Investigation of low temperature thermal stability in bulk nanocrystalline Ni

Materials Science and Engineering A 427 (2006) 7–15

Investigation of low temperature thermal stability in bulk nanocrystalline Ni Manish Chauhan, Farghalli A. Mohamed ∗ Department of Chemical Engineering and Materials Science, University of California, Irvine, 916 Engineering Tower, Irvine, CA 92697, USA Received 14 July 2005; received in revised form 12 October 2005; accepted 12 October 2005

Abstract Grain growth behavior of bulk nanocrystalline Ni, prepared by an electrodeposition technique with average grain sizes of 20 and 15 nm was investigated in the homologous temperature (T/Tm ) range of 0.20–0.40. In studying grain growth, the techniques of X-ray diffraction and transmission electron microscopy were used. The results show that in the temperature range of 0.20–0.30Tm , there is no appreciable grain growth, even after long annealing times. However, in the temperature range of 0.3–0.4Tm , the rate of grain growth was rapid during the initial period of annealing, which 1/n decreases with increase in time. The value of time exponent, n, deduced from the grain growth equation of the general form D1/n − D0 = Kt was found to be approximately 0.1 for both grain sizes of Ni. At temperatures higher than 0.3Tm , an approximate activation energy of 105 ± 3 kJ/mol, which is close to the activation energy for grain boundary diffusion in polycrystalline Ni, was measured. At temperatures lower than 0.3Tm , an approximate activation energy of 11 ± 3 kJ/mol was found. It is suggested that this low activation energy represents the energy for the re-ordering of the nanocrystalline grain boundaries. © 2006 Published by Elsevier B.V. Keywords: Nanocrystalline Ni; Grain growth; XRD; TEM; Impurity segregation; Electrodeposition

1. Introduction Nanocrystalline materials (nc-materials) that are characterized by a grain size in the range of 1–100 nm [1,2] have been found to exhibit unusual physical and mechanical properties. Since these unique properties are closely related to the extremely fine grain size and the large volume fraction of grain boundaries associated with nc-materials, it is of importance to maintain the microstructure at a nanometer scale during the structural applications of these materials at the elevated temperatures. Therefore, it is not surprising that the research on thermal stability of ncmaterials has recently received considerable attention. Gleiter [3] reported that nc-Fe with a starting grain size of approximately 10 nm was thermally stable up to 0.26Tm (where Tm is the melting temperature of pure metal). After increasing the temperature to 0.37Tm and annealing for 10 h, the grain size of nc-Fe increased by five times, while the material became microcrystalline when annealed at 0.42Tm . Annealing studies

∗

Corresponding author. Tel.: +1 949 8245807; fax: +1 949 8242541. E-mail address: [email protected] (F.A. Mohamed).

0921-5093/$ – see front matter © 2006 Published by Elsevier B.V. doi:10.1016/j.msea.2005.10.039

on nc-Ni–1.2 wt.% P [4] revealed that up to 0.28Tm , nanocrystalline structure was retained, whereas the material transformed rapidly into a microcrystalline structure above 0.4Tm . Klement et al. [5] investigated the thermal stability of nc-Ni with an average grain size of 10 and 20 nm by differential scanning calorimetry (DSC). They reported that in nc-Ni, nucleation and abnormal grain growth took place in the temperature range of 0.2 and 0.32Tm , that normal grain growth took place in the range of 0.32 and 0.34Tm , and that grain growth approaches equilibrium in the range of 0.37 and 0.45Tm . Lee et al. [6] studied the grain growth of nc-Ni powders prepared by cryomilling with an average grain size of 22 nm. They reported that a grain size of 150 nm remained unchanged even after very long annealing times when annealed at 0.56Tm due to the presence of dispersions, which acted as a grain growth inhibitor. They also reported that the activation energy for the grain growth in nc-Ni to be 113 kJ/mol, a value that is close to the activation energy for the grain boundary diffusion of polycrystalline Ni. More recently, Kobayashi and Kashikura [7] examined the grain growth behavior of nc-Ni–4.4 mass% P. They reported the grain growth exponent in Ni to be around 4.5–5.9 (time exponent to be around 0.16–0.22).

8

M. Chauhan, F.A. Mohamed / Materials Science and Engineering A 427 (2006) 7–15

The present investigation was carried out in order to study the static grain growth behavior of bulk nc-Ni prepared by an electrodeposition procedure at relatively low homologous temperature (less than 0.4Tm , where Tm is the melting temperature of pure Ni). It is the purpose of this paper to report and discuss the results of the investigation.

neously present, the separation of the size and strain broadening by the method of IB can be presented by the equation given below: 
(δ2θ)2 Kλ δ2θ + 16e2 = (1) L tan θ0 sin θ0 tan2 θ0

2. Experimental procedure

where δ2θ is IB (in radians), λ the wave length of copper K␣ radiations, K a constant taken as 1, L the mean grain size, θ 0 the position of the peak maximum and e is equivalent to one-forth of the root-mean-square strain:

Nc-Ni with average initial grain sizes of 20 and 15 nm that was prepared by an electrodeposition technique was used. The chemical composition of as-electrodeposited Ni is tabulated in Table 1. The metal was provided by two sources: (a) Integran Technologies Inc., Toronto, Canada, in the form of sheet with an average initial grain size of 20 nm and a thickness of 0.5 mm, and (b) the Lawrence Livermore National Laboratory, CA, in the form of sheet with an average initial grain size of 15 nm and a thickness of 0.3 mm. Nc-Ni from the former source was tested to obtain experimental data while nc-Ni from the latter source was selectively used to check experimental trends. The samples were annealed in air at four different temperatures, varying from 0.2 to 0.4Tm , i.e., 393, 493, 593 and 693 K, for five different annealing times ranging from 0.5 to 25 h. After annealing the samples were rapidly cooled in air and characterized using X-ray diffraction (XRD) to calculate the grain size. XRD measurements were conducted using a Siemens D5000 diffractometer equipped with a graphite monochromator using copper K␣ (λ = 0.15406 nm) radiations. General scans with a step size of 0.01 degree (2θ) and with a step time of 2 s were conducted for grain size determination. Following subtraction of the instrumental broadening and K␣2 components, integral breadth for five strong FCC Ni peaks ({1 1 1}, {2 0 0}, {2 2 0}, {3 1 1} and {2 2 2}) was measured. For grain size measurement, the method of integral breadth (IB) [8] was adopted since the results of an analysis of grain size in bulk nc-materials have shown [9] that this method provides the closest approximation to the average grain size determined by transmission electron microscopy (TEM). In the method of IB [8], the average crystallite size and the lattice microstrain were estimated by XRD line broadening. This broadening of Bragg’s peaks is caused by the small size of the diffracting grains and by lattice strain. When the size and strain broadening are simultaTable 1 Chemical composition of as-electrodeposited nc-Ni Elements

Wt.%

C Si P S Cu Co B Ni

0.013 <0.001 0.003 0.058 0.023 0.071 0.0091 Balance

e=

1 2 0.5 ε  4

(2)

By performing a least-square fit to (δ2θ)2 / tan2 θ0 plotted against (δ2θ/(tan θ0 sin θ0 )) for all the measured peaks of a sample, L and e can be estimated from the slope of the straight line and y intercept, respectively. In addition to applying the method of IB, TEM investigation on as-received and post annealed samples was conducted using a Philips CM 20 TEM at 200 kV. The investigation by means of TEM serves two purposes: (a) to check the data on grain growth that are obtained from the IB method, and (b) to provide information on grain size distribution in both as-received and post annealed samples. The TEM sample preparation was carried out in two steps. First, the sample was mechanically mirror polished down to a thickness of less than 50 ␮m and a punch was used to cut 3 mm disc from it. The 3 mm discs were then chemically polished using a twin jet polisher, in which the surface exposed to the twin jets was chemically etched until the thinned area was transparent to the optical sensor. The solution used was a mixture of 75% CH3 OH and 25% conc. HNO3 . The voltage used was 8 V and the ampere reading was 4 mA. The temperature of the chemical solution was maintained below 243 K. Differential scanning calorimetry (DSC) investigation was performed on 20 nm nc-Ni samples in a commercial laboratory, OCM Test laboratories, Anaheim, CA. The scan was performed in Universal TA Instruments using nitrogen as a purge gas. Measurements were done between 323 and 973 K at a heating rate of 10 K/min. 3. Results 3.1. As-received material Fig. 1(a and b) shows the XRD pattern for an as-received electrodeposited nc-Ni specimens of 15 and 20 nm, respectively. The XRD pattern for both 15 and 20 nm specimen shows strong orientation around (1 1 1) followed by (2 0 0), indicating a preference for the planes with the lowest surface free energy to lie in the plane of the specimen [10]. For 15 nm specimens, the orientation around (2 0 0) is much stronger in terms of the relative peak intensity as compared with the standard JCPDS pattern (04-0850), whereas 20 nm specimens show an extra peak of (3 0 0), which is not a standard reflection plane for FCC metals. This observation reveals the presence of strongly preferred texture with the (1 0 0) planes oriented predominantly in the aselectrodeposited samples.

M. Chauhan, F.A. Mohamed / Materials Science and Engineering A 427 (2006) 7–15

9

Fig. 1. XRD profile for as-electrodeposited nc-Ni with an initial average grain size of: (a) 15 nm and (b) 20 nm.

Five FCC diffraction peaks, i.e., (1 1 1), (2 0 0), (2 2 0), (3 1 1) and (2 2 2), were used to calculate the average grain size of ncNi using IB method. The average grain size obtained by this method for 15 nm sample was 15 ± 2 nm and for 20 nm sample was 18 ± 2 nm. Fig. 2(a and b) shows the TEM micrographs for 15 and 20 nm samples, respectively. In order to measure the grain size distribution, a number of representative micrographs (optically magnified by three times) were used. The histogram of distribution for 20 nm samples is shown in Fig. 3. A similar distribution was obtained for 15 nm samples. It is clearly evident from the micrographs and grain size distribution plot that the grain size present in the nc-Ni tend to follow a normal distribution curve. The minimum, maximum and average grain size values obtained from TEM for 15 nm sample were 9, 35 and 14.3 nm with a standard deviation of 5.57 nm. Similarly, for 20 nm sample the minimum, maximum and average grain size values obtained from TEM were 3.8, 53 and 18.1 nm with a standard deviation of 10.71 nm. Comparing the results obtained for the average grain size from XRD and TEM shows that there is good agreement between the two methods.

Fig. 2. TEM micrograph for as-received nc-Ni: (a) 15 nm and (b) 20 nm.

3.2. Isothermal grain growth process To investigate the grain growth behavior of nc-Ni, isothermal annealing was performed at four different temperatures (393, 493, 593 and 693 K) for various annealing times (0.5–25 h). Fig. 4(a) shows the average grain size values determined from the IB method as a function of annealing time at various temperatures for 20 nm samples. The grain size data plotted show that the rate of grain growth is rapid during initial period of annealing, and then later decreases with the increase in annealing time. Fig. 4(a) also indicates that the rate of grain growth was very slow at temperature below 0.3Tm as compared to the grain growth rate near 0.4Tm . These trends noted for 20 nm

10

M. Chauhan, F.A. Mohamed / Materials Science and Engineering A 427 (2006) 7–15

Fig. 3. Grain size distribution in as-received 20 nm nc-Ni, showing the normal grain size distribution.

Fig. 5. TEM micrograph for 20 nm nc-Ni after annealing at 393 K for: (a) 30 min and (b) 25 h.

samples are consistent with those shown for 15 nm samples in Fig. 4(b). Fig. 5(a and b) shows the TEM micrographs of the 20 nm samples annealed at 393 K for 0.5 and 25 h, respectively. The average value of grain size obtained from TEM (shown in Fig. 6(a and b)) and XRD are in good agreement. Inspection of the grain size distribution plots before and after annealing at 393 K for 0.5 and 25 h suggests that with an increase in annealing time, keeping temperature constant, the extent of abnormal grain growth increases and that the grain size distribution becomes increasingly bimodal in nature. Also, TEM observations indicate that the tendency for bimodal distribution increases with increasing temperature above 393 K. 3.3. DSC results Fig. 4. Instantaneous grain size as a function of annealing time for different temperatures for: (a) 20 nm sample and (b) 15 nm sample.

Fig. 7 is the DSC plot for 20 nm nc-Ni samples. An examination of the plot shows two observations. First, two exothermic

M. Chauhan, F.A. Mohamed / Materials Science and Engineering A 427 (2006) 7–15

11

peaks are present: one below 510 K (0.3Tm ) and another above 555 K. Second, the energy released during the former exotherm is much lower than that associated with the latter one. 4. Discussion 4.1. Grain growth kinetics The earliest theoretical considerations of isothermal grain growth kinetics have assumed that for normal grain growth, the variation of grain size with time obey a parabolic kinetic law that may presented by [11]: D = K t n

(3)

where D is the average grain size, t the isothermal annealing time, and n (time exponent) and K (rate constant) is a temperature dependent parameter that characterizes the material under investigation; K is insensitive to the grain size and time. This particular relationship applies best if the initial grain size is small compared with the grain size that is being measured during growth. Since this condition is often not the case, especially in the early stages of growth, Beck et al. [11] established that the relationship should normally be expressed by the following more general form: 1/n

D1/n − D0

Fig. 6. Grain size distribution in 20 nm nc-Ni after annealing at 393 K for: (a) 30 min, showing the normal grain size distribution and (b) 25 h, showing the bimodal grain size distribution.

= Kt

where D0 is the initial pre-growth average grain diameter and K is rate constant. The activation energy for the grain growth, Q, can be calculated using an Arrhenius-type equation represented by:
 Q K = K0 exp − (5) RT where K0 is a frequency term and R is the gas constant. It is clear from the above equations that the kinetics of the grain growth can be described by two significant parameters, namely, activation energy (Q) and time exponent, n. The elementary theories of grain growth, which are either based on the proportionality of the growth rate to the interfacial free energy per unit volume [12] or based on the inverse proportionality of the rate of boundary migration to the boundary curvature [13], predicts a value 0.5 for n. However, experimental data indicate that the value of n, in most cases, is less than 0.5, and that for a given metal or alloy it usually increases with increasing temperature, approaching the limiting value of 0.5 for very pure metals or at very high temperatures. By differentiating Eq. (4), the isothermal rate of normal grain growth can be expressed in the form of the following equation: dD = nK(D)1−(1/n) dt

Fig. 7. DSC plot for the as-received 20 nm Ni samples exhibiting two distinct exothermic peaks.

(4)

(6)

Eq. (6) predicts that the rate of grain growth is related only to the instantaneous grain size D. Such a prediction for the rate of grain growth is found to be consistent with experimental results of Beck et al. [11].

12

M. Chauhan, F.A. Mohamed / Materials Science and Engineering A 427 (2006) 7–15

of 0.17–0.22 and 0.18–0.2, respectively, which increases with increase in temperature. Malow and Koch [16] studied the grain growth behavior in nc-Fe prepared by mechanical attrition. They reported that the value of n remains almost constant at 0.1 below 0.4Tm , similar to the observation made in the present study below 0.4Tm , but it increases from 0.1 to 0.33 with the increase in temperature from 0.4 to 0.45Tm , respectively. Natter et al. [17] also reported an increase in the value of n from 0.1 at low temperature to 0.33 at high temperature in nc-Fe prepared by pulsed electrodeposition. Spassov and Koster [18] studied the grain growth of crystallized FeZr2 and (Fe,Co)Zr2 nanocrystalline samples. They reported that the value of n was independent of temperature over a wide range of temperature (over 200 K) and that its value was equal to 0.33 for both systems. However, based on the measurements on nc-Cu, which were made by the sliding wear technique, Ganapathi et al. [19] concluded that it was difficult to identify a grain growth mechanism on the basis of only n, since they could produce excellent fits for their data regardless of the value of n (whether n is 0.25, 0.3 or 0.5). Similar observation regarding the time exponent was made by Holfer and Averback [20] who studied the grain growth behavior for porous nc-TiO2 . 4.2. Activation energy

Fig. 8. Instantaneous grain growth rate, dD/dt, as a function of instantaneous grain size, D, for different temperature for (a) 20 nm sample and (b) 15 nm sample.

Using the grain growth data shown in Fig. 4, the instantaneous grain growth rate (dD/dt) is plotted as a function of instantaneous grain size (D) on a double logarithmic scale in Fig. 8. The values of n were estimated from the slope of the straight lines that, according to Eq. (6), is equal to 1 − (1/n). The value of time exponent, n, determined using this method remains almost constant and independent of temperature in the entire range of annealing temperature investigated. The average value of n was found to be 0.1 ± 0.01 in the temperature range of 0.20–0.4Tm . The values quoted for n by Higgins [14] for 16 different metals had a range from 0.05 to 0.50 and the majority of the values were in between the range of 0.2–0.3. Values of the time exponent, n, were determined by several investigators who studied the grain growth behavior of nc-materials prepared by various methods. A time exponent value of 0.33 was reported for nc-NbAl3 that was prepared by ball-milling [15]. With some additions of Ti that retard the grain growth of nc-NbAl3 phase, the n value decreased to 0.25 [15]. The values of n determined for Ni and Ni3 P phase in the grain growth study of Ni–4.4 mass% P [7] were found to be in the range

The activation energy of a thermally activated transformation is a very important parameter that can provide insight into the nature of the transformation. To determine the average value of activation energy for grain growth, Q, K1 (=K/n, where n = 0.1 and value of K was determined from the y intercept of Fig. 8 for different temperatures) is plotted against the reciprocal of absolute temperature, 1/T, on a semi-logarithmic scale, as shown in Fig. 9(a) for 20 nm specimens. Inspection of the data plotted in Fig. 9(a) shows that there is a curvature in the Arrhenius plot and that the plot can be fitted with two straight lines, one below 0.3Tm and another above it. Using Eq. (6), the value of Q can be obtained from the slope of straight lines (=−Q/R). Below 0.3Tm , the average value of activation energy for grain growth in nc-Ni determined from the plot was close to 11 ± 3 kJ/mol, whereas above 0.3Tm , the average value of Q was found to be 108 ± 2 kJ/mol. The trend of the plot for 20 nm samples was also present in that for 15 nm samples as shown in Fig. 9(b). In this case, the activation energy above 0.3Tm was found to be 103 ± 2 kJ/mol. A trend similar to that noted for the activation energy for grain growth in nc-Ni was reported in the grain growth study of nanostructured Al 5083 powder [21], ultra-fine grained (UFG) Al 5083 bulk alloy [22], both prepared by cryomilling and UFG Al–3% Mg alloy prepared by equal channel angular pressing (ECAP) [23]. Malow and Koch [16] and Natter et al. [17] also reported two regions of activation energy for low (0.35–0.4Tm ) and high (0.4–0.45Tm ) temperature regions in the nc-Fe prepared by mechanical attrition and pulsed electrodeposition technique, respectively. However, they did not report such a low activation energy value in the low temperature region as was obtained in the present work (below 0.3Tm ). The activation energy values obtained in the low temperature region by Malow and Koch [16] and Natter et al. [17] were close to the value for grain boundary

M. Chauhan, F.A. Mohamed / Materials Science and Engineering A 427 (2006) 7–15

13

Table 2 Activation energy for the rearrangement of the grain boundaries in nc and UFG materials in low temperature region Material under investigation

Average initial grain size, D0 (nm)

Al–3% Mg 200 Al 5083 (bulk) 305 Al 5083 (powder) 16 Ni 20

Fig. 9. Arrhenius plot to estimate the activation energy for grain growth in nc-Ni between K1 (=K/n) and the reciprocal of absolute temperature, 1/T, on semi-logarithmic axis for (a) 20 nm and (b) 15 nm samples.

diffusion in nc-Fe. In high temperature region, the value reported by Natter et al. [17] was close to the value for grain boundary diffusion in coarse grained Fe whereas the value reported by Malow and Koch [16] was close to lattice diffusion in coarse grained Fe. The very low value of activation energy below 0.3Tm indicates the presence of highly unstable grain boundaries and high texture in the material, which need only a small amount of energy for the re-ordering of the microstructure. Gleiter [1] reported that nc-materials are in a thermodynamically non-equilibrium state at room temperature as compared to microcrystalline materials. These materials have highly unstable grain boundaries and large total surface area due to the presence of large volume fraction of intercrystalline region. According to Palumbo et al. [24], as crystal size decreases below 10 nm, total intercrystalline volume fraction of the material increase to 20–50%. At a grain size of ∼5 nm the volume fractions associated with intercrystalline region and perfect crystal are equivalent, i.e., equal to 50%. The activation energy for the rearrangement of the grain boundaries, as listed in Table 2, was reported to be around 30 kJ/mol [23] in Al–3% Mg processed by ECAP, 5.6 kJ/mol in Al 5083 alloy

Activation energy (kJ/mol)

References

30 25 ± 5 5.6 11 ± 3

[23] [22] [21] Present study

powder [21] and 25 kJ/mol in bulk Al 5083 prepared by cryomilling [22]. The results of the DSC investigation as plotted in Fig. 7 are consistent with those obtained for the activation energy in Fig. 9. First, the exothermic peak noted in the range of 473–510 K is in conformity with the concept that re-ordering of the grains and grain boundaries in this range of temperature is the origin for the low activation energy obtained in that region. In their investigation on nanostructured Al, Qin et al. [25] observed a broad exothermic peak at low temperatures. They explained this peak in terms of energy released as a result of the transformation of randomly grain boundaries into ordered boundaries. Also, Wang et al. [23] has attributed an exothermic peak noted in a submicron Al–3% Mg in the temperature range of 420–440 K to the advent of recovery process that is associated with the tendency to make boundaries more delineated. Second, the second higher peak can be considered to correspond to a grain growth exotherm. As shown by the grain growth data and Figs. 4 and 9, increasing the temperature above 555 K will accelerate the grain growth process that is controlled by diffusion of atoms along the grain boundaries. Inspection of the values of the activation energy that have been reported for the re-ordering of the grain boundaries in the low temperature region suggests that these values depends on the initial grain size and the texture present in the material. When the initial grain size is in the range of submicron (∼300 nm) the activation energy for the re-ordering was found to be approximately in the range of 25–30 kJ/mol. However, this range of activation energy decreases to 5–10 kJ/mol as grain size decreases below 20 nm. As discussed above, nc-materials are in a thermodynamically non-equilibrium state and the volume fraction of intercrystalline region increases rapidly below grain size value of 100 nm. Because of the high volume fraction of intercrystalline region, the interfacial stored energy in the nc-materials increases significantly as grain size decreases below 100 nm and, as a result, required activation energy for re-ordering of the grain boundaries decreases with decreasing grain size. The activation energies deduced from the Arrhenius plot for 20 and 15 nm samples between 0.3 and 0.4Tm were about 108 and 103 kJ/mol, respectively. The values obtained for the activation energy are very close to the activation energy for grain boundary diffusion in polycrystalline Ni (115 kJ/mol) [26]. More recently, Lee et al. [6] studied the grain growth behavior of nc-Ni powders with an average grain size of 22 nm prepared by cryomilling. The activation energy reported by those investigators was 113 kJ/mol, which is very close to the activation

14

M. Chauhan, F.A. Mohamed / Materials Science and Engineering A 427 (2006) 7–15

energy found in the present investigation. In addition, several investigators reported the activation energies for grain growth in some nc-oxides, alloys and pure metals [6,16,17,21,23,27,28] to be close to that usually observed for normal grain boundary diffusion in polycrystalline materials. On the above basis, the present finding along with those of other investigations is consistent with the concept that grain growth involves the transport of atoms across or along the grain boundaries. 4.3. Impurity segregation and grain growth In most of the grain growth studies involving nc-materials, the value of n was found to be different from 0.5 that was reported for pure metals or at high temperature. This behavior suggests that various factors may affect the grain growth process in nc-materials. The factors include Zener drag [4] (where a particle interacts with the grain boundary to reduce the energy of the boundary–particle system and restrains the boundary movement), pinning of grain boundaries by pores, solute (impurities) atoms or inclusions [29,30]. These second phase particles (solute atoms, inclusions or dispersions) pin the moving grain boundaries and retard grain growth in nc-materials. In nc-materials prepared by cryomilling, incoherent second phase dispersion particles, such as Al2 O3 , NiO, etc., plays a significant role in pinning the grain boundaries and stabilizing the microstructure. Several investigators [6,21,22] reported very high thermal stability in these cryomilled nc-materials due to the presence of these dispersion particles even at temperature higher than 0.5Tm and after long annealing times. Similar thermal stability was reported for nc-Ni–Si [31] and Ni–P [4,7,32] solid solution alloys, prepared by other techniques such as electrodeposition, where segregation of Si and Ni3 P, respectively, to the grain boundaries have been found to be responsible for preventing grain growth in nc-phases. In addition to observing excellent thermal stability in many nc-materials, abnormal grain growth behavior has also been reported for pure nc-Cu, Ag, Pd [33,34] and Ni [5,35–37]. It has been suggested that this type of anomalous grain growth is due to the segregation of impurities along the grain boundaries in nc-materials. If the impurity distribution is spatially non-uniform, enhanced grain growth may occur in regions of lowest impurity content, whereas other grain grows in a normal way where grain boundaries are pinned by the impurity atoms. In the present investigation, abnormal grain growth is observed in 20 nm Ni samples. It has been found that grains start growing normally at 393 K and after 30 min of annealing. However, after leaving the sample for 25 h at the same temperature, it yields an abnormal grain size distribution and microstructure becomes bimodal in nature. Based on an examination of the microstructure of nc-Ni prepared by an electrodeposition, Wang et al. [35] reported that many nanometer sized crystals within the clusters with small misorientations were present in the metal. These crystals need very small amount of energy to rotate slightly and join each other within a cluster to form a large grain. This could be the origin of abnormal grain growth at very low temperature. Fig. 10 shows a scanning electron microscopy (SEM) micrograph of as-electrodeposited nc-Ni with an average

Fig. 10. SEM micrograph of as-electrodeposited nc-Ni with an average grain size of 15 nm. Arrow “A” indicates the domain boundary; “B” indicates the grain boundary for nanometer sized crystals within the domain or cluster; “C” indicates the coalescence of two domain boundaries.

grain size of 15 nm. This micrograph indicates the presence of several nanometer size crystals inside the clusters. According to present TEM observations, the degree of abnormal grain growth increases with increasing temperature. A possible explanation for this trend may be related to boundary segregation. It has been reported (also shown in Table 1) that some amount of S and C atoms are present in the form of impurities in nc-Ni prepared by an electrodeposition. These impurities may segregate along the grain boundaries during the grain growth process [5,36,37]. The process by which impurity atoms segregate at boundaries is associated with two characteristics. First, the segregation of impurity atoms would be more significant at low than at high temperatures; this can be generally explained on the basis of the relation: C = C0 exp(E/kT), where C is the concentration of impurity atoms in the boundary, C0 the average concentration of impurity atoms, E the binding energy between an impurity atom and the boundary and k is Boltzmann’s constant. Second, the segregation of impurities to boundaries may occur preferentially. This property may arise from possible effects such as those associated with boundary orientation. It is quite possible that a number of grains with certain orientations may be able to avoid the pinning effect of impurities, a condition that leads to abnormal grain growth behavior. 5. Conclusions Thermal stability of nc-Ni with an average grain size of 20 and 15 nm was investigated in relatively low temperature range and the following conclusions can be made: 1. Impurities play a significant role in thermal stability of ncmaterials. Segregation of these impurities along the grain boundaries reduces the system energy and assist in providing enhanced thermal stability if the amount of impurities is sufficient to inhibit the grain growth. This segregation process is important in nc-Ni since the metal contains some amount

M. Chauhan, F.A. Mohamed / Materials Science and Engineering A 427 (2006) 7–15

of S and C atoms introduced as a result of preparation by electrodeposition. 2. The activation energy for grain growth obtained at temperatures below 0.3Tm was found to be approximately 11 kJ/mol in nc-Ni samples. This low energy is consistent with those energies reported for re-ordering boundaries in nc-metals. At temperatures above 0.3Tm , the activation energy for grain growth was found to be 103 kJ/mol for 15 nm samples and 108 kJ/mol for 20 nm samples. These values are close to that of grain boundary self diffusion in polycrystalline Ni (115 kJ/mol). Acknowledgments This work was supported by National Science Foundation under Grant No. DMR-0304629. Thanks are also due to Lisa Chan and Anna Torrents, undergraduate students, for their assistance in conducting the experiments and Li-Chung Lai of MC2 for his assistance with TEM. References [1] [2] [3] [4] [5] [6] [7] [8] [9]

H. Gleiter, Acta Mater. 48 (2000) 1. F.A. Mohamed, Y. Li, Mater. Sci. Eng. A 298 (2001) 1. H. Gleiter, Prog. Mater. Sci. 33 (1989) 223. K. Boylan, D. Ostrander, U. Erb, G. Palumbo, K.T. Aust, Scr. Metall. 25 (1991) 2711. U. Klement, U. Erb, A.M. El-Sherik, K.T. Aust, Mater. Sci. Eng. A 203 (1995) 177. J. Lee, F. Zhou, K.H. Chung, N.J. Kim, E.J. Lavernia, Metall. Mater. Trans. A 32 (2001) 3109. S. Kobayashi, Y. Kashikura, Mater. Sci. Eng. A 358 (2003) 76. H.P. Kulg, L.E. Alexander, X-ray Diffraction Procedures, John Wiley & Sons, New York, 1974, p. 661. G. Zhang, F. Zhou, E.J. Lavernia, Metall. Mater. Trans. A 34 (2003) 1349.

15

[10] J.K. Mackenzie, A.J.W. Moore, J.F. Nicholas, J. Phys. Chem. Solids 23 (1962) 185. [11] P.A. Beck, J.C. Kremer, L.J. Demer, M.L. Holzworth, Trans. AIME 175 (1948) 372. [12] F.A. Mohamed, T.G. Langdon, Acta Metall. 23 (1975) 1443. [13] L.E. Burke, Trans. AIME 180 (1949) 73. [14] G.T. Higgins, Met. Sci. J. 8 (1974) 143. [15] S. Kawanishi, K. Isonoshi, K. Okazaki, Mater. Trans. JIM 34 (1993) 49. [16] T.R. Malow, C.C. Koch, Acta Mater. 45 (1997) 2177. [17] H. Natter, M. Schmelzer, M.S. Loffler, C.E. Krill, A. Fitch, R. Hempelmann, J. Phys. Chem. B 104 (2000) 2467. [18] T. Spassov, U. Koster, J. Mater. Sci. 28 (1993) 2789. [19] S.K. Ganapathi, D.M. Owen, A.H. Chokshi, Scr. Metall. Mater. 25 (1991) 2699. [20] H.J. Holfer, R.S. Averback, Scr. Metall. Mater. 24 (1991) 2401. [21] V.L. Tellkamp, S. Dallek, D. Cheng, E.J. Lavernia, J. Mater. Res. 16 (2001) 938. [22] I. Roy, M. Chauhan, F.A. Mohamed, unpublished data. [23] J. Wang, Y. Iwahashi, Z. Horita, M. Furukawa, M. Nemoto, R.Z. Valiev, T.G. Langdon, Acta Mater. 44 (1996) 2973. [24] G. Palumbo, S.J. Thorpe, K.T. Aust, Scr. Metall. Mater. 24 (1990) 1347. [25] X.Y. Qin, X.J. Wu, L.F. Cheng, Nanostruct. Mater. 2 (1993) 99. [26] H.I. Frost, M.F. Ashby, Deformation Mechanism Maps, Pergamon, New York, 1982. [27] N. Wang, Z. Wang, K.T. Aust, U. Erb, Mater. Sci. Eng. A 237 (1997) 150. [28] A. Kunpmann, B. Gunther, H.D. Kunze, Mater. Sci. Eng. A 168 (1993) 165. [29] H. Hahn, J. Logas, R.S. Averback, J. Mater. Res. 5 (1990) 609. [30] R.S. Averback, H.J. Holfer, R. Tao, Mater. Sci. Eng. A 166 (1993) 169. [31] P. Knauth, A. Charal, P. Gas, Scr. Metall. Mater. 28 (1993) 325. [32] K. Lu, Scr. Metall. Mater. 25 (1991). [33] A. Kumpmann, B. Gunther, H.D. Kunze, Mater. Sci. Eng. A 168 (1993) 165. [34] V.Y. Gertsman, R. Birringer, Scr. Metall. Mater. 30 (1994) 577. [35] N. Wang, Z. Wang, K.T. Aust, U. Erb, Acta Mater. 45 (1997) 1655. [36] G.D. Hibbard, J.L. McCrea, G. Palumbo, K.T. Aust, U. Erb, Scr. Mater. 47 (2002) 83. [37] M. Abraham, P. Holdway, M. Thuvander, A. Cerezo, G.D.W. Smith, Surf. Eng. 18 (2002) 151.