Journal of Metastable and Nanocrystalline Materials Vols. 20-21 (2004) pp. 35-40 online at http://www.scientific.net (2004) Trans Tech Publications, Switzerland Crystallization of Pd 40 Cu 30 Ni 10 P 20 Bulk Glass N. Wanderka, E. Davidov, G. Miehe 1, V. Naundorf, M.-P. Macht, J. Banhart Hahn-Meitner-Institut Berlin, Glienicker Str. 100, D-14109, Berlin, Germany 1 Technische Universität Darmstadt, Petersenstr. 23, D-64287 Darmstadt, Germany Keywords: Bulk amorphous alloy, Crystallization, Transmission electron microscopy, Three dimensional atom probe. Abstract. The stability of the Pd 40 Cu 30 Ni 10 P 20 glass as one of the most stable metallic bulk glasses was investigated. The decomposition and crystallization of as-quenched Pd 40 Cu 30 Ni 10 P 20 at temperatures above the glass transition was followed by differential scanning calorimetry (DSC), X-ray diffraction, transmission electron microscopy and by three-dimensional atom probe (3DAP). In the as-cast glass no indication of crystals could be detected. Furthermore, complete chemical homogeneity of the glass even on the nm scale was demonstrated by the 3DAP-analysis. After annealing at 590 K for 30 h a primary crystalline phase Pd 30 Cu 12 Ni 31 P 27 with trigonal structure is formed. After heating in the DSC up to 660 K two crystalline phases with different compositions are found. One of them with nearly the same composition as the bulk glass is also trigonal. The other one of very small size is rich in Pd. After slow cooling of the melt, at least five crystalline phases were found. The first one is Cu 3 Pd, a cubic phase of ordered Cu 3 Au-type. The second phase is tetragonal primitive Pd 50 Cu 35 Ni 2 P 13. The third phase is tetragonal body centered Pd 50 Cu 20 Ni 10 P 20. The fourth and fifth are of composition Pd 40 Cu 39 Ni 2 P 19 and Pd 21 Cu 20 Ni 35 P 24, respectively. Introduction The multicomponent bulk glasses containing Pd, Cu, Ni, P exhibit excellent glass forming ability (GFA). Extremely low critical cooling rates down to 0.1 K/s [1, 2] can be applied for their production. The thermal stability of such glasses against crystallization opens a wide time window for experimental investigations, e.g. of the crystallization behaviour. The early stages of crystallization seem to play an important role for the large GFA and thermal stability. Previously investigations have been reported: in-situ observations of the crystallization of Pd 40 Cu 30 Ni 10 P 20 glass [3, 4], formation of primary crystals in Pd 43 Cu 27 Ni 10 P 20 and Pd 40 Cu 30 Ni 10 P 20 glasses [5, 6], kinetics of phase transformation in Pd 43 Cu 27 Ni 10 P 20 [7], atom probe studies of Pd 46 Cu 36 P 18, Pd 48 Cu 32 P 20 and Pd 40 Cu 30 Ni 10 P 20 glasses [8], and diffusion in Pd 43 Cu 27 Ni 10 P 20 and Pd 40 Cu 30 Ni 10 P 20 glasses [9, 10]. In the present work we report about an amorphous decomposition during annealing of the Pd 40 Cu 30 Ni 10 P 20 glass in the supercooled liquid region, before several crystalline phases are formed. The microstructural evolution is followed by differential scanning calorimetry (DSC) and X-ray diffraction (XRD). The characterization of different phases was carried out by transmission electron microscopy (TEM) and by 3-dimensional atom probe (3DAP). Experiment Amorphous Pd 40 Cu 30 Ni 10 P 20 bulk glass was produced by the following procedure: alloying of the pure elements in a quartz tube; remelting in a HF induction device and quenching; fluxing in B 2 O 3 ; and remelting again in the HF induction device and quenching. The resulting glassy ingots were All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of the publisher: Trans Tech Publications Ltd, Switzerland, www.ttp.net. (ID: 134.30.19.8-15/10/06,16:47:53)
36 Metastable, Mechanically Alloyed and Nanocrystalline Materials 2003 characterized by optical microscopy and scanning electron microscopy to ascertain the absence of primary crystals, which cannot be detected by XRD analysis if the volume fraction is < 5 %. The annealing temperatures and times were chosen with the aim to characterize the different stages of crystallization. Annealing experiments were performed as follows: a) for 12h to 150h at 590 K (above the glass transition); b) heating the glass in the DSC at a rate of 20 K/min up to 660 K (beginning of the first crystallization peak) and subsequent fast cooling; and c) melting the alloy and subsequent slow cooling to room temperature. Samples appropriate for the analysis were cut from the glassy ingots. TEM samples were prepared electrochemically by jet polishing at 293 K with 30 V, using a solution of CH 3 COOH and HClO 4 in the ratio of 9:1 and subsequent thinning by ion beam milling. Two Philips TEM microscopes a EM400 operating at 100 KV and a CM30 operating at 300 kv with an attached EDS spectrometer were used for the microstructural analysis. The beam size for the EDS measurements was about 20 nm. The 3DAP specimens were thin prisms with a size of 0.2 x 0.2 x 10 mm 3. These atom probe samples were prepared by electropolishing one end in a solution of 2% HClO 4 in butoxyethanol at room temperature with 20 V DC. The 3DAP analysis was performed in ultra-high vacuum of 10-8 Pa with a pulse repetition frequency of 1000 Hz. The fraction of pulse voltage to the standing tip voltage was 0.2, and the tip temperature was about 40 K. A 3DAP (TAP, CAMECA) [11] was employed in the present study. Polished slices from the ingots of about 1mm thickness were used for X-ray diffraction analysis using Cu-Kα radiation. Results and Discussion A typical DSC scan (heating rate of 20 K/min) of the as-quenched bulk glass is shown in Fig. 1. The thermal stability of this glass is indicated by the wide temperature range of 90 K between glass transition at 570 K and onset of crystallization at 660 K. Some exothermal and endothermal peaks are observed, which are attributed to crystallization and to phase transition. Above 800 K the crystallization during the DSC run is completed. If the glass is isothermally annealed just above the glass transition at 590 K for 12 h to 150 h the formation of the first crystalline phases requires a long time. The results of the XRD investigation for several annealing times are shown in Fig. 2. Up to 20 h, the diffuse maxima of the as-quenched glass remain unchanged. After 30 h annealing tiny Endothermic heat Flow /arb. units 0,4 0,2 0,0-0,2-0,4-0,6-0,8-1,0-1,2-1,4 T g 590 K -1,6 500 600 700 800 T x Temperature / K Fig. 1 DSC scan of the as-quenched Pd 40 Cu 30 Ni 10 P 20 (heating rate 20 K/min). Fig. 2 X-ray diffractogram of the bulk glass Pd 40 Cu 30 Ni 10 P 20 bulk glass after heat treatment at 590 K for different times.
Journal of Metastable and Nanocrystalline Materials Vols. 20-21 37 Bragg peaks superimpose the diffuse diffractogram of the amorphous phase. In fact the 3DAP analysis of the glass annealed for 12 h (not shown here) does not show any evidence of a compositionally modulated microstructure. All elements of the alloy are distributed homogeneously in the respective analyzed regions of about 500 nm length. However, after 20 h annealing time small regions of about 30 nm size depleted in Ni and enriched in Pd were detected by 3DAP. The three-dimensional reconstruction of Ni atoms within an analyzed volume of 9 x 9 x 65 nm 3 is shown in Fig. 3. Microchemical information is obtained from the concentration depth profiles taken for Ni from the small box of 1.5 x 1.5 x 65 nm 3 marked in the analyzed volume. The concentration depth profiles indicate local heterogeneities which exceed the 2σ limits of statistical scattering. One of the regions is depleted and the other one is enriched in Ni. The compositions of the Ni rich and Ni depleted areas together with the nominal composition of bulk glass are listed in Table 1. It is noted that the concentration of Pd and of Cu are anticorrelated to that of Ni and of P. Such decomposition requires long-range diffusion. 9 nm 65 nm 9 nm Isoconcentration surface 10 at% Ni 1.5x1.5x65 nm 3 Fig. 3 Three dimensional elemental mapping of Ni after annealing the Pd 40 Cu 30 Ni 10 P 20 glass for 20 h at 590 K. The 10 at.% Ni isoconcentration surface is indicated. The marked small box of 1.5 x 1.5 x 65 nm 3 across the interfaces is used for evaluation of the concentration depth profile (below). The microstructure of the sample annealed up to 30 h is investigated using TEM. The crystallization starts with the formation of a starfish -shaped primary phase. Fig. 4 shows a bright field TEM image of such crystals embedded in the amorphous matrix. The characteristics of crystals and matrix are demonstrated by the attached selected area electron diffraction (SAED) pattern. The crystalline phase is trigonal R-centered with a=1.91 nm, c=1.87 nm, the SAED-pattern shows the [210] zone. The composition as measured by EDX is Pd 30 Cu 12 Ni 31 P 27 (see Table 1), i. e. the crystals are enriched in Ni and P and depleted in Pd and Cu. A similar composition (Pd 32 Ni 27 Cu 16 P 25 ) of primary crystals in Pd 43 Ni 10 Cu 27 P 20 glass after annealing for 200 s at 675 K has been reported by [6]. However, the reported cell differs from that in the present work. This may be due to the higher
38 Metastable, Mechanically Alloyed and Nanocrystalline Materials 2003 annealing temperature applied by [6]. The composition of the amorphous phase between the crystal arms is anticorrelated to that of the crystals (see Table I). The average composition is Pd 60 Cu 27 Ni 7 P 6. This is similar to the secondary crystalline phase formed around the primary crystals observed by [6]. These results suggest that the decomposition kinetics of PdCuNiP-based glasses proceeds by the similar concentration changes. The typical time for onset of crystallization derived from the XRD investigations in Fig. 2 is in accordance with the TEM results were no crystalline phase had been observed for annealing times less than 30 h. From a comparison of these results with the 3DAP observations it must be concluded that a long incubation period is necessary to achieve a significant decomposition before the nucleation of crystallites can start. As a long range decomposition is controlled by diffusion, this process will contribute significantly to the thermal stability of the PdCuNiP glasses. A [100] B [210] Fig. 4 Bright field TEM micrograph of a starfish - the shaped primary crystal which has formed at first during annealing the Pd 40 Cu 30 Ni 10 P 20 glass for 30 h at 590 K. The electron diffraction pattern of [210] zone for Pd 30 Cu 12 Ni 31 P 27 phase and diffuse rings corresponding to amorphous matrix are presented. Fig. 5 Bright field TEM micrograph of microstructure formed after heating the Pd 40 Cu 30 Ni 10 P 20 glass in the DSC up to 660 K and subsequent fast cooling to room temperature. Two phases marked by A and B are visible. The electron diffraction pattern of the [100] zone of the crystal B is shown. X-ray diffractograms and TEM observations of the sample annealed up to 150 h (not shown here) exhibit a very complex microstructure which contains more that 5 phases. Fig. 5 shows the microstructure of a sample after annealing in the DSC up to 660 K (beginning of the first crystallization peak). Very small precipitates (A) embedded into large crystals (B) can be seen in this Figure. The composition of the small precipitates is about Pd 61 Cu 19 Ni 4 P 16. Because the size of the precipitates is smaller than the electron beam the contribution of the matrix may falsify the measured composition of the small precipitates. More research is under progress to analyze the structure of this phase. The crystals marked by B have nearly the same composition as the asquenched glass. The unit cell of this phase was determined as trigonal R- centered with a=0.678 nm, c=8.11 nm. The diffraction pattern of the [100] zone of the crystal B is shown in the inset of Fig. 5. In order to determine the stable phases the Pd 40 Cu 30 Ni 10 P 20 alloy was heated up to a temperature of 886 K (liquid melt) and then slowly cooled down to room temperature. An example of the resulting microstructure is shown in the bright field TEM image in Fig. 6. Two of the at least five phases are visible. Small precipitates are embedded into larger crystals of the composition Pd 50 Cu 35 Ni 2 P 13. The cell of this phase was identified as tetragonal primitive with a=0.486 nm, c=1.067 nm. It has a
Journal of Metastable and Nanocrystalline Materials Vols. 20-21 39 strongly developed subcell with c' = c/2 = 0.534 nm. The diffraction symbol of this tetragonal subcell is P 4/ c. The [100] zone diffraction pattern is figured in Fig. 6. The composition of the small precipitates is Cu 3 Pd and the structure is of the ordered L1 2 type (see Fig. 6). Besides the small Cu 3 Pd crystals, the microstructure contains very big grains of Cu 3 Pd crystals with small amount of Ni and P (Cu 75 Pd 21 Ni 1.5 P 2.5 ). The structure of these crystals is also L1 2 ordered with the lattice parameter a = 0.368 nm. The formation of a stable Cu 3 Pd type phase has already been reported by different authors [6, 12, 13]. The presence of lines which are forbidden by F-centring indicates an ordering of Cu and Pd. The intensity of these lines, however, is by far smaller than expected for a complete ordering. 2000 111 + + Cu 3 Pd [100] 300 nm [110] Fig. 6 Bright field TEM micrograph of the Pd 40 Cu 30 Ni 10 P 20 alloy slowly cooled down from the liquid melt to room temperature. Small precipitates of the Cu 3 Pd type with L1 2 structure are embedded in large tetragonal crystals of Pd 50 Cu 35 Ni 2 P 13. The electron diffraction pattern of the [100] zone for Cu 3 Pd and [110] zone for Pd 50 Cu 35 Ni 2 P 13 phase are presented. Intensity / [a.u.] 1500 1000 x 200 + x x Pd 50 Cu 20 Ni 10 P 20 500 x 220 100 x x x x + x x x x + 0 20 30 40 50 60 70 80 2θ / degree Fig. 7 X-ray diffractogram of the Pd 40 Cu 30 Ni 10 P 20 alloy slowly cooled down from the liquid melt to room temperature. The reflections of Cu 3 Pd and Pd 50 Cu 20 Ni 10 P 20 are marked. Another stable phase which forms large grains is Pd 50 Cu 20 Ni 10 P 20. This phase is tetragonal body centered with lattice parameters a=0.97 nm, c=0.46 nm. The similarity of the cell parameters to that of some X 3 P-structures (X=Mo, I-42m [14], Cr, I-4 [15]) suggests a relation between the structures. This relation is confirmed by the x-ray diffractogram. Preliminary calculations prove the presence of a considerable amount of material which obviously displays the corresponding structure type (see XRD). Hence the composition should rather be written (Pd,Cu,Ni) 75 P 25. A phase of similar Table 1 Chemical compositions of the Pd 40 Cu 30 Ni 10 P 20 bulk glass, decomposed areas and the first crystals after annealing at 590 K, measured by 3DAP and TEM/EDX (in at. %) Pd Cu Ni P as-quenched / nom. composition 40 30 10 20 3DAP, 590 K/20 h region 1 41.9 23.9 15.9 18.3 region 2 51 30 1.9 17.1 TEM/EDX, 590 K/30 h primary crystals 30 12 31 27 amorphous regions between crystals 60 27 7 6.0
40 Metastable, Mechanically Alloyed and Nanocrystalline Materials 2003 composition was observed after annealing at 675 K for 200 s [6]. This phase, however, is orthorhombic. Embedded in the tetragonal phase, small Ni rich precipitates of composition Pd 21 Cu 20 Ni 35 P 24 were found. In addition, a phase of composition Pd 40 Cu 39 Ni 2 P 19 was identified by TEM/EDX analysis. The XRD pattern of the material cooled slowly from the liquid melt also indicates the presence of numerous phases (Fig. 7). The diffractogram is governed by lines of Cu 3 Pd and of the tetragonal body centred phase Pd 50 Cu 20 Ni 10 P 20. Several Bragg-peaks cannot unambiguously be attributed to one of the other identified phases because the cells are large and only very few information about the structure and hence about the structure factors is available. Furthermore, it cannot be excluded that still more phases of small volume fraction are present in the alloy. The present results demonstrate that the decomposition of a glass with different nominal composition of Pd, Cu, Ni and P proceeds in a similar way. Depending on the annealing conditions various crystalline phases form which differ significantly in composition and structure. None of these phases equals the equilibrium phases, i.e. the phases formed at the early stages of crystallization of the glass were not found in the microstructure of the slowly cooled liquid melt. Summary The early stages of crystallization in Pd 40 Cu 30 Ni 10 P 20 bulk glass has been investigated by DSC, XRD, 3DAP and TEM. After 20 h annealing time at a temperature of 590 K decomposition of all elements was measured on the nm scale by 3DAP. After 30 h annealing time at 590 K the formation of the primary crystals was observed by TEM/EDX. The first crystalline phase has a trigonal R- centered cell with the lattice parameters: a=1.91 nm, c=1.87 nm. Two further phases are formed after annealing in the DSC at 660 K. One of these phases with about the average glass composition was identified as a trigonal R-centred with a=0.678 nm and c=8.11 nm. The other one of very small grain size is rich in Pd. Cell parameters have not yet been determined. During slow cooling of the liquid melt, at least five crystalline equilibrium phases were found. The first is an ordered cubic Cu 3 Pd phase (L1 2 structure) with small admixtures of P and Ni and with the lattice parameter a=0.368 nm. The second phase Pd 50 Cu 35 Ni 2 P 13 is tetragonal with lattice parameters a=0.486 nm and c=1.097 nm. The third phase Pd 50 Cu 20 Ni 10 P 20 is tetragonal body centered with lattice parameters a=0.97 nm and c=0.46 nm. The fourth and fifth are of composition Pd 40 Cu 39 Ni 2 P 19 and Pd 21 Cu 20 Ni 35 P 24, respectively. References [1] A. Inoue, N. Nishiyama, Mater. Trans. JIM, 37 (1996) 181 [2] A. Inoue, N. Nishiyama and H.M. Kimura, Mater. Trans. JIM. 38 (1997) 179 [3] N. Nishiyama, M. Matsushita and A. Inoue, Scripta Mater. 44 (2001) 1261 [4] N. Nishiyama and A. Inoue, Mat. Sci. Forum Vols. 386-388 (2002) 105 [5] J. F. Löffler, J. Schroers and W. l. Johnson, Applied Phys. Lett. Vol. 77, Nr. 5 (2000) 681 [6] E. Pekarskaya, J. Schoers and W. l. Johnson, Mat. Res. Symp. Proc. Vol. 644 (2001) L12.7.1 [7] J. Schroers, W. l. Johnson and R. Busch, Applied Phys. Lett. Vol. 77, Nr. 8 (2000) 1158 [8] H.G. Read, K. Hono, A.P. Tsai, A. Inoue, Mater. Sci. & Eng. A226-228 (1997) 453 [9] V. Zöllmer, K. Rätzke and F. Faupel, Phys. Rev. B, Vol. 65 (2002) 220201-1 [10]. T. Zumkley, V. Naundorf, M.-P. Macht, G. Frohberg, Ann. Chim. Sci. Mat., 27 (5) (2002) 55 [11] D. Bavette, B. Deconihout, A. Bostel, J.M. Sarrau, M. Bouet, M. Menand, Rev. Sci. Inst. 64 (1993) 2911 [12]N. Nishiyama and A. Inoue, Mat. Trans. JIM, 37 (10), (1996) 1531 [13] I-R. Lu, Ph. Thesis, Universität Ulm, 2002, Germany [14] B.Sellberg, S.Rundqvist, Acta Chemica Scandinavica 19 (1965) 760-762 [15] M.Owusu, H.Jawad, T.Lundstroem, S.Rundqvist, Physica Scripta 6 (1972) 67-70