Computational and Theoretical Chemistry 1053 (2015) 238–244

Contents lists available at ScienceDirect

Computational and Theoretical Chemistry

journal homepage: www.elsevier.com/locate/comptc

Chemical bonding and surface interactions in Bi2Se3 and Bi4Se3

Matthew S. Christian a, Sarah R. Whittleton a, Alberto Otero-de-la-Roza b, Erin R. Johnson a,⇑
a
Chemistry and Chemical Biology, School of Natural Sciences, University of California, Merced, 5200 North Lake Road, Merced, CA 95343, USA
b
National Institute for Nanotechnology, National Research Council of Canada, 11421 Saskatchewan Drive, Edmonton, Alberta T6G 2M9, Canada

a r t i c l e i n f o a b s t r a c t

Article history: Topological insulators are a new class of materials that are insulators in the bulk but have near zero elec-
Received 8 August 2014 tron transport dissipation behavior on the surface. These surface states are topologically robust (resistant
Received in revised form 17 September 2014 to impurities, defects, and geometry deformations), which makes these materials ideal candidates for a
Accepted 18 September 2014
number of technological applications. Well-known three-dimensional topological insulators are bismuth
Available online 30 September 2014
selenide (Bi2Se3), which is composed of five-atom-thick layers that interact non-covalently with each
other, and Bi–Se alloys resulting from combining these quintuple layers with elemental bismuth bilayers.
Keywords:
In this article, we examine the surface sliding and binding energetics of the combinations of quintuple
Chemical bonding
Density-functional theory
layers and bisumth bilayers found in Bi2Se3 and Bi4Se3. In addition, we investigate the nature of the
Topological insulators chemical bonding in these systems and its relation to the surface energetics.
Van der Waals Ó 2014 Elsevier B.V. All rights reserved.
Tribology

1. Introduction electronic structure, the signature of a QSH state is the crossing

Topological insulators (TIs) [1–7] are a new class of materials
characterized by metallic surface states that are ‘‘topologically pro-
tected,’’ that is, resistant to impurities and surface defects, and by
being an insulator in the bulk. The mechanism behind the TI prop-
erties is similar to the quantum Hall (QH) effect. In the QH state,
electrons in a bidimensional material subject to a strong magnetic
field perpendicular to the surface are forced to move in quantized
orbits, making the material an insulator in the bulk, but conducting
on the surface. In these metallic surface states, electrons are forced
to move exclusively in one direction, thus preventing backscatter-
ing at defects and achieving an almost dissipationless electron
flow. Topological insulators show a closely-related behavior, called
the quantum spin Hall (QSH) effect. In the QSH state, as in the QH
state, dispersionless metallic surface states are present, but a
strong external magnetic field is not required, thus making these
materials much more interesting from a practical point of view.
The QSH state arises from strong spin–orbit interactions present
in TIs. In a TI, the electron spin is locked perpendicular to the crys-
tal momentum, and electrons of different spins are forced to move
exclusively in opposite directions on the surface [5]. Backscatter-
ing, which would involve a spin-flip, is forbidden due to the pres-
ence of time-reversal symmetry. The QSH state is resistant to
impurities, surface defects and disorder [5]. In terms of the
⇑ Corresponding author.
E-mail address: ejohnson29@ucmerced.edu (E.R. Johnson).
of bands corresponding to surface states, forming what is called a
Dirac cone [5]. In addition to being theoretically interesting [3],
TIs have great potential for use in future technological applications
[4,6] such as, for instance, spintronics [8] and thermoelectrics
[9,10].
The strong spin–orbit interaction required for the TI state limits
the availability of such materials. Heavy elements, where relativis-
tic effects are important, are usually involved to generate strong
spin–orbit coupling. Bi2Se3 is the most-studied three-dimensional
topological insulator [3,11,12], presenting a Dirac cone composed
of surface states and a relatively large band gap in the bulk. Its
structure, shown in Fig. 1, consists of ABCABC stacks of five-atom
layers, called quintuple layers (QL). The QLs interact with each
other primarily via van der Waals forces [13]. Each Bi atom in a
QL forms close contacts with six Se atoms, in an octahedral
arrangement. The Bi4Se3 material is formed by combining QLs
and elemental bismuth bilayers (Bi2) in a 1:1 ratio (Fig. 1), and also
presents QSH states and topological insulator behavior [14,15]. The
bismuth bilayer is a known 2D topological insulator [16]. Because
of the relatively weak inter-layer binding, the QL and Bi2 can be
stacked to form Bi–Se alloys with variable Bi and Se concentration,
resulting in extremely long cell lengths in the stacking direction
and possibly incommensurate phases [13]. Material properties
could be controlled by varying the stoichiometry, though more
information about the surface energetics and chemical behavior
is needed; [16] this will be addressed in the present work via
density-functional theory (DFT).

http://dx.doi.org/10.1016/j.comptc.2014.09.023
2210-271X/Ó 2014 Elsevier B.V. All rights reserved.
 M.S. Christian et al. / Computational and Theoretical Chemistry 1053 (2015) 238–244
Several experimental and theoretical studies on the surface
behavior of Bi–Se materials have been conducted in recent years.
It is known that the Bi2Se3 surface is relatively inert [17] and resis-
tant to oxidation [18,19]. As expected, Bi2Se3 cleaves preferentially
at the van der Waals inter-QL contacts. It is known experimentally
that the geometry and electronic structure of the surface are sensi-
tive to cleavage and show ageing effects, with the Bi concentration
increasing over time, although the Se termination can be recovered
after annealing [8,20]. He et al., on the other hand, report a prefer-
ence for a surface Bi termination in Bi2Se3 at room temperature
that may be attributed to a bismuth bilayer [21], but the Se termi-
nation is preferred at low temperatures. The position of the Dirac
cone relative to the Fermi level, which is important in the charac-
terization of the TI behavior, is also affected by charge transfer
from surface adatoms [20,22].
The nature of the inter-layer interaction was previously
explored by Lind et al. [13] using local and semi-local density
functionals. However, a functional that includes dispersion attrac-
tion is required to accurately describe the inter-layer van der
Waals interactions. Liu et al. [23] and Luo et al. [24] used non-
local van der Waals density functionals to study the band struc-
ture, equilibrium geometry, and exfoliation of Bi2Se3, although lit-
tle attention has been paid to the surface energetics and the
nature of the inter-layer chemical bonding. In this work, we apply
the exchange-hole dipole (XDM) dispersion model [25], which
has been shown to give accurate exfoliation energies for graphite
and sublimation energies for molecular crystals [26,27], to study
the binding energies and sliding behavior of the QL and Bi2 sur-
faces. Furthermore, the nature of the chemical bonding in Bi2Se3
and Bi4Se3 is relatively unexplored. We present an analysis of
the chemical bonding properties in these materials using a range
of techniques, including the Quantum Theory of Atoms in Mole-
cules (QTAIM) [28–30], Non-Covalent Interactions (NCI) [31,32],
and the Electron Localization Function (ELF) [33,34], to study
the atomic charges and inter-layer interactions in these Bi–Se
materials.
Fig. 1. Structures of bulk Bi2Se3 (left) and Bi4Se3 (right), showing the quintuple
layer (QL) in the former and the QL and intercalated Bi2 layers in the latter. Bi atoms
are represented in violet and Se atoms in yellow. (For interpretation of the
references to color in this figure legend, the reader is referred to the web version of
this article.)
239
2. Computational methods
Common density functionals have several well-known short-
comings [35], of which the lack of dispersion interactions and the
systematic underestimation of the band gap are particularly rele-
vant to the modeling of topological insulators. In recent years,
great progress has been made towards the efficient and accurate
modeling of non-covalent interactions by using dispersion-energy
functionals [36,37]. These methods capture long-range correlation
effects and are coupled to a common density functional (called the
base functional) that, in principle, accounts for the non-covalent
energy terms other than dispersion.
The structures of the topological insulators considered in this
article (Bi2Se3 and Bi4Se3) consist of stacked QLs and bismuth
bilayers that interact with each other via non-covalent interac-
tions (see Fig. 1). In this work, we account for the dispersion
energy by using the exchange-hole dipole moment (XDM) model,
proposed by Becke and Johnson [25,26,38–47]. In XDM, the total
energy from the base functional is supplemented by the
dispersion correction,
E ¼ Ebase þ EXDM ; ð1Þ
that is calculated using an atom-based pairwise expression:
X X C n;ij f ðRij Þ
n
EXDM ¼  : ð2Þ
n¼6;8;10 i<j
Rnij
The sum runs over all pairs of atoms in the system, Rij is the dis-
tance between atoms i and j; f n is a damping function that deacti-
vates the dispersion correction at short range, and the C n;ij are the
dispersion coefficients for each pairwise interaction. The latter are
approximated non-empirically by the electrostatic energy of the
electron distributions and their exchange holes in the two interact-
ing atoms. It has been shown that the XDM model offers excellent
accuracy for gas-phase [46], as well as condensed-matter, systems
[26,27].
In this article, we use the XDM dispersion functional to model
the inter-layer interactions in the Bi–Se materials. Our calculations
were performed under periodic boundary conditions using the
pseudopotentials-plane waves approach as implemented in the
Quantum ESPRESSO program [48]. We used B86b [49] exchange
coupled with the PBE correlation [50] functional. Both Bi2Se3 and

Bi4Se3 crystallize in the rhombohedral (R3m) space group and we
have used the rhombohedral (primitive) and hexagonal (conven-
tional) cells where convenient. The primitive-cell calculations were
carried out using a 4  4  4 k-point grid while the grid for the
hexagonal cell was 4  4  2. We used the Projector-Augmented
Wave method [51] (PAW) and scalar-relativistic pseudopotentials.
The cutoff energies were 60 Ry for the plane-waves and 600 Ry for
the density.
Regarding the surface energetics, we optimized the bulk geom-
etries and found the equilibrium structures of Bi2Se3 and Bi4Se3. In
Bi2Se3, the most stable cleavage occurs between QLs because of
their weak non-covalent binding. The Bi4Se3 crystal presents alter-
nating QL and bismuth bilayers, so cleavage disrupts van der Waals
contacts between a QL and a Bi2 bilayer.
It is also interesting to examine the non-covalent interactions
and energy landscape for sliding, which determines the tribological
behavior of these materials. In Bi2Se3, we considered two QLs slid-
ing on top of each other, and for Bi4Se3 we examine a bismuth
bilayer sliding over a QL. These calculations were performed using
a supercell with a length of 75 bohr atomic units perpendicular to
the slabs. The sliding energetics were computed by displacing the
top layer and fixing the x and y atomic coordinates while relaxing
the vertical coordinate.
240 M.S. Christian et al. / Computational and Theoretical Chemistry 1053 (2015) 238–244

techniques. QTAIM atomic charges were calculated using the Yu–

Trinkle algorithm [52] implemented in the CRITIC2 program
[53,54]. In addition, the electron localization function (ELF)
[33,34] was used to determine the relative importance of the cova-
lent, inter-layer charge transfer, and ionic contributions to the
bonding by mapping the position of the localized electron pairs.
Calculations using norm-conserving pseudopotentials and a larger
cutoff energy were used to compute the ELF.
Finally, we used the non-covalent interaction (NCI) index
[31,32] to map the intra- and inter-layer interactions. The NCI
index is a measure of non-covalent contacts, similar to the topol-
ogy analysis of the electron density in QTAIM. It is based on two
quantities: the electron density and the reduced density gradient
(s, RDG), defined as

1 jrqj
s¼ : ð3Þ
2ð3p2 Þ
1=3 q4=3
NCI plots comprise low reduced-gradient isosurfaces, for a certain
small isovalue of s. The coloring is determined by the electron den-
sity times the sign of the second Hessian eigenvalue [31]. Strong
Fig. 2. Inter-layer and intra-layer distances (in angstrom) in Bi2Se3 and Bi4Se3 at the non-covalent interactions appear as localized blue domains
bulk geometry. The atomic Bader charges for the atoms in each layer are shown in whereas weak molecular contacts are represented by extended
red. The distance between QLs in Bi2Se3 and between QL and Bi2 in Bi4Se3 are
green regions. The NCI plots were generated using CRITIC2 [53,54].
indicated by dashed arrows. (For interpretation of the references to color in this
figure legend, the reader is referred to the web version of this article.)
3. Results and discussion
Table 1
Inter-layer binding energies for different interactions between the QL and the 3.1. Structure and chemical bonding in bulk Bi2Se3 and Bi4Se3
bismuth bilayer (in kcal/mol). The base functional (‘‘Base’’) and dispersion (‘‘Disp’’)
contribution to the binding energy are also given. The inter-layer distances and Bader atomic charges for bulk
Layers Total Base Disp Bi2Se3 and Bi4Se3 are shown in Fig. 2. The inter-QL distance for
Bi2Se3 (2.52 Å) is in close agreement with experimental results
(QL)–(QL) 6.33 0.52 6.86
(QL)(QL)–(QL) 6.36 0.48 6.85 (2.58 Å [55], 2.25 Å [56]) and previous DFT results: C09-vdW [24]
(QL)–(QL)–(QL) 12.70 1.01 13.71 (2.58 Å), PBE-D2 with spin–orbit coupling [24] (2.57 Å), and
(QL)–(Bi2) 8.60 0.07 8.67 optPBE-vdW [23] (2.41 Å). Non-dispersion-corrected functionals
(Bi2)(QL)–(Bi2) 8.68 0.26 8.94 such as PBE with spin–orbit coupling give very long inter-layer
(Bi2)–(QL)–(Bi2) 17.28 0.33 17.61 separations [24] (3.30 Å), as does the non-local van der Waals
(QL)(Bi2)–(QL) 8.99 0.15 9.14 functional coupled with revPBE [24] (3.72 Å).
(QL)–(Bi2)–(QL) 17.60 0.22 17.81
The inter-layer distances between two QLs and between one QL
and the bismuth bilayer are significantly longer than the intra-
In order to examine the nature of the chemical bonds in these layer distances, as expected because of the non-covalent character
bismuth selenide systems, we employed a number of different of the bonding. However, the intra-QL distances differ depending

Fig. 3. Non-covalent interaction (NCI) plots within one QL (top left), the bismuth bilayer (top right), and the between one QL and Bi2 in Bi4Se3 (bottom left), and the inter-
layer interaction in Bi2Se3 (bottom right). The isovalue is s ¼ 0:3. Bi atoms are represented in violet and Se atoms in yellow. Strong non-covalent interactions are represented
in blue and weak interactions are in green. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
 M.S. Christian et al. / Computational and Theoretical Chemistry 1053 (2015) 238–244 241

Fig. 4. Contour plots of the electron localization function (ELF) in the plane bisecting two QLs in Bi2Se3 (left) and two QLs and the intercalated bismuth bilayer in Bi4Se3
(right). The green dotted lines correspond to the paths used in Fig. 5. (For interpretation of the references to color in this figure legend, the reader is referred to the web
version of this article.)

on whether the QL is adjacent to a Bi2 bilayer or to another QL. The

1
QL intra-layer distances in Bi2Se3 (1.58 and 1.92 Å) are closer to the
0.9
equivalent distances in an isolated QL (1.59 and 1.93 Å) than to
Electron localization function
0.8
those in Bi4Se3 (1.54 and 1.83 Å), indicating that the donor– 0.7
acceptor interactions between two QLs are much smaller than 0.6
between a bismuth bilayer and a QL. Likewise, the Bi2 intra-layer 0.5
distance adjacent to a QL in Bi4Se3 (1.76 Å) is shorter than its value 0.4
in vacuum (1.79 Å). In Table 1, discussed later, this inter-layer 0.3
donor–acceptor interaction results in a higher binding energy 0.2
between a QL and Bi2 than between QLs themselves. 0.1
0
These geometric differences can be explained readily by inter-
layer charge-transfer effects. The Se atoms at the center or edges Se Bi Se Bi Se Se
of a QL have different charges, with the center seleniums being
Se Bi Se Bi Se Bi Bi
Bi2Se3
more ionic than the edge seleniums because they are coordinated Bi4Se3
to a higher number of electropositive Bi atoms, which essentially 0 2 4 6 8 10 12 14 16 18 20
behave as +1 cations. The bismuth bilayer, however, is formally Interatomic distance (Å)
composed of elemental Bi and our results indicate that there is a
charge transfer of 0.13 electrons from the Bi2 to each of the adja- Fig. 5. Electron localization function (ELF) along the paths connecting bonded
cent QLs. This charge transfer increases the ionicity of the QL and atoms in successive atomic layers (green paths in Fig. 4). The intra-QL ELF profiles
are represented first, while the inter-layer regions and the intra-Bi2 region are
decreases the intra-layer distances, while at the same time weak-
shown at the end of the plot. The profiles for Bi2Se3 (red) and Bi4Se3 (blue) are
ening the intra-layer bonding of Bi2. These observations point to superimposed. The atomic positions are shown below the plot (red for atoms in
the ionic nature of the binding in each QL and the metallicity of Bi2Se3 and blue for Bi4Se3). (For interpretation of the references to color in this
the bismuth bilayer. figure legend, the reader is referred to the web version of this article.)
The atomic charges in Fig. 2 can be compared to the QL charges
in the vacuum: 1.03 for Bi, 0.84 (Se, center), and 0.61 (Se, edge).
The charge of Bi2 in vacuum is naturally zero. These results indicate the surface of the QL may be related to the metallic conductance.
that charge transfer from Bi2 to the QL increases its ionicity, short- The charge transfer from the Bi2 bilayer to the QL in Bi4Se3 induces
ens the intra-layer distances, and improves the intra-layer binding. changes in the bonding restricted to the edge Se-Bi bond. As seen in
The strength and spatial extent of the inter- and intra-layer Fig. 5, the ELF value at the bond midpoint decreases by about 0.1,
interactions can be visualized using the NCI plots shown in indicating an increase in ionicity.
Fig. 3. Again, the intra-layer interactions are significantly stronger, The inter-layer ELF is consistently smaller than 0.5, but is higher
leading to small, localized high-density (blue) NCI domains, for the QL–Bi2, than the QL–QL, non-covalent interaction, consis-
whereas the inter-layer contacts are weaker (green). The NCI plots tent with the inter-layer charge transfer behavior in the former.
also indicate that the QL–Bi2 interaction is stronger than the QL–QL The Bi–Bi ELF along the bond path is higher than the equivalent
interaction. However, in both cases the inter-layer interactions are Bi–Se intra-layer values in the QL, indicating relatively-strong Bi–
resolved into localized atom–atom contacts, instead of extended Bi bonding.
contact regions as are common in molecular crystals [32].
The ELF maps the degree of electron localization in real space. 3.2. Binding energies
Fig. 4 shows a colormap of the ELF values in bulk Bi2Se3 and Bi4Se3,
and Fig. 5 represents the same quantity along a path connecting Table 1 shows the inter-layer binding energies for different
bonded atoms in successive atomic layers. The figures show that, combinations of QLs and Bi2 bilayers. The interaction between
within a QL, bonds between the edge Se and Bi are more covalent two QLs is weaker than that between a QL and Bi2 by about
than between the central Se and Bi, which according to the charges 2.3 kcal/mol, as predicted by the NCI plots and the atomic charges
in Fig. 2 are more ionic. The increased electron delocalization on in the previous section.
242 M.S. Christian et al. / Computational and Theoretical Chemistry 1053 (2015) 238–244

The ability to bind to a single QL is mostly unaffected by adja- 3.3. Sliding

cent layers. The change in the QL–QL binding energy caused by a
third QL is only 0.03 kcal/mol. It is known experimentally that
the intra-layer phonon frequencies change only slightly (a few
cm1) with increasing QL thickness in thin films [57], which is con-
sistent with our results. Our observations also rule out any kind of
inter-layer cooperative effects in multi-QL thin films, suggesting
that the onset of the TI state with thickness [58] comes from the
physical separation of the surfaces on both sides of the slab.
Contrary to what happens between QLs, the binding energy of
QL–Bi2 is affected if the bismuth bilayer is in contact with another
QL. This is not surprising given the charge transfer between these
layers, seen in Figs. 2 and 5. A QL adjacent to the Bi2 increases
the binding energy of the QL on the opposite side of the bismuth
bilayer by 0.39 kcal/mol. These inter-layer cooperative effects are
not expected to propagate in the Bi4Se3 bulk. In all cases studied,
the inter-layer binding comes almost exclusively from the disper-
sion energy contribution, consistent with the dominant van der
Waals interaction between layers.
Another way of examining the inter-layer interactions in Bi2Se3
and Bi4Se3 is to study the effect of displacing combinations of the
QL and Bi2 layers with respect to each other. In addition to provid-
ing further insight regarding the bonding, study of their tribologi-
cal behavior has relevance to the processing and manufacturing of
these materials. The potential energy surfaces (PES) for sliding of
either a QL or Bi2 bilayer over a QL are shown in Fig. 6, together
with the dispersion and base-functional contributions and the
inter-layer distances.
The sliding energetics of the QL–Bi2 model are different from
the QL–QL PES. The inter-layer contact between two QLs is through
negatively-charged Se atoms and, as a consequence, the configura-
tion where two Se atoms are on top of each other is strongly disfa-
vored, having a binding energy of only 0.5 kcal/mol. The Se atoms
at the edge of the QL prefer to be positioned over a void in the sur-
face below and voids with an underlying Bi cation are preferred to
the empty voids by about 1 kcal/mol.

4.0 3.0 4.0 4.5

BE per slab (kcal/mol)

BE per slab (kcal/mol)

Se Se Se Se
4.0
3.0 2.5 3.0
3.5
2.0
y (Å)
y (Å)

2.0 2.0 3.0

1.5 2.5
1.0 Bi 1.0 Bi
2.0
0.0
(a) Se Se
1.0
0.0 Se Se 1.5
0.5 1.0
−3.0 −2.0 −1.0 0.0 1.0 2.0 3.0 4.0 −3.0 −2.0 −1.0 0.0 1.0 2.0 3.0 4.0
x (Å) x (Å)
BE per slab (kcal/mol, XDM)

BE per slab (kcal/mol, XDM)

4.0 3.5 4.0 4.5
Se Se Se Se 4.0
3.0 3.0 3.0 3.5
2.5 3.0
y (Å)
y (Å)

2.0 2.0
2.5
Bi 2.0 Bi 2.0
1.0 1.0
1.5
0.0
(b) Se Se
1.5
0.0 Se Se 1.0
1.0 0.5
−3.0 −2.0 −1.0 0.0 1.0 2.0 3.0 4.0 −3.0 −2.0 −1.0 0.0 1.0 2.0 3.0 4.0
x (Å) x (Å)
BE per slab (kcal/mol, bare)

BE per slab (kcal/mol, bare)

4.0 −0.2 4.0 0.20

Se Se −0.3 Se Se
3.0 −0.4 3.0 0.10
−0.5
y (Å)
y (Å)

2.0 2.0 0.00

−0.6
Bi −0.7 Bi −0.10
1.0 1.0
−0.8
0.0
(c) Se Se −0.9 0.0 Se Se
−0.20

−1 −0.30
−3.0 −2.0 −1.0 0.0 1.0 2.0 3.0 4.0 −3.0 −2.0 −1.0 0.0 1.0 2.0 3.0 4.0
x (Å) x (Å)
4.0 3.6 4.0 3.6
Se Se Se Se
3.4 3.4
3.0 3.0
3.2
z (Å)
y (Å)

3.2
y (Å)

2.0
z (Å)

2.0
3.0
Bi 3.0 Bi
1.0 1.0 2.8

0.0
(d) Se Se
2.8
0.0 Se Se
2.6
2.6 2.4
−3.0 −2.0 −1.0 0.0 1.0 2.0 3.0 4.0 −3.0 −2.0 −1.0 0.0 1.0 2.0 3.0 4.0
x (Å) x (Å)

Fig. 6. Plots in the left column correspond to the sliding of two QLs relative to each other and the right column presents data for a Bi2 layer on top of a QL. In all the plots, the
origin corresponds to a configuration where the atoms in the top layer are exactly on top of the atoms in the bottom layer. The four rows show, in order: (a) the potential
energy surfaces for sliding represented as binding energy relative to the isolated layers, (b) the dispersion contribution to the binding energy, (c) the base functional
contribution to the binding energy, and (d) the inter-layer distance. All energies are in kcal/mol and all distances are in angstroms.
 M.S. Christian et al. / Computational and Theoretical Chemistry 1053 (2015) 238–244
The geometry of the Bi2 surface contact is roughly the same as
that of the QL, except that the Se atoms are replaced by Bi. As in the
QL–QL case, Bi atoms in the bilayer prefer to be in direct contact
with the voids possessing an underlying Bi atom, but with a stron-
ger binding energy consistent with the results in Table 1. However,
the geometry with the Bi directly over the other QL void is very
unstable and configurations with Bi on top of the Se atoms are
favorable. The energy barrier for sliding is smaller in (QL)2
(1 kcal/mol) than in QL–Bi2 (2 kcal/mol).
For both (QL)2 and QL–Bi2, the dominant features of the poten-
tial energy surface are captured by the dispersion-energy term. The
lowest-energy arrangements occur when the corrugated surfaces
interlock, such that the number of close atomic contacts, and hence
the dispersion energy, is maximized. Interestingly, the base func-
tional gives a contribution to the PES that is roughly an inverted
version of the dispersion-corrected PES. The minimum-energy ori-
entations occur where the number of close atomic contacts, which
are repulsive in the absence of dispersion, is minimized. This con-
firms the dominant dispersion contribution to the inter-layer inter-
actions. In the most stable orientation, the two surfaces approach
closer to each other and the dispersion attraction serves to shift
the position of the Pauli repulsive wall of the potential. This
increases the repulsive contribution from the base functional.
The dispersion contribution is not as favorable in the on-top
configuration and the repulsive wall is encountered at longer
inter-layer distances, as shown in the last two panes of Fig. 6.
4. Conclusions
We have presented in this article a study of chemical bonding
and surface energetics in topological insulators based on Bi–Se
alloys, with particular emphasis on Bi2Se3 and Bi4Se3. The structure
of Bi2Se3 is a stack of five-atom Se–Bi–Se–Bi–Se layers (quintuple
layers, QL) whereas Bi4Se3 comprises a 1:1 ratio of QLs and bis-
muth bilayers (Bi2). It is experimentally known that these two
layer types can liberally amalgamate, so thorough knowledge of
the inter-layer chemistry is essential.
The inter-layer binding is dominated by non-covalent interac-
tions, so description of dispersion contributions to the binding is
key. Using the exchange-hole dipole moment (XDM) dispersion
model combined with the B86bPBE density functional, we calcu-
lated the structure and binding properties of both the bulk and
model slabs containing combinations of QLs and Bi2 bilayers.
Because of the dispersion corrections, the bulk structures are in
agreement with the known experimental results.
The chemical bonding properties were examined using the non-
covalent interaction (NCI) index, the electron localization function
(ELF), and atomic charges from Bader’s Quantum Theory of Atoms
in Molecules (QTAIM). The NCI plots show that inter-layer binding
is weaker than the intra-layer and that binding between QLs is
weaker than between a QL and Bi2. QTAIM charges and the intra-
layer distances explain this observation: the interaction between
two QLs barely affects the intra-layer geometry whereas some
charge is transferred from Bi2 to the QL, increasing the binding
and decreasing the Se–Bi bond lengths at the edge of the QL
directly facing the bismuth bilayer. However, the changes to the
electronic structure induced by Bi2 do not propagate further than
the first two atomic layers of the QL. The ELF and QTAIM charges
indicate that the nature of the intra-layer bonding is ionic, with a
certain covalent component, and that the charge donation from
Bi2 increases the stability of the QL and decreases the bismuth
bilayer binding energy.
Consistent with the chemical bonding picture, the binding ener-
gies are 6.3 kcal/mol between two QLs and 8.6 kcal/mol between a
QL and Bi2 and are independent of the presence of additional layers
243
adjacent to the QL. However, there is a small degree of non-additiv-
ity in the Bi2–QL interaction where the presence of another QL
adjacent to the Bi2 increases the binding energy by 0.4 kcal/mol.
Finally, we calculated the energetics of sliding for two QLs and
for QL–Bi2. In both cases, the orientation where the surface atoms
of one layer face the QL voids containing the underlying Bi atoms is
favored. In the QL–QL system, the second most-stable position on
the PES is for the surface atoms to face the other void, with an
underlying vacancy, while in QL–Bi2, it is the orientation where
the Bi atoms are on top of the Se atoms on the surface of the QL.
The energy barrier for sliding is significantly larger in QL–Bi2
(2 kcal/mol) than in QL–QL (1 kcal/mol). The inter-layer binding
and sliding energetics are dominated by dispersion interactions.
This implies that formation of other Bi–Se alloys, mixing QLs and
Bi2 bilayers with more complex stoichiometry, should be facile.
This could allow systematic tuning of materials properties to meet
specific manufacturing requirements for spintronic and thermo-
electric applications.
Acknowledgements
The authors thank the Spanish Malta/Consolider initiative (No.
CSD2007-00045) and the UC Lab Fees Program, Award number
237789.
References
[1] B.A. Bernevig, T.L. Hughes, S.-C. Zhang, Quantum spin Hall effect and
topological phase transition in HgTe quantum wells, Science 314 (2006)
1757–1761.
[2] M. König, S. Wiedmann, C. Brüne, A. Roth, H. Buhmann, L.W. Molenkamp, X.-L.
Qi, S.-C. Zhang, Quantum spin Hall insulator state in HgTe quantum wells,
Science 318 (2007) 766–770.
[3] J. Maciejko, T.L. Hughes, S.-C. Zhang, The quantum spin Hall effect, Ann. Rev.
Condens. Matter Phys. 2 (2011) 31–53.
[4] D. Kong, Y. Cui, Opportunities in chemistry and materials science for
topological insulators and their nano structures, Nature Chem. 3 (2011)
845–849.
[5] X.-L. Qi, S.-C. Zhang, The quantum spin Hall effect and topological insulators,
Phys. Today 63 (2010) 33–38.
[6] M.Z. Hasan, C.L. Kane, Colloquium: topological insulators, Rev. Mod. Phys. 82
(2010) 3045.
[7] B. Yan, S.-C. Zhang, Topological materials, Rep. Prog. Phys. 75 (2012) 096501.
[8] A.S. Hewitt, J. Wang, J. Boltersdorf, P.A. Maggard, D.B. Dougherty, Coexisting Bi
and Se surface terminations of cleaved Bi2Se3 single crystals, J. Vac. Sci.
Technol. B 32 (2014) 04E103.
[9] S. Urazhdin, D. Bilc, S. Mahanti, S. Tessmer, T. Kyratsi, M. Kanatzidis, Surface
effects in layered semiconductors Bi2Se3 and Bi2Te3, Phys. Rev. B 69 (2004)
085313.
[10] Y. Saeed, N. Singh, U. Schwingenschl, Thickness and strain effects on the
thermoelectric transport in nanostructured Bi2Se3, Appl. Phys. Lett. 104 (2014)
033105.
[11] Y. Xia, D. Qian, D. Hsieh, L. Wray, A. Pal, H. Lin, A. Bansil, D. Grauer, Y. Hor, R.
Cava, et al., Observation of a large-gap topological-insulator class with a single
Dirac cone on the surface, Nature Phys. 5 (2009) 398–402.
[12] H. Zhang, C.-X. Liu, X.-L. Qi, X. Dai, Z. Fang, S.-C. Zhang, Topological insulators
in Bi2Se3, Bi2Te3 and Sb2Te3 with a single Dirac cone on the surface, Nature
Phys. 5 (2009) 438–442.
[13] H. Lind, S. Lidin, A general structure model for Bi–Se phases using a superspace
formalism, Solid State Sci. 5 (2003) 47–57.
[14] T. Hirahara, G. Bihlmayer, Y. Sakamoto, M. Yamada, H. Miyazaki, S.-i. Kimura, S.
Blügel, S. Hasegawa, Interfacing 2D and 3D topological insulators: Bi(1 1 1)
bilayer on Bi2 Te3 , Phys. Rev. Lett. 107 (2011) 166801.
[15] T. Valla, H. Ji, L. Schoop, A. Weber, Z.-H. Pan, J. Sadowski, E. Vescovo, A. Fedorov,
A. Caruso, Q. Gibson, et al., Topological semimetal in a Bi–Bi2Se3 infinitely
adaptive superlattice phase, Phys. Rev. B 86 (2012) 241101.
[16] Q.D. Gibson, L.M. Schoop, A.P. Weber, H. Ji, S. Nadj-Perge, I.K. Drozdov, H.
Beidenkopf, J.T. Sadowski, A. Fedorov, A. Yazdani, T. Valla, R.J. Cava,
Termination-dependent topological surface states of the natural superlattice
phase Bi4Se3, Phys. Rev. B 88 (2013) 081108.
[17] L.V. Yashine, M.R. Sánchez-Barriga, J ad Scholz, A.A. Volykhov, A.P. Sirotina, V.S.
Neudachina, M.E. Tamm, A. Varykhalov, D. Marchenko, G. Springholz, G. Bauer,
A. Knop-Gericke, O. Rader, Negligible surface reactivity of topological
insulators Bi2Se3 and Bi2Te3 towards oxygen and water, ACS Nano 7 (2013)
5181–5191.
[18] O. Tereshchenko, K. Kokh, V. Atuchin, K. Romanyuk, S. Makarenko, V.
Golyashov, A. Kozhukhov, I. Prosvirin, A. Shklyaev, Stability of the (0 0 0 1)
surface of the Bi2Se3 topological insulator, JETP Lett. 94 (2011) 465–468.
244 M.S. Christian et al. / Computational and Theoretical Chemistry 1053 (2015) 238–244
[19] V. Atuchin, V. Golyashov, K. Kokh, I. Korolkov, A. Kozhukhov, V. Kruchinin, S.
Makarenko, L. Pokrovsky, I. Prosvirin, K. Romanyuk, et al., Formation of inert
Bi2Se3 (0 0 0 1) cleaved surface, Cryst. Growth Des. 11 (2011) 5507–5514.
[20] K. Park, C. De Beule, B. Partoens, The ageing effect in topological insulators:
evolution of the surface electronic structure of Bi2Se3 upon K adsorption, New
J. Phys. 15 (2013) 113031.
[21] X. He, W. Zhou, Z.Y. Wang, Y.N. Zhang, J. Shi, R.Q. Wu, J.A. Yarmoff, Surface
termination of cleaved Bi2 Se3 investigated by low energy ion scattering, Phys.
Rev. Lett. 110 (2013) 156101.
[22] I. Rusinov, I. Nechaev, E. Chulkov, Theoretical study of influencing factors on
the dispersion of bulk band-gap edges and the surface states in topological
insulators Bi2Te3 and Bi2Se3, J. Exp. Theor. Phys. 116 (2013) 1006–1017.
[23] W. Liu, X. Peng, X. Wei, H. Yang, G.M. Stocks, J. Zhong, Surface and substrate
induced effects on thin films of the topological insulators Bi2Se3 and Bi2Te3,
Phys. Rev. B 87 (2013) 205315.
[24] X. Luo, M.B. Sullivan, S.Y. Quek, First-principles investigations of the atomic,
electronic, and thermoelectric properties of equilibrium and strained Bi2Se3
and Bi2Te3 including van der waals interactions, Phys. Rev. B 86 (2012)
184111.
[25] A.D. Becke, E.R. Johnson, Exchange-hole dipole moment and the dispersion
interaction revisited, J. Chem. Phys. 127 (2007) 154108.
[26] A. Otero-de-la Roza, E.R. Johnson, Van der Waals interactions in solids using
the exchange-hole dipole moment, J. Chem. Phys. 136 (2012) 174109.
[27] A. Otero-de-la Roza, E.R. Johnson, A benchmark for non-covalent interactions
in solids, J. Chem. Phys. 137 (2012) 054103.
[28] R.F.W. Bader, Atoms in Molecules. A Quantum Theory, Oxford University Press,
Oxford, 1990.
[29] R.F.W. Bader, A quantum-theory of molecular-structure and its applications,
Chem. Rev. 91 (1991) 893–928.
[30] R. Boyd, C. Matta, The Quantum Theory of Atoms in Molecules: from Solid
State to DNA and Drug Design, Wiley-VCH, Weinheim, Germany, 2007.
[31] E. Johnson, S. Keinan, P. Mori-Sánchez, J. Contreras-García, A. Cohen, W. Yang,
Revealing noncovalent interactions, J. Am. Chem. Soc. 132 (2010) 6498–6506.
[32] A. Otero-de-la Roza, E.R. Johnson, J. Contreras-García, Revealing non-covalent
interactions in solids: NCI plots revisited, Phys. Chem. Chem. Phys. 14 (2012)
12165–12172.
[33] A.D. Becke, K.E. Edgecombe, A simple measure of electron localization in
atomic and molecular systems, J. Chem. Phys. 92 (1990) 5397.
[34] B. Silvi, A. Savin, Classification of chemical bonds based on topological analysis
of electron localization functions, Nature 371 (1994) 683–686.
[35] A.J. Cohen, P. Mori-Sánchez, W. Yang, Challenges for density functional theory,
Chem. Rev. 112 (2012) 289–320.
[36] E.R. Johnson, I.D. Mackie, G.A. DiLabio, Dispersion interactions in density-
functional theory, J. Phys. Org. Chem. 22 (2009) 1127–1135.
[37] G.A. DiLabio, A. Otero-de-la Roza, Non-covalent interactions in density-
functional theory, in: K.B. Lipkowitz, (Ed.), Rev. Comput. Chem., Wiley-VCH,
Hoboken, NJ, 2014, (in press, arXiv: 1405.1771).
[38] A.D. Becke, E.R. Johnson, Exchange-hole dipole moment and the dispersion
interaction, J. Chem. Phys. 122 (2005) 154104.
[39] A.D. Becke, E.R. Johnson, A density-functional model of the dispersion
interaction, J. Chem. Phys. 123 (2005) 154101.
[40] A.D. Becke, E.R. Johnson, Exchange-hole dipole moment and the dispersion
interaction: high-order dispersion coefficients, J. Chem. Phys. 124 (2006)
014104.
[41] E.R. Johnson, A.D. Becke, A post-hartree-fock model of intermolecular
interactions: inclusion of higher-order corrections, J. Chem. Phys. 124 (2006)
174104.
[42] E.R. Johnson, A.D. Becke, Van der waals interactions from the exchange hole
dipole moment: application to bio-organic benchmark systems, Chem. Phys.
Lett. 432 (2006) 600–603.
[43] A.D. Becke, E.R. Johnson, A unified density-functional treatment of dynamical,
nondynamical, and dispersion correlations, J. Chem. Phys. 127 (2007) 124108.
[44] F.O. Kannemann, A.D. Becke, Van der Waals interactions in density-functional
theory: rare-gas diatomics, J. Chem. Theory Comput. 5 (2009) 719–727.
[45] F.O. Kannemann, A.D. Becke, van der Waals interactions in density-functional
theory: intermolecular complexes, J. Chem. Theory Comput. 6 (2010) 1081–
1088.
[46] A. Otero-de-la Roza, E.R. Johnson, Non-covalent interactions and
thermochemistry using XDM-corrected hybrid and range-separated hybrid
density functionals, J. Chem. Phys. 138 (2013) 204109.
[47] A. Otero-de-la Roza, E.R. Johnson, Many-body dispersion interactions from the
exchange-hole dipole moment model, J. Chem. Phys. 138 (2013) 054103.
[48] P. Giannozzi, S. Baroni, N. Bonini, M. Calandra, R. Car, C. Cavazzoni, D. Ceresoli,
G. Chiarotti, M. Cococcioni, I. Dabo, et al., Quantum ESPRESSO: a modular and
open-source software project for quantum simulations of materials, J. Phys.:
Condens. Matter 21 (2009) 395502.
[49] A. Becke, On the large-gradient behavior of the density functional exchange
energy, J. Chem. Phys. 85 (1986) 7184.
[50] J. Perdew, K. Burke, M. Ernzerhof, Generalized gradient approximation made
simple, Phys. Rev. Lett. 77 (1996) 3865–3868.
[51] P. Blöchl, Projector augmented-wave method, Phys. Rev. B 50 (1994) 17953.
[52] M. Yu, D.R. Trinkle, Accurate and efficient algorithm for Bader charge
integration, J. Chem. Phys. 134 (2011) 064111.
[53] A. Otero-de-la Roza, M.A. Blanco, A. Martín Pendás, V. Luaña, Critic: a new
program for the topological analysis of solid-state electron densities, Comput.
Phys. Commun. 180 (2009) 157–166.
[54] A. Otero-de-la Roza, E.R. Johnson, V. Luaña, Critic2: a program for real-space
analysis of quantum chemical interactions in solids, Comput. Phys. Commun.
185 (2014) 1007–1018.
[55] R. Wyckoff, Crystal Structures, Interscience Publishers, New York, 1960.
[56] S. Nakajima, Crystal structure of Bi2Te3-xSex, J. Phys. Chem. Solids 24 (1963)
479.
[57] J. Zhang, Z. Peng, A. Soni, Y. Zhao, Y. Xiong, B. Peng, J. Wang, M.S. Dresselhaus,
Q. Xiong, Raman spectroscopy of few-quintuple layer topological insulator
Bi2Se3 nanoplatelets, Nano Lett. 11 (2011) 2407–2414.
[58] S.S. Hong, W. Kundhikanjana, J.J. Cha, K. Lai, D. Kong, S. Meister, M.A. Kelly,
Z.-X. Shen, Y. Cui, Ultrathin topological insulator Bi2Se3 nanoribbons exfoliated
by atomic force microscopy, Nano Lett. 10 (2010) 3118–3122.