Scripta Materialia 50 (2004) 1209–1213 www.actamat-journals.com
Residual stresses and magnetic properties of alumina–nickel nanocomposites O. Aharon a
a,b
, S. Bar-Ziv b, D. Gorni b, T. Cohen-Hyams a, W.D. Kaplan
a,*
Faculty of Materials Engineering, Technion––Israel Institute of Technology, Haifa 32000, Israel b Rafael Ltd., P.O. Box 2250, Haifa 31021, Israel
Received 14 January 2004; received in revised form 27 January 2004; accepted 4 February 2004
Abstract Nanocomposites were produced by infiltration of alumina preforms with a nickel-nitrate solution, followed by reduction and sintering. Multiple infiltration and reduction steps resulted in nanocomposites with varying Ni particle concentrations. The strain in the nickel particles was evaluated from X-ray diffraction, and correlated with the magnetic properties of the nanocomposites. 2004 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Nanocomposites; Al2 O3 ; Ni; Residual stresses; Magnetic properties
1. Introduction The concept of ceramic–metal and ceramic–ceramic nanocomposites has been explored in detail in the last decade [1]. Ceramic matrix nanocomposites are usually based on a micron-sized ceramic matrix reinforced with sub-micron or nanometer sized particles of a second phase. The second phase can be located inside the matrix grains, and/or at grain boundaries. Numerous studies have been conducted on nanocomposites, with the goal of improving the mechanical properties [2–6], or attempts to understand the strengthening mechanisms [7,8]. Initial processing methods for Ni reinforced alumina were based on conventional consolidation followed by reduction and sintering in a controlled atmosphere [4,5]. Here an alternative method to produce Ni–alumina nanocomposites is discussed. Partially sintered green preforms were spontaneously infiltrated with a Ninitrate salt solution, followed by drying, reduction, and sintering. This method was used to produce Ni reinforced alumina nanocomposites with a fixed Ni particle concentration [3]. In the present work, this concept is extended to include multiple infiltration/reduction steps, as a method to vary the Ni particle content within the *
Corresponding author. Tel.: +972-482-94580; fax: +972-48294580. E-mail address:
[email protected] (W.D. Kaplan).
nanocomposite, as well as an additional sintering stage including hot isostatic pressing (HIP). The present study has two main objectives. First to understand the nucleation of secondary Ni particles within an existing microstructure of porous alumina containing initial Ni particles. In addition, the stress field in the vicinity of the reinforcing phase was measured to understand its influence on the magnetic properties of the nanocomposite.
2. Experimental methods 2.1. Processing Alumina slips were prepared by mixing (ball-milling for 12 h) 81.3 wt.% alumina powder (Alcoa, SG A-16, >99.8% purity, mean particle size 0.3–0.6 lm) with 17.9 wt.% distilled water, 0.73 wt.% deflocculant (Darvan) and 0.073 wt.% ammonia. After slip-casting, the preforms were dried for 3 h at 150 C, and then fired for 2 h at 900 C to remove the deflocculant and to form a green body with minimal strength for handling. The fired preforms were infiltrated with a Ni-nitrate salt solution: 83 wt.% Ni(NO3 )2 · 6H2 O, (Fluka Chemicals, >98% purity) with 17 wt.% distilled water. The infiltration process was conducted under vacuum (0.3 Torr), at room temperature for 30 min. The infiltrated specimen dimensions were 50 · 50 · 10 mm.
1359-6462/$ - see front matter 2004 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.scriptamat.2004.02.006
1210
O. Aharon et al. / Scripta Materialia 50 (2004) 1209–1213
The sintering process included three stages: reduction, pressureless sintering and hot isostatic pressing (HIP). Reduction was conducted at 900 C for 2 h in hydrogen (99.96%) at 1 atm. The desired Ni concentration was achieved by setting the number of process cycles of infiltration/reduction. (Up to six cycles of infiltration and reduction were conducted during this research, where each cycle added approximately 1.4 wt.% Ni.) Multiple infiltration/reduction was followed by pressureless sintering at 1440 C for 2 h in argon (99.99%) in order to achieve mostly closed pores, which is essential for the HIP stage. HIP was conducted at 1440 C for 2 h at a pressure of 2000 atm argon. The sintering and HIPing processes were conducted at 1440 C, which is below the melting point of pure Ni (1453 C). 2.2. Characterization X-ray diffraction (XRD) was conducted using a Philips X’Pert Automatic Diffractometer (Philips PW-3020 Goniometer) included a long fine-focus copper X-ray tube (k ¼ 0:15406 nm), operated routinely at 40 kV and 40 mA with 1 divergent and anti-scattering slits coupled with 0.2 mm receiving slits. XRD quantitative phase analysis was used to determine the relative Ni–alumina phase content in the nanocomposite. Several mixtures of alumina and Ni powder with varying contents of Ni (5, 7.5, 10, and 15 wt.%) were prepared and measured by XRD to create a calibration curve, based on the integrated intensity ratio of the largest Ni and alumina reflections (Ni: (1 1 1), (0 0 2), a–Al2 O3 : (1 0 1 4), (1 1 23), (1 1 2 6)). Microstructural characterization of polished and thermally etched specimens (1300 C/2 h/Ar) was conducted using high resolution scanning electron microscopy (HRSEM––LEO Gemini 982 equipped with a Schottky electron source). Microstructural characterization of dimpled and ion-milled specimens was conducted using transmission electron microscopy (TEM––JEOL 2000FX operated at 200 kV). Density measurements were conducted using the Archimedes method with iso-propanol as the soaking media for better infiltration into the residual open porosity. Mechanical characterization was conducted using a fully articulated three point bending jig under a tension/compression testing machine (Lloyd LR-10K) with a cross-head speed of 0.1 mm/min. The specimen dimensions were 1.2 · 2 · 50 mm with a surface quality of 20 lm. Magnetization behavior measurements at 3C were conducted using an Oxford Teslatron magnetometer. The residual stress in the nanocomposite was determined using the calculated directional modulus of elasticity and the measured strain. The strain was measured using XRD peak shifts. This method can be used only
when the measured phase is under a homogenous strain field, within the XRD sampling volume. In the Ni–alumina nanocomposite the roughly spherical Ni particles are embedded inside the alumina matrix. We therefore assume that they are under a relatively homogeneous strain field, which evolves during cooling of the nanocomposite from the sintering temperature, due to the larger thermal expansion coefficient of nickel versus alumina. Consequently, the Ni is expected to be under tensile strain. In order to measure the exact 2h angle of the Ni diffraction peaks, the specimen must be in the focal plane of the Bragg–Brentano XRD focal circle. The exact positioning of the specimen was done by placing a small amount of stress relived molybdenum powder over part of the specimen. The molybdenum peaks (in comparison to the Mo JCPDS peak positions) were used as an internal standard for 2h. The strain for the different lattice planes was calculated from the shift between the calibrated and the JCPDS Ni d-space (Eq. (1)). The stress was calculated by multiplying the directional strain with the directional modulus of elasticity [9] (Eq. (2)) which was calculated using known compliance constants [10–12]. eh k l ¼
ðdh k l dh0 k l Þ dh0 k l
ð1Þ
dh k l ––measured Ni d-space, dh0 k l ––JCPDS Ni d-space Eh1k l ¼
ðc11 þ c12 Þ ½ðc11 c12 Þðc11 þ 2c12 Þ 1 2 þ ðn2x n2y þ n2x n2z þ n2y n2z Þ c44 ðc11 c12 Þ 2
2
ð2Þ 2
where n2x ¼ ðh2 þkh 2 þl2 Þ, n2y ¼ ðh2 þkk 2 þl2 Þ, and n2z ¼ ðh2 þkl 2 þl2 Þ, c11 ¼ 246:5 GPa [10–12], c12 ¼ 147:3 GPa [10–12], c44 ¼ 124:7 GPa [10–12].
3. Results and discussion One of the research objectives was to characterize the influence of Ni content on density, both after sintering and after HIP. As evident from Fig. 1, the density of seven samples with varying Ni contents from 0–8.7 wt.% is relatively constant. HIP processing improved the average density by about 2%, and also reduced the standard deviation in the density of the specimens after sintering. The density was calculated using the volume rule of mixtures between alumina and Ni, i.e. the amount of Ni was taken into account in the density measurements. In order to calculate the theoretical density of the nanocomposite, the Ni content must be known, which was measured by two different methods; weight gain of the fired alumina preform after infiltration, reduction
O. Aharon et al. / Scripta Materialia 50 (2004) 1209–1213
Fig. 1. The relation between relative shrinkage and Ni concentration at the end of the sintering and HIP processes.
and sintering, and XRD quantitative analysis, where the concentration was determined using the calibration method. The first technique is not sensitive to the distribution of particles within the sample, while XRD quantitative analysis measures the Ni content in the vicinity of the surface. (The X-ray penetration depth for the reflections used is 20 lm, compared to the sample thickness of 10 mm.) This is best shown in Fig. 2, where the correlation between the two methods is presented, and where for each bulk Ni concentration measurement there are several Ni concentrations measured from XRD quantitative analysis (from sliced specimens). Thus the data in Fig. 2 also indicates the relatively homogeneous distribution of the Ni particles throughout the bulk
Fig. 2. A comparison of measured Ni concentration by XRD quantitative phase analysis versus weight gain.
1211
specimen. The Ni concentration analysis assumed that the samples were composed of only two components; Ni and alumina, which was confirmed by XRD. HRSEM was used to evaluate the alumina grain size and morphology. The micrographs presented in Fig. 3a and b are examples of alumina–Ni nanocomposites with Ni concentrations of 4.6 and 8.7 wt.%, respectively. In Fig. 3a alumina grains of 8 lm are evident, while in Fig. 3b there are no alumina grains larger than 4 lm. This indicates that Ni plays some role in limiting grain growth of the matrix, most likely by pinning grain boundaries. TEM micrographs of the above samples (Fig. 4a and b) revealed that most of the Ni particles are located at the alumina grain boundaries, and that most of the triple junctions are occupied by Ni particles. As the Ni concentration rises after every cycle of infiltration and reduction, two mechanisms of Ni particle evolution in the matrix can occur. The Ni particles can grow, due to deposition of new Ni on existing particles, or new Ni particles can nucleate, thus increasing the total number of particles and maintaining a small particle size distribution. The size of Ni particles in the nanocomposites
Fig. 3. Secondary electron HRSEM micrograph of: (a) a nanocomposite with 4.6 wt.% Ni, and (b) a nanocomposite with 8.7 wt.% Ni. Some Ni particles are visible, due to their slightly brighter contrast compared to the alumina grains.
1212
O. Aharon et al. / Scripta Materialia 50 (2004) 1209–1213
Fig. 4. Bright field TEM micrographs of a nanocomposite containing (a) 4.6 wt.% Ni and (b) 8.7 wt.% Ni.
was evaluated from TEM micrographs from ten different areas of each TEM sample. Most Ni particles are in the 150–250 nm range, and there are no particles larger 450 nm. Ni particles smaller than 100 nm were not measured. An average Ni particle size of 220 nm was found, regardless of the total Ni particle concentration. From Fig. 4a and b there is clear evidence that multiple infiltration and reduction stages introduce new Ni par-
ticles. Fig. 4a presents a good correlation to Fig. 3a, where a mix of smaller and larger alumina grains can be seen. The larger alumina grains are in areas where the local Ni content is low, thus again suggesting that the presence of Ni particles results in alumina grain refinement by grain boundary pinning. Three point bending tests were conducted on sets of four nanocomposite specimens with 2.8, 4.6, 7.2 and 8.7 wt.% Ni, as well as monolithic alumina samples sintered and HIPed at 1440C under the same conditions used to produce the nanocomposite specimens. No significant difference was found in the average bending strength (657–687 MPa) or in the Weibull modulus (8.6–11.5), between the samples sintered at 1440 C. This result stands in contrast to the results of Niihara et al. [2] who showed that the fracture strength of alumina increased from 450 to 1070 MPa after the addition of 10 vol.% Ni. The main objective of the research was to measure the strain field in the vicinity of the reinforcing phase (Ni) and to understand its influence on the magnetic properties of the nanocomposite as a whole. As explained in the previous section, the strain was measured using XRD peak shifts and the stress was calculated using Eqs. (1) and (2). The results for the nanocomposite sample with 8.7 wt.% Ni are presented in Table 1. The Ni particles are under residual tensile stress and the alumina matrix is under a small compressive stress, as expected. The Ni residual stresses for nanocomposites with different particle concentrations are presented in Table 2, which shows a rather constant stress on the (0 0 2) planes and a large deviation for other planes. The magnetization behavior of a nanocomposite specimen with 8.7 wt.% Ni is presented in Fig. 5. The nanocomposite exhibited a ferromagnetic behavior; a hysteresis loop with saturation magnetization of 38.9 emu/gNi and a coercive field of 46 Oe. Saturation in
Table 2 Residual stresses for nanocomposites with varying Ni contents (MPa)
Table 1 Residual stresses measured for a nanocomposite with 8.7 wt.% Ni hk l Measured d-space (A) JCPDS d-space (A) Ni Ni Ni Ni Al2 O3
111 200 220 311 (1 0 1 4)
2.039 1.766 1.248 1.0651 2.549
2.034 1.762 1.246 1.062 2.551
hk l
Ni content (wt.%) 1.43
4.6
7.23
8.7
111 200 220 311
465 259 480 409
708 255 401 345
459 259 334 570
720 272 406 470
Strain (%)
E (GPa)
Stress (MPa)
0.237 0.2 0.175 0.255 )0.053
304 136 232 184
720 272 406 470
O. Aharon et al. / Scripta Materialia 50 (2004) 1209–1213
1213
4. Summary and conclusions
Fig. 5. Magnetic hysteresis loop of a nanocomposite containing 8.7 wt.% Ni. The inset is a magnified view near the cross-axis point.
magnetization was achieved at 2500 Gauss. The coercive field is at least one order of magnitude larger than that of monolithic Ni, 0.7 Oe [13]. It is well known that there is a dependence of the coercive field on the size and shape of the magnetic particles [14], where below a critical radius, a single magnetic domain will exist. The largest coercivity is measured when a single domain is first created. Below a certain size, thermal fluctuations reduce the coercivity and above this size the movement of domain walls will also reduce the coercivity. In addition, if fine particles are separated by a non-magnetic media, the critical radius is expected to grow according to the change in the magnetostatic energy of the separated grains [13]. The increase in the coercive field may be due to the grain size. However, most of the Ni particles are in 150–250 nm range, such that there is a minor possibility for a dominant single domain behavior. Residual stresses also cause high coercive fields. An increase in the coercive field from the monolithic value of 0.7 Oe to about 30 Oe was observed in work-hardened nickel under internal stresses of about 100 MPa [13]. As was shown here, large residual tensile stresses exist in the Ni particles (between 272 and 720 MPa). These stresses lower the energy of the Bloch-walls between domains, and thus make their movement difficult. Thus, it seems that the internal stresses are the major cause for a high coercive field.
A process has been developed to produce ceramicmatrix nanocomposites with varying metal particle concentrations, based on a multiple infiltration/reduction process (prior to sintering) of metal-salts. The process leads to a microstructure containing a larger number of Ni particles as a function of process cycle. This method has a number of advantages compared to other methods, in that it does not require the use of submicron Ni oxide particles, which can cause serous health problems; the consolidation of the preform can be conducted by either cold pressing or slip casting; and this method could be used to create functionally graded nanocomposites and/or nanocomposites with several different reinforcing metals within the ceramic matrix. The residual stresses in the Ni grains were found to vary from 250 MPa in (2 0 0) to 720 MPa in (1 1 1). Large internal stresses in the Ni particles are the main cause for a high coercive field, one order of magnitude larger than for monolithic nickel.
Acknowledgements The authors wish to express their gratitude to M. Lieberthal and A. Tzetkov for assistance with the processing. This research was partially supported by the Technion Fund for the Promotion of Research.
References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14]
Sternitzke M. J Eur Ceram Soc 1997;17:1061. Niihara K, Nakahira A. Ann Chim 1991;16:479. Lieberthal M, Kaplan WD. Mater Sci Eng A 2001;A302:83. Oh ST, Sando M, Niihara K. J Am Ceram Soc 1998;81:3013. Sekino T, Nakajima T, Ueda S, Niihara K. J Am Ceram Soc 1997;80:1139. Tuan WH, Lin MC, Wu HH. Ceram Int 1995;21:221. Levin I, Kaplan WD, Brandon DG, Layyous AA. J Am Ceram Soc 1995;78:254. Kolhe R, Hui CY, Ustundag E. Acta Mater 1996;44:279. Heerden DV, Zolotoyabko E, Shechtman. J Mater Res 1996;11:2825. Alers GA. Phys Rev 1960;119:1532. Slagle OD, Mckinsky HA. J Appl Phys 1967;38:437. Hearmon RFS. Landolt-B€ orenstin 1966;1:1. Chikazumi S. Physics of magnetism. New York: John Wiley & Sons; 1964. p. 19, 287. O’handley RC. Modern magnetic materials. New York: John Wiley & Sons; 2000. p. 435.
