1: \documentclass[twocolumn,showpacs]{revtex4}
2: \usepackage[dvips]{graphicx}
3: \usepackage{dcolumn}
4: \usepackage{amsmath}
5: \makeatletter
6: \def\btt#1{\texttt{\@backslashchar#1}}%
7: \DeclareRobustCommand\bblash{\btt{\@backslashchar}}%
8: \makeatother
9: \begin{document}
10:
11: \title{Dynamics of a dislocation bypassing an impenetrable precipitate: the Hirsch mechanism revisited}
12: \author{Takahiro Hatano}
13: \affiliation{
14: Earthquake Research Institute, University of Tokyo, Tokyo 113-0032, Japan.}
15: \date{\today}
16:
17: \begin{abstract}
18: Dynamical process where an edge dislocation in fcc copper bypasses an impenetrable precipitate
19: is investigated by means of molecular dynamics simulation.
20: A mechanism which is quite different from the Orowan mechanism is observed,
21: where a dislocation leaves two prismatic loops near a precipitate: i.e. the Hirsch mechanism.
22: It is found that the critical stress for the Hirsch mechanism is almost the same as the Orowan stress,
23: while the spatial inhomogeneity of the shear stress is essential to the Hirsch mechanism.
24: We also find that the repetition of the Hirsch mechanism does not increase the critical stress.
25: \end{abstract}
26:
27: \pacs{81.40.Cd, 61.72.Bb, 62.20.Fe}
28: \maketitle
29:
30: Plastic flow in crystalline materials is mainly dominated by the movement of dislocations.
31: They are curvilinear defects which can move at much lower stress levels than
32: the theoretical strength of a perfect crystal \cite{friedel}.
33: A dislocation interacts with other lattice defects such as voids or precipitates,
34: which obstruct the dislocation motion to result in hardening.
35: Such dislocation-obstacle interactions dominate plasticity of crystalline materials,
36: and hence they are one of the major concerns in materials science.
37: Until very recently, they were considered exclusively by plausible continuum models
38: \cite{bacon,scattergood}, which deal with a single glide plane and neglect atomistic discreteness.
39: The recent computer development enables direct molecular dynamics (MD) simulation
40: on dislocation systems, which consist of multi-million atoms.
41: MD simulations have revealed dynamical properties and atomistic details of dislocations.
42: For example, edge dislocations absorb vacancies when they interact with voids \cite{osetsky}
43: or stacking fault tetrahedra \cite{wirth}.
44: They are good illustrations of the usefulness of MD simulation in the field of dislocation physics.
45: Along the line of these studies, the interaction with an impenetrable precipitate,
46: which has not been investigated by MD simulation, is explored in this Letter.
47:
48: An interaction between a dislocation and a precipitate is characterized by their shear moduli.
49: When the shear modulus of a precipitate is larger than the bulk's, the interaction is repulsive
50: so that the precipitate can be impenetrable \cite{incoherency}.
51: In this case, the extent of hardening becomes significant.
52: This is the principle of so-called "particle strengthening", which is widely utilized in processing
53: stronger materials \cite{argon}.
54: There is a mechanism by which a dislocation bypasses impenetrable obstacles,
55: where a dislocation largely bows out to leave a dislocation loop around the obstacle \cite{friedel}.
56: Note that the loop is on the original glide plane.
57: This process and the resultant dislocation loop are referred to as the Orowan mechanism
58: and an Orowan loop, respectively.
59: However, because the interaction between an Orowan loop and a dislocation is strongly repulsive,
60: theoretical calculations in which Orowan loops are assumed to be persistent
61: predict unreasonably strong work-hardening \cite{ashby,brown1}.
62: To avoid this contradiction, some alternative bypass mechanisms have been presented.
63: Hirsch \cite{hirsch1,humphreys} had pointed out the possibility of prismatic loops formation in a bypass mechanism.
64: This mechanism is named after Hirsch and referred to as the Hirsch mechanism.
65: On the other hand, Brown and Stobbs presented another mechanism where a dislocation loop
66: of a secondary Burgers vector nucleates on the precipitate surface \cite{brown2}.
67: In both cases, the resultant structure is a row of prismatic loops which was experimentally observed
68: by transmission electron microscopy (TEM) \cite{humphreys,brown2,sarnek}.
69: However, since the dynamical process cannot be observed by TEM,
70: the postulated mechanisms have not been directly tested.
71: In this Letter, we wish to clarify the dynamics of prismatic loops formation
72: and to find out quantitative conditions which determine resultant dislocation loops:
73: Orowan loops or prismatic loops. Along the line of thought, an MD simulation is performed.
74:
75: %%% THE MODEL %%%
76:
77: %%%% computational model %%%%
78: Let us describe the computational model.
79: We consider fcc copper utilizing a many-body interatomic potential of Finnis-Sinclair \cite{finnis}
80: and adopt parameters which are determined by Ackland et al \cite{ackland}.
81: The lattice constant $a=3.615$ \AA.
82: The dimensions of the model system are $x=23\times[11\bar{2}]$ ($20$ nm),
83: $y=79\times[1\bar{1}0]$ ($40$ nm), and $z=27\times[111]$ ($17$ nm).
84: This system consists of approximately $1.2$ million atoms.
85: (In addition to this system, we prepare some smaller systems which have different length in the $x$ direction.)
86: We introduce an edge dislocation parallel to the $x$ axis, setting the Burgers vector to be parallel to the $y$ axis.
87: Note that we focus on an edge dislocation, since a screw dislocation can avoid a precipitate
88: by switching its glide planes via cross-slip \cite{note1}, which is not our interest here.
89: %%% precipitate %%%
90: An impenetrable precipitate is modeled by a set of immobile atoms, which are coherent to the bulk crystal.
91: It mimics a precipitate of the infinite shear modulus, which cannot be sheared by dislocations.
92: %%%% boundary %%%%
93: Periodic boundary conditions are employed in the $x$ and the $y$ directions,
94: and the surfaces exist only in the $\pm z$ directions.
95: In order to cause the shear strain, $(111)$ surface is displaced at a constant velocity.
96: We test two cases, which are referred to as conditions I and II, respectively.
97: In condition I, the upper ($+z$) surface move towards the $-y$ direction
98: while the lower surface is fixed.
99: In condition II, both surfaces move opposite to each other.
100: In both conditions, we set the strain rate $\dot{\epsilon}=7 \times 10^6$ [$\rm{s}^{-1}$].
101: Note that we adopt the condition I unless explicitly indicated.
102: Temperature is set to be $300$ K, where the velocities of copper atoms are given randomly
103: according to the Maxwell-Boltzmann distribution.
104:
105:
106: %%%% results %%%%%%%%
107: Let us turn to the simulation results. First, we focus on the dynamics and the geometry.
108: The configuration is shown in Fig. \ref{snapshots} \cite{li},
109: where a dislocation does not leave an Orowan loop but two prismatic loops.
110: \begin{figure}
111: \includegraphics[scale=0.7]{snapshots.eps}
112: \caption{Successive snapshots of an edge dislocation bypassing an impenetrable precipitate.
113: The radius of the precipitate (blue atoms) is $1.5$ nm.
114: A dislocation (red atoms) is visualized by omitting atoms which have $12$ nearest neighbors.
115: (a) A dislocation bows out to form a screw dipole.
116: (b) The screw dipole cross-slips.
117: (c) The screw dipole undergoes double cross-slip to annihilate each other at the top of the precipitate.
118: (d) The dislocation has bypassed the precipitate with a superjog. A prismatic loop is left behind.}
119: \label{snapshots}
120: \end{figure}
121: At the first stage, a dislocation bows out to form a screw dipole of opposite signs.
122: The main difference between the Orowan mechanism and the mechanism observed here
123: lies in the behavior of this screw dipole.
124: It cross-slips onto a $(11\bar{1})$ plane (see FIG. \ref{crossslip}),
125: and then undergoes double cross-slip onto another $(111)$ plane,
126: on which the screw dipole eventually annihilates each other.
127: \begin{figure}
128: \includegraphics[scale=0.5]{crossslip.eps}
129: \caption{Geometry of the primary cross-slip from $(111)$ to $(11\bar{1})$.
130: (a) A $[1\bar{1}1]$ projection. (b) A $[111]$ projection.}
131: \label{crossslip}
132: \end{figure}
133: In FIG. \ref{snapshots} (d), a prismatic loop is formed on the left side of the precipitate
134: and a superjog is formed on the opposite side.
135: Then the superjog is dragged by the dislocation for a while, and is eventually separated to
136: form the secondary prismatic loop as shown in Fig. \ref{loops} (a).
137: If the primary cross-slip takes place towards the extra atomistic half plane
138: which constitutes the edge dislocation, the prismatic loop created before the precipitate is interstitial
139: and the secondary loop after the precipitate is vacancy.
140: (If the primary cross-slip goes opposite, the order is reversed).
141: Note that the resultant prismatic loops have the same Burgers vector as that of the original edge dislocation.
142: \begin{figure}
143: \begin{center}
144: \includegraphics[scale=0.43]{loops.eps}
145: \caption{(a) Configuration just after a single bypass process.
146: A dislocation leaves two prismatic loops: an interstitial loop and a vacancy loop.
147: The secondary loop is formed relatively far from the precipitate,
148: since it has been dragged by the dislocation as a superjog.
149: (b) A row of prismatic loops after the passage of three dislocations.
150: Note that the loops on the right side of the precipitate do not form an apparent row.}
151: \label{loops}
152: \end{center}
153: \end{figure}
154: Indeed, this mechanism is identical to the Hirsch mechanism \cite{humphreys}.
155: We stress that this is the first case where the Hirsch mechanism is dynamically observed,
156: while TEM observations have detected only the resultant structures.
157: Note that the mechanism of Brown and Stobbs \cite{brown2} has not been observed
158: in the conditions investigated here.
159: More importantly, it is found that the Hirsch mechanism always occurs for the condition I,
160: while the Orowan mechanism is realized for the condition II.
161: To clarify the underlying physics, we have to investigate the quantitative aspect.
162:
163: %%%%% Void size dependence of the critical stress %%%%%%%%%
164: We investigate the critical resolved shear stress for the Hirsch mechanism, which we will call the "Hirsch stress".
165: The shear stress is defined as the area-averaged force acting on the surfaces.
166: The Hirsch stress is defined as the maximum shear stress during a bypass process.
167: It is realized just before the primary cross-slip, by which the Hirsch stress is determined.
168: Behaviors of the Hirsch stress with respect to the precipitate radius and to the precipitate spacing
169: are shown in Fig. \ref{r-crss}, where we find the Hirsch stress is described by the following logarithmic law.
170: \begin{equation}
171: \label{logarithmiclaw}
172: \sigma_{yz}=\frac{A}{L}\log\frac{1}{\left(0.5r^{-1}+L^{-1}\right)B},
173: \end{equation}
174: where $A$ and $B$ denote material-dependent parameters, $r$ is the radius of a precipitate,
175: and $L$ is the precipitate spacing in the $x$ direction.
176: Namely, $L=L_x-2r$, where $L_x$ denotes the system size in the $x$ dimension.
177: Note that $A=6.5$ N/m and $B=0.5$ nm in the present system.
178: \begin{figure}
179: \includegraphics[scale=0.5]{crss.eps}
180: \caption{The critical resolved shear stresses for the Hirsch mechanism (denoted by the red $\times$ symbols).
181: Note that the Orowan stress is almost the same as the Hirsch stress.
182: The green dashed lines represent Eq. (\ref{logarithmiclaw}) with $A=6.5$ N/m and $B=0.5$ nm.
183: (a) Dependence of the critical stress on the precipitate radius, where $L_x$ is $20.4$ nm.
184: (b) Dependence of the critical stress on the spacing of the precipitates, where $r$ is $1.5$ nm.}
185: \label{r-crss}
186: \end{figure}
187: This kind of logarithmic behavior of the critical stress is universal in the context of
188: dislocation-obstacle interactions \cite{osetsky,hatano}, the reason of which is explained as follows.
189: From the balance of forces that are transverse to the dislocation line,
190: the resolved shear stress ($\tau$) and the radius of curvature ($\rho$) of a bowing dislocation are interrelated as
191: \begin{equation}
192: \label{balance}
193: \tau=\frac{\gamma}{\rho b}.
194: \end{equation}
195: Note that $\gamma$ denotes the line tension, which depends on the configuration of a bending dislocation.
196: When an edge dislocation bends to form a screw dipole at a pinning point, the line tension is estimated as \cite{bacon}
197: \begin{equation}
198: \label{gamma}
199: \gamma=\frac{Gb^2}{4\pi}\log \frac{1}{B}\left(\frac{1}{2r}+\frac{1}{L}\right)^{-1}.
200: \end{equation}
201: Since a fully bending dislocation forms a semi-arc between two pinning points, the radius of curvature is $\rho=L/2$.
202: Inserting this relation and Eq. (\ref{gamma}) into Eq. (\ref{balance}) immediately leads to the logarithmic law.
203: \begin{equation}
204: \label{logarithmiclaw2}
205: \tau=\frac{Gb}{2\pi L}\log\frac{1}{B}\left(\frac{1}{2r}+\frac{1}{L}\right)^{-1}.
206: \end{equation}
207: Although Eqs. (\ref{logarithmiclaw}) and (\ref{logarithmiclaw2}) are of the same form,
208: the quantitative discrepancy is considerable.
209: Namely, $Gb/2\pi=1.96$ N/m in Eq. (\ref{logarithmiclaw2}), while the corresponding factor
210: in Eq. (\ref{logarithmiclaw}) is $A=6.5$ N/m, which is three times larger than $Gb/2\pi$.
211: There are at least two reasons for this disagreement.
212: The first is that dissociation of dislocations is not taken into account in deriving Eq. (\ref{logarithmiclaw2}).
213: It takes much more stress to bend two partial dislocations simultaneously \cite{hatano}.
214: The second is the high dislocation density.
215: Since the glide of a single dislocation causes large strain relaxation in small systems,
216: the critical depinning stress may depend on the initial position of a dislocation in MD simulations.
217:
218:
219: Then we discuss the difference between the Hirsch and the Orowan mechanisms,
220: both of which are realized in the present simulation depending on the boundary conditions.
221: It should be remarked that the Orowan stress is almost the same as the Hirsch stress,
222: and therefore we must refine the analysis.
223: Recalling that the Hirsch mechanism is realized by displacing only the upper surface,
224: inhomogeneity of the shear strain may play a key role.
225: In order to see the inhomogeneity more explicitly, we calculate the average atomistic displacements
226: in the $y$ direction for each mechanism. We define
227: \begin{equation}
228: \delta y(z)=\frac{1}{n(z)}\sum_i\left[y_i(t)-y_i(0)\right],
229: \label{deltay}
230: \end{equation}
231: where the subscript $i$ denotes the $i$-th atom.
232: We take the sum regarding all the atoms that belong to the layer of $z<z_i(t)<z+1.89$ \AA.
233: Then $n(z)$ is the number of atoms in the layer, and $t$ is the time just before each bypass mechanism begins
234: (i.e. when the shear stress becomes maximum.)
235: \begin{figure}
236: \includegraphics[scale=0.3]{dyprofile.eps}
237: \caption{Spatial profiles of the average atomistic displacement in the $y$ direction when the shear stress becomes maximum.
238: The red solid line represents $\delta y (z)$ for the Hirsch mechanism,
239: while the green dashed line denotes that for the Orowan mechanism. The gradient of $\delta y(z)$ for the Hirsch mechanism
240: is asymmetric with respect to $z=95$ \AA (the glide plane). System parameters $r=1.5$ nm and $L=20.4$ nm.}
241: \label{dyprofile}
242: \end{figure}
243: FIG. \ref{dyprofile} shows $\delta y(z)$ for each mechanism.
244: For the Hirsch mechanism, the gradient of $\delta y$ is different with respect to the glide plane.
245: Because the cross-slip which precedes the Hirsch mechanism occurs towards the larger stress region,
246: it is concluded that the strong spatial inhomogeneity of the shear stress is essential to the Hirsch mechanism.
247: This spatial inhomogeneity is attributed to the existence of the precipitate.
248: Because the precipitate consists of immobile atoms, it inhibits the deformation caused by the boundary displacement.
249: Therefore, if the upper (or lower) surface is displaced, the elastic strain tends to localize on that side \cite{asymmetry}.
250: When the both surface is equivalently displaced, the elastic strain becomes symmetric to realize the Orowan mechanism.
251: The above discussion further leads us to the speculation that Orowan loops become unstable for the larger shear strain.
252: They can be decomposed into two prismatic loops via double cross-slip of the screw component,
253: which is essentially the same manner as the Hirsch mechanism.
254: For example, in the system shown in FIG. \ref{dyprofile}, the shear strain of $0.103$ is enough to destabilize Orowan loops.
255:
256: %%%% observation of a row of loops %%%%%%
257: While the discussions so far involve a single bypass process, for the rest of this Letter,
258: iteration of the Hirsch mechanism is briefly investigated.
259: Namely, we consider the dynamics of subsequent dislocations which successively interact with a precipitate.
260: It involves the relevance of the Hirsch mechanism to work-hardening.
261: Although prismatic loops can move under an appropriate stress field (prismatic punching),
262: they cannot move in the simple shear condition to be stagnant near the precipitate.
263: However, they do not cause hardening since subsequent dislocations can easily penetrate them.
264: For example, when $L_x=20.4$ nm and $r=1.5$ nm, it takes less than $180$ MPa for dislocations
265: to penetrate prismatic loops, which is much smaller than the Hirsch stress itself, $560$ MPa.
266: Plastic flow is maintained by repetition of the Hirsch mechanism,
267: which yields a row of prismatic loops as is shown in Fig. \ref{loops} (b).
268: This is what has been observed in the experiments \cite{humphreys,brown2,sarnek}.
269: By contrast, the loops on the right side of the precipitate do not form an apparent row.
270: Because the prismatic loops move towards the $+y$ direction (the left in FIG. \ref{loops})
271: when they are sheared by dislocations, the loops on the $-y$ (the right) side of the precipitate
272: cannot move across the precipitate to collide each other there.
273: This phenomenon may be related to plastic cavitation which is often observed near precipitates \cite{sato},
274: and also to the rotational plastic flow around a nondeformable particle \cite{brown3}.
275:
276: %%%%%%%%%%%% discussion %%%%%%%%%%%%%%%%%%%
277: To conclude, the present study indicates that the dynamics of dislocation-obstacle interaction is much richer than considered before.
278: Cross-slip plays the crucial role in a bypass mechanism which is quite different to the Orowan mechanism.
279: We have clarified the condition that determines which bypass mechanism occurs.
280: Namely, the shear stress which is asymmetric with respect to the glide plane causes cross-slip of the screw dipole
281: and the Hirsch mechanism follows.
282: Further development along the line of the present study will be the relevance of the Hirsch mechanism
283: to macroscopic plasticity. A three-dimensional continuum approach, which is referred to as "dislocation dynamics",
284: is now developing to compute the many-body dynamics of dislocations and obstacles \cite{zbib}.
285: It may be friutful that the present results are suitably incorporated to such continuum approaches.
286: In addition, it should be remarked that the dislocation dynamics is closely related to the interface dynamics
287: in the quenched disorder, which is extensively investigated in the field of nonlinear physics \cite{kardar}.
288: Dislocation physics would be much richer in communication with the different research field.
289:
290: The author gratefully acknowledges helpful discussions with Yuhki Satoh, Hideki Matsui, and Hideo Kaburaki.
291:
292:
293: \begin{thebibliography}{99}
294: \bibitem{friedel}
295: J. Friedel, {\it Dislocations} (Pergamon Press, Oxford, 1964).
296:
297: \bibitem{bacon}
298: D. J. Bacon, U. F. Kocks, and R. O. Scattergood,
299: Phil. Mag. {\bf 28}, 1241 (1973).
300:
301: \bibitem{scattergood}
302: R. O. Scattergood and D. J. Bacon, Acta Metall. {\bf 30}, 1665 (1982).
303:
304: \bibitem{osetsky}
305: Yu. N. Osetsky and D. J. Bacon, J. Nucl. Mater. {\bf 323}, 268 (2003).
306:
307: \bibitem{wirth}
308: B. D. Wirth, V. V. Bulatov, and T. Diaz de la Rubia,
309: J. Eng. Mater. Tech. {\bf 124}, 329 (2002).
310:
311: \bibitem{incoherency}
312: Although incoherency is an important ingredient that causes impenetrability in many alloy systems,
313: we do not consider incoherent precipitates here.
314:
315: \bibitem{argon}
316: A. S. Argon, {\it Physics of Strength and Plasticity} (MIT Press, Cambridge, MA, 1969).
317:
318: \bibitem{ashby}
319: M. F. Ashby, in {\it Oxide Dispersion Strengthening}, proceedings of the Second Bolton Landing Conference,
320: Bolton Landing, New York, 1966, edited by G. S. Ansell, T. D. Cooper, and F. V. Lenel
321: (Gordon and Breach, New York, 1968), p. 143.
322:
323: \bibitem{brown1}
324: L. M. Brown and W. M. Stobbs, Phil. Mag. {\bf 23}, 1185 (1971).
325:
326: \bibitem{hirsch1}
327: P. B. Hirsch, J. Inst. Metals {\bf 86}, 7 (1957).
328:
329: \bibitem{humphreys}
330: F. J. Humphreys and P. B. Hirsch, Phil. Mag. {\bf 318}, 73 (1970).
331:
332: \bibitem{brown2}
333: L. M. Brown and W. M. Stobbs, Phil. Mag. {\bf 23}, 1201 (1971).
334:
335: \bibitem{sarnek}
336: A. M. Wusatowska-Sarnek, H. Miura, and T. Sakai,
337: J. Mater. Sci. {\bf 34}, 5477 (1999).
338:
339: \bibitem{note1}
340: Although edge dislocations can also avoid impenetrable precipitates by the local climb motion,
341: this process requires long-range self-diffusion and is irrelevant to materials at room temperatures.
342:
343: \bibitem{finnis}
344: M. W. Finnis and J. E. Sinclair, Phil. Mag. A {\bf 50}, 45 (1984).
345:
346: \bibitem{ackland}
347: G. J. Ackland, D. J. Bacon, A. F. Calder, and T. Harry,
348: Phil. Mag. A {\bf 75}, 713 (1997).
349:
350: \bibitem{li}
351: J. Li, Modelling Simul. Mater. Sci. Eng. {\bf 11}, 173 (2003).
352:
353: \bibitem{hatano}
354: T. Hatano and H. Matsui, Phys. Rev. B {\bf 72}, 094105 (2005).
355:
356: \bibitem{sato}
357: Y. Satoh, T. Yoshiie, H. Mori, and M. Kiritani,
358: Mater. Sci. Eng. {\bf A 350}, 44 (2003).
359:
360: \bibitem{brown3}
361: L. M. Brown, Phil. Trans. R. Soc. Lond. A {\bf 355}, 1979 (1997).
362:
363: \bibitem{asymmetry}
364: Although the model precipitate of immobile atoms is somewhat artificial, the asymmetric shear stress
365: may be realized if other lattice defects exist near the precipitate, such as dislocations or grain boundary.
366:
367: \bibitem{zbib}
368: H. M. Zbib, M. Rhee, J. P. Hirth, and T. D. de La Rubia,
369: J. Mech. Behavior Mater. {\bf 11}, 251 (2000).
370:
371: \bibitem{kardar}
372: M. Kardar, in {\it Fundamental Problems in Statistical Mechanics}, edited by H. van Beijeren,
373: Phys. Rep. {\bf 301}, 85 (1998).
374:
375: \end{thebibliography}
376: \end{document}
377: