R.D.S. Lisboa, M.N.R.V.Perdigão, C.S.Kiminami ABSTRACT

Transcription

1 Crystallization of amorphous Al 84 Ni 8 Co 4 Y 3 Zr 1 melt-spun ribbons R.D.S. Lisboa, M.N.R.V.Perdigão, C.S.Kiminami Departamento de Engenharia de Materiais, Universidade Federal de São Carlos São Carlos, SP, Brazil ABSTRACT Amorphous metallic alloys are interesting engineering materials since for several systems and compositions their mechanical, magnetic and chemical properties are far superior to those of their crystalline counterparts. Aluminum amorphous alloys are by far the most promising candidates to structural applications, since they combine excellent mechanical properties and corrosion resistance with low density. For special compositions all these properties are proven to be further enhanced when nanocrystallization occurs over an amorphous matrix. This work investigates the crystallization of the easy glass former Al 84 Ni 8 Co 4 Y 3 Zr 1 alloy. Amorphous ribbons were produced using a single roller melt-spinner equipment under argon atmosphere. Crystallization annealings were carried out both isochronally and isothermally with the aid of a Netzsch DSC404 calorimeter and oil bath. Structural evaluation was conducted on a Philips CM120 transmission electron microscope operated at 120 kv and a Siemens D5000 X-ray diffractometer using Cu-Kα radiation. Calorimetric analyses revealed four crystallization events, regardless of heating rate. The crystallization sequence proceeds with homogeneous primary crystallization of nanometric α-al crystals over an amorphous matrix, which crystallizes to α-al + Al 9 Co 2 later on during the second stage of crystallization. The two final crystallization reactions correspond to recrystallization and polymorphic crystallization of intermetallic compounds. Keywords: amorphous aluminum alloys, aluminum-nickel-yttrium-cobalt-zirconium alloys, nanocrystallization CONGRESSO BRASILEIRO DE ENGENHARIA E CIÊNCIA DOS MATERIAIS, 14., 2000, São Pedro - SP. Anais 30201

2 1. INTRODUCTION Over the past ten years a number of aluminum-based alloys containing up to 15 at.% transition metals (TM) and/or lanthanide (LM) solutes were found to be amorphizable when rapidly quenched from the melt [1,2]. The amorphous state imparts extremely high tensile yield strength to the alloys, and values higher than 1.1 GPa have been commonly reported; these figures are about 1.5 times higher than those of the best crystalline precipitation hardened aluminum alloys. The excellent mechanical resistance of some aluminum amorphous alloys may be further enhanced when a nanometric-sized dispersion of α-al crystals is precipitated over an otherwise amorphous matrix. The mixed amorphous plus nanocrystalline structure may be obtained by quenching on the margin (composition or quenching rate) of glass formation or in a more controllable manner by annealing amorphous single phase alloys. Such a partial crystallization may promote a rise of about 1.5 times in tensile yield strength compared to the single-phase amorphous state. The hardening effect of nanocrystallization is attributed by Zhong et al. [3] to the enrichment of the remaining amorphous matrix with the slow diffuser TM and/or LM atoms rejected by the growing nanocrystals. This solute partitioning builds up at the interface nanocrystal / amorphous matrix and impedes further growth of the nanometric α-al crystals, favoring the formation of a fine scale microstructure. On the other hand, Kim et al. [4] explain the rise in mechanical properties as an effect of the nanocrystals themselves, which would be small enough to be defect free, thus extremely deformation resistant. Recently, the limitation of reduced volume of amorphous and/or nanocrystalline alloys produced by rapid quenching techniques has been partly overcome by the successful consolidation of amorphous atomized powders into bulky parts, while retaining extremely high mechanical strength [5]. Consolidation techniques often require the starting amorphous material to be warmed up to temperatures above crystallization onset temperatures. Upon crystallization, however, many of the amorphous/nanocrystalline Al alloys become extremely brittle due to intermetallic precipitation. The knowledge of the crystallization is thus extremely important in order to tailor the microstructure and related properties of aluminum amorphous alloys. This work investigates the crystallization reactions of the easy glass former Al 84 Ni 8 Co 4 Y 3 Zr 1 alloy in an attempt to support microstructural evaluation, which is essential for the production of amorphous ribbons and powders and for the production of bulky material by means of warm consolidation of amorphous precursors. CONGRESSO BRASILEIRO DE ENGENHARIA E CIÊNCIA DOS MATERIAIS, 14., 2000, São Pedro - SP. Anais 30202

3 2. EXPERIMENTAL Amorphous Al 84 Ni 8 Co 4 Y 3 Zr 1 ribbons were produced using a single roller melt-spinner equipment under high purity Ar atmosphere. A peripheral wheel velocity of 56 m s -1 was used and the melt was ejected at 1523 K with an overpressure of 20 kpa. Ribbons were obtained with an average thickness of 35 µm and width of 2.0 to 3.0 mm. Isochronous crystallization heat treatments and calorimetric analysis were carried out in a Netzsch DSC 404 calorimeter with heating rates (β) ranging from 5 to 40 K min -1 under flowing high purity Ar atmosphere. Isothermal annealings were conducted on a heated Santovac 5 oil bath with temperatures being controlled with an accuracy of ± 1 K. Annealing times ranged from 1 up to 120 minutes, samples being then rapidly cooled to room temperature by immersion in cold water. Structural evaluation of as-quenched and annealed ribbons was conducted on a Philips CM120 transmission electron microscope (TEM) operated at 120 kv and a Siemens D5000 X-ray diffractometer (XRD) using Cu-Kα radiation. As-quenched amorphous TEM samples were prepared using an ion milling Bal-Tec RES 010 equipment operated at an angle of 13º and 5.0 kv, 0.8 ma for 3.5 h. Crystallized samples were thinned with a Struers Tenupol twinjet electropolishing apparatus using an electrolyte of 25% nitric acid in methanol at 248 K. 3. RESULTS Figure 1 shows DSC thermograms for the as-quenched samples obtained at heating rates ranging from 5 to 40 K min -1. Four exothermic peaks are clearly distinguished irrespective of heating rate, although an expected shift in onset temperatures is observed. The four exotherms indicate a four-stage crystallization sequence. The activation energies (E A ) for each of the exothermic reactions were calculated according to Kissinger s approach [6]. Table 1 summarizes the calorimetric and E A data as obtained from thermograms in figure 1. At temperatures below the first peak the exothermic effect characteristic of structural relaxation is very weak, and no evidence of a glass transition is observed. TEM investigation of the as-quenched ribbons confirmed the amorphous structure, as evidenced by the appearance of a broad diffuse halo in the selected-area diffraction (SAD) analysis. Within the resolution of the TEM technique, which is about 5 nm in the present case, no crystals could be detected in imaging mode, except for a few isolated α-al nanocrystals as shown in figure 2. The amorphous structure of the as-quenched ribbons was also confirmed by XRD experiments, as shown in figure 3. CONGRESSO BRASILEIRO DE ENGENHARIA E CIÊNCIA DOS MATERIAIS, 14., 2000, São Pedro - SP. Anais 30203

4 Amorphous samples were continuously heated in the DSC (40 K min -1 ) up to temperatures past each one of the four exothermic peaks (510, 615, 695 and 835 K) in order to promote the reaction associated with each peak and allow for the growth of the products formed. In doing so one assures that phases form in size and quantity sufficient for precise XRD analysis. Figure 3 shows identified XRD patterns of the as-quenched and annealed samples. Identification of the crystalline phases described below was accomplished by comparison with the JCPDS data. The diffraction patterns indicate a crystallization process occurring by the formation of α-al phase upon the first exotherm followed by the formation of additional α-al phase plus Al 9 Co 2 upon the second exotherm. The last two exotherms correspond to the formation of a mixture of intermetallic compounds and probably recrystallization of Al crystals previously formed. Some unidentified diffraction peaks have also been observed. Figures 4a and 4b show TEM micrographs and corresponding SAD pattern of a sample continuously heated up to the first peak temperature as determined from the 40K min -1 DSC thermogram (493 K). Figures 4c and 4d show TEM micrographs and SAD pattern of a sample continuously heated up to the second exothermic peak temperature (628 K). From figures 4a and 4b it is clear that only the α-al phase is formed after heating to 493 K, while the remaining phase is an amorphous one. This α-al phase is found to have a lattice parameter larger than expected for pure aluminum, indicating the phase to be supersaturated. On the other hand, the diffraction pattern in figure 4c for the sample heated to 628 K lacks the broad diffuse halo typical of amorphous phases and instead presents a myriad of diffraction spots. According to the XRD results, many of the spots may be indexed as belonging to the Al 9 Co 2 phase. Unfortunately, due to limitations on the SAD technique concerning aperture sizes it was impossible to index the electron diffraction pattern without incurring in serious errors caused by the obvious multiple diffraction effect seen. Microdiffraction experiments are in due course in order to solve this problem. Even though, the SAD and the bright- and dark-field images in figures 4c and 4d allow the conclusion that crystallization has consumed the entirety of the material, little or none amorphous phase still remaining. In order to understand the nanocrystallization reaction and the nanocrystalline state in itself, isothermal experiments were carried out at 453 K and 473 K for times ranging from 1 to 120 minutes. Figures 5a through 5e show TEM bright-field (BF) micrographs and corresponding SAD patterns of samples isothermally annealed at 453 K for 1 up to CONGRESSO BRASILEIRO DE ENGENHARIA E CIÊNCIA DOS MATERIAIS, 14., 2000, São Pedro - SP. Anais 30204

5 120 minutes. From the BF and DF images (not shown) nanocrystals size and number were measured, the results indicating that nearly spherical nanocrystals about 5 nm in diameter grow larger and increase in number as annealing time is increased. After 120 minutes at 453 K the average size of the nanocrystals has increased to 40 nm and a strongly dendritic morphology has developed; 3 times more nanocrystals are observed after the longer annealing time. These results indicate both that nanocrystal growth is very sluggish at these relatively high temperatures and that nucleation of new nanocrystals actually occurs over an amorphous matrix. For comparison, figure 5f shows a TEM BF micrograph and corresponding SAD pattern of a sample annealed at 473 K for 120 minutes; note that the number of nanocrystals has not increased in comparison to the sample annealed at 453 K for 120 minutes. After isothermal annealing at 453 K and 473 K, nanocrystallized samples were continuously heated in the DSC (40 K min -1 ) up to 873 K in order to evaluate the effect of annealing time and temperature on the enthalpy release of each crystallization stage. In figure 6a the change in peak area is shown for samples annealed at 473 K. It is interesting to note that peaks have actually become doubled at intermediate times of 5 and 10 minutes, suggesting that a new reaction is taking place promoted by the annealing. Furthermore, the onset as well as the peak temperature has shifted to higher values, suggesting a higher thermal stability of the remaining amorphous phase. In figure 6b, the effect of annealing time and temperature on peak temperatures and transformed fraction of the first reaction is shown for samples. The transformed fraction is calculated as H(t,T)/ H, where H(t,T) is the enthalpy released by a sample annealed for the time t at temperature T and H is the total enthalpy released by an amorphous sample. Enthalpy and peak temperature of the second, third and fourth peaks were not affected by annealing. Figure 7 shows TEM micrographs of a sample isothermally annealed at 628 K for 5 minutes, corresponding to the peak temperature of the second exothermic reaction as determined from the 40 K min -1 DSC curve. Contrary to the first crystallization reaction, no sign of the product of the second reaction, Al 9 Co 2 as determined from XRD experiments, is detected even after annealing at the peak temperature. This suggests a different, slow mechanism of reaction for the formation of the Al 9 Co 2 intermetallic phase. Note the many small precipitates in figure 7b, which according to the SAD pattern are identified as α-al nanocrystals. CONGRESSO BRASILEIRO DE ENGENHARIA E CIÊNCIA DOS MATERIAIS, 14., 2000, São Pedro - SP. Anais 30205

6 4. DISCUSSION All the evidences gathered from DSC, XRD and TEM indicate that the as-quenched ribbons are amorphous. Although the concomitant observance of an exothermic peak in DSC continuous heating traces and a broad diffuse scattering in XRD and TEM are not sufficient proofs that the ribbons are truly amorphous instead of microcrystalline [7], the TEM images were taken at sufficiently high magnifications to clearly detect the presence of crystals as small as 5 nm, as was the case for micrograph 1. Many reports state that only isothermal DSC runs are indicated to distinguish between a truly amorphous and a microcrystalline structure. The strong transient signals, however, lead many times to the incorrect conclusion that grain growth of fine crystallites occurs instead of nucleation and growth. In this work TEM darkfield imaging at high magnifications was assumed to be more accurate. From the TEM and XRD diffraction observations the first crystallization reaction occurs with primary crystallization of α-al nanocrystals over a remaining amorphous matrix. The nanocrystals first formed assume a nearly spherical morphology, becoming clearly dendritic as temperature is increased during isothermal annealing and/or continuous heating. This may be seen in figures 5 and 7. From the isothermal experiments presented in figure 5 it is noteworthy the sluggish growth rate of the α-al nanocrystals. The low growth rate may be attributed to the combined effect of low diffusivity of the large Y and Zr atoms and to diffusion field impingement of these atoms between growing nanocrystals. This latter effect causes the structure to remain in nanometric sizes. Such a mechanism was observed by Hono et al. [8] in the Al-Y-Fe system. From the results presented, there seems to be a discrepancy between DSC and TEM observations. Although no evidence of a reaction is observed in the DSC traces below the first crystallization onset temperature, TEM micrographs of samples isothermally annealed at 453 K and 473 K show a dispersion of Al nanocrystals that was not present in the asquenched samples. Foley et al. [9] explain this apparent discrepancy as an insufficient heat release during the beginning of nanocrystallization for the DSC to detect, since the diffusion rate and thus the extent of transformation is very low. During continuous heating, on the other hand, the rising temperature increases the diffusivity of atoms and allows for a greater particle growth rate and further nucleation events. These effects result in the exotherm onset that eventually reaches a peak value when adjacent diffusion fields impinge each other. XRD of samples heated past the first exothermic peak after isothermal annealing at 473 K detected only α-al phase, indicating that the double peaks observed in figure 6a are CONGRESSO BRASILEIRO DE ENGENHARIA E CIÊNCIA DOS MATERIAIS, 14., 2000, São Pedro - SP. Anais 30206

7 related to α-al formation only. It is quite reasonable that isothermal annealing has promoted a high density of nucleus to develop, many of which were impeded to grow by the low diffusivity of Y and Zr at the temperature of 473 K. Later on, during the continuous heating experiments, both growth of these nuclei and further nucleation may take place, causing the detection of the two overlapping peaks experimentally observed. Figure 6b further supports this analysis showing that the transformed fraction during isothermal anneal is nearly constant up to 5 minutes, steeply rising after these hold times. TEM analysis of these samples actually revealed a threefold increase in number of aluminum crystals with increasing hold time past 5 minutes, while the accompanying increase in nanocrystals size was less drastic. This effect of isothermal anneal was also observed for samples heated at 453 K. The second crystallization reaction is identified by XRD as being the precipitation of Al 9 Co 2 plus additional α-al phases. Following the classification of Köster [10], crystallization or devitrification reactions of amorphous alloys may be primary, eutectic or polymorphic. Although not confirmed yet in this work, preliminary evidences such as that presented in figure 7 and XRD analysis are inconsistent. According to the XRD results, the second stage crystallization might only proceed with a eutectic transformation of the remaining amorphous matrix into the mixture α-al plus Al 9 Co 2. What figure 7 presents is a high density of Al nanocrystals not present before, and that may reasonably be the starting point of the eutectic transformation. However, such a eutectic transformation of the remaining amorphous matrix does not include Ni, Y or Zr elements that are known to be enriching the amorphous phase. Indeed, preliminary in situ heating TEM investigations (not presented) have shown the remaining amorphous matrix to polymorphically transform into a single phase rich in Al, Ni and Y. Further experiments are now being conducted to solve this apparent inconsistency. 5. CONCLUSIONS Calorimetric analyses revealed four crystallization events, regardless of heating rate, to occur in amorphous Al 84 Ni 8 Co 4 Y 3 Zr 1 alloy. The crystallization reactions proceed with homogeneous primary crystallization of nanometric α-al crystals over an amorphous matrix, which seems to crystallize eutectically to α-al + Al 9 Co 2 later on during the second stage of crystallization. The actual nature of the second crystallization reaction is still being investigated. The two final steps correspond to recrystallization of the aluminum phase and polymorphic crystallization of intermetallic compounds. CONGRESSO BRASILEIRO DE ENGENHARIA E CIÊNCIA DOS MATERIAIS, 14., 2000, São Pedro - SP. Anais 30207

8 REFERENCES [1] A.Inoue, K.Ohtera, A.P.Tsai, T.Masumoto, Japn. J. Appl. Phys., 27 (1988) L479. [2] Y. He, S. Poon, G. Shiflet, Science, 241 (1988) [3] Z.C. Zhong, X.Y. Jiang, A.L. Greer, Phylos. Mag. B, 76(4) (1997) 505. [4] Y.H. Kim, K. Hiraga, A. Inoue, T. Masumoto, H.H. Jo, Mater. Trans. Japan Inst. Metals, 35 (1994) 293. [5] Y. Kawamura, H. Mano, A. Inoue, Proceedings of the Fifth International Conference on Nanostructured Materials, Sendai, Japan, August [6] H.E Kissinger, Analyt. Chem., 29 (1957) [7] L.C. Chen, F. Spaepen, Nature, 336 (1988) 366. [8] K. Hono, Y. Zhang, A. Inoue, T. Sakurai, Mater. Trans. JIM, 36 (1995) 909. [9] J.C. Foley, D.R. Allen, J.H. Perepezko, Scripta Mater., 35(5) (1996) 655. [10] U. Köster, U. Herold, Scripta Metall., 12 (1978) 75. temperature (ºC) K/min exo DSC signal 10 K/min 20 K/min 40 K/min temperature (K) Figure 1. DSC thermograms of as-quenched amorphous ribbons obtained at various heating rates (β), indicated near each trace. CONGRESSO BRASILEIRO DE ENGENHARIA E CIÊNCIA DOS MATERIAIS, 14., 2000, São Pedro - SP. Anais 30208

9 Table 1. Thermal, calorimetric and kinetics information as obtained from figure 1, showing crystallization onset temperature (T X ), peak temperature of the i th exotherm (T pi ), the enthalpy of i th reaction ( H i ) and activation energy of i th step of crystallization (E Ai ). β (K min -1 ) T X T p1 T p2 T p3 H 1 H 2 (K) (K) (K) (K) (K) (J/g) (J/g) T p4 H 3+4 (J/g) E A1 E A2 E A3 E A4 (kj/mol) (kj/mol) (kj/mol) (kj/mol) Figure 2. TEM bright-field image and corresponding SAD pattern of an as-quenched amorphous ribbon. The arrows indicate a few isolated nanocrystals over an amorphous featureless matrix. intensity (cps) fcc-al Al 9 Co 2 β-al 3 Y? unidentified Al 3 Zr?????? 835 K? 695 K 615 K 510 K as-quenched θ (º) Figure 3. XRD patterns of as-quenched amorphous ribbon and annealed samples at temperatures indicated near each pattern. Precipitated phases identified are indicated in the inset. CONGRESSO BRASILEIRO DE ENGENHARIA E CIÊNCIA DOS MATERIAIS, 14., 2000, São Pedro - SP. Anais 30209

10 Figure 4. TEM bright-field (a,c) and dark-field (b,d) micrographs of samples continuously heated to (a,b) 493 K and (c,d) 628 K. CONGRESSO BRASILEIRO DE ENGENHARIA E CIÊNCIA DOS MATERIAIS, 14., 2000, São Pedro - SP. Anais 30210

11 Figure 5. TEM bright-field images and corresponding SAD patterns of samples held at 453 K for (a) 1, (b) 2, (c) 5, (d) 10 and (e) 120 minutes. Figure 5f corresponds to a sample held for 120 minutes at 473 K. 6a temperature (K) exo isothermal temperature 473K as-quenched DSC signal 5' 10' 20' 120' temperature (ºC) CONGRESSO BRASILEIRO DE ENGENHARIA E CIÊNCIA DOS MATERIAIS, 14., 2000, São Pedro - SP. Anais 30211

12 6b 1,0 0,8 isothermal hold time (min) transformed fraction peak temperature 473 K 540 transformed fraction 0,6 0,4 0,2 453 K 520 peak temperature (K) 0, isothermal hold time (min) 500 Figure 6. (a) DSC thermograms obtained from samples previously isothermally annealed at 473 K for the indicated times. In (b) transformed fraction of primary crystallization reaction and temperature shift caused by the isothermal annealing. Figure 7. TEM bright-field (a) and dark-field (b) images of a sample isothermally annealed at 628 K for 5 minutes. Note the small precipitates, identified as aluminum nanocrystals. Acknowledgements: authors thank FAPESP for financial support through grant 98/ CONGRESSO BRASILEIRO DE ENGENHARIA E CIÊNCIA DOS MATERIAIS, 14., 2000, São Pedro - SP. Anais 30212