Structure, bonding nature, and binding energy of alkanethiolate on As-rich GaAs (001) surface: a density functional theory study.

Structure, Bonding Nature, and Binding Energy of Alkanethiolate on As-Rich GaAs (001)
Surface: A Density Functional Theory Study
Oleksandr Voznyy and Jan J. Dubowski*
Department of Electrical and Computer Engineering, Centre de Recherche en Nanofabrication et
Nanocaracte´riasation (CRN2), UniVersite´ de Sherbrooke, Sherbrooke, Que´bec J1K 2R1, Canada
ReceiVed: July 23, 2006; In Final Form: October 5, 2006
Chemisorption of alkanethiols on As-rich GaAs (001) surface under a low coverage condition was studied
using first principles density functional calculations in a periodic supercell approach. The thiolate adsorption
site, tilt angle and its direction are dictated by the high directionality of As dangling bond and sulfur 3p
orbital participating in bonding and steric repulsion of the first three CH2 units from the surface. Small charge
transfer between thiolate and surface, strong dependence of total energy on tilt angle, and a relatively short
length of 2.28 Å of the S-As bond indicate the highly covalent nature of the bonding. Calculated binding
energy of 2.1 eV is consistent with the available experimental data.
Self-assembled monolayers (SAMs) of organosulfur com-
pounds on solid surfaces have attracted a great deal of interest
from both a fundamental perspective and due to potential
applications. While SAMs on noble metals have been exten-
sively studied,1-3 relatively limited data has been available
concerning monolayers of thiols on semiconductor surfaces. The
formation of SAMs of alkanethiols on oxide-free GaAs (001)
surface, together with the potential of this approach for efficient
surface passivation of this material, has been demonstrated some
time ago.4,5 Among other potential applications of SAMs on
semiconductor surfaces and particularly on GaAs are the
development of precursors for the growth of II-VI materials6
and creation of transition layers for ohmic contacts and Schottky
diodes.7 Long-chain SAMs have been applied for speciality
masks used in nanolithography.8 There has also been observed
a steady growing interest in developing GaAs-thiol interfaces
for chemical sensing9 and biosensing applications.10,11 The
structure of a chemisorbed monolayer film of thiol is determined
by the surface chemical bond and the intermolecular van der
Waals forces between the hydrocarbon chains. Theoretical
modeling of the semiconductor-thiol interface can provide
information valuable for understanding the bonding nature of
such a material system and, ultimately, it would help to design
and optimize a semiconductor-thiol interface addressing a
specific application. Modeling of the GaAs-thiol interface
deserves separate attention as the to date available theoretical
calculations of metal-thiol interfaces show markedly different
results, even for very similar surfaces such as gold, silver, and
copper.12-16 Clearly, the adsorption of a thiol is strongly
influenced by the material’s overall chemical reactivity, lattice
constant, crystallographic orientation, etc. Therefore, a little
information available from the well-studied SAMs of thiols on
noble metals can be applied to GaAs. In contrast to alkanethiols
on gold, which are considered a prototype example of SAMs,
theoretical studies of thiols on GaAs appear to be missing in
the literature. In this work, we report the results of ab initio
calculations of alkanethiols adsorption on GaAs (001) under
low surface coverage conditions.
The calculations have been performed using a density
functional theory (DFT) approach based on pseudopotentials
and numerical localized atomic orbitals as basis sets, as
implemented in the SIESTA code.17 We have used the general-
ized gradient approximation (GGA) with the Perdew-Burke-
Ernzerhof exchange-correlation functional (PBE),18 which was
reported19 to reproduce organic molecules properties with the
quality comparable to the results obtained with the Becke-
Lee-Yang-Parr functional (BLYP).20,21 The PBE also provides
better accuracy than local density approximation in description
of polar surfaces,22 such as GaAs (001). We used scalar
relativistic Troullier-Martins pseudopotentials with nonlinear
core corrections (NLCC), not including 3d electrons in valence
configuration for Ga and As. Double œ plus polarization orbital
(DZP) bases were used for Ga and As, and triple œ plus
polarization (TZP) for H, C, S, and As in the topmost layer,
with the cutoff radii optimized variationally using a simplex
method.23 The method of Monkhorst and Pack was used for
Brillouin zone sampling at accuracy equivalent to that obtained
with an 12 Å radius supercell.24 The Hartree and exchange-
correlation potentials were evaluated on a real space mesh with
a 350 Ry equivalent plane wave energy cutoff providing
convergence of binding energies to the 5 meV level.
The periodic supercell approach was employed to model
GaAs (001) surface. Surface reconstruction with full monolayer
of As was used throughout the work, as it has been shown that
standard etching procedures used for thiols deposition on GaAs
result in an excess of As on surface and XPS shows that sulfur
binds to As rather than to Ga.25,26 The slab consisted of eight
atomic layers, with thiol placed on one side and the rear surface
saturated with H atoms. A distance of 12 Å from the topmost
atom of thiol to the rear side of the next slab in the z-direction
was used to avoid spurious interactions between slabs. The
bottommost layer of Ga atoms along with H-layer was fixed
23619
2006, 110, 23619-23622
Published on Web 11/04/2006
10.1021/jp064675l CCC: $33.50 © 2006 American Chemical Society

during geometry optimization, while the rest of the atoms were
allowed to relax. The use of hydrogen termination is necessary
due to the polar nature of the (001) surface. Binding energies
were calculated as a difference between total energy of the slab-
adsorbate system and the energies of bare surface and adsorbate
molecule in gas phase, using the counterpoise method to avoid
the basis set superposition error (BSSE).27 Spin polarization was
included in all binding energy calculations. The reliability of
our calculations was tested by comparison with published
experimental and theoretical results of lattice constant, bulk
modulus, band structure and density of states of bulk GaAs,28
structures and energetics of different GaAs surfaces,22 and
energies of dissociation of thiol and hydrogen molecules.12-14,16
The calculated GaAs lattice constant is 5.756 Å (experimental
value 5.65 Å), H-H bond strength in hydrogen molecule
4.57 eV and S-H bond in thiol 3.78 eV, as compared to
experimental values of 4.75 (with zero-point energy removed
for direct comparison to calculations) and 3.73 eV, respectively.
The observed overestimation of lattice constant and deviations
of binding energies are common for all GGA calculations.
We studied the structure of the thiolate adsorbed on GaAs
(001) surface by using (2 � 4) surface unit cell (8 � 16 Å2),
which corresponds to a low coverage limit. This avoided
interaction between thiols on the surface and potential calcula-
tion errors, as it is known that current DFT functionals do not
describe the van der Waals interactions correctly. Dimerization
of top arsenic atoms was found to stabilize the system by
1.36 eV in comparison with the unreconstructed surface. It has
been reported that short-chain thiols (with 1 or 2 carbon atoms)
adopt different equilibrium geometries and have different
binding energies than longer chain molecules.12,16,29 Since we
are interested in chemisorption of thiol on the surface as a stage
preceding the formation of SAMs, we have carried out calcula-
tions for pentanethiol, a relatively long chain thiol that would
not require excessive computation time.
The equilibrium geometries obtained from molecular dynam-
ics simulations and subsequent conjugate gradient geometry
optimizations starting from a thiolate lying parallel to the surface
and from a thiolate standing upright are shown in Figure 1a
and 1b, respectively. Both geometries have practically identical
energy and simulations starting from different initial conditions
end up in one of them. The sulfur atom is situated almost on
top of arsenic, while structures with sulfur in bridge or hollow
site positions were found to be 1 eV higher in energy.
The structure in Figure 1a has a tilt angle of 44° from surface
normal with the S-C-C plane tilt of 35° (which is sometimes
called “lean”) and S-C bond lying almost parallel to the surface.
The thiol is tilted diagonally, in the direction of the second
nearest neighbor As dimer. In the second structure thiol is tilted
30° from normal in the direction of the farther arsenic of the
adjacent As dimer, with 20° lean and S-C bond is 55° from
surface normal. There are numerous similarities between these
two geometries. In both of them the As-S bond lies along the
arsenic dangling bond and the As-S-C angle is close to the
value of H-S-C angle in free thiol. Geometries with As, S,
and C lying in one line are found to be up to 0.5 eV higher in
energy than the optimal configurations, similar to reported values
for thiols on gold.12 Such a preference for the bond directionality
from both surface and thiol indicates a high covalency and thus
strength of the bond. The length of As-S bond is 2.28 Å, which
is shorter than 2.5 Å for Au-S12,15,16 and 2.35 Å for Cu-S,12,30
suggesting stronger binding of thiol to GaAs than to Au surface.
The value of the As-S-C angle is 106° and 102.5° for the
structures in Figure 1a and 1b, respectively. This is closer to
the ideal tetrahedral angle of 109.5°, rather than to 96° H-S-C
angle in free thiol, suggesting that sulfur s states could hybridize
with p states, similar to the hybridization reported for methylthiol
on gold and silver in the on-top positions15 and for phenylthiol
on gold.31 However, closer examination of the density of states
and shapes of high-lying molecular orbitals for the thiol-GaAs
case does not support this suggestion. Figure 2 shows the
molecular orbitals lying in the 1 eV energy window below the
Fermi level. This is the highest occupied molecular orbital
(HOMO) of the thiol and it coincides in energy with the dangling
bonds of arsenic. The molecular orbital on sulfur has a
pronounced px character, similar to that in free thiol, thus, the
bonds to carbon and to arsenic (or hydrogen) which are formed
by the remaining py and pz orbitals should tend to be at right
angle. The deviation to higher angle values can be explained
by steric repulsion between hydrogen atoms in the first CH2
unit and arsenic in adjacent dimers. In both configurations the
distance from H to the nearest As is about 3.3 Å, which is
comparable to the sum of van der Waals radii of H (1.2 Å) and
As (2 Å). This also explains the tilt direction of the thiol; it
rotates to accommodate the first CH2 unit in the hollow between
As dimers. Other tilt directions are less favorable since one of
the first three CH2 units would always approach the surface
too close. Thus, the observed geometries can be seen as a
compromise between the favorable As-S bond direction and
As-S-C angle, and bigger distance of the first three CH2 units
from surface. Similar conclusions have been made for the
adsorption site of mercapto and methylthiolate on gold.13
Figure 1. Optimized geometries of pentanethiol on As-rich GaAs (001)
surface obtained from relaxation of thiolate lying flat to the surface
(a) and standing upright (b).
Figure 2. High lying molecular orbitals of thiolate chemisorbed on
(001) GaAs surface. Isodensity surfaces correspond to 0.02 a.u.
23620 J. Phys. Chem. B, Vol. 110, No. 47, 2006 Letters

Simulations with removed adjacent As dimer show that thiol
can adopt smaller As-S-C angles and rotate around As-S
bond more easily, supporting our conclusions.
Examination of the lower density isosurfaces of the orbitals
shown in Figure 2 shows that this is a ð-type orbital antibonding
with respect to both S-As and S-C bonds, similar to that found
for ethanethiol on copper.30 Lower energy orbitals (not shown
here) have ó-character. From Figure 2 one can see that the
orbitals of As in the top layer are substantially dehybridized in
comparison with that of bulk and have pz character. For the
surface reconstruction with full monolayer of As used here, this
leads to lowering of two of four dimers in the unit cell to
maximize the overlap of dehybridized py As orbitals with Ga
atoms in the underlying layer at the expense of elevation of
another two dimers. The pz orbitals of one of the dimers (atoms
7 and 8 in Figure 2) have energy higher than Fermi level,
resulting in the formation of undesired surface states. As can
be seen from Figure 2, thiol adsorption removes the energy states
near the valence band edge from underlying arsenic. Although
the overall density of states of the slab have not changed at
low coverage regime, since the rest of As dimers remained
unaffected by thiol, higher coverage with thiols would partially
eliminate the surface states. This could explain, for instance,
the observed increase in photoluminescence intensity of thiol-
treated GaAs surface.4,11,32,33
We have further investigated the charge density redistribution
due to adsorption of thiolate on surface. Figure 3 shows the
regions of loss and gain of electron density induced by the bond
formation for the geometry of Figure 1a. Charge redistribution
for the structure of Figure 1b is essentially the same. Only
electrons around sulfur and the first CH2 unit of the thiolate
are involved in bonding, which confirms our suggestion that
pentanethiol would correctly reproduce the chemisorption of
longer chain thiols on GaAs surface. Electron density is depleted
in the As dangling bond region and gained between S and As,
showing a formation of the covalent bond. Examination of lower
density isosurfaces (not shown here) also reveals accumulation
of charge below As and on the pz orbital of the second As atom
in the dimer (see also Figure 2). A small depletion of electron
density along the S-C bond suggests its weakening upon
adsorption. There is also a redistribution of charge around sulfur
atoms, showing the accumulation of charge on sulfur px orbital,
same as in Figure 2 and similar to that reported for methanethiol
on copper.30 This density gain would be smaller when compared
to thiol rather than to thiolate, since in free thiolate nonbonded
px and py orbitals mix in a homogeneous ring around sulfur,
while in thiol (or thiolate on surface) py becomes involved in a
bond to H (or As) thus making px and py orbitals more
distinguishable. The charge redistribution on surface atoms other
than underlying arsenic is minimal in contrast to that reported
for gold31 and copper sufaces.30 For qualitative description of
the charge transfer, the Mulliken population analysis has also
been performed. Since Mulliken charges are known to be
strongly dependent on the basis set, we have carried out
calculations for a less complete DZP basis as well as for several
shorter bases. All these calculations have shown a small electron
transfer of about 0.05 electrons from surface to thiolate, which
compares with 0.4 and 0.6 electrons for thiolate on gold15 and
copper30 surfaces, respectively. This indicates low ionicity and
high covalency of the bond, consistent with our previous
conclusions from energy dependence on the As-S-C angle
and suggesting strong binding of thiolate to GaAs surface. It
also supports our suggestion that the gain of density on the px
orbital of sulfur is due to the redistribution of its own density
rather than from underlying arsenic.
The calculated binding energy of thiolate to As-rich
GaAs (001) surface is 2.11 eV. This compares to 1.64-2.3 eV
reported for Au (111)12-16,29 and 2.19 eV for Cu (111)12,30
surfaces calculated using a similar theoretical technique. Such
a big binding energy is consistent with the successful use of
high temperatures (100 °C) for growth of self-assembled
monolayers of thiols on GaAs,5 survival of 200 °C annealing,34
and ultrasonic bath,35 confirming that thiols form a robust
interface with GaAs. Unfortunately there are no experimental
measurements of the binding energy of thiol on As-rich GaAs
(001) surface to date, and available temperature programmed
desorption (TPD) data for Ga-rich (001)36 and (110)37 surfaces
shows a very different and complex behavior complicating their
comparison with our results. One of the features of the TPD
data is the absence of thiolate desorption peaks, showing instead
the recombinative desorption of thiol, dithiol, alkane, and
molecular hydrogen with energies around 1.45 eV. This suggests
that hydrogen obtained by the cleavage of S-H bond stays on
the surface and plays an important role in thiol adsorption-
desorption process, and that the energy of As-S bond is higher
than the energy needed for recombinative desorption.
Our results also qualitatively address some important issues.
First, the large energy difference between bridge and on-top
positions provides a substantial barrier for diffusion of thiols
on the GaAs surface, higher than that on copper, silver, and
gold,12 complicating the growth of self-assembled monolayers.26
Second, the chemisorbed molecule tends to go upright from the
lying flat on surface position, similar to reported theoretical
results for gold and copper.12,30 Such a behavior is explained
by steric repulsion of the first CH2 unit from the surface and a
relatively short chemical bond acting as a lever. This implies
that chemisorbed short-chain thiols would be connected to
surface only via the headgroup, liberating the adsorption sites
for other thiols. Longer chain thiols can be still lying on the
surface attracted by van der Waals forces, although strongly
bended near the headgroup.
In summary, we have investigated the adsorption of pen-
tanethiol on an As-rich GaAs (001) surface in DFT using the
PBE functional. The geometry of the thiolate on the surface is
dictated by the high directionality of the As dangling bond,
sulfur 3p orbital and steric repulsion of the first three CH2 units
Figure 3. Regions of loss (light blue) and gain (red) of electron density
induced by adsorption of thiolate on surface of (001) GaAs. Isodensity
surfaces correspond to ( 0.006 a.u.
Letters J. Phys. Chem. B, Vol. 110, No. 47, 2006 23621

from the surface. Thiolate adsorption is shown to partially
eliminate the electronic surface states. Small charge transfer
between thiolate and surface, strong dependence of total energy
on tilt angle, and short S-As bond length (2.28 Å) indicate a
highly covalent nature of the bonding. Calculated binding energy
of 2.11 eV is consistent with the available experimental data.
Acknowledgment. The funding for this research has been
provided by the Canadian Institutes for Health Research and
the Canada Research Chair Program. The calculations presented
in this work have been carried out using the infrastructure
provided by the Re´seau que´be´cois de calcul de haute perfor-
mance (RQCHP).
Supporting Information Available: Pseudopotentials, basis
sets, and 3D structures of optimized geometries. This material
is available free of charge via the Internet at http://pubs.acs.org.
References and Notes
(1) Love, J. C.; Estroff, L. A.; Kriebel, J. K.; Nuzzo, R. G.; Whitesides,
G. M. Chem. ReV. 2005, 105, 1103.
(2) Schreiber, F. Prog. Surf. Sci. 2000, 65, 151.
(3) Ulman, A. Chem. ReV. 1996, 96, 1533.
(4) Lunt, S. R.; Ryba, G. N.; Santangelo, P. G.; Lewis, N. S. J. Appl.
Phys. 1991, 70, 7449.
(5) Sheen, C. W.; Shi, J. X.; Martensson, J.; Parikh, A. N.; Allara, D.
L. J. Am. Chem. Soc. 1992, 114, 1514.
(6) Nishimura, K.; Nagao, Y.; Sakai, K. J. Cryst. Growth 1993, 134,
293.
(7) Nakagawa, O. S.; Ashok, S.; Sheen, C. W.; Martensson, J.; Allara,
D. L. Jpn. J. Appl. Phys. 1991, 30, 3759.
(8) Amro, N. A.; Xu, S.; Liu, G. Y. Langmuir 2000, 16, 3006.
(9) Seker, F.; Meeker, K.; Kuech, T. F.; Ellis, A. B. Chem. ReV. 2000,
100, 2505.
(10) Mohaddes-Ardabili, L.; Martinez-Miranda, L. J.; Salamanca-Riba,
L. G.; Christou, A.; Silverman, J.; Bentley, W. E.; Al-Sheikhly, M. J. Appl.
Phys. 2004, 95, 6021.
(11) Ding, X.; Moumanis, Kh.; Dubowski, J. J.; Frost, E. H.; Escher,
E. Appl. Phys. A 2006, 83, 357.
(12) Cometto, F. P.; Paredes-Olivera, P.; Macagno, V. A.; Patrito, E.
M. J. Phys. Chem. B 2005, 109, 21737.
(13) Gottschalck, J.; Hammer, B. J. Chem. Phys. 2002, 116, 784.
(14) Gronbeck, H.; Curioni, A.; Andreoni, W. J. Am. Chem. Soc. 2000,
122, 3839.
(15) Sellers, H.; Ulman, A.; Shnidman, Y.; Eilers, J. E. J. Am. Chem.
Soc. 1993, 115, 9389.
(16) Yourdshahyan, Y.; Rappe, A. M. J. Chem. Phys. 2002, 117, 825.
(17) Soler, J. M.; Artacho, E.; Gale, J. D.; Garcia, A.; Junquera, J.;
Ordejon, P.; Sanchez-Portal, D. J. Phys. Condens. Matter 2002, 14, 2745.
(18) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. ReV. Lett. 1996, 77,
3865.
(19) Ordejon, P. Phys. Status Solidi B 2000, 217, 335.
(20) Becke, A. D. Phys. ReV. A 1988, 38, 3098.
(21) Lee, C. T.; Yang, W. T.; Parr, R. G. Phys. ReV. B 1988, 37, 785.
(22) Miotto, R.; Srivastava, G. P.; Ferraz, A. C. Phys. ReV. B 1999, 59,
3008.
(23) Junquera, J.; Paz, O.; Sanchez-Portal, D.; Artacho, E. Phys. ReV.
B 2001, 64.
(24) Moreno, J.; Soler, J. M. Phys. ReV. B 1992, 45, 13891.
(25) Adlkofer, K.; Tanaka, M. Langmuir 2001, 17, 4267.
(26) Jun, Y.; Zhu, X. Y.; Hsu, J. W. P. Langmuir 2006, 22, 3627.
(27) Vanduijneveldt, F. B.; Vanduijneveldtvanderijdt, J. G. C. M.;
Vanlenthe, J. H. Chem. ReV. 1994, 94, 1873.
(28) Filippi, C.; Singh, D. J.; Umrigar, C. J. Phys. ReV. B 1994, 50,
14947.
(29) Franzen, S. Chem. Phys. Lett. 2003, 381, 315.
(30) Ferral, A.; Paredes-Olivera, P.; Macagno, V. A.; Patrito, E. M. Surf.
Sci. 2003, 525, 85.
(31) Bilic, A.; Reimers, J. R.; Hush, N. S. J. Chem. Phys. 2005, 122,
094708.
(32) Ding, X.; Moumanis, Kh.; Dubowski, J. J.; Tay, L.; Rowell, N. L.
J. Appl. Phys. 2006, 99, 054701.
(33) Moumanis, Kh.; Ding, X.; Dubowski, J. J.; Frost, E. H. J. Appl.
Phys. 2006, 100, 034702.
(34) Dassa, O.; Sidorov, V.; Paz, Y.; Ritter, D. J. Electrochem. Soc.
2006, 153, G91.
(35) Ding, X.; Moumanis, Kh.; Dubowski, J. J.; Frost, E. H. Proc. SPIE
2006, 6106, 61061L.
(36) Singh, N. K.; Doran, D. C. Surf. Sci. 1999, 422, 50.
(37) Camillone, N.; Khan, K. A.; Osgood, R. M. Surf. Sci. 2000, 453,
83.
23622 J. Phys. Chem. B, Vol. 110, No. 47, 2006 Letters