Alexander V. Lebedev*a,
Irina V. Lebedevab,
Andrey A. Knizhnikac and
Andrey M. Popovd
aKintech Lab Ltd., Moscow 123182, Russia. E-mail: allexandrleb@gmail.com
bNano-Bio Spectroscopy Group and ETSF Scientific Development Centre, Departamento de Física de Materiales, Universidad del País Vasco UPV/EHU, San Sebastian E-20018, Spain. E-mail: liv_ira@hotmail.com
cNational Research Centre “Kurchatov Institute”, Moscow 123182, Russia
dInstitute for Spectroscopy of Russian Academy of Sciences, Troitsk, Moscow 142190, Russia. E-mail: popov-isan@mail.ru
First published on 8th December 2015
The principal characteristics of interlayer interaction and relative motion of hexagonal boron nitride (h-BN) layers are investigated by the first-principles method taking into account van der Waals interactions. The dependence of the interlayer interaction energy on the relative translational displacement of h-BN layers (potential energy surfaces) are calculated for two relative orientations of the layers, namely, for the layers aligned in the same direction and in the opposite directions upon the relative rotation of the layers by 180 degrees. It is shown that the potential energy surfaces of bilayer h-BN can be approximated by the first Fourier components determined by symmetry. As a result, a wide set of physical quantities describing the relative motion of h-BN layers aligned in the same direction including barriers to their relative sliding and rotation, shear mode frequency and shear modulus are determined by a single parameter corresponding to the roughness of the potential energy surface, similar to bilayer graphene. The properties of h-BN layers aligned in the opposite directions are described by two such parameters. The possibility of partial and full dislocations in stacking of the layers is predicted for h-BN layers aligned in the same and opposite directions, respectively. The extended two-chain Frenkel–Kontorova model is used to estimate the width and formation energy of these dislocations on the basis of the calculated potential energy surfaces.
The structural and elastic properties of the ground state of bulk h-BN have been studied experimentally in detail. The interlayer distance,28–36 bulk modulus,31,32,35,36 shear modulus,35 shear mode frequency37,38 (E2g mode with adjacent layers sliding rigidly in the opposite in-plane directions) and frequency of relative out-of-plane vibrations (B1g ZO mode with adjacent layers sliding rigidly towards and away from each other)39 have been measured and stacking of the layers has been established.28,40 The interlayer distance has also been determined for few-layer h-BN,41–43 while for bilayer h-BN, only stacking is known from the experimental studies.42 Rotational stacking faults have been observed for few-layer h-BN.42 However, no experimental data on the interlayer binding energy, corrugation of the potential surface of interlayer interaction energy and barriers to in-plane relative displacement of the layers are available neither for bilayer nor for few-layer or bulk h-BN.
State-of-the-art first-principles methods have been applied to calculate the interlayer distance,39,43–55 interlayer binding energy,43,48,54 bulk modulus,44–48,50,51,53 shear modulus,53 shear mode frequency46,52,53 and frequency of out-of-plane relative vibrations39,45,46,52,53 for bulk h-BN. Much less theoretical data are available for bilayer h-BN, including the interlayer distance,43,48,54–56 interlayer binding energy,43,48,54,56 bulk modulus and frequency of out-of-plane vibrations.43 The dependence of the interlayer binding energy on the layer stacking has been analyzed both for bulk49,51,55 and bilayer h-BN54,55,57 and the barriers to relative sliding of the layers have been calculated.54,55,57 However, the results of these calculations on relative in-plane motion of h-BN layers are rather contradictory and obtained with the aim of studying the balance between van der Waals and electrostatic forces without relation to measurable physical quantities.
Considerably more detailed theoretical and experimental data are available for the interaction and relative motion of graphene layers. The approximation58 of the interaction energy of a single carbon atom in a graphene flake with a graphite layer by the first Fourier components determined by graphene symmetry has been used to derive a simple expression for the potential surface of the interaction energy of graphene layers.9,10,59 The latter has been confirmed both by density functional theory calculations with the dispersion correction (DFT-D) and calculations using empirical potentials with very good accuracy.10,60,61 According to this expression, the shape of the potential energy surface at a given interlayer distance is determined by a single parameter and thus links different physical quantities related to relative motion of the layers such as barriers to relative sliding and rotation, shear mode frequencies and shear modulus. Moreover, the approximation enabled the Frenkel–Kontorova model to be extended to the case of two interacting structures with a slightly different lattice constant62 to calculate the width and formation energy of dislocations in the stacking of graphene layers and to consider the commensurate–incommensurate phase transition in bilayer graphene upon stretching of one of the layers.63 Using the approximation, the barrier to relative motion of the layers in bilayer graphene was estimated, on the basis of the experimentally measured values of the shear mode frequency and width of dislocations in the layer stacking, to be 1.7 meV per atom (ref. 61) and 2.4 meV per atom (ref. 7), respectively (note that all energies for bilayers in the present paper are represented in meV per atom in the top (adsorbed) layer so that they are more or less equal to the exfoliation energy of the bulk material per atom and adsorption energy of a flake on the 2D layer per atom of the flake). These values are in good agreement with those predicted by theory: ∼1 meV per atom (ref. 64, DFT calculations for graphite within local density approximation (LDA) and generalized gradient approximation (GGA)), 2.6 meV per atom (ref. 59), 1.82 meV per atom (ref. 65, LDA calculations for bilayer graphene), 1–1.5 meV per atom (ref. 66, DFT-D calculations for polycyclic aromatic hydrocarbons adsorbed on graphene), 2.6 meV per atom (ref. 59), 2.07 meV per atom (ref. 60, DFT-D calculations for bilayer graphene), 1 meV per atom (ref. 59), 1.92 meV per atom (ref. 60, calculations for bilayer graphene using the vdW-DF functional67). It should also be noted that although there is a considerable discrepancy in the experimental estimates of the barrier to relative motion of graphene layers, its relative value is even less than that for the exfoliation energy of graphite ranging from 31 ± 2 to 52 ± 5 meV per atom (ref. 68–71).
The aim of the present paper is to predict experimentally measurable physical quantities associated with the potential surface of the interaction energy of h-BN layers. We use the vdW-DF2 approach72 and the adequacy of our calculations is confirmed by the agreement in the interlayer interaction energy for symmetric stackings of h-BN layers with the values obtained recently by the local second-order Møller–Plesset perturbation theory (LMP2),55 which is a high-level ab initio method that, different from DFT, fully describes vdW interactions. Analogously to the expression for the potential surface of interaction energy between graphene layers, simple expressions describing potential energy surfaces of h-BN layers aligned in the same and opposite directions are introduced with the parameters fitted to the results of the calculations. Based on these expressions a set of physical quantities related to the interlayer interaction and relative motion of h-BN layers is estimated, including the barriers to relative motion and rotation of the layers, shear mode frequencies and shear modulus for bilayer and bulk h-BN, length and formation energy of a dislocation in stacking of the layers in bilayer h-BN. The implications of the approximation introduced for the h-BN potential energy surface for further calculations of h-BN mechanical properties and simulations of dynamical phenomena related to oscillation of h-BN layers are discussed. The results obtained are also important for understanding the interactions in recently discovered73–76 graphene-h-BN heterostructures (see also ref. 77 for a review) showing the commensurate–incommensurate phase transition.76 The calculations performed here represent a necessary intermediate stage (along with the previous calculations for bilayer graphene60,61) before theoretical consideration of the properties of such heterostructures related to the interlayer interaction.
The paper is organized in the following way. Section 2 is devoted to first-principles calculations of elastic properties of a single h-BN layer and the potential surface of interlayer interaction energy in bilayer and bulk h-BN. The estimates of physical quantities associated with the interaction and relative motion of h-BN layers are presented in Section 3. Our conclusions are summarized in Section 4.
The optimized bond length is l = 1.455 Å, which is in agreement with the experimental data for few-layer40–43 and bulk h-BN28–36 and previous calculations.39,43–53,55 The effect of the interlayer interaction on the structure of h-BN layers is neglected.55
The potential surface of the interlayer interaction energy U(x,y), i.e., the dependence of the interlayer interaction energy U on the relative displacement of two h-BN layers in the armchair (x) and zigzag directions (y) at a fixed interlayer distance d, has been calculated by shifting the layers with respect to each other as rigid. Account of deformations within the layers was shown to have a negligible effect on the potential surface of the interaction energy of the carbon nanotube walls81 and the shells of the carbon nanoparticles.82,83 Taking into account the mirror planes and three-fold rotational symmetry of a single h-BN layer, it is sufficient to explicitly calculate U(x,y) only for 1/6 part of the considered unit cell. The dependence U(x,y) has been obtained for 78 points uniformly distributed on this part of the unit cell. The interlayer distance has been set at the experimental value of d = 3.33 Å for the bulk material.28–36 The measurements for few-layer h-BN are less precise and include this value within the error bars 3.25 ± 0.10 Å (ref. 42). It should also be noted that the LMP2 calculations55 give the same interlayer distance of 3.34 Å for bilayer and bulk h-BN. Though significant efforts have been made recently to improve the description of the long-range vdW interactions in DFT,67,72,84,85 available DFT methods still fail to predict the equilibrium distance between 2D layers with sufficient accuracy.55,59,60,86 However, the magnitude of corrugations of the potential energy surface59,60,64 and the barriers to relative motion of the layers60 depend exponentially on the interlayer distance. Therefore, small deviations in the equilibrium interlayer distance predicted by different DFT methods result in significant errors in the properties related to interaction and relative motion of the layers.55,59,60,86 On the other hand, inclusion of vdW interactions almost does not affect the potential energy surface at a given interlayer distance.54,55,59,60,66 Thus, it is reasonable to study the potential energy surface at the equilibrium interlayer distance known from the experiments rather than to rely on the values provided by the DFT methods.
Let us now discuss the results of calculations of the potential energy surfaces for two orientations of h-BN layers in the simulation cell (Fig. 1). For h-BN layers aligned in the same direction (Fig. 1a and b), the potential surface of the interlayer interaction energy is very similar to that of graphene layers, which is in agreement with previous calculations.55 The energy minima correspond to the AB stacking in which nitrogen (boron) atoms of the top layer are located on top of boron (nitrogen) atoms of the bottom layer and boron (nitrogen) atoms of the top layer are on top of hexagon centers. It is clear that at this relative orientation of the layers all the configurations corresponding to the AB stacking with boron (AB2) or nitrogen (AB1) atoms of the top layer on top of the centers of the hexagons are equivalent. Shifting the layers by half of the bond length in the armchair direction (towards the nearest minimum) brings the systems to the saddle-point (SP) stacking which corresponds to the barrier to relative sliding of the layers. We find that this barrier is about 2 meV per atom (Tables 1 and 2), which is within the range reported for graphene7,59–61,64–66 and close to the LMP2 value ∼2.5 meV per atom for h-BN.55 Shifting the layers by one and a half of the bond length more in the armchair direction results in the AA stacking in which all the atoms of the top layer are located on top of the equivalent atoms of the bottom layer and the interlayer interaction energy reaches its maximum for the given interlayer distance. The energy of the AA stacking relative to the AB one is about 18 meV per atom (Tables 1 and 2), which is close to the DFT data for graphene7,59–61,64–66 and for bilayer55 and bulk49,51,55 h-BN as well as the LMP2 values of about 20 meV per atom55 for bilayer and bulk h-BN.
Fig. 1 Calculated interlayer interaction energy of h-BN bilayer U (in meV per atom of the top layer) as a function of relative displacement of the layers in the armchair (x, in Å) and zigzag (y, in Å) directions at the interlayer distance of d = 3.33 Å: (a and b) h-BN layers aligned in the same direction and (c and d) h-BN layers aligned in the opposite directions. The energy is given relative to the AA′ stacking. (b and d) Black solid lines correspond to the calculated dependences of interlayer interaction energy U on displacement x in the armchair direction (y = 0) along the thick lines indicated in figures (a and c). Curves approximated according to eqn (2) (b) and (3) (d) are shown by red dashed lines. Structures of the symmetric stackings are indicated. Boron and nitrogen atoms are coloured in blue and magenta, respectively. |
Stacking | vdW-DF2 | LMP2 |
---|---|---|
AB1′ | 3.10 | 4.42 |
AB2′ | 15.77 | 16.50 |
SP′ | 3.57 | |
AB | 0.40 | 0.24 |
SP | 2.32 | |
AA | 18.26 | 19.72 |
Stacking | vdW-DF2 | LMP2 |
---|---|---|
AB1′ | 3.02 | 3.76 |
AB2′ | 15.91 | 16.14 |
SP′ | 3.57 | |
AB | 0.38 | 0.44 |
SP | 2.39 | |
AA | 18.47 | 19.89 |
The potential energy surface for h-BN layers aligned in the opposite directions (Fig. 1c and d) is rather different from those for graphene and h-BN layers aligned in the same direction, which is in agreement with previous calculations.54,55 In this case there are two types of inequivalent AB′ stackings. In the AB1(2)′ stacking all of the boron (nitrogen) atoms of the top layer are on top of the boron (nitrogen) atoms of the bottom layer and the rest of the atoms of the top layer are on top of the hexagon centers. Also, the interlayer interaction energy reaches its maximum for the given orientation at the AB2′ stacking, while the local minima correspond to the AB1′ stacking along with the AA′ stacking in which all of the nitrogen and boron atoms of the top layer are on top of the boron and nitrogen atoms of the bottom layer, respectively. The saddle-point stacking corresponding to the barrier to relative sliding of the layers lies on the straight path between the nearest configurations corresponding to the AB1′ and AA′ but not in the middle (0.53 Å from AB1′ and 0.92 Å from AA′) since the minima are of different depth.
All the DFT methods predict that the energies of the AB, AB1′ and AA′ stackings are rather close and differ only by several meV per atom. However, the order of stability of these structures changes depending on the method and chosen interlayer distance.43,49,51,54,55,57 Our results indicate that the most stable stacking for both h-BN bulk and bilayer is AA′ (Tables 1 and 2) and these results are consistent with the experimental observations for bulk h-BN28 as well as with the LMP2 calculations55 and DFT calculations with account of nonlocal many-body dispersion (MBD)57 for bilayer and bulk h-BN. The AB stacking is higher in energy only by 0.4 meV per atom and the relative energy of the AB1′ stacking is 3 meV per atom, which is in agreement with the LMP2 results55 (Tables 1 and 2) and the experimental data42 that both the AA′ and AB stackings are observed for bilayer h-BN. The AB1′ minima in our case are very shallow and are close in energy to the transition state SP′ for relative sliding of the layers. The corresponding barrier for transition from the AA′ to AB1′ stacking is 3.6 meV per atom, somewhat higher than in the case of h-BN layers aligned in the same direction.
It should also be mentioned that switching the vdW contribution off (using only LDA correlation and rPW86 exchange72) does not change the order of the symmetric stackings (Table 3). Their relative energies with respect to the AA′ stacking are changed by 0.4–0.8 meV per atom, which is within 5% of the magnitudes of corrugation of the potential energy surfaces both for h-BN layers aligned in the same and opposite directions. An analogous conclusion has previously been made for graphene bilayer59,60,66 and h-BN bilayer54,55,57 for a set of functionals with the dispersion correction.
Stacking | vdW-DF2 | rPW86(x) + LDA(c) | Difference |
---|---|---|---|
AB1′ | 3.10 | 3.87 | −0.76 |
AB2′ | 15.77 | 15.30 | 0.48 |
SP′ | 3.57 | 4.08 | 0.51 |
AB | 0.40 | 0.85 | −0.45 |
SP | 2.32 | 2.68 | 0.36 |
AA | 18.26 | 17.62 | 0.64 |
Another observation is that the calculated relative energies of different stackings for bilayer and bulk h-BN are nearly identical (Tables 1 and 2), which is in agreement with previous calculations for graphene.60,61 The relative deviation is within 0.2 meV per atom, i.e. about 1% of the magnitudes of corrugation of the potential energy surfaces for h-BN layers aligned in the same and opposite directions, and can be explained by interaction of non-adjacent layers in the bulk material.
U = 0.5(U0 + U1(2cos(k1x)cos(k2y) + cos(2k1x) + 1.5)), | (1) |
UI = U0,tot + 4.5U1NB + (U1NN + U1BB − U1NB)×(2cos(k1x)cos(k2y) + cos(2k1x) + 1.5), | (2) |
For h-BN layers aligned in the opposite directions,
(3) |
The parameters of the approximation are found from the relative energies of the symmetric stackings AA′, AB1′, AB2′ and AB as
(4) |
The values obtained are U1NB = −0.0888 meV per atom, U1BB = 3.328 meV per atom and U1NN = 0.512 meV per atom and allow straightforward physical interpretation. The negative value of U1NB corresponds to the attraction of ions of different sign, while positive values of U1BB and U1NN are related to repulsion of ions of the same sign. The greater value of U1BB compared to U1NN is associated with the larger size of the boron ion as compared to the nitrogen one.87
In addition to corrugations of the potential energy surface for h-BN layers aligned in the opposite directions and relative energy of the co-aligned configuration, which are reproduced exactly, the approximation is very accurate in relative energies of the SP and AA stackings for h-BN layers aligned in the same direction (Fig. 1b and d). The magnitude of corrugation for this orientation of the layers according to the approximation is EAA − EAB = 4.5(U1NN + U1BB − U1NB) = 17.68 meV per atom, which is only 0.18 meV per atom or 1% smaller than the DFT value obtained. The barrier to relative sliding of the layers aligned in the same directions according to the approximation is ESP − EAB = 0.5(U1NN + U1BB − U1NB) = 1.96 meV per atom. This is only 0.05 meV per atom or 2.6% greater then the DFT value. The standard deviation of the approximated potential energy surfaces from the ones obtained by the DFT calculations for h-BN layers aligned in the same and opposite directions are 0.056 meV per atom and 0.014 meV per atom, respectively, which is within 0.3% of their magnitudes of corrugation.
We should note that the possibility to accurately approximate the potential energy surfaces by expressions containing only the first Fourier harmonics has been previously demonstrated for interaction between graphene layers60,61,66,88 and between carbon nanotube walls, both for infinite commensurate walls81,89–92 and in the case where corrugations of the potential surface are determined by the contribution of edges93 or defects.81 Thus, we can expect that analogous expressions can describe potential energy surfaces for other layered materials with the van der Waals interaction between layers or for translational motion of large molecules physically adsorbed on crystal surfaces.
It should also be pointed out that the expression given by eqn (2) for approximation of the potential energy surface for h-BN layers aligned in the same direction is exactly the same as the one for graphene.9,10,59,61 The calculated barrier to relative sliding of h-BN layers aligned in the same direction and the magnitude of corrugation of the potential energy surface (Tables 1 and 2) are within the ranges reported for graphene.7,59–61,64–66 Therefore, it can be expected that these materials have very similar physical properties. This is confirmed in Section 3 by comparison of physical quantities estimated for h-BN on the basis of eqn (2) with the experimental data for graphene.
(5) |
Fig. 2 Calculated shear mode frequencies f (in cm−1) as functions of the number of layers n for h-BN crystals with different stacking of the layers: AA′ — blue solid line, AB1′ — green dashed line, AB — red dash-dotted line. The experimental and theoretical data for bulk h-BN are included: □ — ref. 37 (experiment), • — ref. 38 (experiment), × — ref. 46 (LDA), ◊ — ref. 52 (LDA), ▲ — ref. 53 (LDA), ▼ — ref. 53 (revPBE). The experimental data for few-layer graphene and graphite are also shown for comparison: ■ — ref. 97 and ♦ — ref. 98. |
The curvature of the potential energy surface also determines the shear modulus, which is expected to be weakly dependent on the number of h-BN layers and can be estimated as:
(6) |
Fig. 3 Calculated shear moduli C44 (in GPa) for h-BN crystals (few-layer or bulk) with different stacking of the layers. The experimental data for bulk h-BN are taken from ref. 35 (Exp1) and ref. 99 (Exp2). The range of values from the previous DFT calculations53 for bulk h-BN is indicated. |
Urot = 〈UI〉x,y = 〈UII〉x,y = U0,tot + 1.5(U1NN + U1BB + 2U1NB) | (7) |
Thus, the barriers to relative rotation of h-BN layers from metastable states corresponding to the AA′, AB1′ and AB stackings are Erot(AA′) = Urot − UII(AA′) = 1.5(U1NN + U1BB − 4U1NB) = 6.29 meV per atom, Erot(AB1′) = −1.5U1NN + U1BB + 3U1NB = 2.29 meV per atom and Erot(AB) = 1.5(U1NN + U1BB) − U1NB = 5.85 meV per atom, respectively. The values for the most stable AA′ and AB stackings are of the order of the data for graphene Erot ∼ 5 meV per atom (ref. 61) and 4 meV per atom (ref. 9 and 10). The obtained expressions for the potential energy surface (eqn (2) and (3)) can be also used to estimate the barriers to relative rotation of h-BN layers in configurations corresponding to Moiré patterns analogously to studies performed for graphene.100
To consider dislocations in bilayer h-BN we follow the formalism of the two-chain Frenkel–Kontorova model62 which was applied previously to study dislocations in double-walled carbon nanotubes62,91 and graphene.63 Elementary dislocations in graphene correspond to partial dislocations with the Burgers vector equal in magnitude to the bond length.7,63 Similar dislocations (b = l, Fig. 4c and d) are also possible in h-BN layers aligned in the same direction as the potential surface of interlayer interaction in this case has degenerate energy minima AB1 and AB2 separated by a distance of one bond length l (Fig. 1a). However, for h-BN layers aligned in the opposite directions, the AB1′ energy minima are rather shallow and are much higher in energy than the AA′ minima (Fig. 1c). Therefore, in this case the minimal possible Burgers vector is equal in magnitude to the lattice constant ( Fig. 4a and b), i.e. the elementary dislocations are full. Both full dislocations in h-BN layers aligned in the opposite directions and partial dislocations in h-BN layers aligned in the same direction are considered below. It should also be noted that even for graphene theoretical estimates of the formation energy and width have been limited to the tensile dislocation with the Burger vector normal to the boundary between commensurate domains,63 although experimentally dislocations with different angles between the Burgers vector and boundary between commensurate domains have been observed.7,19–22,24 Here we extend the Frenkel–Kontorova model to describe dislocations with the Burgers vector at an arbitrary angle β to the normal to the boundary between commensurate domains. In particular, the results for shear dislocations (Fig. 4b and d), where the Burgers vector is parallel to the boundary between commensurate domains β = ±π/2, are presented along with the results for tensile dislocations (Fig. 4a and c) with β = 0.
In the limit of large systems and low density of dislocations it can be assumed that dislocations are isolated. This means that the layers are commensurate at large distances from the dislocation center. Correspondingly, the boundary conditions for the relative displacement of atoms of the layers can be formulated as = 0 at x → −∞ and = − at x → +∞, where x is the coordinate along the normal to the boundary between commensurate domains and is the Burgers vector. According to the Frenkel–Kontorova model,62,63 the formation energy of isolated dislocations per unit length is determined by the sum of excessive elastic and van der Waals interaction energies originating from the deformation and incommensurability of the layers, respectively,
(8) |
The dislocation path, i.e. the dependence of the relative displacement of h-BN layers on the coordinate x in the direction perpendicular to the boundary between commensurate domains (Fig. 4e–g) that minimizes the formation energy UD (eqn (8)), should satisfy the Euler–Lagrange equations:
(9) |
This means that the dislocation path is the same as the trajectory of a particle on the inverse potential energy surface −V() with the fixed initial and final positions corresponding to the deepest energy minima of the original potential energy surface. It should be noted that the “particle mass” is anisotropic and equal to G/2 and E/2 in the directions along and across the boundary between commensurate domains, respectively.
For partial dislocations in h-BN layers aligned in the same direction (Fig. 4c and d), the straight line between the adjacent AB minima (Fig. 1a) corresponding to the minimum energy path satisfies eqn (9) and thus gives exactly the dislocation path. For full dislocations in h-BN layers aligned in the opposite directions (Fig. 4a and b), the solution of eqn (9) cannot be found analytically. Moreover, this solution should depend on the angle between the Burgers vector and the boundary between commensurate domains. However, it is clear that strong scattering introduced by rapidly changing potential at slopes of the AB2′ hill (Fig. 1c) should be avoided. Therefore, we assume that the path of full dislocations in h-BN layers aligned in the opposite directions is also roughly described by the minimum energy path AA′–AB1′–AA′ consisting of two straight lines at the angle 2π/3 to each other. More accurate approximations of the dislocation path give a correction to the formation energy within 10%.
Integration of the Euler–Lagrange equations (eqn (9)) provides the analogue of the energy conservation law:
(10) |
(11) |
Let us first consider partial dislocations in h-BN layers aligned in the same direction (Fig. 4c and d). In these dislocations, atoms are displaced along straight lines, θ = π + β and u changes from 0 to b = l. According to eqn (11), the formation energy of partial dislocations is:
(12) |
Based on the approximation of the interlayer interaction energy along the minimum energy path by the cosine function and eqn (10), it was shown analytically that partial dislocations are described by a soliton with a relatively short incommensurate region separating commensurate domains.63 Although in the present paper we use eqn (2) to approximate the potential energy surface of h-BN layers aligned in the same direction, the calculated path of partial dislocations is very close to the analytical solution (Fig. 4g). Since the slope of the dependence of displacement u on position x across the boundary is nearly constant around the dislocation center and close to the maximum value |′|max, the dislocation width can be defined in a similar way to the analytical solution:
(13) |
Fig. 6 Calculated dislocation width lD (in nm) as a function of angle β (in degrees) between the Burgers vector and normal to the boundary between commensurate domains for a full dislocation (solid black line) in h-BN layers aligned in the opposite directions and a partial dislocation (dashed red line) in h-BN layers aligned in the same direction. The experimental data for graphene from ref. 7 are shown by squares (■) with error bars. The DFT result63 for partial tensile dislocations in graphene is indicated by a circle (•). |
In the case of full dislocations in h-BN layers aligned in the opposite directions (Fig. 4a and b), two contributions from the straight pieces of the dislocation path at angles β + π/6 and β − π/6 to the normal to the boundary between commensurate domains need to be taken into account. This can be done by a simple substitution of ϕ(β) by ϕ(β − π/6) + ϕ(β + π/6) in eqn (12) and (13). All the conclusions made above for partial dislocations are also valid for full ones (Fig. 5 and 6) with the only difference being that full tensile and shear dislocations correspond to the boundaries between commensurate domains in the armchair and zigzag directions, respectively (Fig. 4a and b), which is opposite to the case of partial dislocations. It should be especially emphasized that according to Fig. 4e–g, both for partial and full dislocations the slope |′(x)| is nearly constant for distances from the dislocation center comparable to the dislocation width. This gives the possibility of accurate measurements of the dislocation width and, consequently, of experimental estimation of the barrier to relative sliding of h-BN layers by transmission electron microscopy analysis of stacking at the boundary between commensurate domains in the same way as has been done for graphene.7
It is shown that the calculated potential surfaces of interlayer interaction energy for h-BN layers aligned in the same and opposite directions can be fitted by simple expressions containing only the first components of Fourier expansions determined by symmetry of the layers. Analogous approximations have been suggested previously for graphene bilayer60,61,66,88 and double-walled carbon nanotubes81,89–93 based both on DFT calculations and semi-empirical potentials. Thus, it can be expected that for other layered materials the potential surface of interlayer interaction energy can also be reproduced closely by the first Fourier components determined by symmetry.
Recently, a concept of the registry index surface was proposed to predict the qualitative features of the potential surface of interlayer interaction energy of h-BN bilayer.54 The expressions introduced here give the possibility to reproduce the shape and quantitative characteristics of the potential energy surface of h-BN bilayer and can be useful for multiscale simulations of such phenomena as atomic-scale slip-stick motion of an h-BN flake attached to STM tip on the h-BN surface, diffusion of an h-BN flake on the h-BN surface and formation of stacking dislocations in h-BN bilayer (analogous to multiscale simulations of the phenomena observed for graphene4–10,18–26).
According to the proposed approximation, a set of physical properties of h-BN materials related to relative motion of the layers are determined by only one or two independent parameters for the layers aligned in the same and opposite directions, respectively. Namely, the shear mode frequency, shear modulus and barrier to relative rotation of the layers have been estimated for bilayer and bulk h-BN in different stackings. The possibility of partial and full dislocations in stacking of the layers is suggested for h-BN layers aligned in the same and opposite directions, respectively. Their width and formation energy are governed by the same parameters of the potential energy surface. In particular, it is shown that a simple link exists between the dislocation width lD and barrier Vmax to relative sliding of the layers Vmax = C(β)/lD2(β), where the coefficient C(β) = ϕ2(β)kl2/4 is determined by the elastic properties of the layers. Experimental measurements of the dislocation width, shear mode frequency or shear modulus can help to refine the parameters of the functions approximating the potential surfaces of interlayer interaction energy in h-BN and to obtain the barriers and other characteristics of these surfaces in the same way as has been done for graphene.7,61
This journal is © The Royal Society of Chemistry 2016 |