† Corresponding author. E-mail:
Project supported by the National Natural Science Foundation of China (Grant No. 11874093).
For a misfit dislocation, the balance equations satisfied by the displacement fields are modified, and an extra term proportional to the second-order derivative appears in the resulting misfit equation compared with the equation derived by Yao et al. This second-order derivative describes the lattice discreteness effect that arises from the surface effect. The core structure of a misfit dislocation and the change in interfacial spacing that it induces are investigated theoretically in the framework of an improved Peierls–Nabarro equation in which the effect of discreteness is fully taken into account. As an application, the structure of the misfit dislocation for a honeycomb structure in a two-dimensional heterostructure is presented.
Dislocations are structural defects that are central to the understanding of the mechanical properties of materials. With the recent development of optoelectronic devices based on heterostructures, it is crucial to understand the properties of the misfit dislocations that occur at the interface between two materials with different lattice constants. In fact, this is a topic that has received much attention for more than fifty years. To present the basic physical concept of misfit dislocation in the simplest way, Frank and Van der Merwe[1] extended earlier work on the Frenkel–Kontorova (F–K) model,[2] which describes an atomic chain linked by nearest-neighbor harmonic forces and subjected to a periodic substrate potential, taking account of the difference in lattice spacing between the chain and substrate. However, the dislocation obtained by the F–K model exhibits a short-range deformation similar to an exponential decay. On the basis of continuum elasticity theory, Dundurs[3] obtained the Airy stress function for an edge dislocation, using an analogy between concentrated forces and edge dislocations.[4] As in the original Peierls–Nabarro (P–N) model,[5,6] the misfit dislocations result from a nonlinear interaction between two semi-infinite crystals. Based on an extension of the P–N theory by van der Merwe,[7] Yao et al.[8] proposed an equation for an interfacial misfit dislocation that was similar to the P–N equation. By using an exact solution given first by van der Merwe[9] and later recovered by Yao et al.,[8] it is possible to use continuum elasticity theory to describe the long-range deformation of a dislocation at length scales outside the core field of the dislocation line. However, in general, it is clearly unrealistic to view a solid as a continuous medium, ignoring the discrete nature of the crystal lattice. Because the dislocation core structures play an important role in many crystal plasticity phenomena, it is of great interest to describe these structures accurately on the atomic scale.
Recent theoretical analyses of dislocation configurations have been based on molecular dynamics simulations[10] or local density function simulations. It is possible to obtain a misfit P–N equation by solution of the balance problem for a semi-infinite lattice,[11] and the resulting equation is similar to that derived by Yao et al.[8] However, in this approach, following that in the original P–N model, when the upper and lower semicrystals are glued together along their interface, the effect of the lower semicrystal on the bottom surface of the upper semicrystal is ignored, as is the effect of the upper semicrystal on the top surface of the lower semicrystal. Furthermore, the changes in the interfacial spacing introduced by misfit dislocations have also been ignored in previous studies.
In this paper, a correction to the misfit dislocation equation that takes account of discreteness is derived by modifying the displacement fields that satisfy the balance equations. An extra term proportional to the second-order derivative, which originates from a discrete feature, it is used to reflect the deformation that occurs near to the dislocation core. It is reasonable to view the modified equation as a combination of the F–K equation and the P–N equation. The integral term describes the elastic effect of the interior of the semi-infinite lattices regarded as continuous media, while the differential term describes the short-range interaction between atoms at the misfit interface and represents the effect of discreteness. To estimate the change in interfacial spacing induced by misfit dislocation, we consider the variation in the generalized stacking fault energy along the direction of interfacial spacing.
As in the original P–N model, misfit dislocations are created by gluing together two semi-infinite crystals of different materials, with each part being a regular crystal with the cut plane as its surface. As shown in a previous investigation of the balance problem for a semicrystal acted on by external forces on the surface,[12] the equilibrium equation for the atoms on the surface is
In the continuum approximation, it is difficult to distinguish between the surface and the interior of a semi-infinite crystal, and the surface effect is ignored. The second-order derivative describes the lattice discreteness effect that arises from the surface effect. To represent the effect of the discrete crystal lattice and to take account of the short-range interaction between the interfacial atoms, we consider a correction based on a reduced dynamical matrix for the surface of a semi-infinite lattice.[13] Taking into account that a heterostructure is composed of two dissimilar crystals, it is necessary to distinguish between the lattice parameters of materials 1 (upper) and 2 (lower), denoted by a and b, respectively, with
Equation (
In comparison with the equation of Yao et al.[8] and Zhang et al.,[11] Equation (
Using Eq. (
Taking account of the fact that the relative displacements of interfacial atoms
We consider a heterostructure in which the misfit occurs in a single direction, as illustrated by the 2D honeycomb structure in Fig.
In fact, the structure and properties of the dislocations depend on the restoring forces defined as the gradients of the generalized stacking fault energy.[16] Recent first-principles calculations of the generalized stacking fault energy (also generally called the γ-surface) have provided reliable values for the restoring forces. Because the lattice parameters of the materials do not match, as in the original P–N theory, we still assume that the effective interaction between two dissimilar semi-infinite crystals can be approximated using a reference lattice. The adhesion energy of the reference lattice varies with the interfacial distance, and this relationship can be approximated well by[11]
To estimate the size and shape of the change in interfacial spacing caused by the misfit dislocation between the misfit planes, it is necessary to calculate the γ-potential of the reference lattice. These calculations are carried out with the Vienna ab initio simulation package (VASP), which is based on Kohn–Sham density functional theory. The generalized gradient approximation (GGA) as proposed by Perdew, Burke, and Ernzerhof (PBE) for the exchange–correlation functions[17] is used and the core electrons are dealt with using the projector augmented wave (PAW) approach.[18,19] The plane-wave cutoff is 600 eV in all the calculations. The criterion to halt the relaxation of the electronic degrees of freedom is set to be when the total energy change is smaller than 10−6 eV. The convergence iteration is halted when the forces on the ions are smaller than
For an interfacial distance of the reference lattice d0=2.171 Å and an energy of adhesion
The lattice parameter of the upper semi-infinite crystal is smaller than that of the lower semi-infinite crystal, therefore there exists a compressive stress at the interface. The dislocation is characterized by a slip distribution that satisfies the boundary conditions
In the improved P–N equation, the second-derivative term describes the lattice discreteness effect that results from the surface effect. To facilitate comparison of the differences (see table
We can estimate the change in interfacial spacing on the basis of Eq. (
From the lattice parameters and elastic constants, we obtain the dimensionless expansion coefficients in Eqs. (
In fact, the equilibrium slip distribution is determined by the competition between two contributions to the energy of misfit dislocation. The first is the elastic strain energy stored in each semicrystal, which can be obtained through the variational formulation of the P–N theory by adding up the contributions from each infinitesimal dislocation and can be written as
To determine the energy of the misfit dislocation, we introduce the adhesion work, which can be written as
In an extension of the original P–N model, we have considered the effects at the interface when the upper and lower semicrystals are glued together. In general, these effects can be represented by second-order derivative terms to the F–K model. These terms can be viewed as corrections to the misfit dislocation equation to take account of the short-range interaction between atoms at the interface. The corrections presented here allow more accurate theoretical predictions of the interfacial atomic configuration. Using a γ-potential that includes the interaction energy due to the variation in interfacial spacing, together with an equation for the normal formation of the glide plane, we have estimated the change in spacing for an interface exhibiting misfit dislocation. We have presented an application of the improved P–N equation to an interfacial misfit dislocation in a BN/AlN heterostructure. Based on an approximate solution of the misfit dislocation equation, the theoretical result for the interface displacement along the glide direction is close to the numerical result. Although we have taken into account the change in interfacial spacing induced by misfit dislocation, it can be seen that there are some discrepancies between the theoretical and numerical results for the interface displacement along the y direction, and so the agreement between the results of analytical theory and those of numerical calculation needs to be further improved. We have found that the correction for the effects of discreteness reduces the value of the energy of misfit dislocation.
[1] | |
[2] | |
[3] | |
[4] | |
[5] | |
[6] | |
[7] | |
[8] | |
[9] | |
[10] | |
[11] | |
[12] | |
[13] | |
[14] | |
[15] | |
[16] | |
[17] | |
[18] | |
[19] | |
[20] | |
[21] | |
[22] |