Applied Surface Science 219 (2003) 158–166

Electronic structure of the Cu3Au(1 1 1) surface

Ch.E. Lekkaa,*, N. Bernsteinb, M.J. Mehlb, D.A. Papaconstantopoulosa,b
a
School of Computational Sciences, George Mason University, Fairfax, VA, USA
b
Center for Computational Material Science, Naval Research Laboratory, Washington, DC, USA

Abstract

Using the NRL-tight binding method, we investigate the electronic structure of the unrelaxed Cu3Au(1 1 1) surface. The tight
binding parameters were determined by fitting to LDA calculations and found to reproduce the surface energy, bulk elastic
constants and bulk phonon modes. The surface electronic band structure shows electronic states at the center of the Brillouin
zone, within
0.1 Ry of the Fermi level, in agreement with available photoemission data. For these states, we give an
interpretation and characterization of the contributions of the s, p and d electrons of the Cu and Au atoms. Charge calculations on
each atom indicate that upon formation of Cu3Au there is charge transfer from the Au atoms to the Cu atoms in agreement with
previous X-ray experiments. The Cu3Au(1 1 1) surface atoms are found to gain charge compared to the bulk system, and the Au
atoms regain some of the charge that they lost upon formation of bulk Cu3Au. Despite the fact that this tight binding Hamiltonian
was fitted to bulk properties of the Cu3Au system, it is able to accurately describe the main features of the Cu3Au(1 1 1) surface,
suggesting that this method is reliable for studying binary intermetallic alloy surfaces.
# 2003 Published by Elsevier Science B.V.

PACS: 71.15.F; 68.35.B; 71.20.B

Keywords: Tight binding approximation; Crystal structure of surfaces; Electronic structure of alloys

1. Introduction the Au states dominating at the bottom of the d bands

and the Cu at the top. Upon formation of Cu3Au the Au
The binary intermetallic alloy Cu3Au has been the 5d band is depleted [8–10], while the spatial compres-
subject of numerous investigations due to its relation- sion of the 5d shells of Au is most likely due to the sp
ship with materials having improved electronic proper- electrons that favor a smaller atomic volume [12].
ties for technological applications including catalysis, While the situation is rather clear concerning the
coating, and high temperature devices [1–3]. A thor- bulk system, studies that could give an insight on the
ough understanding of any material requires detailed surface electronic structure are rather limited. Angle-
and exhaustive knowledge of its electronic structure. resolved ultraviolet photoemission spectroscopy of the
Theoretical [4–7] and experimental [8–11] studies Cu3Au(1 1 1) surface revealed two sp-like surface
established that in Cu3Au there is hybridization of resonances at 0.4 and 1.0 eV below the Fermi energy
Cu and Au states giving rise to a mix of d bands, with and near the center of the surface Brillouin zone (SBZ)
[13,14]. The surface state (SS) at 0.4 eV is commonly
*
Corresponding author. Tel.: þ1-302-6510-97310;
referred to as a Shockley SS and has been extensively
fax: þ1-302-6510-97034. studied in the cases of the Cu, Ag and Au (1 1 1)
E-mail address: [email protected] (Ch.E. Lekka). surfaces [15–17]. The Shockley SS has served as a

0169-4332/$ – see front matter # 2003 Published by Elsevier Science B.V.

doi:10.1016/S0169-4332(03)00592-0
 Ch.E. Lekka et al. / Applied Surface Science 219 (2003) 158–166 159

prototype for the observation of a variety of phenomena where the sum is over the occupied states and the
[18–32] and it was found to play an important role eigenvalues are shifted to include the non-band struc-
in surface morphology [31], surface band structure ture energy appearing in the DFT formalism
[18,19,21,27–29], adsorbates and steps [32]. There-
FðnðrÞÞ
fore, a Shockley-type SS may also play a significant e0n ¼ en þ (2)
role in phenomena happening on the Cu3Au surface Ne
[33–46], rendering a detailed study of this state very where en are the Kuhn–Sham eigenvalues of the
important. Despite its importance, detailed studies DFT calculations, Ne is the number of electrons in
describing this state and other possible SSs, along with the calculation and F(n(r)) includes corrections to the
the contribution of the s, p and d electrons of the Au and kinetic energy and the Coulomb energy double count-
Cu atoms are lacking. ing terms. The shift F(n(r))/Ne avoids the need to
The aim of the present work is to study the surface choose an arbitrary zero of energy for the band struc-
band structure of the Cu3Au(1 1 1) surface and to ture term and the usual pair-potential term, simplifying
obtain information on the role of the s, p and d the numerical fitting process. The detailed justification
electrons of Cu and Au atoms in selected states. We is explained elsewhere [47–52].
used the NRL-tight binding (TB) method [47–52] in The two-center Slater–Koster formulation of TB
conjunction with density functional theory in the local with a non-orthogonal basis breaks the problem into
density approximation (DFT LDA) to provide the the calculation of the on-site, Hamiltonian and overlap
database needed for the determination of the necessary parameters. In this method, the on-site parameters
TB parameters. The NRL-TB approach is approxi- depend on the local environment and represent the
mately 103 times faster than the various LDA methods energy required to place an electron in a specific
and therefore more feasible for the large number of atomic shell. The Hamiltonian (hopping integral)
atoms needed for surface calculations in the present parameters represent the matrix elements for electrons
work. Because of its quantum-mechanical character hopping from one site to another and the overlap
the NRL-TB scheme is preferable to empirical meth- parameters describe the mixing between non-ortho-
ods such as the embedded atom method (EAM). gonal orbitals on neighbor sites. In all three cases, we
Certainly, the empirical methods produce total ener- must determine pairwise interactions between atoms
gies faster but the TB has more reliable transferability of the same type as well as those between atoms of
to structures that were not fitted. A second advantage different species [51]. This involves an extension of
of the quantum-mechanical basis of the TB calcula- the original work, which only describes monoatomic
tions is that they yield eigenvalues and eigenvectors systems [47–50,59–61].
that allow us to evaluate charge transfer between The environmental dependence of the on-site para-
bulk and surface atoms. We selected the Cu3Au(1 1 1) meters is controlled by a set of atomic-like densities
surface because it has the same stoichiometry as [51]
the bulk system [53] (25% Au concentration) and X
there is a variety of experimental data for both the pði; ~jÞ ¼ expð
l~2i~j jRi
Rj jÞFðjRi
Rj jÞ (3)
order–disorder transition [54–57] and the electronic j2~j
structure of the ordered [13,14] and the disordered
where the ith atom is of type ~i; the jth atom of type ~j;
phases [58].
rði; ~jÞ is the density of atom i due to neighboring atoms
of type ~j; and l~i~j is a fitting constant to be determined.
The cut-off function F takes the form
2. Computational details
yðRc
RÞ
The total energy of the NRL-TB method is FðRÞ ¼ (4)
f1 þ exp½ðR
Rc Þ=Lc þ 5g
expressed as
where y is the step function. For the Cu–Au system,
X
occ
EðnðrÞÞ ¼ e0n (1) we take Rc ¼ 12:5 Bohr and Lc ¼ 0:5 Bohr for all
n interactions.
160 Ch.E. Lekka et al. / Applied Surface Science 219 (2003) 158–166

The on-site terms themselves are polynomial func- LDA calculations. We constructed the database of
tions of r2/3 electronic band structures and total energies for var-
X ious structures at several volumes around the equili-
hk ðiÞ ¼ ak ð~iÞ þ ½bk ð~i; ~jÞrði; ~jÞ2=3 þ ck ð~i; ~jÞrði; ~jÞ4=3 brium density. In all cases, we used a uniform regular
~j
k-point mesh. For the pure Cu and Au that k-point
þ dk ð~i; ~jÞrði; ~jÞ2  (5) mesh corresponds to 146, 55 and 84 k-points in the
irreducible part of the face-centered cubic (fcc) lattice,
where the sum is over all atom types in the system. In body-centered cubic (bcc) lattice and the simple cubic
general, we use k ¼ s, p, d. (sc) zone, respectively. For the CuAu sodium–chloride
The two-center Slater–Koster hopping integrals are (B1), cesium–chloride (B2) and copper–gold (L10)
determined using an exponentially damped polyno- structures, we use a k-point mesh of 89, 84 and 216
mial, and depend only on the atomic species and the k-points, respectively. In addition, we used 84 and 60
distances between the atoms: (or 44) k-points in the Cu3Au (L12) and AlFe3 (DO3)
Hkk0 m ði; j; RÞ structures for the Cu3Au and Au3Cu lattices, respec-
tively. For each structure, we used 5–11 volumes, with
¼ ½Akk0 m ð~i; ~jÞ þ Bkk0 m ð~i; ~jÞR
the largest number of volumes for the experimental
þ Ckk0 m ð~i; ~jÞR2  exp½
D2kk0 m ð~i; ~jÞRFðRÞ (6) equilibrium structure. The total database of eigenva-
lues and total energies contains about 5000 entries. We
where the parameters A, B, C and D are to be fitted. For
used the Levenberg–Marquardt algorithm to adjust our
like-atom ð~i ¼ ~jÞ interactions, there are 10 independent
TB parameters to give the best possible reproduction
Slater–Koster parameters:
of the database. The total energies are usually
sss; sps; pps; ppp; sds; pds; pdp; dds; ddp; ddd: weighted at about 500 times larger than a single band.
We first fitted the total energies and the band structures
When the atoms are of different types, we must
of pure Cu (sc, bcc, fcc) and pure Au (sc, bcc, fcc)
include an additional four parameters:
separately and fixed the corresponding parameters. We
pss; dss; dps; dpp: then fitted the total energies and the band structures of
Since we are using a non-orthogonal basis set, we CuAu(B1, B2), Cu3Au(L12, DO3) and Au3Cu ((L12,
DO3), in order to determine the Cu–Au interactions.
must also parametrize the overlap integrals. These
For these structures, the average rms error of the
have a form similar to that of the hopping integrals
occupied bands is approximately 20 mRy and less
Skk0 m ði; j; RÞ than 1 mRy for the total energies.
¼ ½Okk0 m ð~i; ~jÞ þ Pkk0 m ð~i; ~jÞR After obtaining the TB parameters, we validated
them by calculating the equilibrium lattice constant,
þ Qkk0 m ð~i; ~jgR2  exp½
Tkk 2 ~~
0 m ði; jÞRFðRÞ (7)
elastic constants and the phonon modes in the high
where O, P, Q and T also represent parameters to be symmetry points for bulk Cu3Au as shown in Table 1.
fitted. In principle, the TB parameters obtained with the
For a two-component system with s, p and d orbi- procedure described above are not unique. However,
tals, there are 316 parameters used in the fit, in contrast we find that the validation tests eliminate unsuitable
to 93 for a single-element parameterization [48,51]. parameter sets, and that remaining parameter sets
These parameters are chosen to reproduce the DFT give equivalent results for the properties we study
LDA values of the eigenvalues en 0 and energies E in here. Then, we proceeded with the calculation of
Eq. (1) over a variety of structures. surface properties. For the Cu3Au(1 1 1) surface elec-
The basic aim of our procedure is to represent the tronic structure we performed TB calculations using a
Slater–Koster Hamiltonian using simple functional 15 atomic layer slab and extended periodically in
forms with the parameters chosen to reproduce the directions perpendicular to the (1 1 1) surface. By
first-principles calculations over a wide range of pres- fixing the dimension of the simulation box at a value
sure and structures. The parameters are obtained by twice as large as the thickness of the crystal along the
fitting to band structures and total energies of the DFT z-direction, we simulated an isolated slab.
 Ch.E. Lekka et al. / Applied Surface Science 219 (2003) 158–166 161

Table 1
Bulk properties of Cu3Au

Property Experiment LAPW Present work

A (Bohr) 7.086 6.949 6.975

B (GPa) 152 140 131
C0 (GPa) 52 60 100
C11 (GPa) 187 180 198
C12 (GPa) 135 120 98
C44 (Gpa) 68 – 92
G4 (cm
1) 0 0 0
G4 (cm
1) 125 110 153
G4 (cm
1) 210 200 270
G5 (cm
1) 161 159 195
Equilibrium lattice parameter (a), bulk modulus (B), elastic
constants (C 0 ¼ C11
C12 , C11, C12 and C44), phonon modes at
 point (G4 and G5) along with the corresponding quantities
the G
found with the LAPW method and the available experimental data
[59–61].

3. Results

3.1. Bulk properties

A direct result of the fitting procedure, which Fig. 1. Electronic density of states of the Cu3Au: (a) total DOS,
reproduces the total energies of the Cu3Au system (b) d partial DOS of Cu atoms, and (c) d partial DOS of Au atoms.
at several volumes, is an accurate determination of the In each case, the Fermi level has been set to 0.
equilibrium lattice parameters and the bulk modulus.
In addition, we calculated the elastic constants Cu, C12
and C44 by imposing external strains in the crystal [61] (Fig. 1a) along with the d partial DOS of Cu and Au
and obtaining the energy as a function of strain, from atoms (Fig. 1b and c, respectively). We observe that
the curvature of which we numerically deduce the below the Fermi level, the electronic states have
elastic constants. Furthermore, using the frozen pho- mostly d character while above it they show mainly
non method [47,62] we calculate the phonon modes at s and p character. In Fig. 1b and c, we see that both
the center of the Brillouin zone (G  point). These kinds of atoms contribute to the entire d-band energy
results are summarized in Table 1 along with the range from
0.6 to
0.05 Ry giving rise to a common
corresponding quantities found independently by the d band for the Cu3Au system. Nevertheless, the Au d
LAPW method and from available experimental data. states dominate the bottom of the d band (Fig. 1c)
As we can see, our predictions are in overall agree- while the contribution of the Cu d states is larger near
ment with the LAPW results and with experimental the top (Fig. 1b). These results are in very good
data not only for the lattice parameter [63] and the agreement with available theoretical [4–7] and experi-
bulk modulus which we have essentially fitted, but mental studies [8–11].
also for the elastic constants [64] and the phonon
modes [65], which are not included in the fitting 3.2. Surface energy
database.
The electronic density of states (DOSs) of Cu3Au In order to calculate the surface energy, we evaluate
(L12) was obtained by the tetrahedron method [66], the energy of the bulk Cu3Au (ECu3 Au ) and the
using 165 k-points in the irreducible part of the energy of the Cu3Au(1 1 1) surface slab (Es) sepa-
Brillouin zone. In Fig. 1, we present the total DOS rately for a system of 60 atoms (15 atomic layers) and
162 Ch.E. Lekka et al. / Applied Surface Science 219 (2003) 158–166

331 in-plane k-points. We evaluated the surface energy result is much better than for most empirical potentials
(g) by: (including some based on the second moment approx-
imation, Finnis–Sinclair and EAM), which give values
Es
ECu3 Au
g¼ (8) between 863 and 993 mJ/m2 for the relaxed g
2A [37,38,41,43]. The potential of Bozzolo–Ferrante–
where Es is the energy of the slab with two free Smith (BFS) is the one exception, giving a very good
surfaces, ECu3 Au is the energy of the ordered Cu3Au relaxed g of 1577 mJ/m2 [39,40].
bulk and A is the surface area. Our calculation predicts
a surface energy of 1708 and 1648 mJ/m2 for the 3.3. Band structure
unrelaxed and the relaxed Cu3Au(1 1 1) surface,
respectively. These results are consistent with our The band structures of the Cu3Au bulk system along
LAPW–LDA results which give 1596 mJ/m2 for the !R
the G  direction and the Cu3Au(1 1 1) surface
unrelaxed surface, calculated using 28 atoms (seven along the G  !M  direction are depicted in Fig. 2a
atomic layers) and 14 k-points, based on the WIEN2k and b, respectively. In Fig. 2a, the narrow bands of the
code [67]. The system used in the LAPW–LDA bulk system below the Fermi level are a mixture of the
calculation is small compared to the TB calculation Cu and Au d bands, in agreement with available
because it requires a lot of computational time to theoretical [4–7] and experimental [8–11] data, while
perform an LAPW calculation in 60 atoms and 331 the bands above the Fermi level have mainly s and
k-points. Furthermore, it worth noting that the agree- p character (Section 3.1). The G  !R  direction of the
ment of the TB calculation with the LAPW–LDA bulk system (in k-space (1 1 1) direction), projects

Fig. 2. Band structure of the Cu3Au: (a) bulk system along the GR direction, and (b) (1 1 1) surface along the GM direction. E1 and E2 are the
experimentally determined surface states.
 Ch.E. Lekka et al. / Applied Surface Science 219 (2003) 158–166 163

Fig. 3. Contribution of the s, p and d electrons of Cu (*) and Au (*) as a function of atomic layers for the three energy states (a)
0.090,
(b)
0.026, and (c) 0.004 Ry at the center of the Brillouin zone.

Fig. 4. Number of electrons of the Cu (*) and the Au () atoms for the different atomic layers. The dotted line (labeled pure Cu and pure Au)
corresponds to the number of electrons (11 e
) in the pure Cu and pure Au system, while the b1 and b2 lines refer to the number of electrons of
the bulk Cu and bulk Au atoms in the Cu3Au system, respectively.
164 Ch.E. Lekka et al. / Applied Surface Science 219 (2003) 158–166

onto G  of the (1 1 1) SBZ. For comparison with atoms lose
0.13 e
compared to the pure Au, while
available experimental results, we will focus our the Cu atoms gain þ0.03 e
, in agreement with
attention at the G point and close to the Fermi level. XANES and XPS experimental data [8]. Turning to
We recall here that angle-resolved photoemission the surface behavior, we find that the Cu3Au(1 1 1)
spectroscopy on the Cu3Au(1 1 1) surface shows surface atoms gain charge, and in particular the Au
two sp-like SSs at E1 ¼
0.029 Ry (Shockley SS) atoms regain 57% of the charge lost upon the forma-
and E2 ¼
0.073 Ry [13]. As can be seen in Fig. 2b, tion of Cu3Au.
in this energy region between the main d-character To understand the contribution of different electro-
area (black region) and the Fermi level of the nic states to the overall charge transfer, we present in
Cu3Au(1 1 1) surface band structure there are several Fig. 5 the decomposition of the curves in Fig. 4 into
energy states. Two of these, at
0.026 and
0.072 Ry, contributions from s, p and d electrons. We observe
lie very closely in energy to the experimental SS. With that upon formation of the Cu3Au system, the d and p
the aim of understanding whether these are SSs and
obtaining information concerning the role of s, p and d
electrons of Cu and Au atoms, we perform a detailed
analysis of the energy states close to the Fermi level,
near G of the SBZ. In Fig. 3, we present the contribu-
tions of the s, p and d electrons of the Cu (*) and Au
(*) atoms as a function of the atomic layer for three
states with energy of
0.090 (Fig. 3a),
0.026
(Fig. 3b) and 0.004 Ry (Fig. 3c). In the
0.090 Ry
state (Fig. 3a), in the bulk the majority of the electro-
nic contributions is due to Cu d, Au d and Cu s
electrons and this contribution vanishes as we proceed
towards the surface layer. In contrast, for the same
energy state the role of p electrons in the bulk system is
negligible but becomes important at the surface layer.
The state at
0.026 Ry (very close in energy to the
experimental E1 state) cannot be clearly classified as a
pure SS or a bulk state because there are significant
contributions from both bulk and surface atoms. Simi-
lar behavior was also found for the state that is close
to the E2 experimental state. For the 0.004 Ry state
(Fig. 3c), we observe that near the surface the con-
tribution of the s, p and d electrons of both Au and
Cu atoms is very important, while it becomes negli-
gible when approaching the bulk, suggesting that this
energy state is a pure SS.

3.4. Charge transfer

For the bulk Cu3Au system as well as the unrelaxed

Cu3Au(1 1 1) surface, we calculate the charge on each
Cu and Au atom by summing the occupied DOS Fig. 5. Number of the d (a), p (b) and, s (c) electrons of the Cu (*)
projected on each atom. In Fig. 4, we plot the number and the Au () atoms for the different atomic layers of the
Cu3Au(1 1 1) surface. The dotted line (labeled pure Cu and pure Au)
of electrons of the Cu and the Au atoms for the corresponds to the number of electrons in the pure Cu fcc and pure
different atomic layers of the Cu3Au(1 1 1) surface. Au fcc, while the b1 and b2 lines refer to the number of electrons of
We observe that upon formation of Cu3Au, the Au the bulk Cu and bulk Au atoms in the Cu3Au system, respectively.
 Ch.E. Lekka et al. / Applied Surface Science 219 (2003) 158–166 165

character of the Au bulk atoms is slightly increased Acknowledgements

while the s character is significantly decreased com-
pared with the pure Au. The Cu atoms in Cu3Au gain d We thank Dr. A. Aguayo for his help with the
character and loose p and s character compare to the WIEN2k code.
pure Cu metal. On the surface, both Cu and Au atoms d
and s orbitals gain charge while these atoms undergo a
reduction of their p character compared to the Cu and References
Au bulk Cu3Au atoms. It should be noted that in both
Figs. 4 and 5, the numbers of electrons for the Cu and [1] W. Pfeiler, B. Sprusil, Mater. Sci. Eng. A324 (2002) 34.
[2] T.W. Kim, D.U. Lee, D.C. Choo, Y.S. Lim, H.S. Lee, J.Y. Lee,
Au in the bulk region of the Cu3Au(1 1 1) slab are in
H. Lim, Solid State Commun. 117 (2001) 501.
very good agreement with the corresponding values [3] P. Deurinck, C. Creemers, Surf. Sci. 419 (1998) 62.
for the bulk Cu3Au system, indicating that the number [4] B. Ginatempo, G.Y. Guo, W.M. Temmerman, J.B. Staunton,
of layers used in the slab calculation is sufficient and P.J. Durham, Phys. Rev. B42 (1990) 2761.
that there is no interaction between the two surfaces [5] P. Weinberger, A.M. Boring, R.C. Albers, W.M. Temmerman,
through the bulk. Phys. Rev. B38 (1988) 5357.
[6] H.L. Skriver, H.P. Lengekeek, Phys. Rev. B19 (1979) 900.
[7] J. Davenport, R.E. Watson, M. Weinert, Phys. Rev. B37
(1988) 9985.
4. Conclusions [8] T.K. Sham, Y.M. Yiu, M. Kuhn, K.H. Tan, Phys. Rev. B41
(1990) 11881.
In the present study, we investigate the electronic [9] G. Wertheim, L.F. Mattheiss, D.N.E. Buchanan, Phys. Rev.
B38 (1988) 5988.
structure of the ordered Cu3Au(1 1 1) unrelaxed sur- [10] G. Wertheim, Phys. Rev. B36 (1987) 4432.
face using the NRL-TB method. We first separately [11] Z.Q. Wang, S.C. Wu, J. Quinn, C.K.C. Loc, Y.S. Li, F. Jona,
determine the TB parameters for the pure Cu and pure J.W. Davenport, Phys. Rev. B38 (1988) 7442.
Au system by fitting to LDA calculations the total [12] V. Heine, L.D. Marks, Surf. Sci. 165 (1986) 65.
energy and the band structure for the fcc, bcc and sc [13] R. Courths, M. Lau, T. Scheunemann, H. Gollish, R. Feder,
Phys. Rev. B63 (2001) 195110.
structures. Then we fit the corresponding values for the [14] M. Lau, S. Loubus, R. Courths, S. Halilov, H. Gollisch, R.
CuAu (B1 and B2 structures), the Cu3Au and Au3Cu Feder, Ann. Physik 2 (1993) 450.
(L12 and DO3 structures) in order to fix the Cu–Au [15] S.D. Kevan, R.H. Gaylord, Phys. Rev. B36 (1987) 5809.
interactions. We find that the our TB Hamiltonian [16] P.O. Gartland, B.J. Slagsvold, Phys. Rev. B12 (1975) 4047.
satisfactorily reproduces the elastic constants, the bulk [17] P. Heimann, H. Neddermeyer, H.F. Roloff, J. Phys. C10
(1977) L17.
phonon frequencies at the high symmetry points of the [18] S.D. Kevan, Phys. Rev. Lett. 50 (1983) 526.
L12 Cu3Au system and the surface energy of the [19] J. Tersoff, S.D. Kevan, Phys. Rev. B28 (1983) 4267.
Cu3Au(1 1 1) system. We perform a detailed analysis [20] R. Courths, H. Wern, U. Hau, B. Cord, V. Bachelier, S.
on the surface electronic band structure close to the Houfner, J. Phys. F: Met. Phys. 14 (1984) 1559.
Fermi level and at the center of the Brillouin zone and [21] B.A. McDougall, T. Balasubramanian, E. Jensen, Phys. Rev.
B51 (1995) 13891.
we find two states situated at
0.026 and
0.072 Ry, [22] S. LaShell, B.A. McDougall, F. Jensen, Phys. Rev. Lett. 77
that lie closely in energy to the SS found in photoemis- (1996) 3419.
sion experimental data. Calculation of the charge on [23] J.A. Knapp, F.J. Himpsel, D.E. Eastman, Phys. Rev. B19
each atom reveals that upon formation of Cu3Au there is (1979) 4952.
charge transfer from the Au atoms to the Cu atoms, in [24] Y. Petroff, P. Thiry, Appl. Opt. 19 (1980) 3957.
[25] S.G. Louie, P. Thiry, R. Pinchaux, Y. Petroff, D. Chandesris, J.
agreement with X-ray experiments. In addition, both Lecante, Phys. Rev. Lett. 44 (1980) 549.
species of the Cu3Au(1 1 1) surface gain charge com- [26] T.C. Hsieh, P. John, T. Miller, T.-C. Chiang, Phys. Rev. B35
pared to the bulk system, especially the Au atoms that (1987) 3728.
regain most of the electron charge loss that occurs upon [27] R. Matzdorf, G. Meister, A. Goldmann, Phys. Rev. B54 (14)
formation of the bulk Cu3Au. We conclude that despite (1996) 807.
[28] F. Theilmann, R. Matzdorf, A. Goldmann, Surf. Sci. 420
the fact that we fit the TB Hamiltonian only to bulk (1999) 33.
properties of the Cu3Au system, it describes accurately [29] G. Nicolay, F. Reinert, S. Schmidt, D. Ehm, P. Steiner, S.
the main features of the Cu3Au(1 1 1) surface. Houfner, Phys. Rev. B62 (2000) 1631.
166 Ch.E. Lekka et al. / Applied Surface Science 219 (2003) 158–166

[30] R. Panagio, R. Matzdorf, G. Meister, A. Goldmann, Surf. Sci. [51] S.H. Yang, M.J. Mehl, D.A. Papaconstantopoulos, M.B. Scott,
336 (1995) 113. J. Phys. Condens. Matter 14 (2002) 1985.
[31] Th. Fauster, Ch. Reuss, I.L. Shumay, M. Weinelt, F. [52] H.J. Gotsis, D.A. Papaconstantopoulos, M.J. Mehl, Phys. Rev.
Theilmann, A. Goldmann, Phys. Rev. B61 (2000) 16168. B65 (2002) 134101.
[32] L. Bourgi, L. Petersen, H. Brune, K. Kern, Surf. Sci. 447 [53] U. Bardi, Rep. Prog. Phys. 57 (1994) 939.
(2000) L157. [54] S.W. Bonham, C.P. Flynn, Surf. Sci. 366 (1996) L760.
[33] B. Gans, P.A. Knipp, D.D. Koleske, S.J. Sibener, Surf. Sci. [55] C. Ern, W. Donner, H. Dosch, B. Adams, D. Nowikow, Phys.
264 (1992) 81. Rev. Lett. 85 (2001) 1926.
[34] D.H. Oh, H.J. Kang, K.H. Chae, C.N. Whang, B.V. King, D.J. [56] X.-M. Zhu, R. Feidenhans’l, H. Zabel, J. Als-Nielsen, R. Du,
O’Connor, D.W. Moon, Surf. Sci. 477 (2001) L289. C.P. Flynn, F. Grey, Phys. Rev. B37 (1988) 7157.
[35] L. Houssiau, P. Bertrand, Surf. Sci. 352 (1996) 978. [57] Yi. Huang, M. Gajdardziska-Josifovska, J.M. Cowley, Ultra-
[36] S.M. Foiles, Surf. Sci. 191 (1987) 329. miscoscopy 57 (1995) 391.
[37] M. Hayoun, V. Pontikis, C. Winter, Surf. Sci. 398 (1998) 125. [58] F. Arola, R.S. Rao, A. Salokative, A. Bansil, Phys. Rev. B41
[38] W.E. Wallace, G.J. Ackland, Surf. Sci. 275 (1992) 1685. (1990) 7361.
[39] R.J. Kobistek, G. Bozzolo, J. Ferrante, H. Schlosser, Surf. Sci. [59] M.J. Mehl, J.F. Osburn, D.A. Papaconstantopoulos, B.M.
307–309 (1994) 390. Klein, Phys. Rev. B41 (1990) 10311;
[40] G. Bozzolo, J. Ferrante, R.D. Noebe, B. Good, F.S. Honecy, P. M.J. Mehl, J.F. Osburn, D.A. Papaconstantopoulos, B.M.
Abel, Comp. Mater. Sci. 15 (1999) 169. Klein, Phys. Rev. B42 (1991) 5362.
[41] W. Fei, A. Kara, T. Rahman, Phys. Rev. B61 (2000) 16105. [60] M.J. Mehl, Phys. Rev. B47 (1993) 2493.
[42] Ch.E. Lekka, N.I. Papanicolaou, G.A. Evangelakis, Surf. Sci. [61] M.J. Mehl, B.M. Klein, D.A. Papaconstantopoulos, in: J.H.
479 (2001) 287. Westbrook, R.L. Fleischer (Eds.), Intermetallic Compounds:
[43] Ch.E. Lekka, N.I. Papanicolaou, G.A. Evangelakis, Surf. Sci. Principles and Applications, Wiley, London, 1994.
488 (2001) 269. [62] R.M. Klein, R.E. Cohen, Phys. Rev. B45 (1992) 12405.
[44] Ch.E. Lekka, G.A. Evangelakis, Surf. Sci. 473 (2001) 39. [63] C. Kittel, Introduction to Solid State Physics, 4th ed., Wiley,
[45] Ch.F. Lekka, D.G. Papageorgiou, G.A. Evangelakis, Surf. Sci. 1971.
518 (2002) 111. [64] G. Simmons, H. Wang, Single Crystal Elastic Constants and
[46] N.I. Papanicolaou, G.C. Kallinteris, G.A. Evangelakis, D.A. Calculated Aggregate Properties: A Handbook, 2nd ed., MIT
Papaconstantopoulos, M.J. Mehl, J. Phys. Condens. Matter 10 Press, Cambridge, MA, 1971.
(1998) 10979. [65] S. Katano, M. Iizumi, Y. Noda, J. Phys. F: Met. Phys. 18
[47] R.F. Cohen, M.J. Mehl, D.A. Papaconstantopoulos, Phys. (1988) 2195.
Rev. B50 (1994) 14694. [66] O. Jepsen, O.K. Andersen, Solid State Commun. 9 (1971)
[48] M.J. Mehl, D.A. Papaconstantopoulos, Phys. Rev. B54 (1996) 1763.
4519. [67] P. Blaha, K. Schwarz, G. Madsen, D. Kvasnicka, J. Luitz, in:
[49] N. Berstein, M.J. Mehl, D.A. Papaconstantopoulos, N.I. K. Schwarz (Ed.), Computer code WIEN2k. An Augmented
Papanicolaou, M.Z. Bazant, E. Kaxiras, Phys. Rev. B62 Plane Wave and Local Orbitals Program for Calculating
(2000) 4477. Crystal Properties, Technical University of Wien, Vienna,
[50] F. Kirchhoff, M.J. Mehl, N.I. Papanicolaou, D.A. Papacon- Austria.
stantopoulos, F.S. Khan, Phys. Rev. B63 (2001) 195101.