.inzr 7 Cu 3, c is less than the commonly used heating rate around 1 C/min. Thus, the glass transition is b Zr 7 Ni 3 5 C/min Zr 7 Ni 3 2 C/min θ (deg) Fig. 4 X-ray diffraction profiles of (a) Zr 7 Cu 3 and (b) Zr 7 Ni 3 crystallized in DSC measurements at ¼ 5 and 2 C/min. MoK radiation was used. Reduced Temperature, T / T l Zr 7 Cu 3 (T l ~ 1 C) T x Zr 7 Ni 3 (T l ~ 11 C) T x Heating rate, β ( C/min) Fig. 5 Reduced glass transition and crystallization temperatures T rg (¼ =T l ) and T rx (¼ T x =T l ) are plotted as a function of log, where T l is 1 C for Zr 7 Cu 3 and 11 C for Zr 7 Ni 3. ð4þ normally observed in Zr 7 Cu 3. c for Zr 7 Ni 3 is just in the range of the heating rates commonly used. Thus, in Zr 7 Ni 3, at less than c in the DSC measurement, the amorphous phase is crystallized due to the atomic diffusion, and by heating at more than c the diffusion of atoms is suppressed up to a relatively high temperature, and the glass transition is observed. Consequently, a small c value denotes better thermal stability of the amorphous state. T rx of Zr 7 Cu 3 is always higher than that of Zr 7 Ni 3 and the slope of T rx (i.e., b x in eq. 4) of Zr 7 Cu 3 is larger than that of Zr 7 Ni 3. Reminding that T rg s for both of the amorphous alloys are comparable at every, we can conclude that suppression of crystallization is essential to stabilize the amorphous state in the present alloys. The differences of the atomic distances between the amorphous and resulting crystalline phases are shown in Table 1. The differences of Zr Zr and Zr Cu pairs are less than 2%. The difference of Cu Cu pairs, however, reaches 24%. In Zr 7 Ni 3, the differences of every pairs are 2.2 to 3.9% that is slightly larger than those of Zr Zr and Zr Cu pairs in Zr 7 Cu 3, but much less than that of Cu Cu pairs. Therefore, a more than 2% change of the Cu Cu distance on average is required for the formation of the Zr 2 Cu crystalline phase in the Zr 7 Cu 3 amorphous alloy. This structural difference between the amorphous and crystalline phases makes rearrangements of atoms, especially in the Cu Cu pair correlations, more complicated to retard the crystallization in the Zr 7 Cu 3 amorphous alloy. By contrast, however, the structure of the Zr 7 Ni 3 amorphous alloy resembles that of Zr 2 Ni in terms of the interatomic distances. Thus, it is expected that the Zr 7 Ni 3 amorphous alloy is easily crystallized to form the Zr 2 Ni phase. It is noted on the crystallization in Fig. 4 that the peak width of Zr 2 Cu is much narrower than that of Zr 2 Ni at every. This indicates that the size of the Zr 2 Cu precipitates evaluated from the peak width at the half maximum is a few times larger than that of Zr 2 Ni. Thus, we consider that the rate-determining process for the crystallization in Zr 7 Cu 3 is the nucleation rather than the growth. In the Zr 7 Ni 3 amorphous alloy, as stated earlier, there are many CSRO clusters similar to Zr 2 Ni in terms of the local structure. Thus, it is natural to consider that there is not so large barrier for the nucleation of Zr 2 Ni. Actually, very fine Zr 2 Ni precipitates are formed. Typical thermally-stable glassy solids, such as oxide and polymer glasses, have a random network structure consisting of certain structural units (molecules). In these glasses, atoms composing a molecule are strongly coupled together with covalent bonds. Thus, the structural units are formed from regular polyhedra (for example, SiO 4 tetrahedra in a silicate glass), and such glasses show extremely good GFA and reversible glass transition. In contrast, in the metallic glasses mainly composed of metallic elements, atoms are mostly linked with the metallic bonds much weaker than the covalent bonds and, therefore, they can diffuse in the amorphous matrix at a relatively low temperature. The atoms in the metallic glasses form densely packed and isotropic clusters like icosahedral clusters in Zr-based metallic glasses. In the stable oxide or polymer glasses, both glass transition and crystallization proceed with rearrangement of the

5 2286 T. Ichitsubo, E. Matsubara, J. Saida and H.-S. Chen structural units (molecules), in which the atomic diffusion is not so significant. In the crystallization of the metallic glasses, however, atomic diffusion plays a significant role as it was shown in the present study. This characterizes the glass transition and crystallization in metallic glasses. Consequently, the thermal stability of the metallic glasses may depend on making the crystallization difficult by reducing the mobile atoms or making a significant difference between amorphous and crystallized structures. 5. Conclusion In the Zr 7 Ni 3 amorphous alloy, the glass transition is observed only at an extremely high heating rate in the DSC measurements. In contrast, the Zr 7 Cu 3 amorphous alloy always shows the glass transition. The heating-rate dependence of T x and in the Zr 7 Cu 3 and Zr 7 Ni 3 amorphous alloys and the structures in both amorphous and crystallized states were investigated. Both the experimental reduced glass transition and crystallization temperatures, T rg and T rx, show linear relations against log. The heating rate c at the intersections of the two curves, i.e. 1. C/min for Zr 7 Cu 3 and 17 C/min for Zr 7 Ni 3, provides us with a measure of GFA and thermal stability of the metallic glasses. Since c of Zr 7 Ni 3 is just in the same order of commonly used in the DSC measurements, the glass transition is not usually observed. However, at larger than c, the glass transition is revealed even in Zr 7 Ni 3. In contrast, for Zr 7 Cu 3, the glass transition is observed at every. The structural difference between the amorphous and crystalline phases, especially in the Cu Cu pair correlations, has been revealed in the Zr 7 Cu 3 amorphous alloy. The similarity of the structure of the Zr 7 Ni 3 amorphous alloy to that of the Zr 2 Ni crystalline phase has also been detected. It is concluded that the different thermal stabilities of these metallic glasses originate from these different structural characteristics in the two amorphous alloys. Acknowledgement This work was partly supported by Grant-in-Aid for Scientific Research on the Priority Area Investigation of Materials Science of Bulk Metallic Glasses (No ) from the Ministry of Education, Science, Sports and Culture, Japan. The high energy X-ray diffraction measurements at SPring-8 were carried out at BL4B2 of SPring-8 under the approval of the SPring-8 Program Advisory Committee. REFERENCES 1) H. S. Chen and D. Turnbull: J. Chem. Phys. 48 (1968) ) H. S. Chen and D. Turnbull: Acta Metall. 17 (1969) ) H. S. Chen: Acta Metall. 22 (1974) ) S. R. Elliott: Physics of Amorphous Materials, (Longman, London and New York, 1984). 5) M. Telford: Materials Today 7 (24) 36. 6) A. Inoue: Acta Mater. 48 (2) ) T. Ichitsubo, E. Matsubara, H. Numakura, K. Tanaka, N. Nishiyama and R. Tarumi: Phys. Rev. B 72 (25) ) J. Saida, M. Kasai, E. Matsubara and A. Inoue: Annals de Chimie- Science des Materiaux 27 (22) 77. 9) Z. Fu-Qian: Mater. Sci. Eng. 97 (1988) ) Y. Waseda: The Structure of Non-Crystalline Materials (McGraw-Hill, New York, 198) 27 41, ) International Tables for Crystallography, Vol. C, edited by A. J. C. Wilson (Kluwer Acad. Pub, 1995), pp ) E. Matsubara, S. Tanaka, A. Makino and T. H. Chiang: Mater. Trans. 45 (24) ) M. Lasocka: Mater. Sci. Eng. 23 (1976) 173.