Materials Science and Engineering A281 (2000) 148 – 155 www.elsevier.com/locate/msea On the disclination-structural unit model of grain boundaries A.A. Nazarov a, O.A. Shenderova b,*, D.W. Brenner b a b Institute for Metals Superplasticity Problems, Russian Academy of Science, 39 Khalturin str., 450001 Ufa, Russia Department of Materials Science and Engineering, North Carolina State Uni6ersity, Raleigh, NC 27695 -7907, USA Received 4 August 1999; received in revised form 22 October 1999 Abstract The relation between two elastic continuum approaches to grain boundary structure, the dislocation and disclination models, is discussed. It is shown that the disclination model has two advantages: a well-behaved expression for the elastic energy of disclination dipole walls, which describes the elastic energy over a wide interval of misorientations, and a continuous misorientation angle dependence of the elastic energy of grain boundaries in an interval between two delimiting boundaries. The elastic energy of the most general, faceted disclination wall is calculated. For cases in which both the energies of delimiting boundaries and elastic constants are available from atomic simulations (001 and 111 tilt boundaries in copper and 001 and 011 tilt boundaries in diamond) quantitative agreement between the disclination model and simulation results is obtained. © 2000 Elsevier Science S.A. All rights reserved. Keywords: Dislocation; Disclination; Grain boundaries; Structural units 1. Introduction In 1950, Read and Shockley [1] derived their classical formula for the elastic energy of low-angle dislocation tilt boundaries gel = eh Gb 2 ln 4p(1− n)h 2pr0 (1) where gel is the energy per unit area of the boundary, G and n are the shear modulus and Poisson ratio, respectively, h is the period of the boundary, b is the magnitude of the Burgers vector of the dislocations, and r0 is the core radius. Since then, numerous attempts have been made to describe the energy of more general, high-angle grain boundaries in terms of the continuum theory of dislocations. The main difficulty encountered by these attempts is that it is impossible to uniquely define a value for the dislocation core energies that are added to the elastic energy given by Eq. (1) and that plays an important role in high angle boundaries. In the 1980s, insights from atomistic computer simulations lead to the structural unit/dislocation model of grain * Corresponding author. E-mail address: [email protected] (O.A. Shenderova) boundaries [2–4]. Based on this model, Wang and Vitek [5] proposed a modification of Eq. (1) that includes the core energy of grain boundary dislocations, and obtained energy versus misorientation angle (g(u)) curves for 001 and 111 tilt boundaries in copper that reproduce the results of computer simulations relatively accurately. However, the entire misorientation range for each axis had to be divided into small subintervals in order to get meaningful elastic energies from Eq. (1). As Sutton and Balluffi [6] pointed out, this shortcoming is not a result of the structural unit model, but rather is due to the logarithmic term in Eq. (1) that imposes a severe constraint on the misorientation range covered. Another way to describe grain boundaries in terms of linear defects is the use of the disclination model [7,8]. A general grain boundary with a misorientation angle u can be considered to consist of alternating segments of two special boundaries uA and uB, which are delimited by partial disclinations of strength 9 v= 9 (uB −uA). Thus the elastic energy of the boundary is given by the energy of a disclination dipole wall [8] gel = Gv 2h f(l) 32p 3(1−n) 0921-5093/00/$ - see front matter © 2000 Elsevier Science S.A. All rights reserved. PII: S 0 9 2 1 - 5 0 9 3 ( 9 9 ) 0 0 7 2 7 - 3 (2) A.A. Nazaro6 et al. / Materials Science and Engineering A281 (2000) 148–155 where h is the period of the wall, l = pLB/h, LB is the length of the segment of special boundary B, and 2l 6 d6 f(l)= 2 sin u du 0 0 cosh 6 −cos u & = − 16 & & l (l− 6) ln 2 sin 6d6 (3) 0 Eq. (2) produces reasonable results for any deviation from the special misorientation, and is better suited for calculating the high angle grain boundary energies than Eq. (1). In Refs. [9,10], the disclination model has been considered as a direct consequence of the structural unit model. The lines at which different types of structural units meet are disclinations. This allows the minority structural units to be considered as disclination dipoles. To emphasize the origin from the structural unit model, the new disclination model has been called the disclination-structural unit model of grain boundaries. This model has recently been used to calculate the energies of 001 symmetrical tilt grain boundaries in diamond, and a reasonably good fit to simulation data has been obtained [11]. In the present paper, the relationship between the dislocation and disclination models is discussed. It is demonstrated that despite the more widespread application of the former, the disclination model is more convenient, requires fewer parameters, and can be used Fig. 1. Atomic structures of the S = 25 [001] (170) symmetric tilt boundary with a misorientation angle u = 73.74° in diamond: (a) the lowest energy configuration with coupled Az units and the corresponding disclination representation; (b) the higher energy zig-zag configuration with separated Az units. Fig. 2. A schematic rendering of the step associated with an Az unit inserted in a (010) plane in diamond cubic crystals. 149 to obtain a reasonably good approximation of the energy of high angle tilt boundaries. The elastic energy of the most general, faceted disclination dipole wall is then calculated. Energy versus misorientation angle curves are calculated for tilt grain boundaries in copper and diamond, systems for which elastic constants and grain boundary energies are available from computer simulations. 2. Disclinations and dislocations in the structural unit model Grain boundary structures in face-centered cubic (f.c.c.) metals are usually characterized by straight arrangements of two types of structural units that can be modeled by flat disclination walls. More complicated structures appear, however, in covalent materials having tetra-coordinated atoms. For example, computer simulations show that the stable configurations of some 001 and 011 tilt boundaries in silicon and diamond have zig-zag arrangements of units [12–15]. Zig-zag tilt boundaries have also been observed by high-resolution electron microscopy in silicon and germanium [16,17]. Illustrated in Fig. 1 are two structures of the boundary S= 25 [001] (170) with a misorientation angle u=73.74° in diamond that have been obtained by simulations using an analytic potential [14]. Both structures are composed of structural units of two types: two units of the S=5 (130) u= 53.13° boundary (we denote them type Az units), and four units of the (010) plane of the perfect lattice, or S= 1/u = 90° ‘boundary’ (F% units). In the first structure, which has the lowest energy, the Az units are coupled into pairs. These pairs constitute the cores of dislocations with Burgers vector b= a0[010]. In the second structure, the Az units are split and constitute the cores of lattice dislocations with alternating Burgers vectors b= 12a0[110] and b= 12a0[1( 10]. It is easily seen that the pairs of Az units in the first type of structure can be considered as disclination dipoles whose arm and strength are equal to 2dA = 10/2 and v=90− 53.13°=36.87°, respectively (Fig. 1(a)). In the second structure Az units shift the planes of F% units (Fig. 1(b)), and the corresponding disclination wall is faceted, i.e. the disclinations lie on two different planes with spacing a. The exact value of the distance a can be obtained using the condition that the Burgers vector of the dislocation calculated from the disclination model equals the Burgers vector known from the coincidence site lattice geometry. A single Az unit inserted between units of the ideal lattice plane (010) as well as a corresponding disclination dipole with an arm L and a strength equal to v= uF% − uA = 36.9° are illustrated in Fig. 2. The Burgers vector of a dislocation equivalent to this dipole can be calculated as b= 2L sin[(uF% −uA)/2] = 2L/ 10 [18]. This must be equal to the magnitude 150 A.A. Nazaro6 et al. / Materials Science and Engineering A281 (2000) 148–155 values, the elastic energy of the disclination dipole wall approaches that given by the dislocation model. For example, at uuA (l=pLB/h0) the function f(l) behaves as f(l): 8l 2(3/2 − ln 2l). Substituting this approximation into Eq. (2) yields gel = Fig. 3. The dependence of the elastic energy of 001 symmetric tilt grain boundaries in an f.c.c. crystal calculated by the disclinationstructural unit (solid line) and dislocation models. The long-dash line corresponds to Eq. (4), short-dash curve to Eq. (5). of the lattice dislocation Burgers vector b = a0/ 2; therefore, L= 5a0/2. This is larger than the dimension of the Az unit along the facet it forms, and is equal to the length of a complex unit that includes an additional unit of the perfect lattice. This complex unit is shown filled in Fig. 2. Components of the Burgers vector normal and tangential to the grain boundary plane are both equal to a0/2, i.e. the dipole must be inclined at an angle of 45° to the grain boundary plane. Thus, the height of the step associated with the dipole is equal to a=L/ 2 = 5a0/2 2. The facetted disclination dipole wall having this separation of planes will be geometrically identical to the corresponding dislocation network. As one can see from a comparison of Fig. 1(b) and Fig. 2, the faceting of the intermediate boundaries in the disclination model is apparently different from that in the structural unit representation. However, both representations are virtually equivalent, since they describe the same atomic structure. Thus, even in the case of grain boundaries in which structural dislocations have a component tangential to the boundary plane, it is possible to construct an appropriate disclination model. The corresponding disclination dipole wall will be faceted, with the step height determined from the condition of the geometrical equivalence of dislocation and disclination models. 3. Comparison of the disclination and dislocation models Shih and Li [8] showed that for the cases when the misorientation angle deviates slightly from the special GbA e 3/2v (u−uA) ln 2p(u− uA) 4p(1− n) (4) where bA = LBv and u− uA = bA/h. When starting from the structural unit model, the exact value of the Burgers vector is calculated as bA = 2dB sin (v/2) [18], dB being equal to the dimension of the non-deformed structural unit of the delimiting boundary B. Eq. (4) is identical to the Read–Shockley Eq. (1) if r0 =dB/e 1/2 is assumed. For the other end of the misorientation interval, the elastic energy in the dislocation model is also given by Eq. (4), where u−uA is replaced by uB −u, dB by dA and bA by bB. Wolf [19] recently proposed an empirical relationship for the energy of high angle grain boundaries in the whole misorientation range derived from the dislocation model. In this relationship, for the case of 001 symmetrical tilt boundaries, for example, u −uA in Eq. (4) is replaced by 2 sin[(u − uA)/2]. By adjusting the coefficients before and in the logarithm, the results of computer simulations for the whole misorientation angle could be fit fairly well. This fact, according to Wolf, ‘suggests that the underlying physics may be the same for low- and high-angle grain boundaries’ [19]. This physics, by the authors’ opinion, is based on the disclination model. Wolf’s equation is also an approximation to the energy of disclination dipole walls. It differs from Eq. (4) only by the use of the more exact geometrical relationships b/h= 2 sin[(u − uA)/2] and L2/ h= sin[(u − uA)/2]/sin(v/2), but uses the same first order approximation to the function f(l). Thus, the elastic energy of a disclination dipole wall in Wolf’s approximation can be calculated from the following equation: gel = n GbA u− uA e 3/2 sin(v/2) 2 sin ln 4p(1− n) 2 2psin[(u − uA)/2] (5) and from a similar equation for the misorientation angles close to uB. Plotted in Fig. 3 are the results of calculations of the elastic energy of 001 tilt boundaries using Eqs. (2), (4) and (5), taking as delimiting ‘boundaries’ the two symmetry planes of the f.c.c. lattice, (110) and (010). In this case dA = a0/ 2, bA = a0/ 2 and dB = a0/2, bB = a0. The solid curve was calculated using the disclination model, with disclination dipole arms equal to dB at the left of the misorientation angle 36.87° and to dA at the right of the misorientation angle. It should be noted that different structural units are considered as disclina- A.A. Nazaro6 et al. / Materials Science and Engineering A281 (2000) 148–155 tion dipoles at the left and right of the AB type structure for calculational convenience. As a matter of fact, the energy of a wall of disclinations does not depend on the way they are coupled into dipoles. For simplicity, the spacing of the dipoles is assumed to change continuously, though this contradicts the structural unit model. This results in a smooth curve having only one cusp at the misorientation angle S= 5/36.87° that corresponds to a structural units stacking AB. The long-dashed curve in Fig. 3 has been calculated from Eq. (4) for the left branch starting from the boundary A and its analogue for the right branch starting at the boundary B. Similarly, the short-dashed curve corresponds to Eq. (5) and its right-branch counterpart. Comparing the results of the three approaches yields the following. First, the dislocation model, which is considered here as a limiting case of the disclination model, provides positive elastic energy values even for very large misorientation angles. This occurs because of the fact that the disclination model predicts that core radii for grain boundary dislocations equal r0 = dB/e 1/2 or r0 =dA/e 1/2, these are less than those usually accepted [5]. The energy can assume negative values, however, even in this case, if the boundaries are composed of structural units of very different dimensions. In such cases, the AB structure occurs at a misorientation angle close to one end of the range between the A and B type boundaries. When approaching this structure from one side, a negative energy may be given by Eqs. (4) and (5), because the dislocation spacing closely approaches the dislocation core diameter. The use of 2 sin(u/2) instead of u does not improve the general behavior of the elastic energy. In both cases of the 151 dislocation model, different elastic energies are given for the transition structure AB when approaching it from the two sides. This is due to the fact that the first-order approximation to f(l) results in the loss of the property f(p − l)= f(l). In Ref. [19], the equality of the energies is achieved by fitting the coefficients multiplying the logarithm. Such fitting is, however, somewhat arbitrary because these coefficients depend on the elastic constants and cannot be changed independently. The conventional dislocation model includes the core energy of dislocations in the logarithmic term [20]. In this way, r0 can be fit for each branch such that the total grain boundary energy assumes the same value at the AB type structure. Similar to the disclination-structural unit model, this requires a single parameter, the energy of the boundary with an AB structure. However, the partitioning of the elastic and core energies in this case is not as clear as it is in the disclination model. Thus, not only can the dislocation model give physically meaningless negative energy values at certain misorientations, but also different elastic energies are obtained for the same grain boundary structure at which the transformation between two dislocation descriptions occurs (the AB type structure). On the other hand, the disclination-structural unit model always yields physically meaningful positive elastic energies and the same elastic energy for the transition structure. (Actually, there is no transition at all, and one changes the disclination dipoles at the AB structure only for convenience.) We conclude, therefore, that the disclination-structural unit model, being geometrically equivalent to the dislocation model, is more convenient for calculations of grain boundary energies over wide misorientation ranges. 4. Elastic energy of a general disclination wall Fig. 4. A faceted disclination wall representing the structure of zig-zag tilt grain boundaries. Consider now the disclination wall depicted in Fig. 4. This wall is representative of the disclination structure of zig-zag grain boundaries, an example of which has been presented in Fig. 1(b). The segments l1 and l2 of the grain boundary consist of one type of structural units, while the other type of units between them shifts the planes of these facets to a distance 9 a with respect to each other. A disclination wall corresponding to a straight arrangement of the structural units will be a particular case of the wall under consideration when a= 0. To calculate the elastic energy of the wall, it is convenient to couple the disclinations into dipoles as illustrated in Fig. 4 so that the complex wall is represented as a pair of single dipole walls each of period h separated by the distance a along the x-axis and the distance y along the y-axis. The elastic energy of the A.A. Nazaro6 et al. / Materials Science and Engineering A281 (2000) 148–155 152 system is a sum of self energies of the single dipole walls g ps (p=1, 2), and their interaction energy gint. The former are given by the equation g ps = Kev h f(lp ) (p = 1, 2) 32p 3 N−1 (6) K v 2h gint = e 3 [g(d,z +l2) − g(d,z) − g(d,z +l2 −l1) 32p + g(d,z −l1)] & (7) 2d x dx 0 & 2t 0 sin y dy cosh x −cos y f(t)−q(d,t) (8) Similar to the function f(t), the function q(d,t) can be expanded by the series q(d,t) =4d 2 ln ! ( −1)kB2k 22k + 2 d 2 +t 2 d 2 +t 2 + 4t 2 ln + % 2 2 d t k(2k)! k=1 2t 2k + 2 − (t+ id)2k + 2 −(t −id)2k + 2 2k + 2 (t+ id)2k + 1 +(t −id)2k + 1 −2t 2k + 1 2k + 1 +t −2(−1)k " d 2k + 2 2k + 2 N % [g(x̃p − x̃k,ỹp − ỹk + lp ) k=1p=k+1 " − g(x̃p − x̃k,ỹp − ỹk )− g(x̃p − x̃k,ỹp − ỹk +lp −lk ) + g(x̃p − x̃k,ỹp − ỹk − lk )] (10) with x̃k = pxk /h and ỹk = pyk /h. Note that in Eq. (10), x̃p − x̃k is equal to either 0 or d, since the p-th and i-th dipoles can share the same plane or be displaced by a. The total energy of the zig-zag grain boundary which is represented by such a disclination dipole wall is a sum of gel, the additive surface energy of structural units, mdAgA + ndBgB h (11) and the disclination core energies [9–11] where z=py/h, d =pa/h, and g(d,t)= f(t)−2 ! Kev 2h N % f(lk ) 32p 3 k − 1 + % 2 where lp =plp/h. Note that the factor G/(1 −n) determining the energy of the disclination dipole walls in the isotropic theory of elasticity (Eq. (2)) has been replaced in Eq. (6) by the energetic factor Ke for edge dislocations that can be calculated in the anisotropic theory of dislocations [20]. The interaction energy is calculated as the work done by interaction forces when moving one wall to infinity in the shear stress field sxy of the other. The use of the dislocation model of disclination dipoles [21] and conventional calculations yield the following expression for the interaction energy: × gel = (9) where i= − 1. The series in Eq. (9) converges relatively quickly in the range 0Bt B p/2 for d 5 2.6. This maximum value of d is well above those that appear in the structures of 001 and 011 grain boundaries [14,15]. Taking account of the fact that q(d, p −t)=q(d, t), Eqs. (8) and (9) allow easy computation of the function g(d, t) for all relevant values of d and t. This result can be generalized to a more complex structure of zig-zag boundaries which is represented by a set of more than two single dipole walls. Let the grain boundary period be composed of N disclination dipoles with the arms lp, ordinates yp and x-coordinates xp; the latter assume values of 0 or a. The elastic energy of this wall is given by a sum of the self energies of N single dipole walls, Eq. (6), and pair interaction energies, Eq. (7), added up over N(N −1) pairs. Hence, Kea0v 2N 32p 3h a (12) In Eqs. (11) and (12), m and n are the numbers of the A and B type of structural units in a disclination dipole wall, gA and gB are the energies of the delimiting grain boundaries composed of these units, and a is a parameter determining the disclination core energy contribution. An analysis of Eq. (10) shows that the elastic energy of a complex disclination dipole wall with given ordinates of the dipoles yi (i= 1,2,…N) is minimum, when all dipoles share the same plane, i.e. a=0. The fact that in the diamond cubic lattice, some grain boundaries have a zig-zag structure is associated with geometric restrictions resulting from the tetra-coordinated bonding of atoms. Due to these restrictions, a flat disclination dipole can be formed from Az type units only by coupling them into pairs, that is increasing the arms of disclination dipoles, as for example, in the case of S= 25/u =73.74° boundary (Fig. 1). The elastic energy of a faceted wall containing single Az units is less than the elastic energy of a straight wall which has a larger disclination dipole arm. Nevertheless, as seen from the results of simulations, the boundary structure with a straight arrangement of units is energetically more favored. This is explained by a lower number of disclinations per period and consequently a lower contribution of the disclination core energy (Eq. (12)) for the straight boundary than for the zig-zag one. 5. Energies of tilt grain boundaries in copper and diamond The disclination-structural unit model requires the energies of delimiting boundaries and bulk elastic prop- A.A. Nazaro6 et al. / Materials Science and Engineering A281 (2000) 148–155 Fig. 5. Misorientation angle dependence of the energies of 001 symmetric tilt boundaries in copper. The dashed line has been obtained by using two delimiting ‘boundaries’, S = 1/0° and 90°. The solid curve is obtained with two additional delimiting boundaries, S = 5/36.9° and 53.1°. The dots represent the data from computer simulations [5]. Fig. 6. Energy versus misorientation angle curves for 111 symmetric tilt boundaries in copper. The dashed line is obtained with no intermediate delimiting boundary, and the solid line corresponds to one intermediate delimiting boundary, S= 7/38.2°. The curves fit the data from Ref. [5]. erties as input. It is convenient to use the energies calculated by atomistic simulations for the former, and either experimental or calculated properties for the latter. In this case the set of atoms described by a given potential can be formally considered as a model material for which the validity of the disclination approach can be tested. 153 Vitek and co-authors obtained a suitable set of simulation data for 001 and 111 symmetrical tilt boundaries in copper [4,5]. The energy factor Ke for the tilt axis 001 calculated from the elastic constants reported in Ref. [5] is Ke = 6.81× 1010 Pa. The energy versus misorientation angle curves for this axis are plotted in Fig. 5. The dashed line was derived using the minimum number of two delimiting ‘boundaries’, S= 1/0° and 90°, with the core energy parameter a=24.1. The solid curve was obtained using two additional delimiting boundaries, S= 5/36.9° and 53.1°, and corresponding core energy parameter values of a1 =12, a2 = 4, and a3 = 48. Note that there are cusps in the g(u) curves not only at the misorientation angles of delimiting boundaries, but also at some intermediate misorientations. These cusps correspond to grain boundaries which have relatively short periods (for instance, structural unit stacking of type AB, AAB, ABB, etc.). Actually, the energy curve consists of an infinite number of cusps the most of which are very small. These local minima occur due to the fact that the elastic energy of a disclination dipole wall increases when slightly deviating from the misorientation angle of any short period boundary, since this deviation is associated with a largeperiod perturbation of the arrangement of disclination dipoles. A rough estimate of grain boundary energies over the entire range of misorientation angle is given even without intermediate delimiting boundaries. Introducing two intermediate delimiting boundaries results in reasonably good agreement with the simulation data. Plotted in Fig. 6 are the results for the 111 tilt axis in copper, for which Ke = 8.24× 1010 Pa [5]. Again, a good approximation is obtained with one intermediate delimiting boundary, S= 7/38.21°. The corresponding a values are a1 = 14 and a2 = 40. The g(u) curves for 011 tilt boundaries in diamond are plotted in Fig. 7. These curves fit simulation data from Ref. [22], where the Tersoff analytic potential and tight binding calculations were used. The solid line is the fit to the Tersoff potential calculations, for which Ke : G/(1−n)= 6.06× 1011 Pa and a1 = 22.9. The other parameter values, a2 = 18, a3 = 12 are taken only as an example. The dashed line fits the tight binding calculations, for which Ke : G/(1−n)= 4.86×1011 Pa, a1 = 22, and again sample values of a2 = 18 and a3 =12 are used for the other intervals. Figs. 8 and 9 present the results of the disclinationstructural unit model calculations of the energies of 001 and 011 tilt boundaries in diamond using as input, results from atomic simulations using Brenner’s many-body bond-order potential [14,15]. For the 001 axis intermediate delimiting boundaries are S=5/ 36.87° and 5/53.13°. Left of the latter, the stable grain 154 A.A. Nazaro6 et al. / Materials Science and Engineering A281 (2000) 148–155 boundary structures have a straight arrangement of structural units. In the range 53.13 Bu B 90° the grain boundaries contain coupled Az units (Fig. 1(b)). The energy factor for the 001 axis calculated from the elastic constants [11] is Ke =5.83 ×1011 Pa. A reasonable fit is obtained with parameter values a1 = 21 for 0 B u B 36.9°, a2 =36 for 36.9B u B 53.1° and a3 = 27 for the range 53.1B u B90°. For 011 tilt boundaries, the following delimiting boundaries have been used [15]: S =1/0°, 19/26.53°, 9/38.94°, and 3/70.53°. In the range 0 Bu B26.53° a straight arrangement of structural units occurs, while in other sub-intervals the zig-zag structures yield faceted Fig. 9. Energies of symmetric 011 tilt boundaries in diamond with input from the simulation results reported in Ref. [15]. Fig. 7. Energies of symmetric 011 tilt boundaries in diamond fitting the data of simulations [22] with Tersoff potential (solid line) and in tight binding approach (dashed line). disclination walls. It can be shown [23] that in all these cases separation of the disclination dipole planes is a= 3a0/4. A calculation of the anisotropic energy factor for edge dislocations, the lines of which are parallel to 011, yields K xe = 5.96× 1011 Pa for the Burgers vector component along the 100 direction and K ye =6.50× 1011 Pa for the Burgers vector parallel to 011. The former value is used for lattice dislocations composing low-angle grain boundaries with median plane (100) that correspond here to misorientation angles near 180°. The latter is used for low-angle boundaries with (01( 1) median plane, i.e. for misorientation angles near 0°. To take into account the change of the energy factor with grain boundary misorientation angle u, an effective factor is calculated as Kx = [(K ye cos u/2)2 + (K ye sin u/2)2]1/2. The following values of the core parameter have been chosen to obtain a good fit to the simulation data for the minimum energy configurations: a1 = 40 for 0B uB26.53°, a2 = 36 for 26.53BuB 38.94°, and a3 = 22 for 38.94B uB70.53°. 6. Discussion Fig. 8. Misorientation angle dependence of the energies of symmetric 001 tilt boundaries in diamond, to fit the data of simulations with the bond-order analytic potential [14]. Two models describing grain boundary structure in terms of elastic continuum theory, the dislocation and disclination-structural unit models, have been compared. Both of these models are based on the same structural unit model and are geometrically equivalent. However, the elastic energy of grain boundaries and, therefore, the partitioning of the elastic and core energies are expressed quite differently. The dislocation model traditionally uses a cut-off, or core radius r0. Due to the logarithmic term including this parameter, the model can yield physically meaningless, negative energies for large misorientation angles of the dislocation wall. If the dimensions of the two types of struc- A.A. Nazaro6 et al. / Materials Science and Engineering A281 (2000) 148–155 tural units composing intermediate boundaries in a given misorientation interval are similar, the elastic energy given by Eq. (4) is positive in this interval. If the dimensions differ greatly, the core diameter of dislocations equivalent to large sized units can closely approach the dislocation spacing, thus leading to negative values of the elastic energy. This is not the only shortcoming of Eqs. (1), (4) and (5). The dislocation model gives two branches of the energy versus misorientation angle curve that start from the two delimiting boundaries A and B (Fig. 3). These branches yield different elastic energy values for the same misorientation angle at which a transformation from one dislocation representation to the other occurs (that is, at which the grain boundary has a structural unit stacking of the type AB). In contrast, the disclination-structural unit model always gives meaningful elastic energies, and the elastic energy versus misorientation angle curve is continuous at the AB-type structure. Therefore, the disclination model allows for a more straightforward division of the grain boundary energy into elastic and core contributions. Due to the increase of allowed misorientation subintervals, the disclination-structural unit model enables a reduction in the number of key structures (delimiting boundaries) whose energies are used as input to calculate the grain boundary energies in the whole misorientation range compare to the dislocation approach. This feature is crucial for a multiscale modeling approach recently introduced [11]. The main idea of this approach is to use first-principles density functional methods to calculate energies of the delimiting structures, and use these energies (and elastic properties) as input into the disclination-structural unit model to yield accurate energies for intermediate grain boundaries. Sufficiently large misorientation intervals covered by the disclination-structural unit model enables the proposed multiscale scheme. A comparison of the results for model materials described by analytic potentials made in the present paper demonstrates the efficiency and accuracy of this multiscale approach. . 155 Acknowledgements A.A. Nazarov has been supported by NCSU through a Subcontract No. 95-0012-01 as a part of the Prime Grant No. N00014-95-1-0270 from the Office of Naval Research through which O.A. Shenderova and D.W. Brenner were supported. References [1] W.T. Read, W. Schockley, Phys. Rev. B 78 (1950) 275. [2] A.P. Sutton, V. Vitek, Phil. Trans. R. Soc. Lond. A309 (1983) 1,37,55. [3] A.P. Sutton, R.W. Balluffi, V. Vitek, Scripta Metall. 15 (1981) 989. [4] G.-J. Wang, A.P. Sutton, V. Vitek, Acta Metall. 32 (1984) 1093. [5] G.-J. Wang, V. Vitek, Acta Metall. 34 (1986) 951. [6] A.P. Sutton, R.W. Balluffi, Interfaces in Crystalline Materials, Clarendon, Oxford, 1995, p. 305. [7] J.C.M. Li, Surf. Sci. 31 (1972) 12. [8] K.K. Shih, J.C.M. Li, Surf. Sci. 50 (1975) 109. [9] V.Y. Gertsman, A.A. Nazarov, A.E. Romanov, R.Z. Valiev, V.I. Vladimirov, Phil. Mag. A 59 (1989) 1113. [10] R.Z. Valiyev, V.I. Vladimirov, V.Y. Gertsman, A.A. Nazarov, A. Romanov, Phys. Metals Metallogr. 69 (3) (1990) 30. [11] O.A. Shenderova, D.W. Brenner, A.A. Nazarov, A.E. Romanov, L.H. Yang, Phys. Rev. B 57 (1998) R3181. [12] M. Kohyama, Phys. Stat. Solids (b) 141 (1987) 71. [13] M. Kohyama, R. Yamamoto, M. Doyama, Phys. Stat. Solids (b) 137 (1986) 11. [14] O.A. Shenderova, D.W. Brenner, L.H. Yang, Phys. Rev. B 60 (1999) 7043. [15] O.A. Shenderova, D.W. Brenner, Phys. Rev. B 60 (1999) 7053. [16] A. Bourret, J.L. Rouviere, in: J.H. Werner, H.J. Möller, H.P. Strunk (Eds.), Polycrystalline Semiconductors, Springer, Berlin, 1989, p. 8. [17] J. Thibault, J.L. Putaux, A. Jacques, A. George, H.M. Michaud, X. Baillin, Mater. Sci. Eng. A164 (1993) 93. [18] A.A. Nazarov, A.E. Romanov, Phil. Mag. Lett. 60 (1989) 187. [19] D. Wolf, Scripta Metall. 25 (1989) 1713. [20] J.P. Hirth, J. Lothe, Theory of Dislocations, Wiley, New York, 1982. [21] A.E. Romanov, V.I. Vladimirov, in: F.R.N. Nabarro (Ed.), Dislocations in Solids, vol. 9, North-Holland, Amsterdam, 1992, pp. 191 – 402. [22] J.R. Morris, C.L. Fu, K.M. Ho, Phys. Rev. B 54 (1996) 132. [23] A.A. Nazarov, O.A. Shenderova, D.W. Brenner, Phys. Rev. B (2000) in press.
© Copyright 2026 Paperzz