1: \documentstyle[12pt]{article}
2: \begin{document}
3: \textwidth 16.cm
4: \textheight 21.cm
5: \pagestyle{empty}
6: \renewcommand{\baselinestretch}{1.2}
7: \newcommand{\be}{\begin{equation}}
8: \newcommand{\ee}{\end{equation}}
9: \newcommand{\bc}{\begin{center}}
10: \newcommand{\ec}{\end{center}}
11: \newcommand{\vs}{\vspace{1.0 cm}}
12: \newcommand{\vm}{\vspace{.5 cm}}
13: \newcommand{\hs}{\hspace{1.0 cm}}
14: \begin{titlepage}
15: \bc
16:
17: {\Huge A STUDY OF NANOCRYSTALLINE COBALT PREPARED BY BALL MILLING \\ }
18:
19: \vs
20: \vm
21: {\large J.C. de Lima, V. H. F. dos Santos, T. A. Grandi and R.S. de Biasi$^{a}$}
22: \vm
23:
24: {\em
25: Departamento de F\'{\i}sica, \\
26: Universidade Federal de Santa Catarina, \\
27: UFSC, Florian\'{o}polis CEP 88040-900, SC, Brazil. }
28:
29: {\em
30: $^a$ Departamento de Engenharia Mec\^anica e de Materiais, Instituto Militar de
31: Engenharia, Rio de Janeiro, Brazil \\ }
32:
33: \ec
34: \end{titlepage}
35:
36: \newpage
37:
38: \bc
39: {\Large {\bf Abstract}}
40: \ec
41:
42: \vspace{1.0 cm}
43:
44: The hcp-fcc transformation induced in cobalt powder by ball milling at room
45: temperature was studied using the excess Gibbs free energy for the metals, in
46: nanometric form, to calculate the activation barriers that the atoms must
47: overcome for the phase transformation to occur. The crystalline component of
48: the nanocrystaline cobalt produced in our laboratory, via ball milling, was
49: analized using the radial distribution function technique. The results suggest
50: that the crystalline component has the same structure as the bulk, contradicting
51: results already reported.
52:
53: \newpage
54: \bc
55: {\bf 1. Introduction}
56: \ec
57: \vs
58:
59: The advent of nanostructured materials has made it possible to imagine a wide
60: range of new technologies. This is associated with the presence of atomic
61: arrangements located at defect centers, such as grain boundaries and interphase
62: boundaries, allowing one to produce solids with new atomic structures and
63: physical properties.
64:
65: From the structural point of view, nanostructured materials can be regarded as
66: being made up of two components, one crystalline, with dimensions on the order
67: of some nanometers, that preserves the structure of bulk crystal, and another
68: interfacial, composed of defects (grain boundaries, interphase boundaries,
69: dislocations, etc.). The latter can be regarded as a highly disordered phase,
70: lacking the short-range order present in glasses as well as the long-range order
71: found in the crystalline state$[1,2]$. This component has caused controversy in
72: the literature. Gleiter${'}$ analysis$[3]$ is based on a gaseous model, while
73: other authors$[4]$ think this is not the best approach. The number of atoms in
74: the two components is about the same$[5]$.
75:
76: Recently, Huang et al.$[6]$ reported a study of phase transformation induced in
77: cobalt powder by ball milling at room temperature. The main conclusions reached
78: by the authors are: for a milling intensity below a lowe threshold value, the
79: starting mixture of hcp and fcc phases transforms to a single hcp phase; for an
80: intermediate range of milling intensities there is a second transformation of the
81: hcp phase to a mixture of the fcc and hcp phases; finally, for milling intensity
82: above an upper threshold value, there is a third transformation of the fcc and
83: hcp mixture to a single fcc phase. Previously, a detailed thermal and structural
84: study of the same phase changes had been performed by Mazzone$[7,8]$; the results
85: are not in agreement with those of Ref.$[6]$. According to Mazzone, the structural
86: transformations induced by ball milling of cobalt powder are more subtle because
87: of two factors: $(1)$ the progressive formation of an increasingly more faulted
88: hcp phase which eventually transforms into a structure based on an almost random
89: stacking of close-packed planes; $(2)$ the progressive contamination from the
90: steel milling tools, which for an iron concentration of the order of a few atoms
91: per cent stabilizes the fcc phase, as shown by the Fe-Co phase diagram.
92:
93: There is in the literature a study performed by Babanov et al.$[9]$ applying
94: the EXAFS technique to nanocrystalline cobalt. These investigators measured the
95: EXAFS oscillations of polycrystalline, nanocrystalline and grain boundary
96: components of cobalt powder. Using the regularization method, they found the
97: corresponding pair correlation functions. For nanometric cobalt, they reported
98: a strong reduction in the nearest-neighbor coordination number (6.35 atoms)
99: of the crystalline component in comparison with polycrystalline cobalt (13.1
100: atoms) while 3.65 atoms located at 0.252 nm were found for the grain boundary
101: component.
102:
103: According to the basic idea that the crystalline component of nanocrystalline
104: materials preserves the structure of the bulk crystal, a strong reduction in
105: the nearest-neighbor coordination number of this component, as the one observed
106: by Babanov, is not expected, since it would imply that its structure is different
107: from that of the bulk crystal. A recent study performed by us $[5]$ on nanocrystalline
108: nickel, using the radial distribution function (RDF), shows that the nearest-
109: neighbor coordination number of this component is similar to that of the bulk
110: crystal.
111:
112: In this paper, we investigate the hcp-fcc transformation in nanocrystalline
113: cobalt obtained via ball milling in our laboratory, using the excess Gibbs
114: free energy for the nanometric cobalt and RDF analysis to obtain the nearest-
115: neighbor coordination number of the crystalline component.
116:
117:
118:
119: \vs
120: \bc
121: {\bf 2. Experimental Procedure}
122: \ec
123: \vs
124:
125: Cobalt (purity, $99.5\%$) was milled for 62 h in argon atmosphere in a Spex
126: 8000 shaker mill, using spheres and a cylindrical steel vial. The mass ratio
127: between the ball and the powder mass (BPR) was 4:1. A ventilation system was used
128: to keep the temperature of the vial close to room temperature. The milled
129: powder samples were analyzed by X-ray diffraction using a Siemens difractometer
130: working with the Cu K$\alpha$ line ($\lambda = 0.15418 nm$).
131:
132:
133: \newpage
134: \bc
135: {\bf 3. Results and Discussion}
136: \ec
137: \vs
138:
139: Fig. 1 shows the X-ray diffraction patterns (XRD) for the cobalt as-received
140: (spectrum A) and after 62 h of milling (spectrum B). From spectrum A, using
141: the integrated intensities of the peaks, we obtained $64.9\%$ and $35.1\%$
142: of hcp and fcc phases, respectively. Spectrum B still shows both phases, but the
143: fraction of the fcc phase is larger than in spectrum A. This result agrees
144: quite well with those of Huang et al.$[6]$ for an intermediate range of milling
145: intensities. Spectrum B does not show other peaks besides those associated
146: with the fcc and hcp phases, indicating that if there is contamination from
147: the steel tools, it must be very low. In recent study $[5]$, using a Spex 8000
148: shaker mill, the same number of steel balls and the same BPR, we measured by
149: X-ray energy dispersive analysis a contamination of less than $3\%$ for nickel
150: powders after 30 h of milling.
151:
152: In order to explain the hcp-fcc transformation of nanocrystalline cobalt upon
153: milling, we will use the thermodynamic properties and the stability of grain
154: boundaries in nanometric metals and the concept of lattice stability.
155:
156: Ball milling has been used as an alternative technique to obtain some metals,
157: such as Ni, Fe, Pd, Co and Cu, in nanometric form. In the first stage of milling,
158: there is a drastic reduction in crystallite size. Further milling leads to ultrafine
159: crystallites. X-ray analysis shows that these crystallites have a few nanometers
160: in diameter. The reduction of crystallite size to a few nanometers cannot be
161: accomplished without creating an interfacial component.
162:
163: From the thermodynamic point of view, a nanometric powder is in a metastable
164: state whose Gibbs free energy is greater than that of the crystalline state. Without
165: annealing, there is no spontaneous transition to the crystalline state, suggesting
166: the existence of an activation energy .
167:
168: Using an universal equation of state proposed by Rose et al.$[10]$ to study the
169: thermodynamic properties and the stability of grain boundaries in nanometric
170: metals, Fecht showed$[11]$ that the excess Gibbs free energy of the interfacial
171: component grows with the excess volume but has a local minimum due to the maximum
172: entropy that this component can reach. The entropy of sublimation of a gas was
173: taken as this limit. The local minimum is responsible for the stability of the metals
174: in nanometric form. From Fechet${'}$ paper is clear that for a nanocrystalline-
175: to-crystalline trasition to occur, the atoms located at the interfacial component
176: must overcome an activation energy whose value can be calculated from the values of
177: the maximum and minimum excess Gibbs free energies. Thus, in order to promote
178: the transition, it is necessary to introduce a certain minimum amount of external
179: energy; this can be done by annealing the material at an appropriate temperature.
180: As an illustration, we show in Fig. 2 the excess Gibbs free energy, given by equation
181: $\Delta$G(V,T) = $\Delta$H(V,T) - T$\Delta$S(V,T), for nanocrystalline Co (hcp)
182: at 300 K. The $\Delta$V = 0.0 value corresponds to the crystalline state. As
183: can be seen, there is a local minimum at $\Delta$V = 0.410; the value of the
184: activation energy is 1.05 x 10$^{-20}$ J/at or about 0.066 eV/at.
185:
186: Unfortunately, we did not find in the literature enough data to calculate the
187: excess Gibbs free energy and the excess volume for nanocrystalline Co in the
188: fcc phase. However, we can use the concept of lattice stability to estimate the
189: Gibbs free energy of this phase. The lattice stability parameter of an element
190: is taken as the difference between the free energy of the phase that is being
191: described and the free energy of the other phase, used as a reference$[12,13]$.
192: The lattice stability parameter for cobalt (hcp) with respect to cobalt (fcc) is
193:
194: Co$^{fcc \rightarrow hcp}$ = [-460 + 0.628 T] (J mol$^{-1}$)
195:
196: At 300 K, the value of the lattice stability parameter is about -271.6 J mol$^{-1}$
197: or -28 x 10$^{-4}$ eV/atom, i.e., a very small value. As the temperature increses,
198: the lattice stability parameter becomes even smaller. The Gibbs free energies of the
199: hcp and fcc phases are thus almost equal. On the other hand, the results shown
200: in Fig. 1 seem to suggest that nanocrystalline Co in the fcc phase has an activation energy
201: greater than that for nanocrystalline Co in the hcp phase, since the fcc-hcp
202: transformation is not observed.
203:
204: In order to transform to the fcc phase, the Co atoms belonging to the interfacial
205: component of the hcp phase have to overcome an activation energy ; the required
206: energy is provided by collisions with the balls. In a Fritsch Pulverisette 5
207: planetary ball mill, the local temperature can reach up to 680 K$[14]$ and
208: several alloys have been formed using this type of mill.In a Spex mill, the
209: local temperature can be even larger. Assuming that T = 680 K in the expression
210: $E_{b}$ =$K_\beta$T = 8.617 x 10$^{-5}$ T eV/K, we obtain an average energy of
211: about 0.059 eV/atom, which agrees quite well with the activation energy calculated
212: at 300 K for nanometric cobalt (hcp). However, based on recent studies$[15]$,
213: we believe that the value of the activation energy depends on the average temperature
214: at the time when the nanometric powder was formed. We also believe that other
215: processes may contribute to the energy balance, as the following arguments
216: suggest: (1) During ball milling of nickel, even after the crystallite size has
217: stabilized at about 15 nm for 30 h of milling, the crystalline component continues
218: to store energy, reaching a saturation value of about 120 J g$^{-1}$ (about
219: 0.073 eV/at) after 120 h of milling$[5]$. The same must occur during ball milling
220: of cobalt. (2) If part of the energy comming from the balls is used for the
221: annealing of defects in both hcp and fcc crystalline phases, the energy thus
222: released is available for the transformation process; in the next collision, new defects
223: are created, generating a continuous cycle. Thus, these two processes (collisions
224: with the balls and annealing of the defects) are the necessary ingredients for
225: the energy balance which defines the driving force of hcp-fcc transformation of
226: cobalt upon milling. Of course, when the milling time is increased, the amount
227: of the hcp phase decreases, reducing the energy associated with the annealing
228: ofdefects of this phase. This reduction could explain the presence of a mixture
229: of the fcc and hcp phases after 62 h of milling (see Figure 1, spectrum B) as
230: observed by Huang et al.$[6]$. Thus, since the lattice stability parameter
231: calculated above is negligible, the fcc phase will be stable.
232:
233: The influence of mechanical milling on the formation of nanocrystalline cobalt
234: was investigated using radial distribution function (RDF) analysis$[16]$. The
235: structure factors, I(K), for as-received and milled powders as a function of
236: the wave vector K = 4$\pi$ sin$\theta/\lambda$, were obtained from XRD data
237: using the method described in Ref.$[17]$; the results are shown in Fig. 3. The
238: anomalous dispersion terms f${'}$ and f${''}$ for Co atoms$[18]$ were taken into
239: account. A comparison between the I(K) functions for a sample milled for 62 h
240: and for as-received cobalt powder shows a reduction in the highest peak
241: intensity of about $49\%$. This reduction is attributed to the interfacial
242: component (containing about $49\%$ of the total number of atoms) whose contribution
243: to the XRD pattern is diffuse. Fig. 4 shows the radial distribution functions
244: (RDF) obtained from the Fourier transform of the function F(K)=K[I(K)-1]
245: multiplied by a damping factor of the form exp(-$\alpha^{2}$K$^{2}$), with
246: $\alpha^{2}$ = 0.01 nm$^{2}$, in order to impose the convergence of F(K). On
247: the RDF curve, the average interatomic distance and the coordination number for
248: each atomic shell are given by the position of the maximum and the area under
249: the peak, respectively. In spite of termination effects due to the limited range
250: of the data (K$_{max}$ = 0.6225 and 0.7050 nm$^{-1}$ for the structure factor
251: corresponding to as-received and milled cobalt powder, respectively), we observe
252: that the RDF${'}$s are of good quality. For as-received cobalt, the RDF (solid line)
253: shows that the first neighbors shell is asymmetrical and not well isolated.
254: A fit (dotted line) can be made assuming two Gaussian distribution functions of
255: 12 Co neighbors centered at 0.248 nm and 3 Co neighbors centered at 0.326 nm.
256: In order to do this, we explicitly take into account the K dependence by
257: Fourier transforming into the K space the Gaussian distributions of distances
258: and back Fourier transforming over the same K range as the one used to calculate
259: the RDF${'}$s. Using the same procedure, the RDF (crossed line) obtained for
260: cobalt milled for 62 h was fitted assuming one Gaussian distribution function
261: of 12.9 Co neighbors centered at 0.258 nm. The slight shift of the interatomic
262: distance to larger values and the increase of the coordination number after
263: milling are probably due to strain fields extending from the boundary regions
264: to the crystallites, displacing the atoms from their ideal lattice sites$[15]$.
265:
266: Our results for nanocrystalline cobalt do not agree with those of Babanov et
267: al.$[9]$, who found a nearest-neighbor coordination number of 6.35 atoms for
268: the crystalline component of nanocrystalline cobalt. The results obtained in
269: this study confirm that the crystalline component of nanocrystalline materials
270: preserves the structure of bulk crystal, as expected.
271:
272: \vs
273: \bc
274: {\bf Acknowledgements}
275: \ec
276: \vs
277:
278: We thank the Conselho Nacional de Desenvolvimento Cient\'{\i}fico e
279: Tecnol\'ogico (CNPq) and the Ag\^encia Financiadora de Projetos (FINEP)
280: , Brazil, for financial support.
281:
282: \newpage
283: \bc
284: {\bf References}
285: \ec
286: \vs
287:
288: [1] X. Zhu, R. Birringer, U. Herr and H. Gleiter, Phys. Rev. B, 35 (1987) 9085.
289:
290: [2] H.J. Fecht, Acta Metall., 38 (1990) 1927.
291:
292: [3] H. Gleiter, NanoStruct. Mater., 1 (1992) 1.
293:
294: [4] E.A. Stern, R.W. Siegel, M. Newville, P.G. Sanders and D. Haskel,
295: Phys.Rev. Lett., 75 (1995) 3874.
296:
297: [5] T.A. Grandi, V.H.F. dos Santos and J.C. De Lima, Solid State Commun.,
298: 112 (1999) 359.
299:
300: [6] J.Y. Huang, Y.K. Wu and H.Q. Ye, Appl. Phys. Lett., 66 (1995) 308.
301:
302: [7] F. Cardellini and G. Mazzone, Philos. Mag. A67 (1993) 1289.
303:
304: [8] G. Mazzone, Appl. Phys. Lett., 67 (1995) 1944.
305:
306: [9] Yu. A. Babanov, I.V. Golovshchikova, F. Boscherini, T. Haubold and S.
307: Mobilio, Physica B 208-209 (1995) 140.
308:
309: [10] J.H. Rose, J.R. Smith, F. Guinea and J. Ferrante, Phys. Rev. B 29 (1984) 2963.
310:
311: [11] J.H. Fecht, Acta Metall. Mater., 38 (1990) 1927.
312:
313: [12] L. Kaufman and H. Bernstein, Computer Calculation of Phase Diagrams,
314: Academic Press, New York and London, (1970).
315:
316: [13] N. Saunders and A.P. Miodownik, J. Mater. Res., Vol. 1, No. 1 (1986) 38.
317:
318: [14] L. Schultz, J. Less-Common Met., 145 (1988) 233.
319:
320: [15] J.C. De Lima, T.A. Grandi, V.H.F. dos Santos, P.C.T. D${'}$Ajello and
321: A. Dmitriev, Phys. Rev. B 62 (2000) 8871.
322:
323: [16] Y. Waseda, Novel Application of Anomalous (Resonance) X-ray Scattering
324: for Structural Characterization of Disordered Materials, Springer-Verlag, Berlin,
325: edited by H. Araki, J. Ehlers, K. Hepp, R. Kippenhaln, H.A. Weindenmuller and
326: J. Zittariz, (1984).
327:
328: [17] C.N.J. Wagner, Liquid Metals, edited by S.Z. Beer, Marcell Dekker, Inc.,
329: New York, (1972) pp. 258-329.
330:
331: [18] S. Sasaki, Anomalous Scattering Factors for Synchrotron Radiation Users,
332: Calculated using Cromer and Liberman${'}$s Method, National Laboratory for
333: High Energy Physics, Japan, (1984).
334:
335: \newpage
336:
337: \vspace{1.0 cm}
338:
339: \bc
340: {\bf Figure Captions}
341: \ec
342:
343: \vspace{1.0 cm}
344:
345: Figure 1 : Experimental X-ray diffraction patterns of the cobalt powders:
346: before milling (A) and after milling for 62 h (B).
347:
348: \vspace{1.0 cm}
349:
350: Figure 2 : Excess Gibbs free energy as a function of excess volume for
351: nanocrystalline Co (hcp) at 300 K.
352:
353: \vspace{1.0cm}
354: Figure 3 : Structure factors of cobalt powders: before milling (dotted line) and
355: after milling for 62 h (solid line).
356:
357: \vspace{1.0 cm}
358: Figure 4 : Radial distribution functions of cobalt powders: before milling
359: (solid line) together with a fitting (dotted line) assuming two Gaussian
360: distributions (see text) and after milling for 62 h (crossed line).
361:
362:
363: \end{document}
364: